‘7
ORLGSION Metal/ Environment . Reactions L
third edition
CORROSION Volume I
MetaVEnvironment Reactions Edited by ...
826 downloads
4278 Views
65MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
‘7
ORLGSION Metal/ Environment . Reactions L
third edition
CORROSION Volume I
MetaVEnvironment Reactions Edited by L.L. Shreir, PhD, CChem, FRIC, FIM, FICorrT, FIMF, OBE R.A. Jarman, MSc, PhD, CEng, MIEE, FIW G.T. Burstein, MSc, PhD, M A
1 E I N E M A N N
Butterworth-Heinemann Linacn House, Jordan Hill, Oxford OX2 8DP 225 Wildwood Avenue, Wobum, MA 01801-2041 A division of Reed Educational and Professional Publishing Ltd
-&member
of the Reed Elsevkr plc group
OXFORD AUCKLAND BOSTON JOHANNESBURG MELBOURNE NEW DELHI First Published 1963 Second edition 1976 Third edition 1994 Reprinted 1995, 1998,2000
0 The several contributors listed on pages xviii-xxii, 1963, 1976, 1994 All rights reserved. No part of this publication may be reproduced in any material form (including photocopying or storing in any medium by electronic means and whether or not transiently or incidentally to some other use of this publication) without the written permission of the copyright holder except in accordance with the provisions of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Coun Road, London, England W IP OLP. Application for the copyright holder's written permission to reproduce any part of this publication should be addressed to the publishers British Library Cataloguing in Publication Data Corrosion - 3Rev.ed I. Shreir. L. L. 11. Jarman, R. A. 111. Burstein, G. T. 620. I623 Library of C o n g m s Cataloguing in Publication Data Corrosionledited by L. L. Shreir, R. A. Jarman. G . T. Burstein p. cm. Includes bibliographical references and index. Contents: v. I . MetaVenvironmental reactions - v. 2. Corrosion control I . Corrosion and anti-corrosives. 1. Shreir, L. L. 11. Jarman, R. A. 111. Burstein, G. T. TA462.C6513 1993 93- I3859 620.1 ' 1223-dc20 ISBN 0 7506 1077 8 (for both volumes) CIP
Printed and bound in Great Britain
FOR E V E N T m B THAT WE PUBLISH, B ~ O R T H N E W M A N N W U PAY POR BTCV TO W A N D CAR8 FOR A T I W .
CONTENTS Volume 1 . MetaVEnvironment Reactions L. L. Shreir, OBE Preface to the third edition Preface to the first edition List of contributors
1. Principles of Corrosion and Oxidation 1.1
Basic Concepts of Corrosion
1.1A
Appendix - Classification of Corrosion Processes
1.2
Nature of Films, Scales and Corrosion Products on Metals
1.3
Effects of Metallurgical Structure on Corrosion
1.4
Corrosion in Aqueous Solutions
1.5
Passivity and Localised Corrosion
1.6
Localised Corrosion
1.7
Bimetallic Corrosion
1.8
Lattice Defects in Metal Oxides
1.9
Continuous Oxide Films
1.10
Discontinuous Oxide Films
1.11
Erosion Corrosion
2. Environments 2.1
Effect of Concentration, Velocity and Temperature V
vi
CONTENTS
2.2
The Atmosphere
2.3
Natural Waters
2.4
Sea Waters
2.5
Soil in the Corrosion Process
2.6
The Microbiology of Corrosion
2.7
Chemicals
2.8
Corrosion by Foodstuffs
2.9
Mechanisms of Liquid-metal Corrosion
2.10
Corrosion in Fused Salts
2.11
Corrosion Prevention in Lubricant Systems
2.12
Corrosion in the Oral Cavity
2.13
Surgical Implants
3. Ferrous Metals and Alloys 3.1
Iron and Steel
3.2
Low-alloy Steels
3.3
Stainless Steels
3.4
Corrosion Resistance of Maraging Steels
3.5
Nickel-Iron Alloys
3.6
Cast Iron
3.7
High-nickel Cast Irons
3.8
High-chromium Cast Irons
3.9
Silicon-Iron Alloys
3.10
Amorphous (Ferrous and Non-Ferrous) Alloys
CONTENTS
4. Non-Ferrous Metals and Alloys 4.1
Aluminium and Aluminium Alloys
4.2
Copper and Copper Alloys
4.3
Lead and Lead Alloys
4.4
Magnesium and Magnesium Alloys
4.5
Nickel and Nickel Alloys
4.6
Tin and Tin Alloys
4.7
Zinc and Zinc Alloys
5. Rarer Metals 15.1
Beryllium
5.2
Molybdenum
5.3
Niobium
5.4
Titanium and Zirconium
5.5
Tantalum
5.6
Uranium
5.7
Tungsten
6. The Noble Metals 6.1
The Noble Metals
7. High-Temperature Corrosion 7.1
Environments
7.2
The Oxidation Resistance of Low-alloy Steels
7.3
High-temperature Corrosion of Cast Iron
vii
viii
CONTENTS
7.4
High-alloy Steels
7.5
Nickel and its Alloys
7.6
Thermodynamics and Kinetics of Gas-Metal Systems
8. Effect of Mechanical Factors on Corrosion 8.1
Mechanisms of Stress-corrosion Cracking
8.2
Stress-corrosion Cracking of Ferritic Steels
8.3
Stress-corrosion Cracking of Stainless Steels
8.4
Stress-corrosion Cracking of High-tensile Steels
8.5
Stress-corrosion Cracking of Titanium, Magnesium and Aluminium Alloys
8.6
Corrosion Fatigue
8.7
Fretting Corrosion
8.8
Cavitation Damage
8.9
Outline of Fracture Mechanics
8.10
Stress-corrosion Test Methods
8.10A Appendix
- Stresses in Bent Specimens
Volume 2. Corrosion Control Introduction to Volume 2 9. Design and Economic Aspects of Corrosion 9.1
Economic Aspects of Corrosion
9.2
Corrosion Control in Chemical and Petrochemical Plant
9.3
Design for Prevention of Corrosion in Buildings and Structures
9.4
Design in Marine and Offshore Engineering
9.5
Design in Relation to Welding and Joining
9.5A
Appendix - Terms Commonly Used in Joining
10. Cathodic and Anodic Protection 10.1
Principles of Cathodic Protection
10.2
Sacrificial Anodes
10.3
Impressed-current Anodes
10.4
Practical Applications of Cathodic Protection
10.5
Stray-current Corrosion
10.6
Cathodic-protection Interaction
10.7
Cathodic-protection Instruments
10.8
Anodic Protection
ix
CONTENTS
X
11. Pretreatment and Design for Metal Finishing 11.1
Pretreatment Prior to Applying Coatings
11.2
Pickling in Acid
11.3
Chemical and Electrolytic Polishing
11.4
Design for Corrosion Protection by Electroplated Coatings
11.5
Design for Corrosion Protection by Paint Coatings
12. Methods of Applying Metallic Coatings 12.1
Electroplating
12.2
Principles of Applying Coatings by Hot Dipping
12.3
Principles of Applying Coatings by Diffusion
12.4
Principles of Applying Coatings by Metal Spraying
12.5
Miscellaneous Methods of Applying Metallic Coatings
13. Protection by Metallic Coatings 13.1
The Protective Action of Metallic Coatings
13.2
Aluminium Coatings
13.3
Cadmium Coatings
13.4
Zinc Coatings
13.5
Tin and Tin Alloy Coatings
13.6
Copper and Copper Alloy Coatings
13.7
Nickel Coatings
13.8
Chromium Coatings
13.9
Noble Metal Coatings
14. Protection by Paint Coatings 14.1
Paint Application Methods
CONTENTS
14.2
Paint Formulation
14.3
The Mechanism of the Protective Action of Paints
14.4
Paint Failure
14.5
Paint Finishes for Industrial Applications
14.6
Paint Finishes for Structural Steel for Atmospheric Exposure
14.7
Paint Finishes for Marine Application
14.8
Protective Coatings for Underground Use
14.9
Synthetic Resins
14.10
Glossary of Paint Terms
15. Chemical Conversion Coatings 15.1
Coatings Produced by Anodic Oxidation
15.2
Phosphate Coatings
15.3
Chromate Treatments
16. Miscellaneous Coatings 16.1
Vitreous Enamel Coatings
16.2
Thermoplastics
16.3
Temporary Protectives
17. Conditioning the Environment 17.1
Conditioning the Atmosphere to Reduce Corrosion
17.2
Corrosion Inhibition: Principles and Practice
17.3
The Mechanism of Corrosion Prevention by Inhibitors
17.4
Boiler and Feed-water Treatment
xi
xii
CONTENTS
18. Non-Metallic Materials 18.1
Carbon
18.2
Glass and Glass-ceramics
18.3
Vitreous Silica
18.4
Glass Linings and Coatings
18.5
Stoneware
18.6
Plastics and Reinforced Plastics
18.7
Rubber and Synthetic Elastomers
18.8
Corrosion of Metals by Plastics
18.9
Wood
18.10
The Corrosion of Metals by Wood
19. Corrosion Testing, Monitoring and Inspection 19.1
Corrosion Testing
19.1A Appendix - Removal of Corrosion Products 19.1B Appendix - Standards for Corrosion Testing 19.2
The Potentiostat and its Applications to Corrosion Studies
19.3
Corrosion Monitoring and Inspection
19.4
Inspection of Paints and Painting Operations
20. Electrochemistry and Metallurgy Relevant to Corrosion 20.1
Outline of Electrochemistry
20.2
Outline of Chemical Thermodynamics
20.3
The Potential Difference at a Metal/Solution Interface
20.4
Outline of Structural Metallurgy Relevant to Corrosion
CONTENTS
21. Useful Information 21.1
Tables
21.2
Glossary of Terms
21.3
Symbols and Abbreviations
2 1.4
Calculations Illustrating the Economics of Corrosion
Index
...
Xlll
1.1. SHREIR, OBE 1914-1992
Lionel Louis Shreir OBE, died on 5th November 1992 after a lifetime devoted to the science and technology of corrosion and education. His industrial career spanned a period of 19 years, from 1929 to 1948, during which time he was employed by the Mond Nickel Company, Baker Platinum Ltd and Plessey Ltd. At the same time he continued his higher education on a part time basis at the Chelsea and Battersea Polytechnicsand Sir John Cass College in London. In 1948, he joined the staff of Battersea Polytechnic, subsequently renamed the University of Surrey, eventually attaining the position of Reader in Corrosion. In 1%2 Lionel became Head of Metallurgy of the Sir John Cass College (now London Guildhall University), a post he enjoyed by greatly expanding the Department, its research and general reputation until his retirement in 1979. Lionel’s contribution to corrosion was outstanding. In addition to his encyclopedic work Corrosion, the present edition being dedicated to his memory, he was author of more than 70 papers and was editor of Corrosion Science for many years. He was engaged as a consultant to a number of organisations up to the time of his death whilst his research initiatives covered many fields, including hydrogen in metals, anodic oxidation and electrodeposition. He was the third recipient of the U.R. Evans Award in 1978 and was awarded the OBE in 1982 in recognition of his services to corrosion. In this context, one of his most notable activities was to advise on the protection of the Thames Barrier. In the past year a Lionel Shreir Award was awarded for the first time by the Institute of Corrosion Science and Technology. Regardless of his achievements Lionel was kind, modest and a very caring man. He will be affectionately remembered for his boundless energy and infectious enthusiasm by his peers, colleagues, friends and the countless past students privileged to have made his acquaintance during what was a R A JARMAN remarkably active life.
xv
The huge success of the first two editions of Corrosion has inevitably created the demand for a third edition. Corrosion science and technology, like most of the physical sciences, has progressed and advanced significantly in the seventeen years since the second edition was published. Such knowledge requires transferral from the laboratory and the journal literature to the wider audience: the student, the teacher, the engineer, the metallurgist and workers in other fields who require knowledge and understanding of the interactions of materials with their environments. The previous two editions, the fruits of Lionel Shreir’s hard labours, have fulfilled this multiple role admirably and the new editors hope that this new edition will continue to do so. The fact that Lionel worked so hard on producing the third edition but did not live to see its publication, is a personal and deeply poignant sorrow for us, as it must be for the many readers of Corrosion who knew and respected him as scientist and friend. The ever-increasing research into corrosion, and the knowledge that this produces is driven to a small part by the corrosion scientist him- or her-self in seeking a detailed understanding of the intricacies of the interfacial processes driving corrosion and passivation. Such a self-fulfilling drive cannot of itself however, be indefinitely sustainable, despite the fascination that this science engenders, since research is costly. Such advances are led primarily by the continuing need to predict, control and prevent corrosion as an engineering imperative. Corrosion science, multidisciplinary in itself, is probably unique in crossing the borders of almost all the technologies: environmental stability of all components of those technologies remains a prime requirement for their success. New technologies, new engineering practices, new materials and new processes can succeed only if the behaviour of their components with the environment is satisfactory, and predictably so. The eighties and nineties, and beyond, see a further need to underpin research and development into corrosion and protection - the growing awareness of the necessity for conservation, of materials and of energy, the so-called green issues. Most materials and components made from them require large energy resources to produce; clearly the quest for longevity and reliability of structures is a significant and worthy contribution towards conserving energy and materials, quite additional to minimising the heavy cost of corrosion failures. As with the second edition, the new volumes have been revised according xvi
PREFACE TO THE THIRD EDITION
xvii
to the general format and structure of their antecedents. Some sections have been completely rewritten to bring them up to date, while others have been altered and extended. New sections have been included to cover areas not previously treated. The incorporation of new authors to carry out such revisions and additions is the inevitable consequence of the fact that thirty years have elapsed since Corrosion first appeared. The multiplicity of authors for the new edition leads (as with previous editions) to a variety of styles of writing and variation in treatment and emphasis of subject matter. One hopes this is beneficial to the work in providing a broader cross-section of corrosion science and technology as a whole: it is for the reader and casual user to judge. One hopes too, that the third edition remains a tribute to the man who initiated Corrosion.
GTB Cambridge
The enormous scope of the subject of corrosion follows from the definition which has been adopted in the present work. Corrosion will include all reactions at a metal/environment interface irrespective of whether the reaction is beneficial or detrimental to the metal concerned -no distinction is made between chemical or electropolishing of a metal in an acid and the adventitious deterioration of metal plant by acid attack. It follows, therefore, that a comprehensive work on the subject of corrosion should include an account of batteries, electrorefining, chemical machining, chemical and electrochemical polishing, etc. The fact remains, nonetheless, that the environmental reaction of a metal used as a construction material is the most important type of corrosion reaction, and the one of most concern to the engineer. The technological and economic consequences of the wastage of metals by corrosion are now fully appreciated, and figures have been published which show the enormous financial losses, both to the individual organisation and to the economy of the country as a whole, resulting from the deterioration of metals. The need for conserving metals has been publicised by Dr U. R. Evans, Dr J . C. Hudson, Mr T. H. Turner, Professor H. H. Uhlig, Dr W.H. J. Vernon, and others, and the ‘corrosion consciousness’ which prevails today is largely due to their efforts. In the light of what has been said above, little further explanation of the implications of the title of the present work is required. Its treatment of the subject of corrosion will centre round the control of the environmental interactions of metals and alloys used as materials of construction. The effective control of corrosion reactions must be based on an understanding of the mechanism of such reactions and on the application of this knowledge to practical problems. The work, regarded as a whole, represents an attempt, therefore, to present the subject of corrosion as a synthesis of corrosion science and corrosion engineering. Thus in the planning of the content an attempt has been made to strike a suitable balance between the primarily scientific and the primarily practical aspects, and so the nature of individual sections ranges from the fundamental and theoretical to the essentially practical. It is hoped that this approach has resulted in a work that will be of some xviii
PREFACE TO THE FIRST EDITION
xix
value to the student, the corrosion worker, and the engineer in the field of corrosion. Corrosion represents the joint effort of over 100 authors, all of whom have been free, within the necessary limitations of length, to express their own views. Grateful acknowledgements are made to the individual authors from Great Britain, the United States, and Canada for their valuable and enthusiastic co-operation. The task of the editor in finding suitable authors for various topics was considerably lightened by the fact that the majority of corrosion specialists in this country belong either to the Corrosion Group of the Society of Chemical Industry or to the Institute of Metal Finishing, and acknowledgements are made to Mr s. C. Britton (then Secretary, Corrosion Group) and to Dr S. Wernick and Mr I. S. Hallows (Hon. Secretary and Assistant Editor respectively of the I.M.F.). The editor wishes to express his appreciation of the considerable assistance received from Dr E. C. Rhodes and Dr G. L. J. Bailey of INCO (Mond) in providing authors from this organisation. The editor also acknowledges with pleasure the encouragement and assistance he has received from Mr L. W. Derry (Head of Department of Metallurgy) and Dr D. M. A. Leggett (Principal) of the Battersea College of Technology, and from Dr A. M. Ward (Principal) of Sir John Cass College. Throughout the course of this work the content and subject matter have been discussed with various workers in the field, and the editor would like to take this opportunity to thank Mr S. C. Britton, Professor C. W. Davies, Dr T. P. Hoar, Dr E. C.Potter, and others for their advice and constructive criticism. He would also like to take this opportunity to express his appreciation to Dr U. R. Evans and Dr W.H. J. Vernon for assistance given when he first contemplated entering the field of corrosion, and for their encouragement and advice in connection with the present work. Finally, grateful acknowledgements are made by the editor to Mr T. F. Saunders and Mrs N.E. Orna, M.A., of George Newnes (Technical Books) for their kind co-operation at all stages of the work. L.L.S. 1963
CONTRIBUTORS
K G Adamson", AMCST, LIM Development Oficer, Magnesium Elektron Ltd, Manchester
T A Banfield*, PhD, DIC, ARCS, CChem, FRIC, FICorr, FTSC Deputy Manager, Group Research Laboratory, Berger Jenson and ivicholson Ltd.
J C B Alcock*, ARCS, DSc, PhD, CChem, FRIC Professor and Chairman, Dept. of Metallurgy and Materials Science, University of Toronto, Canada
P J Barnes, BSc, MRSC, CChem, ATSC Consultant E W Beale*, A R K Senior Scientific Ofleer, Materials Quality Assurance Directorate. Ministry of Defence
M D Allen, CEng, MIM, MICorrST
Penspen, London
D Arne*, MICorr Associate, Spencer and Partners, Consulting Engineers
J Bentley", BSc, DipChemEng, CEng, MIChemE Principal Chemist, Wastes Division Directorate General, Water Engineering, Dept. of the Environment
K F Anderson' Morganite Carbon Ltd. (Formerly) B Angel1 Section Head, Corrosion and Protection, Defence Research Agency, Poole, Dorset
W Betteridge", DSc, FInstP, FIM Consultant, Formerly of International Nickel Ltd.
J E Antill*, PhD, BSc Head, Chemical Metallurgy Group, Materials Development Division, UKAEA, Harwell
P J Boden*, PhD, CEng, CChem, FRSC, MIM, FlCorr Senior Lecturer, Dept. of. Metallurgy and .
Materials Science, Nottingham University
V Ashworth Global Corrosion Consultants Ltd Shifnal, Shropshire
C J L Booker', BSc, PhD, A R K , FICorr Formerly Senior Lecturer in Corrosion Science, Dept. of Metallurgy and Materials, City of London Polytechnic
D J Astley, BSc, ARCS, PhD, PGCE
Formerly Senior Technical Oficer IMI Research and Development, Birmingham J C Bailey*, BSc, FIM Formerly Deputy Director (Technical), Aluminium Federation
J W L F Brand*, MITE, TEng(CEI), MICorr Divisional Manager, Corrosion Control Division, Corrosion and Welding Engineering Ltd.
W E Ballard', CChem, FRIC, FIM Consultant, Formerly Managing Director, Metallisation Lid. *Contributor to earlier editions
xx
CONTRIBUTORS C F Britton*, LRIC, FICorr, FInstPet Corrosion Consultant Formerly vfAEA Technology and Rohrback Instruments Lid
J B Cotton*, CChem, AMCT, ARIC, FICorr Industrial Consultant
S C Britton*, MA, CChem, FRIC, FIM, IMF, FICorr Tin Research Institute (Retired)
R A Cottis Senior Lecturer in Corrosion Science and Engineering, Corrosion and Protection Centre, UMIST
J A Brydson*, FPRI, ANCRT Technical Consultant
R N Cox, BSc, CEng, MIM Building Research Establishment, Garston, Watford
T R Bullett, BSc, CPhys, FInstP, FICorrST
G W Currer, CEng, MIEE, MICorrST Consultant
W Bullough*, BSc, ARIC Principal Research Ofleer, BSC Research Centre, Strip Mills Division
G T Burstein, MSc, PhD, MA Aflliated Lecturer, Department of Materials Science and Metallurgy University of Cambridge V E Carter*, FICorr, FIMF Corrosion and metal finishing consultant, Formerly of BNFRA
J E Castle', BSc, PhD, CChern, FRSC FICorr Head of Department, University of Surrey, Guildford
K A Chandler*, BSc, ARSM, FICorr Head, Corrosion Advice Bureau, BSC A R L Chivers', MA Senior Technical Oflcer, Zinc Development Association, London
M Clarke*, BSc, PhD, DSc, CChem, FRIC, FIM, FICorr, FIMF Consultant, formerly Principal Lecturer, Dept. of Metallurgy and Materials, City of London Polytechnic R J Clarke*, MA, CEng, FIChemE, FIFST Won. Visiting Lecturer in Food Engineering, Queen Elizabeth College, London H G Cole', BSc, FIMF, FICorr Principal Scientific Oflcer, Ministry of Defence (Procurement Executive) H H Collins*, BSc, CChem, FRIC Superintendent, Chemistry Research, Stanton & Staveley/BSC
J Congleton, BSc, PhD, FIM, CEng Senior Lecturer Department of Mechanical, Materials and Manufacturing Engineering University of Newcastle upon Tyne
xxi
D P Dautovich*, MSc, PhD Corrosion Engineer, Research Division, Ontario Hydro, Canada K Julyan Day*, FICorr, FTSC, MBIM Anti-Corrosion Consultant J Dodd*, BSc, FIM, FIBF
Metallurgical Consultant Dodd and Associates, Colorado, USA P D Donovan*, MSc, ARIC, FIM Principal Scientific Oflcer, Ministry of Defence C W Drane*, BSc, CChem, FRIC Technical Manager, Water Specialities and Services, Industrial Chemicals Division, Albright and Wilson Ltd. F G Dunkley*, FICorr Consultant, Formerly of British Rail, Derby
E J Easterhrook*, BSc(Eng), ARSM, AMIMM, MIM Formerly Principal Lecturer, Dept. of Metallurgy and Materials, City of London Polytechnic J Edwards, BSc, PhD Consultant Formerly of The British Non-Ferrous Metals Research Association and International Nickel Limited T E Evans', BSc, ARIC, FICorr Principal Technologist, International Nickel Ltd., Birmingham D Eyre, BSc, MSc, PhD, MICorr Principal Corrosion Engineer Spencer and Partners, London G N Flint
von Fraunhofer, PhD, MSc, FRSC Director, Laboratory of Molecular and Materials Science, School of Dentistry, University of Louisville, Kentucky JA
P C Frost, Senior Research Scientist, Cookson Group plc Yarnton, Oxfordshire
xxii
CONTRIBUTORS
D Fyfe", MA, PhD Senior Marketing Engineer, Chemetics Ltd, Montreal, Canada
M H a * , CChem. FRIC, FIMF, FTSC Member of the Association of Consulting Scientists Principal of Manfred Hess, Consulting Chemist and Paint Technologist
D R Cabe, BSc, MMet, PhD, CEng, FIM, FICorr Institute of Polymer Technology and Materials Engineering, Loughborough
G L Higgins*, Bsc, MIMF
P J Gay", BSc, FTSC, FICorr Consultant
Chemetall Ltd, Aylesbury
D K Hill*, DSc,.PhD, FSGT Formerly Technical Manager, British Indestructo Glass Ltd. (Retired) J Hines*, MA, PhD Formerly North West Region Materials Group Manager, ICI Lid.
J S Gerrard*. AMIEE Formerly Joint General Manager, Metal and Pipeline Endurance Ltd.
T P Hoar
G N J Gilbert
R A E Hooper, BMech, FIM, CEng, FICorr Group Technical Manager, Authur Lee & Sons plc. S h m l d
P T Gilbert', BSc, PhD, CChem, FRIC, FIM, FIMarE, FICorr, CEng Metallurgical Consultant Formerly BCIRA
T B Grimley*, BSc, PhD Reader in Theoretical Chemistry, Dept. of Inorganic, Physical and Industrial Chemistry, University of Liverpool
B H Hanson, BSc Consultant Formerly IMI Titanium J 0 Harris*, PhD Professor of Bacteriology, Kansas State University
S J Harris, MSc, PhD, CEng, FIM, FIMF Department of Metallurgy and Materials Science, University of Nottingham A C Hart*, DTech, BSc, CChem, MRSC, FIMF
Managing Director, Hart Coating Technology
K Hashimoto, DSc Professor Institute for Materials Research Tohoku University. Sendai. Japan
H Howarth*, AMICorr, AMet Investigator, Production Metallurgy Section, Special Steels Division, BSC
J C Hudson*, DSc, DIC, ARCS, FIM Consultant, Formerly of BISRA
D E Hughes*, MA, DSc, PhD, CBiol, FIBiol Professor Emeritus (Microbiology) University of Wales R S Hullcoop Ray Hullcoop and Associates, High Wycombe D Inman*, BSc, ARCS, FRIC, PhD, DSc, DIC, MIMM Reader in Chemical Metallurgy, Nufield Fellow in Extraction Metallurgy, Dept. of Metallurgy and Materials Science. Imperial College, London R A Jarman*, MSc, PhD, CEng, MIEE, FWeldI, FIM Consultant, formerly School of Engineering, University of Greenwich L Kenworthy*, MSc, ARCS, CChem, FRIC, FIM. FICorrT Consultant, Formerly Navy Dept. (Ministry of Defence)
B T Kelly, MSc, ChP. InstP Consultant Physicist
CONTRIBUTORS
E G King, CEng, BSc, MIM, MIWeld Consultant
G N King, MSc Department of Metallurgy University of Nottingham
D G Kingerley, MSc, BSc, CChem, MRSC, CEng, MInstE, FICorr Dept of Materials Engineering and Materials Deign. University of Nottingham D Kirkwood, PhD Senior Lecturer School of Mechanical and Offshore Engineering The Robert Gordon University, Aberdeen F LaQue', BSc, LLD. Past President, Nat. Assoc. Corrosion Eng., Am. Soc. Test and Mat., Electrochemical Society Senior Lecturer, Scripps Institution of Oceanography, University of California D N Layton*, PhD, MSc, ARCS, DIC, MInstP, FIMF Managing Director, Fredk. Mountford (Birmingham) Ltd. EurIng M F Leclerc, BSc, PhD, MIM, mng Technical Executive Biome? Ltd D A Lewis*,BSc(Eng). FICorr Partner of Spencer and Partners, Consulting Engineers
E L Littauer', BSc, PhD, MIM, AMIMM Manager, Electrochemistry and Environmental Sciences, Lockheed Missiles and Space Co., California, USA
xxiii
J Mackowiak, BSc, PhD, CEng, MIM Retired Senior Lecturer, University of Surrey, Guildford Consultant in high temperature corrosion C A May*, MSc, PhD Lecturer, School of Engineering, University of Greenwich
J E 0 Mayne*, DSc, ARCS, DIC, CChem, FRIC, FICorr Dept. of Materials Science and Metallurgy UniveGity of Cambridge
P Mclntyre, BSc, PhD, CEng, FIM Technology Consuitant National Power plc, Research and Engineering, Swindon, Wiltshire A D Mercer", BSc Principal Scientific Ofleer National Physical Laboratory N S C Millar*. CChem, FRIC, FICeram, MICorr, MBIM General Manager, Thermovitrine Ltd.
W G O'Donnell, BSc, MSc, CChem. MRSC. APRI Plascoat Systems Ltd, Surrey J W OldBeld, BSc, PhD Managing Director Cortest Laboratories Ltd, Sheffeld R J Oliphant, BA, MSc, PhD, AWIEM Technical Specialist WRc plc, Swindon
D S Oliver*, BSc, PhD, FIM, FInstP Group Director of Research and Development, Pilkington Bros. Ltd. M W O'Reilly*. Dip Tech, LRIC Decorative Paints Market Team Leader, ICIPaints Division
C 0 Lloyd*, BSc Principal Scientific Oflcer, Division of Materials Applications. National Physical Laboratory
S Orman*, BSc, PhD, FICorr, CChem, FRIC Senior Principal Scientific Oficer, A WRE, Aldermaston
N A Lockington*, MA, PhD, A R K , FIM Metallurgist. Director, The Chrome-Alloying Co. Ltd.
R N Parkins', BSc, PhD, DSc, FIM Professor and former Head of Department, Metallurgy and Engineering Materials, University of Newcastle upon Tyne
C L Long*, PhD, CChem, FRIC Principal Scienti@cOflcer, Energy Technology Support Unit. UKAEA, Harwell
W A Lure, BMetEng Retired P Lydon Roxby Engineering International Ltd, Kent
A W Pearson*, MIM Research Division, British Aluminium Co. Ltd.
J S Picard, D-k-Sc Research Director, Centre National de la Recherche Scientifique, Laboratoire d'ELectroshime Analytique et Appliqude, Ecole Nationale Supdriere de Chimie de Paris, France
xxiv
CONTRIBUTORS
L W finder*, BSc, MICorr Research Oflcer PowerGen L Pinion, BSc Consultant R Pinner', BSc, FICorr, FIMF Consultant
J S Pitman, FIM Research Scientist Servicised Ltd F C Porter*, MA, FIM, FICorr, FIMF Zinc Development Association, London
B S Poulson, B k . PhD, FICorr Chief Technologist (Joining and Surface Engineering) International Research and Development Ltd. IK Prall*, PhD, BSc, CChem, FRIC
Section Manager, Unilever Research Laboratory J T Pringle ICI Paints, Slough
B A Proctor, DSe, FIP Formerly Manager, Fibres and Glass, Pilkington Group Research
R P M Procter*, MA, PhD, CEng, FIM, FICorr Vice Principal The University of Manchester Institute of Science and Technology. UMIST M J Pryor E F Redknap*, BSc Retired F H Reid*, BSc, CChem, FRIC, FIMF Consultant, Formerly of International Nickel Ltd.
EurIng C E D Rowe. BSc, CEng, MIM Manager, Technical Services/Quality Assurance, Climax Special Metals Fabrications Ltd, Brentwood
J C Rowlands*, FICorr Defence Research Agency, Holton Heath ISadowska-Mazur*, MSc (Gdansk) Formerly Research Assistant, City of London Polytechnic
S R J Sauders, BSc, PhD, DIC National Physical Laboratory M J Sehofiefd, PhD, MICorr Technical Manager, Cortest Laboratories Ltd. I R Scholes, BSc, CChem, FRSC, FICorr Formerly Manager, IMI Research and Development Wilton, Birmingham B A Scott*, ARCS, BSc, PhD, CChem, FRIC Deputy Information Oficer, Group Technical Information Service, British Aluminium Co. Ltd. P M Scott, BSc, PhD Framatome, Paris J C Scully*, MA, PhD, CEng, FIM, FICorr Senior Teaching Fellow, School of Materials, University of Lee&
H J Sharp*. PhD, MSc. CChem? FIM, FPRI Director of International Associates R E Shaw*, BSc, FIM, FIMF ICI Paints Division (Retired)
P G Sheasby, BSc, FIMF Alcan International Ltd, Banbury
J A Richardson, BSc, PhD, MIM, MRSC, CChem, CEng Manager, Materials, ICI Engineering, Cleveland
L Shenvood Formerly Global Corrosion Consultants Ltd Shifnal, Shropshire
M 0 W Richardson, BTech, PhD, CChem, FRSC. FPRI IPTME Loughborough University of Technology
G S Shipley", FICeram Technical Advisor, Hathernware Ltd.
R G Robson', BSc(Eng). MIEE Chartered Engineer
N R Short, BSc. PhD Department of Civil Engineering, Aston University
D van Rooyen*, BSc, PhD Advisory Scientist, WestinghouseBettis Atomic Power Labs., USA M Roper
L L Shreir*, PhD, CChem, FRIC, FIM, FICorr, FIMF Former Head, Dept. of Metallurgv and Materials, City of London Polytechnic
CONTRIBUTORS
XXV
E W Skerrey', BSc, CChem. FRIC,
R Walker., BSc. DipEd. MSc, MSc(Eng), PhD LeGurer, Metallurgy and Materials Technology Dept. University of Surrey, Guildford
FICorrT, AIM Assistant Manager, Application Technology Department, Research Division, British Aluminium Co. Ltd.
G W Walkiden', BSc, CChem, FRIC, MIM Consultant, Ever Ready Central Laboratories
H Silman*, BSc, CEng, CChem. FRIC. FIChemE, FIM. FIMF Consultant
R A Smith*, BSc, PhD, FIM Manager, Research Laboratory. International Nickel Ltd. J F Stanners., BSc. FICorr Head of Corrosion Research, Inter-Services Laboratory. BSC
D R A Swynnerton T N Tate
DRA Swynnerton, Defence Research Agency, Nr Stone, Staffs
H Tatton', ARIC, FIMF, FTSC Technical Oflcer, British Standards Institution
W
D S Tawil, BSc
Technical Marketing Manager, Magnesium Elektron Inc, Lakehurst. New Jersey J G N Thomas*, BSc, PhD, ARCS, DIC Formerly Corrosion Section, Division of Materials Applications, National Physical Laboratory
A W Thoriey, BSc. AIM Consultant Formerly UK Atomic Energy Authority . I E Truman", AMet
Consultant metallurgist Jessop Saville Ltd, Sheffield S Turgoose, MA, PhD, MICorrST Lecturer in Corrosion Science and Engineering, UMlST
G P A Turner, MA Formerly Industrial Paints Research Manager, IC1 Paints, Slough
J R Walters* Consultant lo British Post Ofice
R B Waterhouse', MA, PhD, FIM, FICorr Reader in Metallurgy, Dept. of Metallurgy and Materials Science. University of Nottingham K 0 Watkias'. FIM. FlCorr Corrosion Advice Bureau, BSC
S A Watson', BSc. PhD. CChem, FRIC, FlMF Senior Development Oflcer, Internatiorral Nickel Ltd.
H C Wesson*. MA, BSc, CChem, FRIC Formerly Technical Manager, Lead Development Association (Retired) E E White*, CChem, FRIC, FIM, CEng, MIMM, FCS, MIInfSc, FICorrT, FIMF Consultant Inter-Services Laboratory, BSC
NR
Whitehouse, BSc PhD The Paint Research Association. Teddington, Middlesex
C Wilson Escol Products Ltd. Huntingdon
R W Wilson*, MA, PhD, CEng. FICorr, FIM Senior Consultant CAPCIS, Manchester
PA
Woods*, BSc
HM Inspector of Factories
K H R Wright', BSc, PhD, MInstP Senior Principal Scientific Oflcer, Materials Group, National Engineering Laboratory
1
PRINCIPLES OF CORROSION AND OXIDATION
1.1 Basic Concepts of Corrosion 1 . 1A Appendix-Classification of Corrosion Processes 1 .2 Nature of Films, Scales and Corrosion Products on Metals 1.3 Effects of Metallurgical Structure on Corrosion 1.4 Corrosion in Aqueous Solutions 1.5 1.6 1.7
Passivity and Localised Corrosion Localised Corrosion Bimetallic Corrosion
1.8 1.9 1.10 1.11
Lattice Defects in Metal Oxides Continuous Oxide Films Discontinuous Oxide Films Erosion Corrosion
1: 1
1:3 1:16 1 :22
1:36 1:55 1:118 1:151 1:213 1:244 1:254 1:268 1:293
1.1 Basic Concepts of Corrosion
Modern technology has at its disposal a wide range of constructional materials -metals and alloys, plastics, rubber, ceramics, composites, wood, etc. and the selection of an appropriate material for a given application is the important responsibility of the design engineer. No general rules govern the choice of a particular material for a specific purpose, and a logical decision involves a consideration of the relevant properties, ease of fabrication, availability, relative costs, etc. of a variety of materials; frequently the ultimate decision is determined by economics rather than by properties, and ideally the material selected should be the cheapest possible that has adequate properties to fuIfil the specific function. Where metals are involved, mechanical, physical and chemical properties must be considered, and in this connection it should be observed that whereas mechanical and physical properties can be expressed in terms of constants, the chemical properties of a given metal are dependent entirely on the precise environmental conditions prevailing during service. The relative importance of mechanical, physical and chemical properties will depend in any given case on the application of the metal. For example, for railway lines elasticity, tensile strength, hardness and abrasion resistance will be of major importance, whereas electrical conductivity will be of primary significance in electrical transmission. In the case of heat-exchanger tubes, good thermal conductivity is necessary, but this may be outweighed in certain environments by chemical properties in relation to the aggressiveness of the two fluids involved -thus although the thermal conductivity of copper is superior to that of aluminium brass or the cupronickels, the alloys are preferred when high velocity sea water is used as the coolant, since copper has very poor chemical properties under these conditions. While a metal or alIoy may be selected largely on the basis of its mechanical or physical properties, the fact remains that there are very few applications where the effect of the interaction of a metal with its environment can be completely ignored, although the importance of this interaction will be of varying significance according to circumstances; for example, the slow uniform wastage of steel of massive cross section (such as railway lines or sleepers) is of far less importance than the rapid perforation of a buried steel pipe or the sudden failure of a vital stressed steel component in sodium hydroxide solution. 1:3
1:4
BASIC CONCEPTS OF CORROSION
The effect of the metal/environment interaction on the environment itself is frequently more important than the actual deterioration of the metal (see Section 2.7). For instance, lead pipes cannot be used for conveying plumbo-
solvent waters, since a level of lead > 0.1 p.p.m. is toxic; similarly, galvanised steel may not be used for certain foodstuffs owing to the toxicity of zinc salts (see Section 2.8). In many chemical processes selection of a particular metal may be determined by the need to avoid contamination of the environment by traces of metallic impurities that would affect colour or taste of products or catalyse undesirable reactions; thus copper and copper alloys cannot be used in soap manufacture, since traces of copper ions result in coloration and rancidification of the soap. In these circumstances it will be essential to use unreactive and relatively expensive metals, even though the environment would not result in the rapid deterioration of cheaper metals such as mild steel. A further possibility is that contamination of the environment by metals’ ions due to the corrosion of one metal can result in the enhanced corrosion of another when the two are in contact with the same environment. Thus the slow uniform corrosion of copper by a cuprosolvent domestic water may not be particularly deleterious to copper plumbing, but it can result in the rapid pitting and consequent perforation of galvanised steel and aluminium that subsequently comes into contact with the coppercontaining water (Sections 4.1, 4.2 and 4.7). Finally, it is necessary to point out that for a number of applications metals are selected in preference to other materials because of their visual appearance, and for this reason it is essential that brightness and reflectivity are retained during exposure to the atmosphere; stainless steel is now widely used for architectural purposes, and for outdoor exposure the surface must remain bright and rust-free without periodic cleaning (Section 3.3). On the other hand, the slow-weathering steels, which react with the constituents of the atmosphere to form an adherent uniform coating of rust, are now being used for cladding buildings (Section 3.2), in spite of the fact that a rusty surface is usually regarded as aesthetically unpleasant. The interaction of a metal or alloy (or a non-metallic material) with its environment is clearly of vital importance in the performance of materials of construction, and the fact that the present work is largely confined to a detailed consideration of such interactions could create the impression that this was the sole factor of importance in materials selection. This, of course, is not the case although it is probably true to say that this factor is the one that is the most neglected by the design engineer.
Definitions of Corrosion In the case of non-metallic materials, the term corrosion invariably refers to their-deterioration from chemical causes, but a similar concept is not necessarily applicable to metals. Many authorities consider that the term metallic corrosion embraces all interactions of a metal or alloy (solid or liquid) with its environment, irrespective of whether this is deliberate and beneficial or adventitious and deleterious. Thus this definition of corrosion, which for convenience will be referred to as the transformation definition,
1:5
BASIC CONCEPTS OF CORROSION
will include, for example, the deliberate anodic dissolution of zinc in cathodic protection and electroplating as well as the spontaneous gradual wastage of zinc roofing sheet resulting from atmospheric exposure. On the other hand, corrosion has been defined’ as ’the undesirable deterioration’ of a metal or alloy, i.e. an interaction of the metal with its environment that adversely affects those properties of the metal that are to be preserved. This definition-which will be referred to as the deterioration definition-is also applicable to non-metallic materials such as glass, concrete, etc. and embodies the concept that corrosion is always deleterious. However, the restriction of the definition to undesirable chemical reactions of a metal results in anomalies which will become apparent from a consideration of the following examples. Steel, when exposed to an industrial atmosphere, reacts to form the reaction product rust, of approximate composition Fe,O, HzO, which being loosely adherent does not form a protective barrier that isolates the metal from the environment; the reaction thus proceeds at an approximately linear rate until the metal is completely consumed. Copper, on the other hand forms an adherent green patina, corresponding approximately with bronchantite, CuSO, 3Cu(OH), , which is protective and isolates the metal from the atmosphere. Copper roofs instalIed 200 years ago are still performing satisfactorily, and it is apparent that the formation of bronchantite is not deleterious to the function of copper as roofing material-indeed, in this particular application it is considered to enhance the appearance of the roof, although a similar patina formed on copper water pipes would be aesthetically objectionable. The rapid dissolution of a vessel constructed of titanium in hot 40Vo H, SO, with the formation of Ti4+aquo cations conforms with both definitions of corrosion, but if the potential of the metal is raised (anodic protection) a thin adherent protective film of anatase, TiO,, is formed, which isolates the metal from the acid so that the rate of corrosion is enormously decreased. The formation of this very thin oxide film on titanium, like that of the relatively thick bronchantite film on copper, clearly conforms with the transformation definition of corrosion, but not with the deterioration definition, since in these examples the rate and extent of the reaction is not significantly detrimental to the metal concerned. Again, magnesium, zinc or aluminium is deliberately sacrificed when these metals are used for the cathodic protection of steel structures, but as these metals are clearly not required to be maintained as such, their consumption in this particular application cannot, according to the deterioration, be regarded as corrosion. Furthermore, corrosion reactions are used to advantage in technological processes such as pickling, etching, chemical and electrochemical polishing and machining, etc. The examples already discussed lead to the conclusion that any reaction of a metal with its environment must be regarded as a corrosion process irrespective of the extent of the reaction or of the rates of the initial and subsequent stages of the reaction. It is not illogical, therefore, to regard passivity, in which the reaction product forms a very thin protective film that controls rate of the reaction at an acceptable level, as a limiting case of a corrosion reaction. Thus both the rapid dissolution of active titanium in 40% H,SO, and the slow dissolution of passive titanium in that acid must be
-
-
1:6
BASIC CONCEPTS OF CORROSION
regarded as corrosion processes, even though the latter will not be detrimental to the metal during the anticipated life of the vessel. It follows that in deciding whether the corrosion reaction is detrimental to a metal in a given application, the precise form of attack on the metal (general, intergranular, etc.), the nature of the reaction products (protective or non-protective), the velocity and extent of the reaction and the location of the corrosion reaction must all be taken into account. In addition, due consideration must be given to the effect of the corrosion reaction on the environment itself. Thus corrosion reactions are not always detrimental, and our ability to use highly reactive metals such as aluminium, titanium, etc. in aggressive environments is due to a limited initial corrosion reaction, which results in the formation of a rate-controlling corrosion product. Expressions such as ‘preventing corrosion’, ‘combating corrosion’ or even ‘fighting corrosion’ are misleading; with the majority of metals corrosion cannot be avoided and ‘corrosion control’ rather than ‘prevention’ is the desired goal. The implication of ‘control’ in this context is that (a) neither the form, nor the extent, nor the rate of the corrosion reaction must be detrimental to the metal used as a constructional material for a specific purpose, and (b)for certain applications the corrosion reaction must not result in contamination of the environment. The scope of corrosion control is considered in more detail in the Introduction to Volume 2 , but it is relevant to mention here that it must involve a consideration of materials, availability, fabrication, protective methods and economics in relation to the specific function of the metal and its anticipated life, At one extreme corrosion control in certain environments may be effected by the use of thick sections of mild steel without any protective system, at the other the environmental conditions prevailing may necessitate the use of platinum. The scope of the term ‘corrosion’ is continually being extended, and Fontana and Staehle have stated3 that ‘corrosion will include the reaction of metals, glasses, ionic solids, polymeric solids and composites with environments that embrace liquid metals, gases, non-aqueous electrolytes and other non-aqueous solutions’. Vermilyea, who has defined corrosion as a process in which atoms or molecules are removed one at a time, considers that evaporation of a metal into vacuum should come within the scope of the term, since atomically it is similar to other corrosion processes4. Evans’ considers that corrosion may be regarded as a branch of chemical thermodynamics or kinetics, as the outcome of electron affinities of metals and non-metals, as short-circuited electrochemical cells, or as the demolition of the crystal structure of a metal. These considerations lead to the conclusion that there is probably a need for two definitions of corrosion, which depend upon the approach adopted: 1. Definition of corrosion in the context of Corrosion Science: the reaction of a solid with its environment. 2. Definition of corrosion in the context of Corrosion Engineering: the reaction of an engineering constructional metal (material) with its environment with a consequent deterioration in properties of the metal (material).
BASIC CONCEPTS OF CORROSION
1:7
Methods of Approach to Corrosion Phenomena The effective use of metals as materials of construction must be based on an understanding of their physical, mechanical and chemical properties. These last, as pointed out earlier, cannot be divorced from the environmental conditions prevailing. Any fundamental approach to the phenomena of corrosion must therefore involve consideration of the structural features of the metal, the nature of the environment and the reactions that occur at the metal/environment interface. The more important factors involved may be summarised as follows: 1. MetaZ- composition, detailed atomic structure, microscopic and macroscopic heterogeneities, stress (tensile, compressive, cyclic), etc. 2. Environment -chemical nature, concentrations of reactive species and deleterious impurities, pressure, temperature, velocity, impingement, etc. 3 . Metd/environment interface - kinetics of metal oxidation and dissolution, kinetics of reduction of species in solution; nature and location of corrosion products; film growth and film dissolution, etc. From these considerations it is evident that the detailed mechanism of metallic corrosion is highly complex and that an understanding of the various phenomena will involve many branches of the pure and applied sciences, e.g. metal physics, physical metallurgy, the various branches of chemistry, bacteriology, etc. although the emphasis may vary with the particular system under consideration. Thus in stress-corrosion cracking (see Section 8.1) emphasis may be placed on the detailed metallurgical structure in relation to crack propagation resulting from the conjoint action of corrosion at localised areas and mechanical tearing, while in underground corrosion the emphasis may be on the mechanism of bacterial action in relation to the kinetics of the overall corrosion reaction (see Section 2.6). Although the mechanism of corrosion is highly complex the actual control of the majority of corrosion reactions can be effected by the application of relatively simple concepts. Indeed, the Committee on Corrosion and Protection6 concluded that ‘better dissemination of existing knowledge’ was the most important single factor that would be instrumental in decreasingthe enormous cost of corrosion in the U.K. Corrosion as a Chemical Reaction at a MetaVEnvironment lntertace
As a first approach to the principles which govern the behaviour of metals in specific environments it is preferable for simplicity to disregard the detailed structure of the metal and to consider corrosion as a heterogeneous chemical reaction which occurs at a metalhon-metal interface and which involves the metal itself as one of the reactants (cf. catalysis). Corrosion can be expressed, therefore, by the simple chemical reaction: . . .(l.l) aA + bB = CC+ dD where A is the metal and B the non-metal reactant (or reactants) and C and D the products of the reaction. The nonmetallic reactants are frequently
1:8
BASIC CONCEPTS OF CORROSION
referred to as the environment although it should be observed that in a complex environment the major constituents may play a very subsidiary role in the reaction. Thus in the ‘atmospheric’corrosion of steel, although nitrogen constitutes approximately 75% of the atmosphere, its effect, compared with that of moisture, oxygen, sulphur dioxide, solid particles, etc. can be disregarded (in the high-temperature reaction of titanium with the atmosphere, on the other hand, nitrogen is a significant factor). One of the reaction products (say, C) will be an oxidised form of the metal, and D will be a reduced form of the non-metal- C is usually referred to as the corrosion product, although the term could apply equally to D. In its simplest form, reaction 1.1 becomes
aA e.g.
4Fe
+ bB = CC
. . .(1.2)
+ 30, = 2Fe20,
where the reaction product can be regarded either as an oxidised form of the metal or as the reduced form of the non-metal. Reactions of this type which do not involve water or aqueous solutions are referred to as ‘dry’ corrosion reactions. The corresponding reaction in aqueous solution is referred to as a ‘wet’ corrosion reaction, and the overall reaction (which actually occurs by a series of intermediate steps) can be expressed as 4Fe + 2 H 2 0 + 3 0 2 = 2Fe20,.H,O
. .(1.3)
Thus in all corrosion reactions one (or more) of the reaction products will be an oxidised form of the metal, aquo cations (e.g. Fe2+(as.), Fe3+(as.)), aquo anions (e.g. HFeOAaq.), FeOi- (as.)), or solid compounds (e.g. Fe(OH),, Fe,O.,, Fe,O,-H,O, Fe,O, .H,O), while the other reaction product (or products) will be the reduced form of the non-metal. Corrosion may be regarded, therefore, as a heterogeneous redox reaction at a metalhonmetal interface in which the metal is oxidised and the non-metal is reduced. In the interaction of a metal with a specific non-metal (or non-metals) under specific environmental conditions, the chemical nature of the non-metal, the chemical and physical properties of the reaction products, and the environmental conditions (temperature, pressure, velocity, viscosity, etc.) will clearly be important in determining the form, extent and rate of the reaction. Environment
Environments are considered in detail in Chapter 2, but some examples of the behaviour of normally reactive and non-reactive metals in simple chemical solutions will be considered here to illustrate the fact that corrosion is dependent on the nature of the environment; the thermodynamics of the systems and the kinetic factors involved are considered in Sections 1.4 and 1.9.
Gold is stable in most strong reducing acids, whereas iron corrodes rapidly, yet finely divided gold can be quickly dissolved in oxygenated cyanide solutions which may be contained in steel tanks. A mixture of caustic soda and sodium nitrate can be fused in an iron or nickel crucible, whereas this melt would have a disastrous effect on a platinum crucible.
BASIC CONCEPTS OF CORROSION
1:9
Copper is relatively resistant to dilute sulphuric acid but will corrode if oxygen or oxidising agents are present in the acid, whereas austenitic stainless steels are stable in this acid only if oxygen or other oxidising agents are present. Iron will corrode rapidly in oxygenated water but extremely slowly if all oxygen is removed; if, however, oxygen is brought rapidly and simultaneously to all parts of the metal surface the rate will become very slow, owing to the formation of a protective oxide film. Lead will dissolve rapidly in nitric acid, more slowly in hydrochloric acid, and very slowly in sulphuric acid. These examples show that the corrosion behaviour of a metal cannot be divorced from the specific environmental conditions prevailing, which determine the rate, extent (after a given period of time) and form of the corrosion process. Metal
Heterogeneities associated with a metal have been classified in Table 1.1 as atomic (see Fig. 1. l), microscopic (visible under an optical microscope), and macroscopic, and their effects are considered in various sections of the present work. It is relevant to observe, however, that the detailed mechanism of all aspects of corrosion, e.g. the passage of a metallic cation from the lattice to the solution, specific effects of ions and species in solution in accelerating or inhibiting corrosion or causing stress-corrosion cracking, etc. must involve a consideration of the detailed atomic structure of the metal or alloy. The corrosion behaviour of different constituents of an alloy is well known, since the etching techniques used in metallography are essentially corrosion processes which take advantage of the different corrosion rates of phases as a means of identification, e.g. the grain boundaries are usually etched more rapidly than the rest of the grain owing to the greater reactivity of the disarrayed metal (see Sections 1.3 and 20.4). Table 1.1 Heterogeneities in metals 1. Atomic (as classified by Ehrlich and Turnbull’, see Fig. 1.1). (a) Sites within a given surface layer (‘normal’ sites); these vary according to the particular crystal plane (Fig. 1.2).
(b) Sites at edges of partially complete layers. (c) Point defects in the surface layer: vacanies (molecules missing in surface layer), kink sites (moleculesmissing at edge of layer), molecules adsorbed on top of complete layer. ( d ) Disordered molecules at point of emergence of dislocations (screw or edge) in metal surface. 2. Microscopic (a) Grain boundaries-usually, but not invariably. more reactive than grain interior. (b) Phases-metallic (single metals, solid solutions. intermetallic compounds), nonmetallic, metal compounds. impurities, etc. -heterogeneities due to thermal or mechanical causes. 3 . Macroscopic (a) Grain boundaries. (b) Discontinuities on metal surface- cut edges, scratches, discontinuities in oxide films (or
other chemical films) or in applied metallic or non-metallic coatings. (c) Bimetallic couples of dissimilar metals. (d) Geometrical factors-general design, crevices, contact with non-metallic materials. etc.
1:10
BASIC CONCEPTS OF CORROSION
Fig. 1.1 Surface imperfections in a crystal (after Erlich and Turnbull ’)
Fig.l.2 Hard-sphere model of face-centred cubic (f.c.c.) lattice showing various types of sites. Numbers denote Miller indices of atom places and the different shadings correspond to differences in the number of nearest neighbours (courtesy Erlich and Turnbull’)
BASIC CONCEPTS OF CORROSION
Fig.l.3
1:11
Environments in corrosion
Macroscopic heterogeneities, e.g. crevices, discontinuities in surface films, bimetallic contacts etc. will have a pronounced effect on the location and the kinetics of the corrosion reaction and are considered in various sections throughout this work. Practical environments are shown schematically in Fig. 1.3, which also serves to emphasise the relationship between the detailed structure of the metal, the environment, and external factors such as stress, fatigue, velocity, impingement, etc.
Types of Corrosion Corrosion can affect the metal in a variety of ways which depend on its nature and the precise environmental conditions prevailing, and a broad classification of the various forms of corrosion in which five major types have been identified, is presented in Table 1.2. Thus an 18Cr-8Ni stainless steel will corrode uniformly during polishing, active dissolution or passivation, but wiII corrode Iocally during intergranular attack, crevice corrosion or pitting; in certain circumstances selective attack along an ‘active path’ in conjunction with a tensile stress may lead to a transgranular fracture. Types of corrosion are dealt with in more detail in Appendix 1.1A.
1: 12
BASIC CONCEPTS OF CORROSION
Tabk 1.2 Types of corrosion Type 1. Uniform (or
almost uniform)
2. Localised
3. Pitting
4. Selective dissolution
5 . Conjoint action of
corrosion and a mechanical factor
Characteristic All areas of the maal corrode at the same (or similar) rate
Certain areas of the metal surface corrode at higher rates than others due to ‘heterogeneities’in the metal, the environment or in the geometry of the structure as a whole. Attack can range from being slightly localised to pitting Highly localised attack at specific areas resulting in small pits that penetrate into the metal and may lead to perforation One component of an alloy (usually the most active) is selectively removed from an alloy Localised attack or fracture due to the synergistic action of a mechanical factor and corrosion
Examples Oxidation and tarnishing; active dissolution in acids; anodic oxidation and passivity; chemical and electrochemical polishing; atmospheric and immersed corrosion in certain cases Crevice corrosion; filiform corrosion; deposit attack; bimetallic corrosion; intergranular corrosion; weld decay
Pitting of passive metals such as the stainless steels, aluminium alloys, etc., in the presence of specific ions, e.g. C1- ions Dezincification; dealuminification; graphitisation Erosion -corrosion, fretting corrosion, impingement attack, cavitation damage; stress corrosion cracking, hydrogen cracking, corrosion fatigue
Ideally, the metal selected, or the protective system applied to the metal, should be such that no corrosion occurs at all, but this is seldom technologically or economically feasible. It is necessary, therefore, to tolerate a rate and a form of corrosion that will not be significantly detrimental to the properties of the metal during its anticipated life. Thus, providing the corrosion rate is known, the slow uniform corrosion of a metal can frequently be allowed for in the design of the structure; for example, in the case of a metal that shows an active/passive transition the rate of corrosion in the passive region is usually acceptable whereas the rate in the active region is not. It follows that certain forms of corrosion can be tolerated and that corrosion control is possible, providing that the rate and form of the corrosion reaction are predictable and can be allowed for in the design of the structure. Pitting is regarded as one of the most insidious forms of corrosion, since it frequently leads to perforation and to a consequent corrosion failure. In other cases pitting may result in loss of appearance, which is of major importance when the metal concerned is used for decorative architectural purposes. However, aluminium saucepans that have been in service for some time are invariably pitted, although the pits seldom penetrate the metal, i.e. the saucepan remains functional and the pitted appearance is of no significance in that particular application.
BASIC CONCEPTS OF CORROSION
1: 13
These considerations lead to the conclusion that the relationship between corrosion and deterioration of properties of a metal is highly complex, and involves a consideration of a variety of factors such as the rate and form of corrosion and the specific function of the metal concerned; certain forms of corrosion such as uniform attack can be tolerated, whereas others such as pitting and stress corrosion cracking that ultimately lead to complete loss of function, cannot. The implications of the terms predictab/e and unpredictable used in the context of corrosion require further consideration, since they are clearly dependent on the knowledge and expertise of the engineer, designer or corrosion designer who takes the decision on the metal or alloy to be used, or the procedure to be adopted, to control corrosion in a specific environmental situation. On this basis a corrosion failure (Le. failure of the function of the metal due to corrosion within a period that is significantly less than the anticipated life of the structure) may be the result of one or more of the following possibilities: 1. Predictable. (a)The knowledge and technology are available but have not been utilised by the designer; this category includes a wide variety of design features such as the wrong choice of materials, introduction of crevices and bimetallic contacts etc., and is the most frequent cause of corrosion failures. (b) The knowledge and technology are available, but have not been applied for economic reasons; e.g. inadequate pretreatment of steel prior to painting and the use of unprotected mild steel for silencers and exhaust systems of cars. 2. Unpredictable.(a)The design has been based on specific environmental conditions, which have subsequently changed during the operation of the process; in this connection it should be noted that small changes in the chemical nature of the environment, temperature, pressure and velocity may lead to significant changes in the corrosion rate and form: the catastrophic oxidation and failure of steel bolts in nuclear reactors in the U.K.resulting from an increase in the temperature of the carbon dioxide is an example of an unpredictable failure due to a change in environmental conditions. (b) There is insufficient knowledge and experience of the metal, alloy or the environment to predict with certainty that failure will not occur; examples could be quoted of new alloys that have been subjected to an extensive series of carefully planned corrosion tests, but have failed in service. Professor M. Fontma’ has made the statement that “Virtually all premature corrosion failures these days occur for reasons which were already well known and these failures can be prevented”. It is apparent from this statement, and from the conclusions reached by the Committee on Corrosion and Protection, that category 1 is responsible for the majority of incidents of corrosion failure that could have been avoided if those responsible were better informed on the hazards of corrosion and on the methods that should have been used to control it.
Principles of Corrosion It has been stated that metallic corrosion is an art rather than a science and that, at present, insufficient knowledge is available to predict with any
1: 14
BASIC CONCEPTS O F CORROSION
certainty how a particular metal or alloy will behave in a specific environment4. It should be appreciated that the decision to use a particular metal or alloy in preference to others in a given environment or to employ a particular protective system is based usually on previous experience and empirical testing (see Chapter 19) rather than on the application of scientific knowledge- the technology of corrosion is without doubt in advance of corrosion science and many of the phenomena of corrosion are not fully understood. Thus the phenomena of passivity which was first observed by Faraday in 1836 is still a subject of controversy, the specific effect of certain anions in causing stress-corrosion cracking of certain alloy systems is not fully understood, and dezincification of brasses can be prevented by additions of arsenic (or other elements such as antimony or phosphorus) but no adequate theory has been submitted to explain the action of these elements (see Section 4.2). An understanding of the basic principles of the science of metallic corrosion is clearly vital for corrosion control, and as knowledge of the subject advances the application of scientific principle rather than an empirical approach may be used for such purposes as the selection of corrosion inhibitors, formulation of corrosion-resisting alloys, etc.
Terminology
The classification given in Table 1.2 is based on the various forms that corrosion may take, but the terminology used in describing corrosion phenomena frequently places emphasis on the environment or cause of attack rather than the form of attack. Thus the broad classification of corrosion reactions into ‘wet’ or ‘dry’ is now generally accepted, and the nature of the process is frequently made more specific by the use of an adjective that indicates type or environment, e.g. concentration - cell corrosion, crevice corrosion, bimetallic corrosion and atmospheric corrosion, Table 1.3 Terminology in corrosion Type of attack
general (uniform) localised pitting (or intense) intergranular transgranular selective parting catastrophic layer filiform
Environmental Wet*
dry atmospheric immersed underground sea water chemical fused-salt flue-gas biochemical bacterial high-temperature liauid-metal
‘See Appendix to this section.
Cause of attack concentration cell bimetallic cell active-passive cell stray current (electrolysis) hydrogen evolution oxygen absoption impingement hydrogen embrittlement caustic embrittlement
Mechanical factors
Corrosion product
stress fretting fatigue cavitation erosion impingement
rusting tarnishing scaling green rot tin pest
BASIC CONCEPTS OF CORROSION
1:15
high-temperature corrosion, sea-water corrosion, etc. Alternatively, the phenomenon is described in terms of the corrosion product itself tarnishing, rusting, green rot. The terminology used in corrosion is given in Table 1.3 and is considered in more detail in Appendix 1.1A. L. L. SHREIR REFERENCES 1. Hoar, T.P.. J. Appl. Chem., 11. 121 (l%l); Vernon, W. H. J.. The Conservation of Natural Resources, Instn. of Civil Engrs., London, 105 (1957); Potter, E. C., Electrochemistv, Cleaver-Hume, London, 231 (1956) 2. Uhlig, H. H.(Ed.), The Corrosion Handbook. Wiley. New York and Chapman and Hall, London (1948);Uhlig, H.H.,Corrosion and Corrosion Control, Wiley, New York (1971); Fontana, M.G.and Greene, N. D., Corrosion Engineering, McCraw-Hill (1967) 3. Fontana, M. G. and Staehle, R. W., Advances in Corrosion Science and Technology, Plenum Press, New York (1990) 4. Vermilyea, D.A., Proc. 1st International Congress on Metallic Corrosion, London, 1961, Butterworths, London, 62 (1962) 5. Evans, U.R., The Corrosion and Oxidation of Metals. Arnold, London, 12 (1960) 6. Report of the Committee on Corrosion and Protection, Department of Trade and Industry, H.M.S.O.(1971) I . Ehrlich, G. and Turnbull, D., Physical Metallurgy of Stress Corrosion Fracture, Interscience, New York and London, 47 (1959) 8. Fontana, M.G., Corrosion, 27, 129 (1971)
1.IA
Appendix- Classification of Corrosion Processes
Existing Classifications A logical and scientific classification of corrosion processes, although desirable, is by no means simple, owing to the enormous variety of corrosive environments and the diversity of corrosion reactions, but the broad classification of corrosion reactions into ‘wet’ or ‘dry’ is now generally accepted, and the terms are in common use. The term ‘wet’ includes all reactions in which an aqueous solution is involved in the reaction mechanism; implicit in the term ‘dry’ is the absence of water or an aqueous solution. These terms are evidently ambiguous; for example, it is not always clear whether ‘wet’ is confined to aqueous solutions-the ‘wetting’ of solids by mercury indicates that liquid-metal corrosion should be classified as ‘wet’. Even if the term is restricted to aqueous solutions, the difficulty arises that the mechanism of growth of magnetite scale during the reaction of the interior of a boiler drum with dilute caustic soda at high temperatures and pressures is best interpreted in terms of a ‘dry’ corrosion process. Similar considerations apply to the reactions of aluminium and zirconium with hightemperature water. Considering oxidation as a typical ‘dry’ reaction it follows from Fig. 1.Ala that at the interfaces: M + (Mz+O/O)
+ z(eO/O)
where Mz+0 is an interstitial metal ion, e 0 an interstitial electron and /O indicates the metal/oxide interface (Section 1A). If the metal dissolves to enter a vacant site, then M
* (Mz+O/O + z e O / O )
where M Z + Orepresents a cation vacancy and e 0 a positive hole. At the gadoxide interface the O2gas ionises (fO,/ads.)
+ 2(e/X)
(O*-/ads.)
where /X indicates the gadoxide interface. By definition, these interfaces can be considered as anodes and cathodes respectively. 1: 16
1: 17
APPENDIX-CLASSIFICATION OF CORROSION PROCESSES Metal
Metal
ze
Metal
02
ML* 4
ze -c
Oxide film Anode
Anode
Cathode
Electrolyte solution
. Cathode
Fig.l.Al Anodes and cathodes in corrosion processes. (a)‘Dry’ corrosion and (a)‘wet’ corrosion
The corresponding ‘wet’ corrosion half-reactions (Fig. 1.Alb) are:
[Mz+ + ~e],,,~, + HzO Mz+aq. --t
and or
+ +
+ +
O2 2H20 4e = 4 0 H O2 4 H + 4e = 2Hz0
‘Dry Corrosion
These are generally metal/gas or metal/vapour reactions involving nonmetals such as oxygen, halogens, hydrogen sulphide, sulphur vapour, etc. and oxidation, scaling and tarnishing are the more important forms. A characteristic of these reactions is that the initial oxidation of the metal, reduction of the non-metal, and formation of compound must occur at one and the same place at the metahon-metal interface. Should the compound be volatile or discontinuous, further interaction at the interface (or through a thin film of constant thickness) is possible and in most cases the reaction rate will tend to remain constant with time (linear law). If the film is continuous it will present a barrier to the reactants and further interaction will necessitate passage of the reactants through the film by (a) diffusion of the non-metal or (a) diffusion and migration of ions of the reactants. The detailed mechanisms of these reactions are considered in Sections 1.8-1.10, but it is appropriate to observe that the formation of a continuous film of reactant product at a metalhon-metal interface will result in a growth rate which, when the film becomes sufficiently thick to be rate determining, decreases as the film thickens, Le. parabolic, logarithmic, asymptotic, cubic, etc. ‘Wet’ Conosion
In ’wet’ corrosion the oxidation of the metal and reduction of a species in solution (electron acceptor or oxidising agent) occur at different areas on the
1: 18
APPENDIX- CLASSIFICATION OF CORROSION PROCESSES
metal surface with consequent electron transfer through the metal from the anode (metal oxidised) to the cathode (electron acceptor reduced); the thermodynamically stable phases formed at the metalholution interface may be solid compounds or hydrated ions (cations or anions) which may be transported away from the interface by processes such as migration, diffusion and convection (natural or forced). Under these circumstancesthe reactants will not be separated by a barrier and the rate law will tend to be linear. Subsequent reaction with the solution may result in the formation of a stable solid phase, but as this will form away from the interface it will not be protective -the thermodynamically stable oxide can affect the kinetics of the reaction only if it forms a film or precipitates on the metal surface (see Sections 1.4 and 1.5). Further points which distinguish ‘wet’ from ‘dry’ corrosion are: 1. In ‘wet’ corrosion the metal ions are hydrated- the hydration energy of most metal ions is very large and thus facilitates ionisation (see Section 1.9). 2. In ‘wet’ corrosion ionisation of oxygen to hydroxyl must involve the hydronium ion or water. 3. In ‘dry’ corrosion the direct ionisation of oxygen occurs.
Corrosion in Organic Solvents Corrosion reactions in aggressive organic solvents are becoming a more frequent occurrence owing to developments in the chemical and petrochemical industries, and these reactions can lead to the deterioration of the metal and to undesirable changes in the solvent. This aspect of corrosion has recently been the subject of an extensive review by Heitz’ who has considered the mechanisms of the reactions, the similarities between corrosion in organic solvents and in aqueous solutions, the methods of study and the occurrence of the phenomenon in industrial processes. Ethanol
Acetone
Acetic acid
Water
Days
Fig.1.M Corrosion of nickel in different solvents containing 0.05 wt.% H2S0, at various temperatures (after Heitz’)
-
APPENDIX CLASSIFICATION OF CORROSION PROCESSES
1: 19
Figure 1.A2 shows the weight loss against time curve for nickel in various solvents containing 0.05 wt. 9'0H,S04 at various temperatures, and illustrates the unpredictable nature of corrosion in organic solvents. Thus the corrosion rates in ethanol are far greater than those in the aqueous acid whereas in acetone the rate is practically zero; even more surprising is the fact that in acetic acid the addition of 0.05% &SO, actually decreases the corrosion rate. Heitz classifies corrosion reactions in organic solvents into 1. Electrochemical reactions, which follow a similar mechanism to those
in aqueous solution. 2. Chemical reactions, which involve direct charge transfer between the metal atom in the lattice of the metal and the oxidising species. In the case of electrochemical reactions the partial anodic reaction results in the formation of a solvated metal cation M:&. , a charged or uncharged metal complex MX- or a solid compound MX,, where X is a halogen ion, organic acid anion, etc. The cathodic partial reactions are as follows: (a) Reduction of a solvated proton to Hz gas
H,+,,", + e + fH2 (b) Reduction of acidic hydrogen of a proton donor
HA+e+fH,+Awhere A - is a carboxylic acid anion, alcoholate ion, etc. (c) Reduction of an oxidising gas Y
Y, + zme zY"where Y can be O , , Cl,, F,, Br,, O,, N 2 0 4 ,etc. +
( d ) Reduction of oxidising ions such as Fe3+,Cuz+,M n O , Clog etc. It is evident from the above that in many systems the reaction of a metal with an organic solvent follows a mechanism that is similar to the electrochemical mechanism of corrosion in aqueous solution. Non-electrochemical processes may be represented by the general equation
where X is a halogen and M is a divalent metal, e.g. the Grignard reaction Mg + CH3C1-+ CH3MgCl A further type of chemical process, which is analogous to hightemperature corrosion, is the reaction of metals with organic sulphur compounds, which follow the equation 2 M + 2RSH
+
2MS
+ H, + R ,
Heitz quotes a number of case stydies of corrosion of metals in organic solvents and concludes that the phenomenology indicates no specific differences from that experienced in aqueous corrosion. Thus general corrosion, pitting, crevice corrosion, intergranular corrosion, erosion-corrosion cracking, hydrogen embrittlement, etc. can all occur in organic solvents. The methods of control also follow that used for corrosion in aqueous
1:20
APPENDIX-CLASSIFICATION OF CORROSION PROCESSES
solutions, although there are certain differences. Thus cathodic and anodic protection are seriously limited by the resistivity of the solvent, and paint coatings deteriorate rapidly in contact with the solvent. Suggested Classification and Nomenclsture
On a basis of the preceding discussion, the classification and nomenclature outlined in Table 1.Alis suggested as a possible alternative to the accepted classification of corrosion reactions into ‘wet’ and ‘dry’. Table 1.Al Classification of spontaneous corrosion reactions 1. Film-free Chemical Interaction
(a) MetaVgas- oxide or compound volatile (e.g. reaction of molybdenum with oxygen. reaction of iron or aluminium with chlorine). (b) MetalAiquid. Reactions of solid metals with liquid metals (e.g. dissolution of aluminium in mercury) Dissolution of metal in their fused halides (e.g. lead in lead chloride). Dissolution of metals in non-aqueous solutions (e.g. reaction of aluminium with carbon tetrachloride). 2 . Electrochemical (a) Inseparable anode/cathode type (insep. A / C ) Reactions with aqueous solutions. Uniform dissolution or corrosion of metals in acid, alkaline or neutral solutions (e.& dissolution of zinc in hydrochloric acid or in caustic soda solution; general corrosion of zinc in water or during atmospheric exposure). Reactions with non-aqueous solution (e.g. dissolution of copper in a solution of ammonium acetate and bromine in alcohol). Reactions with fused salts. (b) Separable anode/cathode type (sep. A/C) All reactions of metals in aqueous or non-aqueous solutions or in fused salts where one area of the metal surface is predominantly anodic and the other is predominantly cathodic so that the sites are physically identifiable. (c) Interfacial anode/cathode type in which the metal surface is filmed (i) Metallgas and metalhapour reactions All reactions in which charge is transported through a film of reaction product on the metal surface-the film may or may not be rate determining (e.g. parabolic, logarithmic, asymptotic, etc. or linear growth laws, respectively). (ii) Metalholution reactions All reactions involving the uniform formation and growth of a film of reaction product (e& reaction of metals with high-temperature water, reaction of copper with sulphur dissolved in carbon disulphide).
It is considered that the main types of corrosion reactions can be classified as follows: 1. Film-free chemical interaction in which there is direct chemical reaction of a metal with its environment. The metal remains film-free and there is no transport of charge. 2. Electrochemical reactions which involve transfer of charge across an interface. These electrochemical reactions can be further subdivided into: (a) Inseparable anode/cathode type (insep. A/C). The anodes and cathodes cannot be distinguished by experimental methods although their presence is postulated by theory, i.e. the uniform
APPENDIX- CLASSIFICATION OF CORROSION PROCESSES
1 :21
dissolution of metals in acid*, alkaline or neutral aqueous solutions, in non-aqueous solution, or in fused salts. (b) Separable anode/cathode type (sep. A/C). Certain areas of the metal can be distinguished experimentallyas predominantly anodic or cathodic, although the distances of separation of these areas may be as small as fractions of a millimetre. In these reactions there will be a macroscopic flow of charge through the metal. (c) Interfacial anode/cathode type (interfacial A/C). One entire interface will be the anode and the other will be the cathode. Thus in Fig. 1.Ala the metal/metal oxide interface might be regarded as the anode and the metal/oxygen interface as the cathode. It is apparent that, in general, 2(a) and 2(b) include corrosion reactions which are normally classified as ‘wet’, while 2(c) includes those which are normally classified as ‘dry’. The terminology suggested can be illustrated by reference to the corrosion behaviour of iron: 1. Reaction of iron with oxygen at room temperature or with oxygen or
water at high temperatures -interfacial A/C type.
2. Reaction of iron with oxygenated water or with reducing acidsinseparable A/C type.
3. Reaction of iron containing a discontinuous magnetite scale with oxygenated water, crevice corrosion, water-line attack, ‘long-line’ corrosion of buried iron pipes, etc. -separable A/C type.
Although it is realised that this classification and terminology has certain limitations, it represents a preliminary attempt to provide a more rational classification of corrosion processes than that based on ‘wet’ and ‘dry’. Acknowledgement
Grateful thanks are due to Dr. W. B. Jepson, Dr. M. Pryor and Mr. J. N. Wanklyn for helpful discussions during the preparation of this Appendix. L. L. SHREIR REFERENCES 1 . Heitz, E., ‘Corrosion of Metals in Organic Solvents’, Advunces in Corrosion Science und Technology (ed.M. G. Fontana and R. W. Staehle), Vol. 4, Plenum Press, 149 (1974)
*Dr.Pryor considers that in certain cases of uniform dissolution of metals in acids (e.g. A1 in hydrochloric or sulphuric acid) or alkalis a thin film of oxide is present on the metal surface- the film is not rate-determining but its presence would indicate that reactions of this type should be classified under 2 (c).
1.2 Nature of Films, Scales and Corrosion Products on Metals
The study of corrosion is essentiallythe study of the nature of the metal reaction products (corrosion products) and of their influence on the reaction rate. It is evident that the behaviour of metals and alloys in most practical environments is highly dependent on the solubility, structure, thickness, adhesion, etc. of the solid metal compounds that form during a corrosion reaction. These may be formed naturally by reaction with their environment (during processing of the metal and/or during subsequent exposure) or as a result of some deliberate pretreatment process that is used to produce thicker films or to modify the nature of existing films. The importance of these solid reaction products is due to the fact that they frequently form a kinetic barrier that isolates the metal from its environment and thus controls the rate of the reaction; the protection afforded to the metal will, of course, depend on the physical and chemical properties outlined above. In general, reaction products (films*, scales and corrosion products) may be formed under the following environmental conditions.
+
H, O,, H2S,etc.) at temperatures that range from ambient to very high (1 000-2 000OC). (b) Direct reaction with an aqueous solution with the formation of a thin invisible film (passivation) or of a thick visible corrosion product (protective or non-protective). (c) By the deliberate formation of thick oxide films (e.g. anodising) at elevated potentials or by changing the nature of existing films by chemical treatments (e.g. chromating or phosphating). (a) Direct reaction with a gas (0,, CO,, CO,
For example, in a dry atmosphere a reactive metal such as aluminium may carry a natural protective oxide film of only some 3 nm thickness, while for increased corrosion resistance aluminium may be anodised to give a coating lo4times thicker (see Section 15.1). However, thickness alone does not provide a criterion of protection; and although a thick protective layer of millscale is formed on iron and steel during processing it is not continuous owing to spalling, and the attack on the exposed substrate at the discontinuities is far greater than if the surface was bare. Thus the kinetics of attack *The distinction between a film and scale is not well defined, but it is usual to use the former when referring to a thin continuous layer of reaction product (visible or invisible) whilst the latter is normally used for thick high-temperature layer (always visible).
1:22
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
1 :23
will be related to a variety of other factors such as composition, structure, continuity, adhesion to the substrate, cohesion, mechanical properties, etc. of the film or scale of reaction products. This section describes in general terms the variation in the nature of very thin films originating in the initial reaction of a metal with its environment and their progression to the thicker overgrowths that control the kinetics. Recent developments in instrumental techniques have led to significant advances in the characterisation of these film- and scale-forming systems, and a summary of the experimental approaches available is provided at the end of the section. It is appropriate to consider first the products of reaction formed by a gaseous oxidising atmosphere and then to proceed to a consideration of the effect of water and aqueous systems. Initial Surface Reaction States
The application of ultra-high vacuum techniques to low-energy electron diffraction (L.E.E.D.) studies of very clean metal surfaces in low-pressure oxidising and sulphidising atmospheres over a range of temperatures above ambient has provided detailed information on the initial states of interaction’,’. The following sequence of events is generally observed in the case of exposure to oxygen: 1. Rapid physical adsorption of molecular oxygen. 2. Chemisorption of atomic oxygen to form a partial or complete monolayer. 3. Further chemisorption of atomic oxygen into a second layer and/or further physical adsorption of 0,. In Stage 2 a distinct structural modification to an expanded lattice at submonolayer coverages has been observed on nickel, indicating that the oxygen ions become progressively incorporated into the metal lattice. These twodimensional crystals then gradually transform into a three-dimensional nickel oxide lattice as more oxygen becomes incorporated. Subsequent exposure to high-temperature conditions (> 1 OOOOC) has confirmed the extreme stability of the Stage 2 state. Similarly, under low-temperature conditions ( X < f . Table 1.4 Spinel phases encountered in alloy oxidation n-type
MgFe,O, NiFe20, ZnFe20, ZnCr, O4 CoA12O4 NiAI, 0,
1 :26
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
It should be noted that single metal oxides such as Fe304and c0304 are inverse spinels, while Mn304is a normal spinel. The spinel structure is prominent in the oxides on iron and The oxides M,O, (and also the hydroxides and oxy-hydroxides M(OH), and MO-OH) exist in the a and y forms. Corundum and haematite represent the isostructural a forms, while they forms have cubic spinel-like structures deficient in metal ions. For example, in y-Fe,03 there are only 21 f Fe3+ ions per unit cell of 320’ions, and these are randomly distributed among the eight tetrahedral and 16 octahedral ‘available’ sites. In magnetite, represented as Fe3+(FeZ+ Fe3+)04, one third of the cations are Fez+ and continuous interchange of electrons between Fez+ and Fe3+ ions in the 16-fold positions accounts for its extremely high electronic conductivity. Careful oxidation of FegO4 yields y-Fe,O,, which may be converted back into Fe304by heating in vacuo at 250°C. Because wustite (FeO) ideally has the NaC1-type structure (f.c.c. anion lattice), with four Fez+ and four 0’-ions per unit cell, deviations from stoichiometrylead to not every octahedral site being filled in the metal deficient lattice (e.g. at 57OOC Fe,.9,0 contains cation vacancies and compensating Fe3+ions). At lower temperatures disproportionation occurs: 4FeO
a-Fe
+ Fe304
Therefore the relationship between these interconvertible structures originates from a cubic anion lattice of 320’- ions in the cell. With 32 Fez+ ions in the octahedral holes stoichiometric FeO is formed. Replacement of a number of Fez+ ions with two-thirds of their number of Fe3+ions maintains electrical neutrality but provides non-stoichiometric Fe, - xO. Continual replacement in this way to leave 24 Fe atoms in the cubic cell produces Fe304, and further exchange to an average of 21fFe3+ ions leads to y-Fe203
Fe, -,O
-+
Fe3044 y-Fe203
In actual oxidation, the cubic anion lattice becomes extended by the addition of new layers of close-packed 0’ions into which Fe atoms migrate to give rise to the appropriate stable structures. The defect y-structures may be stabilised by the presence of Li+ or H+ ions (e.g. LiFe,O,). Cation diffusion rates in these and other lattices developed on metal surfaces play an important r61e in governing corrosion behaviour . Surface Reaction Products Formed in Aqueous Environments
Whereas a film formed in dry air consists essentially of an anhydrous oxide and may reach a thickness of 3 nm, in the presence of water (ranging from condensed films deposited from humid atmospheres to bulk aqueous phases) further thickening occurs as partial hydration increases the electron tunnelling conductivity’. Other components in contaminated atmospheres may become incorporated (e.g. H,S, SO2, CO,, Cl-), as described in Sections 2.2 and3.1. Films may thus range from thin transparent oxides (passive films on Al, Cr, Ti and Fe-Cr alloys), or thin visible sulphides (on Cu and Ag) to thicker
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
1 :27
Table 1.5 Variations in the nature and thickness of the product formed on aluminium under different conditions Formation conditions
Nature of oxide film
Dry air or Oz Humid atmosphere Boiling water Chemical conversion Anodic oxidation (barrier films)
Amorphous A I 2 0 3 AlOOH + AlzO3.3H2O MOOR (or AI203.HzO AlOOH + anions of solution Amorphous + crystalline N,O, + anions of solution
Thickness (nm) 1-2 50-100
500-2000
1oO0-so0O 1000-3000
'visible films, which may be compact, adherent and protective (anodic oxide films on Al and Ti, PbSO, films on Pb, etc.) or bulky, poorly adherent and non-protective (rust on steel, 'white rust' on Zn). In some cases, fairly precise limits can be placed on the nature and thickness of the products formed under different conditions, as with aluminium illustrated in Table 1.5. In other cases, the undesirable wastage of the basis metal (e.g. the rusting of steel) is of more significance than the thickness of the corrosion product, although the nature of the latter may provide information useful in interpreting the mechanism of its formation. Thus in industrial atmospheres the presence of FeSO, .4Hz0 has been identified in combination with a-and y- FeO.OH, and the two latter incorporate free water in excess of the composition Fez03. H 2 0 . Furthermore, although some of the corrosion product may be adherent, most of it is not'2 (Sections 3.1 and 3.2). In the fully immersed situation where the corrosion product is produced by a secondary reaction such as M 2 + + 2 H 2 0 + M(OH)z + 2H+, as in the case of iron or zinc in dilute aqueous aerated chloride solutions, the sites of the anodic and cathodic processes are separated, and widely so in the partially immersed condition. Thus OH- ions are formed at the cathode and Mzl ions at the anode, giving rise to dispersed M(OH)z where they meet and react; under these circumstances the corrosion product cannot influence the kinetics. If chloride or sulphate is present, a basic compound M,(OH),(X), may form whose range of stability will depend upon the concentration of the anion pX and the pH of the solution; diagrams with axes pX and pH have been constructed that show the range of stability of these basic compounds. In the case of iron, the Fe(OH), formed initially is subsequently oxidised to yellow FeO(0H) or Fe203.HzO, or in low oxygen conditions black Fe, 0,is formed containing green reduced corrosion products. Vertical surfaces allow ready detachment of the products formed, while they may settle on a horizontally corroding surface and provide some blanketing action, restraining access of oxygen to the surface. Precise identification of the products and a knowledge of the pH at their location on the surface may provide information on the conditions of formationt3. Thin Passive Films
In considering passivity and passivation (Sections 1.4 and 1.9, the nature of the surface product (the passivating film) entering into the process between
1 :28 NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
the curve for active dissolution and that for the onset of film breakdown or oxygen evolution, assumes considerable significance. As the system passes from the active to the passive state the initial interaction depends on the composition of the aqueous phaseL4.An initial chemisorbed state on Fe, Cr and Ni has been postulated in which the adsorbed oxygen is abstracted from the water molecules’. This has features in common with the metal/gaseous oxygen interaction mentioned previously. With increase in anodic potential a distinct ‘phase’ oxide or other film substance emerges at thicknesses of 1-4nm. Increase in the anodic potential may lead to the sequence
M-M-OH
monolayer
+M(OH)2+ multilayer
MO phase oxide
which has been suggested for Ni in acid solutions, and Cd and Zn in alkaline solutions. On the other hand, Fe in strong H2S04first forms a layer of FeSO, crystals, which at higher potentials is replaced by an Fe203film, the normal product formed during anodic polarisation in dilute acid 15. In nearneutral solutions the passive film on Fe (2-6nm thick) has been characterised as the so-called cubic oxide y-Fe,O, overlying a thin film of Fe,O, on the metal surfaceI6. The nature of y-Fe,O, in passive films is very significant and has been reviewed in detail”. Here again a spinel structure is prominent (derived from magnetite). Its structure is considered to be cation defective with protons (H+) progressively replacing Fez+ions in the Fe304spinel, and leading to a continuous series of solid solutions of which Fe,O, and Fe,O, are the end products. In some cases an HFe,O, composition is indicated in which some Fez+ ions have been replaced by protons. The implication of this mechanism of replacement of Fez+ ions is that water is incorporated into the passive film by a process of oxidative hydrolysis of the initial Fe, 0, substrate as the potential of the metal is progressively raised. An important feature of such films is their low ionic conductivity that restricts cation transport through the film substance. Electronic semiconduction, however, permits other electrode processes (oxidation of H 2 0 to 0,) to take place at the surface without further significant film growth. At elevated anodic potentials adsorption and entry of anions, particularly chloride ions, may lead to instability and breakdown of these protective films (Sections 1.5 and 1.6). Thick Anodic Films
Where the electronic conductivity of the film substance is low, as in the case of the ’valve’metals (Al, Nb, Ta, Zr, Ti), an increase in anode potential gives rise to a high electric field across the passive layer. Under these circumstances ion transport occurs and film growth continues to several hundred volts with thicknesses rising to hundreds of nanometres. At low voltages an amorphous or microcrystalline ‘barrier’ oxide is formed, which may recrystallise thermally or by the action of a high field to y-Al,O, , /3-Ta20, or TiO,, etc. A ‘mosaic’ structure has been attributed to these amorphous films” to account for their high field conduction properties. In the case of
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
1 :29
a valve metal with variable valency a number of anodic oxides may form over a range of anodic potential, e.g. Ti in strongly oxidising conditions gives TiO,, while anodic passivation at lower potentials leads to Ti,03, 3-4 TiO,, or even Ti,05. Furthermore, different structural modifications can be produced depending on the precise conditions of formation. For example, with AIMand Ti21 high temperatures and high formation voltage tend to favour crystalline modifications as compared with the more commonly observed amorphous oxides. While, in general, anodic films produced represent those expected from thermodynamic data, significant free-energy gradients may exist across the film substance. Such situations may lead to complex geometrical arrays of different compounds as shown by BurbankZ2 Potential IV;S.H.E.)
> 1.8
1.77
1.69 1.66 I - 51
0.25
0.21
0.00
-0.30
Fig. 1.4 Schematic representation of the reaction products formed on lead in su$huric acid and their distribution over a range of anodic potentials (after Burbank )
Table 1.6
Schematic representation of experimental techniques and their range of application (extended from the table of Wood l o )
Technique (electrochemica,!) Cyclic voltammetry (adsorption, monolayers) Potentiodynamicpolarisation (passivation, activation) Cathodic reduction (thickness) Frequency response analysis (electrical properties, heterogeneity) Chronopotentiornetry (kinetics) Chronoamperometry(kinetics) Photoelectrochemical methods (electronic properties, heterogeneity)
Technique Ellipsometry (kinetics) Electrometric reduction (kinetics; thickness) Interference colours and spectrophotometry (kinetics; thickness) A.C. impedance (thickness; conduction mechanisms and profiles; compactness; crystallinity) Electrical methods (kinetics; thickness) Manometric and volumetric methods (kinetics) Thermogravimetry (kinetics from very thin films to thick scales; stoichiometry) Electrical conductivity of oxides and allied methods (defect structures; conduction mechanisms; transport numbers) Radioactive tracers and allied methods (kinetics; self diffusion; markers) Inert markers (transport mechanisms) Gas adsorption (surface area)
Thickness range (approximate) of technique I nm
ADSORBED LAYERS, VERY THIN FILMS AND NUCLEI
Technique
I spectroscopy X-ray photo-electron spectroscopy (composition, Secondary ion mass thickness) spectrometry Ion scattering
[FILMS
I
Stress measurements
v
Adhesion
200 nm
Stress/strain characteristics
1
1 pm
1
100 pm
Creep Hardness (oxide mechanical properties; oxygen solution in metal) Thermal cycling tests
T
2 E
Surface-enhanced Raman spectroscopy (chemistry) Laser microprobe mass spectrometry (composition) X-ray fluorescence analysis (composition; thickness) X-ray diffraction (structure; grain size; preferred orientation; stress) Scanning laser microscopy Optical microscopy (local thickness; topography; nucleation; general morphology; internal oxidation) I.R. spectroscopy (specialised analysis and applications) Spectrographic analysis (trace element analysis) Chemical analysis (analysis; stoichiometry) Vacuum fusion analysis (oxygen solubility in metal)
b
z
tY
1:32
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
for anodic films of lead in sulphuric acid (Fig. 1.4) in which it can be seen that the nature and thickness of the oxidation products are highly dependent on the anodic potential. In the particular case of aluminium in acid electrolytes, an initially formed thin barrier film breaks down to give a porous coating which can be grown to a considerable thickness (Table 1.5). The voltage remains low as the porous anodic coating continues to thicken. Significant amounts of the acid anion (SO:-, PO:-, CrOi-) may be incorporated into the oxides so produced, together with protons to provide a degree of hydration (see Section 15.1). These features can significantly influence the structure and properties of the coatings obtained. Techniques of Examination
This limited survey has indicated the wide range of chemical compounds, particularly oxides, which may be formed on a metal surface as a result of a corrosion process. The nature of such films and scales needs to be carefully characterised. Fortunately, a wide spectrum of experimental techniques is now available to provide such valuable information, and others are under development. A convenient summary is provided in Table 1.6. In this scheme the nature of the surface product is arbitrarily divided into (a) adsorbed layers, very thin films and nuclei (1-200 nm thickness); (b) thin films (200 nm-1 pm), and (c) scales (above 1 pm). The principal techniques are located as appropriately as possible to indicate their areas of useful application. The spectrum thus ranges from the regime of very clean metal surfaces to grossly thick scales which may result from exposure to industrial oxidising atmospheres. Initial interaction may be studied by field-ion or fieldemission spectroscopy and low energy electron diffraction, after which time the kinetics of the growth process may be followed by such techniques as ellipsometry, thermogravimetry, or electrometric reduction, while the structure may be examined by electron microscopy, electron diffraction or X-ray microanalysis. Stoichiometric and defect characteristics may be examined by a number of electrical methods. As the thickness approaches scale dimensions less sensitive techniques become applicable. Information on stress distribution, hardness, porosity, adhesion as well as thermal cycling characteristics also become accessible. Chemical analysis, scanning electron microscopy, X-ray diffraction techniques and gas adsorption data may provide further information on the composition, structure and porosity of thick scales, while electron probe microanalysis permits detailed examination of the concentration profiles across specimen sections. Many of these techniques are equally applicable to films formed under aqueous electrochemical conditions.
Recent Developments In recent years the number of techniques available for analysis of metal surfaces has proliferated greatlyz3-”. Many of the new methods are ultrahigh vacuum (UHV) techniques suitable for analyses of films ranging in
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON
METALS 1 :33
thickness from a single monolayer to around a micrometre23-28. These techniques are still being improved and updated and many of them have attained a high degree of accuracy and sensitivity. Most noteworthy and probably most widely spread are X-ray photoelectron spectroscopy (XPS) and Auger electron spectroscopy (AES). These highly sensitive UHV techniques provide quantitative chemical analyses of surfaces and are sensitive to even submonolayer levels of atoms. They are sensitive to all atoms except hydrogen (and helium for AES). Even here, XPS can be used to provide some information on the presence of H+ in oxide films by analysis of the oxygen signal. A E S has the great advantage over XPS of being highly spatially resolved, enabling chemical 'maps' to be generated; these show the distribution of elements across the surface. XPS, although less spatially resolved (recent developments of the technique have improved this significantly), has the advantage over AES of being sensitive to the chemical state of the atoms; the technique can distinguish readily atoms in different oxidation states. Both techniques can be used to generate depth-profiles of the composition. Secondary ion mass spectrometry (SIMS) and ion scattering spectroscopy (ISS) fulfil a similar function to AES and XPS. They are less widely available, but can be used to great sensitivity (sub-monolayer up to around a micrometre, with depth profiling) and can be used for elemental mapping. To date, they are less quantitative than AES and XPS. The composition of surface films can be determined as a function of depth using these UHV techniques. Such depth profiles are usually provided by sequential removal and analysis of layers of the surface films, removal being achieved by sputtering with an ionized noble gas beam. XPS can alternatively achieve a depth/composition profile by angular resolution, a nondestructive technique, successful for films up to the escape depth of the photoelectrons, typically around 1 to 3 nm in thickness. The technique finds widespread use in the analysis of the very thin passivating films formed electrolytically on metals such as stainless steels, for which it is very powerful indeed. These UHV methods generally provide ex-situ analyses, that is to say, the surface must be removed from the environment in which the film was formed and transferred to a UHV chamber; some features of the surface films may be altered by the analytical technique itself, particuIarly with very thin films which are formed electrochemically. The same is true of laser microprobe mass spectrometry (LAMMS), a very rapid method of producing a spot elemental analysis of a surface to a depth of around a micrometre, but not yet fully quantitative. LAMMS operates by transient ablation of the surface with an intense focused laser beam, and issues a mass spectrum of the ablated fragments. Because AES uses a primary electron beam as a probe, the technique can be more destructive to the surface than XPS, which employs a beam of soft X-rays. Several UHV techniques which have been developed have not found such wide use in corrosion analysis, despite potential applicability. Ultraviolet photoelectron spectroscopy (UPS) is one of these, operating in a similar fashion to XPS (but using an ultraviolet excitation), and probing the valence electrons, rather than the core electrons of the atoms. Because the energies of the valence electrons are so very sensitive to the precise state of the atom, the technique is in principle very informative; however exactly this high sensitivity renders the data difficult to interpret, particularly as a routine
1:34
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
analytical procedure. By and large the techniques which find application in corrosion are those which are relatively easy to use and easy to interpret. Electrical characteristics of surface films formed electrochemically can be analysed using frequency response analysis (FRA) (sometimes called elec29*30*31. This technique is trochemical impedance spectroscopy, or EIS)23-zs. capable of detecting separate components of films by resolving their separate resistance and capacitances in situ, for which most other electrochemical techniques are blind. The method has found wide application in the analysis of the passive state. It is also widely used to yield useful information on the state of applied surface coatings, such as paints. Measurement of photocurrents generated by illuminating the surface while it is polarized in solution is increasingly being used to probe electronic By focusing properties of surface films generated electrochemi~ally~~*~*~~. the light source and scanning the probe over the electrochemically polarised surface, this technique can be used to yield a photocurrent map of the surface. Other in situ measurements employing illumination of thin surface films generated electrochemically also yield characteristic information on passivating oxide films; these include ellipsometry, infrared spectroscopy and surface-enhanced Raman spectroscopy (SERS)23*24s30331. The very new techniques of scanning tunnelling microscopy (STM) and atomic force microscopy (AFM) have yet to establish themselves in the field of corrosion science. These techniques are capable of revealing surface structure to atomic resolution, and are totally undamaging to the surface. They can be used in principle in any environment in situ, even under polarization within an electrolyte. Their application to date has been chiefly to clean metal surfaces and surfaces carrying single monolayers of adsorbed material, rendering examination of the adsorption of inhibitors possible. They will indubitably find use in passive film analysis. C. J. L. BOOKER G. T. BURSTEIN
REFERENCES 1. Benard, J., ‘Adsorption of Oxidant and Oxide Nucleation’. in Oxidation of Metals and Alloys, Seminar. 1970; American Society for Metals, Ohio, 1 (1971)
2. Uhlig, H. H., Proceedings of the Third International Congress on Metallic Corrosion, Moscow, 1966, Vol. 1, 25 (1%9); aIso Corros. Sci., 7, 325 (1%7) 3. Fehlner, F. P. and Mott, N. F., ‘Oxidation in the Thin-film Range’, as Reference 1, 37 (1971) 4. Fehlner, F. P. and Mott, N. F.. Oxid. Meiak. 2, 59 (1970)
5. Cathcart, J. V., ‘The Structure and Properties of Thin Oxide Films’. as Reference 1, 17 (1971) 6. Grauer, R. and Feitknecht, W., Corrosion Sei., 6. 301 (1966) 7 . Hauffe, K.. Metallobeg7ache. 8, 97 (1954) 8. Kubaschewski, 0. and Hopkins, B. E., Oxidation of Meials and Alloys, Butterworths, London, 114 (1967) 9. Wood, G. C., ‘The Structures of Thick Scales on Alloys’, as Reference 1, 201 (1971) IO. Wood, G. C., in Techniquesin Metals Research, Rapp. R. A. (Assoc. Ed.), Vol. 4, Interscience, New York, 494 (1970) 11. Douglas, D. L.. ‘Exfoliation and the Mechanical Propertiesof Scales’. as Reference 1. 137 (1971)
NATURE OF FILMS, SCALES AND CORROSION PRODUCTS ON METALS
1:35
12. Evans, U. R.. Corrosion and Oxidation of Metals, First Supplementary Volume, Arnold, London, 194 (1968) 13. Feitknecht, W. and Keller, G . , Z . Anorg. Chem., 262, 61 (1950); Feitknecht, W., Weidman, H. and Haberli, E., Helv. Chim. Acta., 26, 1911 (1943), 32,2294 (1949) and 33,922 (1 950) 14. Brusic, V., ‘Passivation and Passivity’, in Oxides and Oxide Films (Ed. Diggle, J. W.), Marcell Dckker, New York (1972) 15. Evans, U. R., as Reference 12, 98 (1%8) 16. Bloom, M. C. and Goldberg, L., Corros. Sci., 5 , 623 (1965) 17. Dignam, M. J., as Reference 14, 91 (1972) 18. Wells, A. F., Structural Inorganic Chemistry, 3rd edn, Clarendon Press, Oxford (1962) 19. Greenwood, N. N., Ionic CvstuLs, Lattice Defects and Nonstoichiometry, Butterworths, London, 92, 101 (1968) 20. Diggle, J. W.,Downie, T. C. and Goulding,. C. W., Chem. Rev., 69,365 (1969) 21. Aladjem, A., J. Mat. Sci., 8, 688 (1973) 22. Burbank, J., J. Electrochem. Soc., 106, 369 (1959) 23. Froment, M. (Ed.), Passivity of Metals and Semiconductors. Elsevier, Amsterdam, (1983) 24. MacDougall, B. R., Alwitt, R. S. and Ramanarayanan, T.A. (Eds.). Oxide Films on Metak and Alloys. Proceedings, 92-22, The Electrochemical Society, Pennington, New Jersey (1992) 25. McCafferty, E. and Brodd, R. J. (Eds.), Surfaces, Inhibition andPassivation. Proceedings, 86-7, The Electrochemical Society, Pennington, New Jersey (1986) 26. Rapp, R. A. (Ed.), High Temperature Corrosion. NACE, Houston, Texas (1983) 27. Bennett, M. J. and Lorimer, G. W. (Eds.), Microscopy ofoxidation. Institute of Metals, London (1991) 28. Augustynski, J. and Balsenc, L., in Modern Aspects of Electrochemistry. No. 13, (Eds. Conway. B. E. and Bockris, J. OM.), 251. Plenum Press. New York (1979) 29. Macdonald, D. D. and McKubre, M. C. H.. in Modern Aspects of Electrochemistry, 14, (Eds. Bockris, J. O’M.,Conway, B. E. and White, R. E.), 61, Plenum Press, New York (1982) 30. Ferreira. M. S. G. and Melendres, C. A. (Eds.), Electrochemical and Optical Techniques for the Study and Monitoring of Metallic Corrosion, NATO AS1 series, Kluwer Academic Publishers, Dordrecht (1991) 31. Efrirna, S., in Modern AspectsofElectrochembtry. 16,(Eds. Conway, B. E., White, R.E. and Bockris, J. O’M.).253, Plenum Press, New York (1985)
I .3 Effects of Metallurgical Structure on Corrosion*
The objective of this section is to show by means of specific examples how the various crystalline defects and structural features described in Section 20.4 can affect the form, location and kinetics of the corrosion of metals and alloys.
Effect of Crystal Defects on Corrosion- General Considerations Before considering specific examples it is appropriate to note that there are, in principle, two quite distinct ways in which crystal defects can affect corrosion behaviour . Firstly, they might be expected to have an effect when corrosion occurs under conditions of active (film-free) anodic dissolution and is not limited by the diffusion of oxygen or some other species in the environment. However, if the rate of active dissolution is controlled by the rate of oxygen diffusion, or if, in general terms, the rate-controlling process does not take place at the metal surface, the effect of crystal defects might be expected to be minimal. Secondly, crystal defects might be expected to affect the corrosion behaviour of metals which owe their corrosion resistance to the presence of thin passive or thick protective films on their surface. The crystal defects and structural features discussed in Section 20.4 might, in principle, be expected to affect the thickness, strength, adhesion, porosity, composition, solubility, etc. of these surface films, and hence, in turn, the corrosion behaviour of the filmed metal surfaces. Clearly, this is the more common situation in practice. Finally, it should be noted that in both cases the effect of crystal defects and microstructural features must, in general, be to tend to make the corrosion less uniform and more localised.
* The basic concepts of physical metallurgy are considered in Section 20.4, which should be regarded by those who are not conversant with the subject as an introduction to this section. Some of the diagrams referred to here will be found in Section 20.4. 1:36
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1 :37
Active Dissolution and Crystal Defects- Energy Considerations The crystal defects described in Section 20.4 are all regions of higher energy than the adjacent perfect crystal lattice; they are therefore all inherently more chemically active and hence are potential sites for preferential attack under conditions of active dissolution. This preferential attack is, however, masked in highly aggressive environments, when there is very rapid dissolution and severe general corrosion, and it is not observed when the corrosion rate is controlled either by oxygen diffusion or some other process not occurring at the metal/environment interface. Furthermore, although the energy associated with the various defects may be quite large in metallurgical terms, when converted to a potential difference it is quite small in electrochemical terms, being not more than a few millivolts, at the most.
Etching of Single Crystals and Polycrystals There is no evidence that any particular crystal structure is more readily corroded than any other. For example, the difference in the corrosion behaviour of austenitic and ferritic stainless steels is, of course, due to compositional rather than structural differences. Using single crystals it has been shown that different low-index crystal faces (see Section 20) exhibit different corrosion rates. However, the relative corrosion rate of the different faces varies with the environment and these structural effects are of little practical significance. On the other hand, the fact that polycrystal grains of different crystallographic orientation may corrode at different rates, is of some importance. A freshly polished metal surface appears quite featureless even when viewed at high magnification, while on etching different grains are attacked to differing degrees, as shown in Fig. 20.28 (bottom). The surface of grain B in Fig. 20.36a probably corresponds to a low-index low-energy plane while the surfaces of grains A and C correspond to high-index high-energy planes. In fact, the surfaces of grains A and C actually consist of low-index terraces separated by ledges with kinks in them. Since dissolution occurs most readily from kinks and ledges, owing to the lower co-ordination number of atoms at such sites, grains A and C will be attacked more rapidly than grain B, as illustrated schematically in Fig. 1 . 5 ~ .It must be emphasised, however, that this is primarily a laboratory effect, albeit an important one. In practice, preferential corrosion of grains of a particular crystallographic orientation is not generally a problem. One possible exception to this is the etching of coarse-grained brass door-handles by sweaty hands!
Dislocations, Etch Pits and the Effect of Cold Work on Corrosion Preferential corrosion or attack at many other types of crystal defect may also be best illustrated during the etching of metallographically polished
1:38
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
I I 1
I
A
o
Apparent width f grain-boundary
I
I I
(bl Fig. 1.5(u) Grain boundary intersecting an etched metallographic surface and (b)etch pit at a dislocation intersecting an etched metallographic surface
surfaces. Thus emergent dislocations intersecting metal surfaces may be revealed by the use of appropriate etchants. There is preferential attack at each dislocation, and small crystallographic etch pits are produced’ as shown schematically in Fig. 1.5b. However, this effect is again of little practical significance, and the development of etch pits is used primarily as a research technique in the study of dislocations. Moreover, the technique must be used with some caution with metals, since there is often doubt as to whether there is a one-to-one correlation between etch pits and emergent dislocations, and because etch pits can also develop at other defects in surface films. Finally, there are many instances where it is thought that etch pits are produced only as a result of segregation of impurities to the dislocations. ‘Clean’ dislocations may not result in etch pits.
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1 :39
Potentially of somewhat greater practical significanceis the effect of cold work on the corrosion of metals. When an annealed material is heavily cold worked, something of the order of 8-80 kJ/kg mol of energy may be stored in the material, as a result of the increased dislocation density, etc. This energy difference is, however, only equivalent to a potential difference of a few millivolts or so between the annealed and the cold-worked material. There is thus at the most only a small difference in the driving force for corrosion in the two cases. However, it is possible that the kinetics of the various anodic and cathodic processes could nevertheless be quite different on annealed and cold-worked surfaces; this would also result in annealed and cold-worked metals exhibiting significantly different corrosion rates. Certainly it has been experimentally observed that cold work markedly increases the corrosion rate of steel and aluminium in acids. The interpretation of this effect is, however, still not clear. Several authors suggest that the increased corrosion rate is due to the increased dislocation density per se, possibly as a result of an increased number of kink sites on the surface increasing the anodic exchange current density. On the other hand Foroulis and Uhlig’ suggest that the increased corrosion rate is due to the segregation of carbon and nitrogen to dislocations, and that the cathodic (hydrogen evolution) reaction is kinetically easier at these sites; this is supported by their observation that cold work does not increase the corrosion rate of highpurity iron. In natural waters, cold-worked commercial carbon steels of the same composition corrode at more or less the same rate as annealed steels, presumably because the corrosion rate in this case is controlled by the diffusion of oxygen. Unprotected carbon steels are sometimes exposed to natural waters, and it is this latter situation which is of greater practical importance than the behaviour of steels in acids, since steels should never be used in these environments unless they are protected.
Etching of Grain Boundaries and Intergranular Corrosion During metallographic etching, twin and grain boundaries are preferentially attacked, as is apparent in Fig. 20.28 (bottom). Shallow grooves develop at these boundaries, and they therefore appear, in the microscope, as . best dark lines of finite width, as illustrated schematically in Fig. 1 . 5 ~ The experimental evidence available indicates that even the grain boundaries in very high purity metals are slightly grooved by appropriate etchants. This is dueto the grain boundaries being inherently more active than the adjacent crystal lattice, as implied by the energy associated with grain boundaries in metals. However, the grain boundaries in impure metals and alloys are generally much more readily etched, primarily as a result of segregation to them of the impurities and alloying additions. In this context it is important to note that grain-boundary regions may be preferentially attacked either because segregation makes them more base or because segregation makes them more noble; in the latter case the grain boundary itself acts as a local cathode, and the region immediately adjacent to the grain boundary
1 :40
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
is preferentially attacked. The subject of segregation and preferential attack at grain boundaries has been reviewed by Aust and Iwao3. Again it must be emphasised that preferential etching of twin and grain boundaries is predominantly a laboratory effect. There are no practical instances of significant corrosion problems resulting from the preferential attack of twin boundaries. In practice, grain-boundary effects in metals and alloys are usually of little or no consequence in the corrosion of metals. Severe intergranular corrosion (in the absence of tensile stress) is generally observed as a practical problem only when there is very gross segregation or solute depletion at grain boundaries, or, in certain instances, when there is marked intergranular precipitation, as discussed below.
Intergranular Corrosion of Austenitic Stainless Steels (Section 3.3)
1200
1 1 00
/-
0.02 ' I . c
1000
Y
-2?
900
n
5
800
700
I
1
I
Time
I
I
.(SI
Fig. 1.6 Curves of the effect of temperature on the time required to sensitise two austenitic stainless steels of different carbon content
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1 :41
e
Fig. 1.7 Light micrograph showing intergranular corrosion of a sensitised austenitic stainless steel; x 200
As is well known, certain austenitic stainless steels may be ‘sensitised’ by certain heat treatments and made highly susceptible to intergranular corrosion. Sensitisation occurs when the alloys are held in, or slowly cooled through, the temperature range 1120-820 K. Quenching through the critical temperature range does not result in sensitisation. The degree of sensitisation, and therefore the susceptibility to intergranular corrosion, depend critically on the time at temperature, the temperature within the critical range or on the cooling rate through the critical temperature range, as well as on alloy composition, in particular the carbon content. These effects are illustrated schematically in Fig. 1.6. The intergranular corrosion, an example of which is shown in Fig. 1.7, is observed in a wide variety of environments in which austenitic stainless steels would normally be expected to have good corrosion resistance. The generally accepted mechanism for sensitisation and the resultant intergranular corrosion was first proposed by Bain, et aL4 and is basically as follows. During sensitisation, thin feathery precipitates of a chromiumrich carbide (Mac, where M = Feo.z-o.,Cro.,-o.,) nucleate and grow in the austenite grain boundaries. These carbide particles, which can only be seen using electron microscopy, are only stable below about 1120K; at higher temperatures they do not form, or if already present, tend to dissolve. On the other hand, below about 820 K, the diffusion rate of chromium in steels is too low for precipitation of the chromium-rich carbide to occur within a practical time scale. During precipitation of the carbide, which contains 70-8Owt.Vo Cr, the austenite matrix adjacent to the grain boundaries becomes depleted of chromium. In particular, the chromium level in these regions falls below the approximately 12% Cr required in solid solution to confer corrosion resistance, i.e. to permit the formation of a complete and protective passive film on the steel surface. The regions adjacent to the grain boundaries are therefore no longer passive and hence corrode preferentially. This mechanism is illustrated schematically in Fig. 1.8. The preferential attack of the non-passive chromium-depleted regions adjacent to the grain boundaries will be accelerated by the fact that these regions will be less noble than both the carbide precipitates in the grain boundary and the passive grain
1 :42
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
Chromium carbide precipitate
Chromiumdepleted zone
boundary
n
Approx.
'1. Cr
Chromium
carbide
Matrix
zone
0 (b)
boundary (C 1
Fig. 1.S(u) Intergranular precipitation of chromium carbide particles in a sensitised austenitic stainless steel and the consequent chromium-depleted zones adjacent to the grain boundaries, (b)variation of the chromium content across a grain boundary in a sensitisedausteniticstainless steel (18Cr) and (c) intergranular corrosion of a sensitised austenitic stainless steel
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1 :43
interiors. This effect is, of course, further exacerbated by the unfavourable high cathodelanode area ratio. It should be emphasised, however, that there will generally be little or no potential difference between the carbide precipitates and the passive grain interiors. Intergranular corrosion is therefore only observed when there is chromium depletion of the grain boundary regions and not when there is carbide precipitation without chromium depletion. In practice, three methods are available for preventing sensitisation and intergranular corrosion of austenitic stainless steels: 1. Quenching through the critical temperature range (if necessary after heat treating well above 1120K to dissolve any existing chromium carbides). 2. The use of very low carbon (usually 1 500 K) thus becomes sensitised and susceptible to intergranular corrosion. The remedy is to heat the fabricated structure or component to about 1340K after welding so that the chromium carbide dissolves and titanium (or niobium) carbide forms; following this solution treatment the rate of cooling is not important.
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1:45
Intergranular Corrosion of Aluminium Alhys
A number of aluminium alloys may also, depending on their metallurgical structure, be susceptible to severe intergranular corrosion. The alloys concerned are primarily the precipitation-hardenable Al-Cu and Al-Zn-Mg based alloys and the work-hardenable AI-Mg alloys containing more than 3% Mg. Since there is much less inherent difference in the corrosion resistance of the grain-boundary regions and the grain centres than there is in the case of sensitised austenitic stainless steels, the mechanism of the intergranular corrosion in these alloys is primarily electrochemical, involving local cell action between grain-boundary precipitates and the adjacent matrix. Aluminium alloys in which intergranular precipitation is not observed (e.g. commercial purity AI and AI-Mn alloys) or in which there is little or no potential differencebetween the matrix and any intergranular precipitates (e.g. balanced AI-Mg-Si alloys with Mg, Si intergranular precipitates) are generally not markedly susceptibleto severe intergranular corrosion. On the other hand, aluminium alloys in which the intergranular precipitates are markedly more noble than the matrix phase (e.g. AI-Cu base alloys with C u d , intergranular precipitates), or alloys in which the precipitates are markedly more base (e.g. AI-Mg alloys and Al-Zn-Mg base alloys with Mg, AI, and MgZn, intergranular precipitates, respectively) may be susceptible to severe intergranular corrosion. The latter precipitates corrode preferentially, while the former stimulate preferential corrosion of the adjacent matrix. The degree of susceptibilityto intergranular attack depends on the nature, amount, size, distribution, etc. of the infer-granular precipitates (and to a lesser extent of the infragranular precipitates), and hence on the heat treatment of the alloy (see Figs 20.31 and 20.34). In general, the precipitation-hardenable alloys are more likely to be susceptible to intergranular corrosion when aged to peak hardness and less likely to be susceptible in the overaged condition. In the work-hardenable AI-Mg alloys the tendency to intergranular precipitation (and hence to intergranular corrosion) increases with increasing Mg contents, with increasing cold work and with increased ageing times at temperatures below about 400 K. Chloridecontaining environments, in particular, are liable to cause severe intergranular corrosion of susceptible aluminium alloys. Intergranular Corrosion in Other AHoy Systems
A number of other alloy systems may also be susceptible to intergranular corrosion. For example, zinc die-casting alloys containing aluminium may be susceptible to intergranular attack in steam- and chloride-containing environments. Stray currents often result in intergranular corrosion of lead cable sheaths. These instances are, however, relatively unimportant compared to the intergranular corrosion of sensitised stainless steels and, to a lesser extent, to intergranular corrosion of intermediate- and high-strength aluminium alloys.
1 :46
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
Effect of Grain Structure on Corrosion The grain structure of alloys, as well as intergranular precipitation, can also markedly affect their corrosion behaviour. For example, the corrosion resistance of certain wrought metals may be less on surfaces perpendicular to the hot-or cold-working direction than on surfaces parallel to this direction. Typically there may be severe localised corrosion starting on the faces perpendicular to the working direction and proceeding into the metal in the working direction, while the surfaces parallel to the working direction remain relatively unattacked. Such end-grain attack, which is basically the result of the grain structure being elongated in the working direction, has been observed in austenitic stainless steels, titanium alloys and mild steel.
Layer Corrosion
The most marked effect of grain structure on corrosion is observed in wrought aluminium alloys. These alloys generally do not recrystallise during heat treatment after rolling, extrusion, etc. mainly because their grain boundaries are pinned by inclusions; they therefore exhibit the elongated pancake-shaped grain structure shown in Fig. 20.34. As a result of this structure, these alloys may be susceptible to eMoliation (also known as layer or lamellar) corrosion. The attack proceeds along a number of narrow planar paths (usually but not necessarily intergranular) parallel to the working direction. The corrosion products formed force the layers apart and cause the metal to swell and, in severe instances, to disintegrate into separate sheets of metal (i.e. to exfoliate). Exfoliation is most common and severe in AI-Cu, Al-Zn-Mg and A1-Mg based alloys, but mild exfoliation also occurs in Al-Mg-Si alloys. Since the exfoliation is normally intergranular it is clear that exfoliation and intergranular corrosion are associated, and exfoliation is usually affected by intergranular precipitation and hence by heat treatment. However, aluminium alloys that are susceptible to intergranular attack will not be susceptible to exfoliation corrosion if they have an equiaxed grain structure. Transgranular exfoliation is thought to be the result of segregation in the original ingot persisting in the wrought alloy.
Stress-corrosion Cracking
Grain structure also affects the stress-corrosion behaviour of high-strength age-hardenable aluminium alloys. Cracking in these alloys is always exclusively intergranular. When they are stressed in the short transverse direction (a in Fig. 1.10) their highly elongated, pancake-shaped grain structure ensures that an easy path for crack propagation is readily available. On the other hand, when stressed in the long-transverse or the longitudinal direction (b and c respectively in Fig. 1.10) the possible intergranular crack paths are clearly complex and difficult. Many high-strength aluminium alloys are therefore quite susceptible to stress-corrosion cracking when stressed in the short-transverse direction but quite resistant or immune when stressed in the long-transverse or longitudinal directions. This result is of considerable
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1 :47
Fig. 1.10 Grain structureof a wrought high-strengthprecipitation-hardeningaluminiumalloy showing potential crack growth paths
practical importance: high-strength aluminium alloys can often be used in sheet form (when the short-transverse tensile stresses are generally negligible) in tempers in which they cannot normally be used in thick sections of forgings (when the tensile stresses in the short-transverse direction may be high). The distinction is, for example, between an aircraft’s skin and its wing spars; with a susceptible alloy stress-corrosion cracking is likely to be a problem in the latter instance, but less so in the former.
Corrosion of Impure Metals and Single-phase Alloys Many of the forms of corrosion already discussed have been caused or affected not only by metallurgical structure but also by the segregation of impurities or alloying additions to dislocations, grain boundaries, precipitates, etc. However, the presence of impurities or alloying elements in homogeneous solid solution can also markedly affect corrosion behaviour, without any segregation effects. In the context of this section it is the deleterious effects of soluble impurities and alloying additions that are relevant, rather than the beneficial effects, such as the addition of chromium and nickel to iron to produce stainless steels, or the addition of nickel and small concentrations of iron to copper to give cupronickels. As is well known, high-purity zinc corrodes much less rapidly in dilute acids than commercial purity material; in the latter instance, impurities (particularly copper and iron) are exposed on the surface of the zinc to give local cathodes with low hydrogen overpotentials; this result is of practical significance only in the use of zinc for sacrificial anodes in cathodic protection or for anodes in dry cells. In neutral environments, where the cathodic
1:48
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
reaction is oxygen reduction, there is very little difference in the corrosion rates of pure and impure zinc. In contrast, the selective dissolution or leaching-out by corrosion of one component of a single-phase alloy is of considerable practical importance. The most common example of this phenomenon, which is also referred to as ‘parting’, is dezincification, Le. the selective removal of zinc from brass (see Section 1.a). Similar phenomena are observed in other binary copperbase alloys, notably Cu-AI, as well as in other alloy systems.
Corrosion and Selective Dissolution in Two-phase Alloys In principle the selective dissolution of the less noble component of a singlephase alloy would perhaps be expected and is in fact observed (dezincification of an a-brass, etc.) even though the details of the mechanism by which it occurs is not yet fully understood. In contrast, the preferential attack of the less noble phase of a two-phase alloy is not only expected and observed -the mechanism by which it occurs in practice is also quite clear. Selective dissolution of the more active phase of a two-phase alloy is best exemplified by the graphitic corrosion (or graphitisation) of grey cast iron. Graphitisation
Cast irons, although common, are in fact quite complex alloys. The ironcarbon phase diagram exhibits a eutectic reaction at 1 420 K and 4.3 wt.%C (see Fig. 20.44). One product of this eutectic reaction is always austenite; however, depending on the cooling rate and the composition of the alloy, the other product may be cementite or graphite. The graphite may be in the form of flakes which are all interconnected (although they appear separate on a
Fig. 1.11 Light micrograph of the microstructure of a pearlitic grey cast iron; x 720
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1:49
metallographic section), or spheroids, which are all separate. The resultant alloys are known as white, grey and spheroidal graphite cast irons, respectively. During subsequent cooling the austenite may transform to ferrite, pearlite or martensite, or it may, in the case of high-nickel irons, be retained to room temperature. The microstructure of a grey cast iron with a predominantly pearlitic matrix is shown in Fig. 1.11. Graphite is more noble than any of the other phases in cast iron and is a very good cathode material; highly effective galvanic cells therefore exist between the graphite and the surrounding, less noble matrix. In grey cast irons, the matrix therefore corrodes preferentially, leaving behind a network of interconnected graphite flakes which is very porous and weak. The attack is often not readily apparent on superficial inspection. White cast irons are not susceptible to graphitisation since they contain no graphite; spheroidal graphite cast irons are also not susceptibleto graphitisation, since although they do contain graphite it is in the form of discrete spheroids which have a limited effect, instead of the interconnected graphite flakes in grey cast iron. Thus not only the existence but also the distribution of a cathodic phase is important (see also Section 3.6).
Influence of Structure on Surface Films-Pitting Corrosion Metals which owe their good corrosion resistance to the presence of thin, passive or protective surface films may be susceptible to pitting attack when the surface film breaks down locally and does not reform. Thus stainless steels, mild steels, aluminium alloys, and nickel and copper-base alloys (as well as many other less common alloys) may all be susceptible to pitting attack under certain environmental conditions, and pitting corrosion provides an excellent example of the way in which crystal defects of various kinds can affect the integrity of surface films and hence corrosion behaviour. In general, pitting corrosion may be divided into two stages, pit initiation and pit propagation. During pit initiation the passive film breaks down and does not reform. During pit propagation, the small active sites formed during the initiation stage propagate, often very rapidly, to form pits. The most recent ideas on the mechanism of pit initiation and propagation are dealt with in some detail in Reference 6. The propagation of pits is relatively well understood and is comparatively insensitive to the structure of the metal (see Sections 1.5 and 1.6). On the other hand, pit initiation which is the necessary precursor to propagation, is less well understood but is probably far more dependent on metallurgical structure. A detailed discussion of pit initiation is beyond the scope of this section. The two most widely accepted models are, however, as follows. Heine, el d 7suggest that pit initiation on aluminium alloys occurs when chloride ions penetrate the passive oxide film by diffusion via lattice defects. McBee and Kruger' indicate that this mechanism may also be applicable to pit initiation on iron. On the other hand, Evans' has suggested that a pit initiates at a point on the surface where the rate of metal dissolution is momentarily high, with the result that more aggressive anions
1 :50
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
are attracted to the point and produce a local environment that is favourable to further dissolution, Le. an autocatalytic process similar to that operative in pit propagation. This view has recently found increasing support, since there is now evidence that pits initiate at flaws or discontinuitiesin the passive film which result from mechanical, geometrical or compositional inhomogeneities in the metal surface'0.''. The latter model, in particular, predicts a strong influence of metallurgical structure on the integrity of the passive film and hence on susceptibility to pitting corrosion. In practice many metallurgical factors do appear to affect pitting corrosion. For example, severe cold work increases the pitting susceptibility of austenitic stainless steels, while molybdenum and nitrogen alloying additions, in particular, reduce it. Pitting is less likely to occur on smooth, polished surfaces than on rough, etched, ground or machined surfaces. Austenitic stainless steels are more susceptible to pitting if they have been held briefly in the sensitising temperature range. Pure aluminium is much more resistant to pitting than impure metal and alloys, particularly those containing copper. In general, the more homogeneous a metal surface the better is the resistance of passive films on that surface to pitting. In austenitic stainless steels, pits have been observed to initiate at grain boundaries and also at certain sulphide inclusions. These effects are all evidence of the fact that crystal defects and metallurgical structure and composition affect the thickness, strength, solubility, porosity, etc. of passive films, and hence the susceptibility of those films to localised breakdown and pitting.
Effect of Mechanical Stresses on Corrosion The presence of stresses does not usually affect the general corrosion behaviour of metals and alloys to any very significant extent. However, two extremely important forms of localised corrosion may occur when metals are simultaneously exposed to stress and a corrosive environment. Metals subjected simultaneously to alternating stresses and any corrosive environment may be subject to corrosion fatigue, while certain alloys exposed simultaneously to tensile stresses and fairly specific environmental conditions may fail by stress-corrosion cracking. Other sections (see Chapter 8) deal specifically with the mechanism and phenomenology of corrosion fatigue and stress-corrosion cracking of various alloy systems, and it is not the intention to duplicate that material in this section. However, the susceptibilityof many alloys to stress-corrosion cracking is determined not only by the presence of tensile stresses and specific environmental conditions, but also by the metallurgical structure of the metals. These instances will be discussed briefly, by way of further examples of the effect of structure on the corrosion of metals. Stress-corrosion Cracking of Copper-base Alloys
Single-phase a-brasses are susceptible to stress-corrosion cracking in the presence of moist ammonia vapour or certain ammonium compounds 12. Here the predominant metallurgical variable is alloy composition, and in
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1:5 1
practice brasses containing less than 10-15% Zn seldom fail by stress corrosion; above about 15% Zn, the stress-corrosion susceptibilityincreases with zinc content. Other structural factors are secondary: cold-worked brass, in practice, is more likely than annealed material to fail by stress-corrosion crocking, but this is probably only a reflection of the fact that the residual stresses are likely to be higher in cold-worked than in annealed alloys. The stress-corrosion susceptibility of a-brasses increases with increasing grain size. There is also evidence that decreasing stacking-fault energy results in a transition from inter- to transgranular cracking in a number of binary copper-base alloy^'^ and that the presence of order (see Section 20.4) increases the susceptibility of certain complex copper-base alloys 14. It must be emphasised, however, that these are only secondary factors which merely tend to increase or decrease fairly marginally the stress-corrosion susceptibility. Stress-corrosion Cracking of Aluminium-base Alloys
In contrast to brasses, metallurgical structure plays a predominant rdle in determining the susceptibility of high-strength aluminium alloys to stresscorrosion cracking in the presence of tensile stresses and moist chloridecontaining environments. Under these conditions these alloys may vary from highly susceptible, to practically immune, to intergranular stress-corrosion cracking, depending on their microstructure, as determined by heat treatment l 5 (see Section 8.5). The effect of grain shape on the stress-corrosion behaviour of aluminium alloys has already been discussed. The effect of heat treatment on the stress-corrosion susceptibility of high-strength precipitation-hardenable Al-Zn-Mg alloys is illustrated schematically in Fig. 1.12. In the solutionheat-treated and quenched condition, these alloys are very resistant to stresscorrosion cracking but they are also too weak to be of much use in this condition. On ageing, the alloys become progressively stronger (see Section 20.4), but also increasingly susceptible to stress corrosion, as shown in Fig. 1.12. Maximum stress-corrosion susceptibility is observed in the intermediatestrength, under-aged condition; thereafter the alloys become increasingly more resistant to stress-corrosion cracking. Thus, as shown in Fig. 1.12 the highest-strength, peak-aged condition is moderately susceptibleto stress corrosion, while the intermediate-strength, over-aged condition is relatively resistant. In practice, therefore, there is a choice between maximum strength alloys with moderate stress-corrosion susceptibility and somewhat lower strength alloys with little stress-corrosion susceptibility. As suggested above, the former condition or temper might be selected for thin sheet applications while the latter heat treatment would be specified for thick sections. There is some doubt as to whether the effect of heat treatment on the stresscorrosion susceptibilityof precipitation-hardenable aluminium alloys results from variations in the precipitate-@ee zone width and the intergranular precipitate morphology, or from variations in the interaction of intragranular precipitates with dislocations. The effect of microstructure on the stress-corrosion susceptibility of AI-Cu, although again very substantial, is somewhat less straightforward than in the case of Al-Zn-Mg alloys 15.
1 :52
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
As-quenched condition
Under- aged condition
Peak-aged condition
Over-aged condition h
VI
U L
In
0 .c w
L
f L
0 L 01
.--a3 L
I
0
-E .-
L
0
2
-.In In
:
C
.-0 VI 0 L
0
u
I In
w
L
-
5
Logarithm isothermal ageing time Isochronal
ageing tcmpcratwc
Fig. 1.12 Curves showing the relationship between strength, stress-corrosion susceptibility and heat treatment for a high-strength precipitation-hardening aluminium alloy
Microstructure also plays a predominant r61e in determining the stresscorrosion susceptibility of the work-hardenable AI-Mg alloys. The AI-Mg system, like the AI-Cu system, exhibits decreasing solubility with decreasing temperature, and on ageing a solution-heat-treated and quenched alloy, precipitation of Mg,Al, is observed. However, the alloy is not strengthened by this precipitation, as it occurs either as very coarse, widely dispersed intragranular precipitates (which do not interact with dislocations), or as a more or less continuous intergranular film. These alloys can therefore only be strengthened by cold working. Nevertheless, the precipitate morphology controls the stress-corrosion susceptibility; alloys which exhibit continuous films of intergranular precipitate are highly susceptible to stress-corrosion cracking (and to exfoliation and intergranular corrosion), while those which exhibit coarse intragranular precipitates (or no precipitation at all) are generally much less susceptible or resistant. The higher the magnesium level, the greater the degree of cold work, and the lower the ageing temperature, the more likely is the formation of a continuous intergranular precipitate and therefore the greater is the potential stress-corrosion susceptibility.
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
1 :53
Stress-corrosion Cracking of Steels
High-strength low-alloy quenched and tempered steels (i.e. steels with yield strengths greater than about 900 MN/m2) may be susceptible to stresscorrosion cracking in the presence of moisture. The crack path is usually intergranular with respect to the prior-austenite grain boundaries, and the mechanism of cracking is generally accepted as involving some form of hydrogen embrittlement. The major metallurgical variable in this instance of environmentally induced cracking is the strength level- the stronger the steel the greater is its susceptibility. However, at constant strength level, steels with martensitic structures are considerably more susceptibleto cracking than steels with bainitic structures (see Section 20.4). Again, at constant strength level, it has been shown that the crack growth rate decreases and the time-to-failure increases as the prior-austenite grain size is reduced 16. In practice, by far the most common case of stress corrosion is that occurring when austenitic stainless steels are simultaneously exposed to tensile stresses and hot, aqueous, aerated, chloride-containing environments. In this case the major variable is alloy composition and structure; virtually all austenitic stainless steels are more or less susceptible to stress-corrosion cracking in these environments, while ferritic and ferritic/austenitic stainless steels are highly resistant or immune. Stress-corrosioncracking of all types of steels formed the topic of a recent conference”, the proceedings of which deal in some detail with the effect of structure on the stress-corrosion susceptibility of these alloys.
Conclusions This discussion on the relationships between structure and corrosion should not be taken as exhaustive. For example, the stress-corrosion cracking behaviour of titanium-base alloys in a variety of environments is affected to differing degrees by the microstructure of the alloys la. Again, the effect of the changes in structure produced by welding on the corrosion behaviour of metals and alloys is of great practical importance19.This topic has been considered above in relation to stainless steels, but it is also of considerable importance in the welding, brazing and soldering of other alloy systems. In certain instances, the corrosion resistance of a weld is markedly affected by the structure of the weld metal and the adjacent heat-affected zone. Further examples of the effect of metallurgical structure on corrosion phenomena are provided by (a) the possible rale of emergent dislocations and slip-steps in the mechanism of stress-corrosion cracking of austenitic stainless steels, (b) by the rale of &ferrite in the corrosion of certain austenitic stainless steels, (c) by the r6le of local spheroidisation in ‘ringworm’ corrosion of mild steel and (d)by the r6le of manganese sulphide and other inclusions in the pitting of mild steel. Since corrosion is essentially a reaction between a metal and its environment, the very significant effect of crystal defects and metallurgical structure on certain corrosion phenomena is to be expected. It is no more possible to
1 :54
EFFECTS OF METALLURGICAL STRUCTURE ON CORROSION
neglect the metallurgical aspects of a corrosion problem than it is to overlook the environmental and electrochemical factors. R. P. M. PROCTER REFERENCES
M.B.. in Roc. U.R. Evans Internat. Con$ on Localised Corrosion, N.A.C.E. (Houston) Foroulis, Z. A. and Uhlig, H. H.,J. Electrochem. Soc.. 111, 522 (1%) Aust. K. and lwao. O., in Roc.U.R. Evans Internat. Conf. on Localised Corrosion. N.A.C.E. (Houston) Bain, E. C., Aborn, R. H. and Rutherford, J. B., Trans. Amer. Soc. Steel Treating. 21, 481 (1933) Cowan. R. L. and Tedmon. C. S.. in Advances in Corrosion Science and Technology, Vol. 111, Plenum Press,New York (1973) Proc. U. R. Evans Internat. COP$ on Localised Corrosion, N.A.C.E. (Houston) Heine, M. A., Keir. D. S. and Pryor, M. J., J. Electrochem. Soc., 112, 29 (1965) McBee, C. L. and Kruger, J., in Proc. U. R. Evans Internat. Conf. on Localised Corrosion, N.A.C.E. (Houston) Evans, U. R., Corrosion, 7 , 238 (1951) Richardson, J. A. and Wood, G. C., Corr. Sci., 10, 313 (1970) Ashworth, V., Boden, P. J., Leach, J. S. L1. and Nehru, A. J., Corr. Sci., 10,481 (1970) Pugh, E. N., Craig, J. V. and Sedriks, A. J., in Proc. Internat. Conf. on Fundamental Aspects of Stress-corrosion Cracking, N.A.C.E., Houston (1969) Ohtani, N. and Dodd, R. A., Corrosion, 21, 161 (1965) Popplewell, J. M., Procter, R. P. M. and Ford, J. A,, Corr. Sci., 12, 193 (1972) Speidel, M. 0. and Hyatt, M. V., in Advances in Corrosion Science and Technology, Vol. 11, Plenum Press, New York (1972) Procter, R. P. M. and Paxton, H. W., Trans. A.S.M., 62, 989 (1969)
1. Ives,
2. 3. 4. 5.
6. 7. 8.
9. 10. 11. 12. 13. 14. 15. 16.
17. Proc. Internat. Conf. on Stress-corrosion Cracking and Hydrogen Embrittlement of Iron Base Alloys, N.A.C.E. (Houston) 18. Blackburn, M. J., Feeney, J. A. and Beck, T. R., in Advances in Corrosion Science and Technology, Vol. 111, Plenum Press, New York (1973) 19. Lancaster, J. F., Metallurgy of Welding, Brazing andsoldering, George Allen and Unwin, London (1970)
1.4 Corrosion in Aqueous Solutions
In this section the interaction of a metal with its aqueous environment will be considered from the viewpoint of thermodynamics and electrode kinetics, and in order to simplify the discussion it will be assumed that the metal is a homogeneous continuum, and no account will be taken of submicroscopic, microscopic and macroscopic heterogeneities,which are dealt with elsewhere (see Sections 1.3 and 20.4). Furthermore, emphasis will be placed on uniform corrosion since localised attack is considered in Section 1.6. Aqueous environments will range from very thin condensed films of moisture to bulk solutions, and will include natural environments such as the atmosphere, natural waters, soils, body fluids, etc. as well as chemicals and food products. However, since environments are dealt with fully in Chapter 2, this discussion will be confined to simple chemical solutions, whose behaviour can be more readily interpreted in terms of fundamental physicochemical principles, and additional factors will have to be considered in interpreting the behaviour of metals in more complex environments. For example, iron will corrode rapidly in oxygenated water, but only very slowly when oxygen is absent; however, in an anaerobic water containing sulphatereducing bacteria, rapid corrosion occurs, and the mechanism of the process clearly involves the specific action of the bacteria (see Section 2.6). All corrosion reactions in aqueous solutions are characterised by the following features: 1. The electrified interface between the metal and the electrolyte solution
(the metal surface may be film-free or partially or completely covered with films or corrosion products). 2. Transfer of positive charge from the metal to the solution with consequent oxidation of the metal to a higher valency state. 3. Transfer of positive charge from the solution to the metal with consequent reduction of a species in solution (an electron acceptor) to a lower valency state. 4. Transfer of charge through the solution and corroding metal. It follows that corrosion is an electrochemical reaction in which the metal itself is a reactant and is oxidised (loss of electrons) to a higher valency state, whilst another reactant, an electron acceptor, in solution is reduced (gain of electrons) to a lower valency state. This may be regarded as a concise expression of the ‘electrochemical mechanism of corrosion’. 1:55
1:56
CORROSION IN AQUEOUS SOLUTIONS
Thermodynamics and Kinetics of Corrosion Reactions Thermodynamics provides a means of predicting the equilibrium state of a system of specified components, but provides no information on the detailed course of the reaction nor of the rate at which the system proceeds to equilibrium. With the aid of suitable catalysts the reaction may be made to proceed by different mechanisms and at different rates, but the final position of equilibrium, which can be predicted unequivocally by thermodynamics, remains unchanged. At elevated temperatures reactions proceed to equilibrium far more quickly than at ambient temperatures, and for this reason metal/gas equilibria at elevated temperatures can be defined entirely in terms of the thermodynamics of the systems under consideration (see Section 7.6). The position is frequently quite different with systems at ambient temperatures, since although a reaction may have a very pronounced tendency to proceed in a given direction this may be entirely nullified, by the very slow rate at which it proceeds to its equilibrium state. Thus although the interaction of hydrogen and oxygen to form water at 25°C is accompanied by a large decrease in the standard free energy (see Section 20.4)
+ tO,(g.)
AG& = -237 kJ H,O(g.) the rate of reaction is so slow that the reaction may be regarded as not occurring at all; however, in the presence of a catalyst, e.g. platinum black, the reaction proceeds instantaneously with explosive violence. It follows that although the change in free energy AG provides a quantitative measure of the tendency of a reaction to proceed in a given direction, it gives no information on the rate of the reaction. It is indeed fortunate that the rates of the majority of corrosion reactions are very slow, since it is this factor that permits metals to be used as constructional materials. Considering metals in contact with oxygen at room temperatures, it follows from the standard free energies of formation of metal oxides that, with the exception of gold, the oxide of the metal is the thermodynamically stable state; the free energy decreases with temperature but the common metals are thermodynamically stable in an oxygen atmosphere only at very elevated temperatures (Fig. 7.55, Section 7.6). However, in practice, highly reactive metals like aluminium, magnesium, tantalum, niobium, etc. are relatively stable in oxygen whereas other reactive metals such as the alkali and alkaline-earth metals oxidise rapidly and completely. In these examples, the kinetics of the reaction is determined by the nature of the oxide film formed on the metal surface, which forms a protective barrier in the former group of metals and a non-protective one in the latter. Nevertheless, it should be noted that when the surface area is very large compared to the volume, the thermodynamic tendency for oxidation will outweigh kinetics, and metals such as aluminium, magnesium, titanium and iron in the form of very fine powders, can ignite spontaneously at ambient temperatures. Most metals (copper, silver, gold, mercury and the platinum metals are the exceptions) are found in nature in the combined form as minerals, which are then reduced to the metal by the expenditure of sufficient energy (chemical, electrical or thermal) to reverse the natural spontaneous reaction. It follows H,(g.)
+
CORROSION IN AQUEOUS SOLUTIONS
1:57
that most metals are thermodynamically unstable and will tend to revert back to the thermodynamically stable combined state when exposed to the atmosphere, but the rate at which the reaction proceeds may be so slow that for all practical purposes the metal is stable. It is of some significance that very reactive metals such as tantalum, niobium and titanium, which have high affinities for oxygen and are reduced from their oxides with the expenditure of considerable chemical energy, are highly stable in the majority of environments; in fact, it is unlikely that these metals will ever revert back to oxides during atmospheric exposure owing to the highly protective nature of the thin oxide film. However, similar considerations do not apply to the less reactive metals iron and mild steel, since the oxide formed on these metals (rust) is not so protective as that formed on certain of the more thermodynamically reactive metals. These examples show that the thermodynamic tendency of metals to corrode is frequently outweighed by kinetic factors that control the rate of the process so that the final position of equilibrium, Le. complete conversion to corrosion product, is attained only slowly or not at all. Although the formation of oxide films, or films of other corrosion products, are of vital importance in controlling the rate of a corrosion reaction in aqueous solution, it must be emphasised that in certain circumstancesthe activation energy of the process or the rate of diffusion of species to and from the metal surface may be more significant than film formation. Thus the slow rate of corrosion of pure zinc in sulphuric acid, in which the oxide film is thermodynamically unstable, is dependent on the high activation energy needed for the reduction of hydrogen ions to hydrogen gas. In neutral solutions, on the other hand, diffusion of oxygen to the metal surface may be the rate-determining factor. The magnitude of AG, the free energy change, of a specific corrosion reaction provides a measure of the spontaneity of the reaction and of the extent to which it will proceed before equilibrium is attained; if AG 0 (positive) the metal is stable and no further consideration need be given to kinetics; if AG = 0 the system is at equilibrium and will not proceed in either direction. The relationship between the free energy decrease -AG and the rate of the reaction aA
+ bB = CC + dD
where A is the metal, B is a reactant in solution, and C and D are the reaction (or corrosion) products, is illustrated in Fig. 1.13, which shows that although the thermodynamic tendency is the same for L, M and N the rates are different. This illustration may be exemplified by regarding L as an unprotected steel pipe buried in a corrosive soil, M the pipe protected by a bituminous coating and N the pipe coated and also cathodically protected; in each case the free energy change for the corrosion reaction
+
+
2Fe H 2 0 to, -,Fe20,.H,0 is the same, but the rate at which the reaction proceeds to equilibrium (complete conversion of iron to Fe, O,.H,O) varies significantly.
Fig. 1.14 Block on inclinedplane illustratingthe significance of (0)corrosion, (b)passivity and (c) immunity
Figure 1.14illustrates the concepts implicit in the terms corrosion, passivity and immunity, which are of particular significance in both the thermodynamic and kinetic approach to corrosion reactions. In this analogy the metal A is in the form of a cube that rests on the top of an inclined plane and is in contact with the environment B. The slope of the plane a may be regarded as a measure of the tendency of the metal to slide down the plane with progressive formation of corrosion product ( -AG) until equilibrium is attained at the bottom of the plane when all the metal is converted to corrosion product. The rate at which the block slides down the plane, which in this analogy corresponds to the rate of the reaction, will depend both on 01 and on the frictional resistance between the metal, or the metal filmed with corrosion product, and the plane. In Fig. 1.14a the frictional resistance of both metal and corrosion product is assumed to be low, and the metal proceeds rapidly down the plane with a correspondingly high corrosion rate. In Fig. 1.14b the corrosion product is assumed to form a continuous adherent film on the metal surface that has a high frictional resistance; it follows that progress of the block down the plane will be decreased to a very low value as soon as a thin film of oxide is formed, and although some corrosion will occur it will not affect the bulk properties of the metal block significantly.
CORROSION IN AQUEOUS SOLUTIONS
1:59
In this analogy Fig. 1,14arepresents a corrosion rate that could not be tolerated in practice, whereas Fig. 1.14b represents the phenomenon of passivity in which the rate of corrosion is controlled at an acceptable level by virtue of the kinetic barrier of corrosion product. It should be noted, however, that the metal in the passive state is still thermodynamically unstable, and any environmental factor that results in the removal of the oxide film, either locally or generally, will result in rapid corrosion; a characteristic of passive metals is that in certain environmental situations localised breakdown of the oxide film occurs with consequent pitting of the exposed metal, the filmed surface surrounding the pit remaining unattacked. Finally, no corrosion will occur if the block can be maintained at the top of the plane by a continuous supply of energy of a magnitude that is equal to or greater than the energy liberated when the reaction proceeds spontaneously. The analogy in Fig. 1.14~represents the phenomenon of immunity that forms the basis of the important practical method of corrosion control known as cathodic protection in which the potential of the metal is depressed sufficiently to prevent it oxidising (see Chapter 10).
Thermodynamic Approach to Corrosion Reactions* The standard electrode potentials I E ', or the standard chemical potentials p e , may be used to calculate the free energy decrease -AG and the equilibrium constant K of a corrosion reaction (see Appendix 20.2). Any corrosion reaction in aqueous solution must involve oxidation of the metal and reduction of a species in solution (an electron acceptor) with consequent electron /~~ transfer between the two reactants. Thus the corrosion of zinc ( E & Z + = -0.76V) in a reducing acid of pH = 4 (aH+= may be represented by the reaction: Zn
+ 2H+(a,+ =
= Zn2++ H2(pH,= 1)
, , .(1.4)
which consists of the two half reactions and
Zn = Zn*+(aq.) + 2e(oxidation of Zn + Zn2+) 2H+ + 2e = H, (reduction of H++ H,)
. . .(1.5) . . .(1.6)
Since AG = -zFE and since 1 F (F is a Faraday) = 96 500 C AGzn~+/z, = -2 x 96 500 x - ( -0.76) J mol-' Zn
and
AGH+,",= -2 x 96500 x ( -4 x 0-059)Jmol-'H,
. . .(1.7) . . .(1.8)
and for the overall reaction AGEdo, = -2 x 96500 x (0.76 - 0.24) = -100.3 kJmol-'Zn which shows that the reaction of zinc with hydrogen ions at a,+ = 104g ion/l is spontaneous and proceeds in the direction as written in equation 1.4. However, in spite of the large free-energy decrease for equation 1.4 the rate of corrosion, which in this particular system is controlled by the rate of hydrogen evolution (equation 1.6), is extremely slow. (At this point it should be noted that the free energies of half reactions can be added algebraically to evaluate AGreaction and similar considerations *Thermodynamic data are given in Chapter 21, Tables 21.5 and 21.6.
1:60
CORROSION IN AQUEOUS SOLUTIONS
apply to the evaluation of E,,,, providing the number of electrons z in the two half reactions and in the complete reaction are the same (see example just given). However, should z differ, then it is essential to sum the free energies in order to evaluate Emion,and this can be illustrated by calculatfrom = 0.346 V, and E&+,cu= 0.522 V: ing the value of Cu2++ 2e = Cu, -AG" = 2 F x 0.346 = 0-792F
. , 41.9)
Cu+ + e = Cu, -AG" = 1 F x 0.522 = 0.522F
. . .(l.lO)
Subtracting equation 1-10 from equation 1.9 Cu2++ e = Cu', -AG" = (0.792F - 0-522F) = 0-170F
... 1 FE&z+/cu+= 1 F x 0.170, mdE&z+/Cu+= 0.17 V Direct subtraction of the standard electrode potential would have given the incorrect value = 0.346 - 0.522 = -0- 176V.) In this example of the corrosion of zinc in a reducing acid of pH = 4, the corrosion product is Zn2+ (as.), but at higher pHs the thermodynamically stable phase will be Zn(OH), and the equilibrium activity of Zn2+will be governed by the solubility product of Zn(OH), and the pH of the solution; at still higher pHs Zn0;anions will become the stable phase and both Zn2+ and Zn(OH), will become unstable. However, a similar thermodynamic approach may be adopted to that shown in this example. Table 1.7 Half reactions involving the oxidation of a metal in aqueous soIutions Equation number 1
2(a) or 2(b) 3(a) or 3(b)
Hau reaction
M M M M M
Mz+(aq.) + ze Oxidation to acquo cations M(OH), + zH+ + ze Oxidation to + zOH- M(OH), + ze metal hydroxide + zH,O MO:-(aq.) + 2zH+ + ze Oxidation to + zOH- MO:-(aq.) + zH+ + ze aquo anions
+
+ zHz0
+
+
]
+
+
]
Table 1.7 shows typical half reactions for the oxidation of a metal M in aqueous solutions with the formation of aquo cations, solid hydroxides or aquo anions2. The equilibrium potential for each half reaction can be evaluated from the chemical potentials of the species involved (see Appendix 20.2) and it should be noted that there is no difference thermodynamically between equations 2(a) and 2(b) nor between 3(a) and 3(b) when account is taken of the chemical potentials of the different species involved. In view of the importance of the hydronium ion, H,O+, and dissolved oxygen as electron acceptors in corrosion reactions, some values of the redox potentials E and chemical potentials p for the equilibria . . .(1.11) 2H++ 2e = H2, E = 0.00 - 0.059pH - O.O3010gp", and
to2+ H 2 0 + 2e = 2 0 H - , E = 1-23 - 0-059pH -1- O.lSlogp,,
. . .(1.12)
at 25°C are given in Table 1.8. The following should be noted: 1 . Dissolved oxygen has a higher redox potential than the hydrogen ion at
all values of pH, i.e. it is a more powerful oxidant.
1:61
CORROSION IN AQUEOUS SOLUTIONS
Table 1.8 Potentials of the H+/fH2 and 02/OH- electrodes (pH2 = po2 - 1 atm*)
Activity
Equilibrium
(IH+ = 1
2H+ + 2 e = H 2 (see equation 1.11)
+
(I"+ uH+
+
f0, H,O 2.9 = 20H(see equation 1.12) 1 am = 1.013
UOH-
UOHuOH-
= = =1 = lo-' = 10-l~
pH
E(V)
0
0.00
7 14 14 7
-0.414 -0.828 0.401 0-815 1.229
o
@=2~%500EkJ 0.00 -79.89 -159.8 71.4 157-3 231.5
x I O J Pa (Pascal).
2. In both equilibria E decreases with increase in pH.
3. A plot of E vs. pH for each equilibrium gives a linear curve of slope 0-059 (see curves Land m in Fig. 1.15 (bottom)). 4. A decrease in po2will displace curve rn in Fig. 1.15 (bottom) in the negative direction, whereas a decrease in pH2 will displace curve f in the positive direction (equations 1.1 1 and 1.12). TabIe 1.9 shows the values of AG",K and the equilibrium activities of metal cations and pressures (fugacities) of hydrogen gas for the reaction of typical metals with a reducing acid (aH+= 1) to form metal cations M z + (as.). Equilibria such as Au3+/Au, Ag'/Ag and Cu2+/Cu have redox potentials >O.OV, and the metal (the reduced species of the equilibrium) may be regarded as being thermodynamically stable in a reducing acid, since the equilibrium activities of metal cations are negligible; thus in these systems no further consideration of the rates of the reaction are necessary. On the other hand, the metals M of M z + / M equilibria that are negative to the H+/H2 equilibrium are thermodynamically unstable, and the high values of the equilibrium activities of ML+indicate that the reaction will proceed to completion, although at a rate that cannot be predicted from thermodynamic consideration alone.Thus pure zinc corrodes slowly in hydrochloric acid, more quickly if the zinc contains impurities of lead and more quickly still when the impurities are copper; in each of these three cases the thermodynamic tendency and the final position of equilibrium (see Table 1.9) are the same, although the rates at which equilibrium is achieved are markedly different. (Note that each equilibrium must be written so that -AG" > 0 (positive), i.e. in the direction in which it proceeds spontais always positive although E,$+/,,, may be either neously so that E&aion positive or negative.) T8Me 1.9 Thermodynamics of the reaction of metals with acid solutions ((IH+ = 1, EA+/iH2 = 0 ' ~ v )
Reaction
f Au3++H, =2H++ f Au 2Ag+ H, =2H+ 2Ag CU~++H,=ZH+ +CU Sn +2H+ =HI+ Sn2+ Fe+2H+ = H2 Fez+ Z n +2H+ = Hz+Zn2+ +AI + 2H+ = H, + +AI3+ Mg + 2H' = H, Mg2+
+
+
+
+
1 -50 0.79 0.337 -0.136
1 -50 0.79 0.337 0-136
-0.440
0.440
-0.763 -1.66 -2.37
0.763 1-66 2.37
289.4 152.4 65.0 26.2 84.9 147.2 320.3 457.3
7 . 2 ~ 1 68 ~- 5 ~ 1 0 -1~. 4~~ 1 0 - ~ ' 1.7 x 6 . 0 l~d 6 4.1 x 2-6X10" 3-8X10-'2 3 - 8 X 1 0 - ' 2 4 . 0 104 ~ 4-OX 104 4 . 0 104 ~ 8 . 1 ~ 1 0 ' 8~. 1 ~ 1 0 ' ~8 . 1 ~ 1 0 ' ~ 7 . 2 I~d S 7 . 2 I~d 5 7 . 2 I~d ' 1 . 9 ~ 1 62 ~ .6~10'~1.9~10'~ 2.2x 1080 2.2x 1080 2 . 2 x loso
1:62
CORROSION IN AQUEOUS SOLUTIONS 1
,
Fe%q
1
1
,
l
,
/
1
(
,
1
~
,
-
0 -2
1
,
)
-1 - 6
> ---.
-
0
-
-
lu'
I
-1
I
I
I
I
I
0
I
PH
I
I
I
I
I
,
7
I
1
1C
-1.6
-2
-1
0
2
1
PH
8
10
12
1C
16
Fig. 1.15(top) Equilibrium potential-pH diagram for the Fe-H,O system showing the zones of stability of cations, anions and solid hydroxides (after Deltombe and Pourbaix') and (bottom) simplified version showing zones of corrosion, immunity and passivity (curve L is the H20/H2 equilibrium at p ~ * =1 and curve rn is the 0 , / H 2 0 equilibrium at poz = 1 )
1:63
CORROSION IN AQUEOUS SOLUTIONS
Table 1.10 Thermodynamics of reactions of noble metals with acids and alkaline solutions containing oxygen (pO2= 1 atm) Reaction ~
2Au+ +H?O = 2Au +702 + 2H+ 0 2Ag++O2 2H+ =2Ag++H20 0 2Cu+402 + 2H+ =2Cu+ + H20 0 2Au++ 20H- = O2 + H 2 0 + 2Au 14 2Ag+ 20H- = 0, + H 2 0 + 2 A g 14 2Cu+ +20H- = O2 + H 2 0 + 2 C u 14
+
+
t t
4
~~
~
~
~
~~
1.68
0.45
86.8
1.7 x 10" 2.4X lo-* 2 . 9 ~ 1 0 ' ~
0.79
0.44
84.9
8 . 1 x 1014 208x10'
1.5~10-~'
0,521
0.71
137.0
1 . 1 x 10'' 1.0X10''
8.2~10-~'
1.68
1.28
247.0
2.4 x
5 . 8 ~
0.79
0-39
75-2
1 . 6 ~ 1 0 '2*5X10-' ~ 2~6x10~~
0.521
0.12
23.1
1 . 2 lo4 ~ l-OX10-2 1*4X1O8
2-OX
Table 1.10 shows the effect of raising the redox potential of the acid by means of dissolved oxygen (Po2 = 1 atm), and it can be seen that the noble metals (metals more positive than 0-OV) silver and copper are now thermodynamically unstable, and due consideration must be given therefore to the factors that control the rates of the reaction. Oxygen has been selected in this example because of its omnipresence, but similar considerations would apply to other oxidants of high redox potential, e.g. nitric acid, ferric salts, hydrogen peroxide, perchlorate, persulphate, etc. In this connection it is of relevance to draw attention to the compositions and redox potentials of the various etching solutions used for specific metals and alloys in metallography, which reveal the structure by a controlled corrosion reaction during which different structural features corrode at different rates. These examples show how the tendency of a corrosion reaction to proceed in the direction of oxidation of metal to Mz+(aq.) may be increased by increasing the redox potential of the solution, i.e. by increasing -AG of the reaction. In the data given in Tables 1.9 and 1.10 it has been assumed that the metal is oxidised solely to Mz+(aq.)and that no complex ions have formed, a situation that is more rare than usual. Thus copper is stable in dilute hydrochloric acid (Table 1.9) but will corrode in the hot concentrated acid owing to the formation of CuCl; complexes; it will also form complexes with a variety of chemicals, including NH,, CN-, etc. Ferrous ions form complexes with ammonium salts, sodium ethylene diamine tetraacetates or concentrated solutions of sodium polyphosphates3, and both Fe2+and Fe3+ form complexes in chloride solutions. Table 1.10 also provides thermodynamic data for the reaction of silver, gold and copper with oxygenated alkaline solutions, and from the low equilibrium activity of Ag+ and Au+ it can be assumed that these metals are stable in that environment. However, in the presence of CN- anions, cyano-complexes are formed and there is an equilibrium between M +and M(CN); that is defined by the instability constant (Table l.ll)4, and although the a,+ (which is dependent upon the redox potential and pH of the solution and is unaffected by CN-) is very low, the aMo,,, is significantly high. Thus addition of CNto an oxygenated sodium hydroxide solution will result in the corrosion of
1:64
CORROSION IN AQUEOUS SOLUTIONS
gold, and this reaction is utilised in extractive metallurgy for the leaching of gold from its ores by means of an oxygenated solution of sodium cyanide. The data given in Tables 1.9 and 1.10 have been based on the assumption that metal cations are the sole species formed, but at higher pH values oxides, hydrated oxides or hydroxides may be formed, and the relevant half reactions will be of the form shown in equations 2(a) and 2(b) (Table 1.7). In these circumstances the a,+ will be governed by the solubility product of the solid compound and the pH of the solution. At higher pH values the solid compound may become unstable with respect to metal anions (equations 3(a) and 3(b), Table l .7), and metals like aluminium, zinc, tin and lead, which form amphoteric oxides, corrode in alkaline solutions. It is evident, therefore, that the equilibrium between a metal and an aqueous solution is far more complex than that illustrated in Tables 1.9 and 1.10.Nevertheless, as will be discussed subsequently, a similar thermodynamic approach is possible. Finally, it is necessary to observe that the values of activities and fugacities calculated are thermodynamic quantities that cannot always be realised in practice, e.g. very high activities of metal ions cannot be attained because of solubility consideration and very low activities have no physical significance. Table 1.11 Equilibrium between M + and M(CN)F in cyanide solutions (aOH- = 1 (pH = 14); aCN- = 1; po2 = 1)
Equilibrium Au(CN)F= Au’ + 2CNAg(CN)F= Ag’ 2CNCu(CN)C= CU+ 2CN-
+ +
Instability constant
UM+
10-39
2 x 10-22 2.5 x 10-7 9.1 x 10-8
10-20
10-16
aM(CN)i
10”
1013 io9
Potential-pH Equilibrium Diagrams Pourbaix and his co-~orkers’*~ have calculated the phases at equilibrium for M/H,O systems at 25°C from the chemical potentials of the species involved in the equilibria, and have expressed the data in the form of equilibrium diagrams having pH vs. E, the equilibrium potential (vs. S.H.E.) as ordinates. These diagrams, which are analogous to the compositiontemperature diagrams of alloy systems, provide a thermodynamic basis for the study of corrosion reactions, although, as emphasised by Pourbaix, their limitations in relation to practical problems must be appreciated. Since these diagrams are referred to throughout this work, some emphasis is given to their significance in this section; reference should be made to the numerous publications of Pourbaix and his co-workers for a fuller account of the subject. It should be emphasised that potential-pH diagrams can also be constructed from experimental E,-Z curves, where EDis the polarised potential and Z the current6. These diagrams, which are of more direct practical significance than the equilibrium potential-pH equilibrium diagrams constructed from thermodynamic data, show how a metal in a natural environment (e.g. iron in water of given chloride ion concentration) may give rise
CORROSION IN AQUEOUS SOLUTIONS
1:65
to general corrosion, pitting, perfect or imperfect passivity, or to immunity, depending on the pH and potential (see Section 1.6, Fig. 1.56). Construction of Potential-QH Dhgrarns
Pourbaix has classified the various equilibria that occur in aqueous solution into homogeneous and heterogeneous, and has subdivided them according to whether the equilibria involve electrons and/or hydrogen ions. The general equation for a half reaction is
aA
+ mH+ + ze-
= bB
+ cH,O
. . .(1.13)
which shows that the equilibrium between the reactant A and product B depends on a,+ (pH) and the electrode or redox potential E, if neither hydrogen ions nor electrons are involved then the equilibrium is independent of pH and E. From the Nernst equation (see page 20.69), and substituting for -logs,+ = pH .(1.14)
It follows from equation 1.14 that for any constant ratio of ab,/ai the E vs. pH relationship will be linear with a slope -0*059m/z, and that when a: = a: = 1 the intercept of the curve on the E axis (Le. pH= 0) will be E e, the standard equilibrium potential, which by definition is the potential when the species involved in the equilibrium are at unit activity. Pourbaix has evaluated all possible equilibria between a metal M and H,O (see Table 1.7) and has consolidated the data into a single potentialpH diagram, which provides a pictorial summary of the anions and cations (nature and activity) and solid oxides (hydroxides, hydrated oxides and oxides) that are at equilibrium at any given pH and potential; a similar approach has been adopted for certain M-H20-Xsystems where X is a non-metal, e.g. C1-, CN-, COz, SO:-, PO:-, etc. at a defined concentration. These diagrams give the activities of the metal cations and anions at any specified E and pH, and in order to define corrosion in terms of an equilibrium activity, Pourbaix has selected the arbitrary value of g ion/l, i.e. corrosion of a metal is defined in terms of the pH and potential g ion4 that give an equilibrium activity of metal cations or anions > conversely, passivity and immunity are defined in terms of an equilibrium activity of c g ion/l. (Note that g ion/l is used here because this is the unit used by Pourbaix; in the S.I. the relative activity is dimensionless.) Fe/H,O system Figure 1.15 (top) is a simplified version7 of the Fe-H,O potential-pH equilibrium diagram [the region of stability of magnetite (Fe,O,) is not included] and it is instructive to consider some of the more important equilibria involved: . . .(l.lS) Curve a,Fez+ 2e = Fe; E = -0.440 0.03010gaF2+
+
+ Curve b, Fe(OH)2 + 2H+ + 2e = Fe + 2H,O;
E = -0.047 - 0.059pH
. . .(1.16)
1:66
CORROSION IN AQUEOUS SOLUTIONS
Curvec,Fe(OH),
+ H+ + e = Fe(OH), + H,O;
E = 0.271 - 0.059pH . . .(1.17) Curved, Fe(OH), + e = FeO,H- + H20; E = -0.81 -0.05910gaFa2,. . .(1.18) Curve e, Fe(OH), + 3H+ e = Fez++ 3H20; E = 1.060 - 0.177pH - 0*059logaFez+ . . .(1.19) Since equation 1.15 does not involve H+ it is pH independent, and variation of aFs+will result in a series of curves parallel to the pH axis that extend across the diagram until the pH is sufficiently high to reduce the O,,Z+ to < g ion/l by formation of Fe(OH), . The relevant equilibrium
+
is:
+
+
Fez+ 2H20 = Fe(OH), 2Hf for which values of p e (J) for the species involved are Fe2+(aq.), -84 760; H 2 0 , -236 700; Fe(OH),, -482 OOO, and since
and at 25°C log K = Since
[ -482 OOO - 2 ( -236 700) - ( -84 760) ] = -13.37 2-303 x 8.315 x 298.16
. . .(1.20b)
a$+ K = - - ,then logK = logo;+ - logaFez+
. . .(1.21) .. .(1.22)
OF€?+
and 10gaFez+= 13.27 - 2pH the equilibrium pH will be 9-7 and the o,,z+ will be Thus at u,,z+ = e at any higher pH value; it follows that the formation of a new solid phase Fe(OH), at a sufficiently high pH must limit the zone of corrosion as defined by Pourbaix. Whereas lowering the potential results in a decrease in uFez+,the converse applies when the potential is raised. However, this increase in activity is again limited by the formation of a solid phase. Thus curve e of Fig. 1.15 (top) gives the equilibrium between Fe(OH), and Fe2+ at any At uFez+= predetermined activity of the latter in the range 10' 10-6g-ion/l, E = [ 1.06 + ( -6 x 0.059)] - 0-177pH which defines the boundary between corrosion and passivity at high potentials (equation 1.19). At high pH values and low potentials, Fe, Fe2+, Fe3', Fe(OH), and Fe(OH), ,etc. will be thermodynamically unstable with respect to Fe0,Hand a further limited zone of corrosion will appear on the right-hand side of the diagram. Significance of Zones in Potential-pH Diagrams
The above outline of the method adopted in the construction of the potential-pH diagram of the Fe-H,O system serves to illustrate the essentially
CORROSION IN AQUEOUS SOLUTIONS
1:67
thermodynamic nature of diagrams of this type, which therefore cannot provide any information on the rates of corrosion processes. However, on the basis of certain assumptions that have no thermodynamic significance, it is possible to separate the Fe-H20 diagram into the following zones [Fig. 1.15 (bottom)] Corrosion: activity of Fez+, Fe3+(aq.) or FeOaaq.) > 10-6g ion/] (0.06p.p.m. of Fe in the case of the cations). Passivity: Fe(OH),, Fe,O, or Fe(OH), in equilibrium with metal ions at an activity < 1 W g ion/l. Immunity: Fe metal in equilibrium with Fe2+(aq.)or FeO,H-(aq.) at an ion/l. activity < It should be noted that Fig. 1.15 (top) is based entirely on thermodynamic data and is therefore correctly described as an equilibrium diagram, since it shows the phases (nature and activity) that exist at equilibrium. However, the concepts implicit in the terms corrosion, immunity and passivity lie outside the realm of thermodynamics, and, for example, passivity involves both thermodynamic and kinetic concepts; it follows that Fig. 1.15 (bottom) cannot be regarded as a true equilibrium diagram, although it is based on one that has been constructed entirely from thermodynamic data. In Fig. 1.15 (bottom) curvestand m show the potential-pH relationships for the reversible hydrogen and oxygen electrodes at p H 2= p o 2 - 1 atm respectively. Within the area confined by the curves L and m, H,O is thermodynamically stable and p H Ic 1 and poz < 1; whereas below Land above m, p H 2> 1 atm, and po, > 1 atm, respectively (see equations 1.11 and 1.12). Thus the diagram shows the solid phases of iron, the activities of metal ions and the pressures of hydrogen and oxygen gas that are at equilibrium at any given potential and pH when pure iron reacts with pure water. This can be illustrated by considering the changes that will tend to occur when iron with a coating of rust (Fe,O, .HzO) is immersed in oxygenated water at pHs and potentials that correspond with the various zones in the Fe-HzO diagram6. Immunity Any FezO, on the surface (or any Fez+ in solution) will be reduced to metal, and aFe2+< 10-6g ion/l; water will be reduced to hydrogen and p H 2> 1 atm; any dissolved oxygen present will be reduced to OH-, and po2 10-6g ion/l; water will be reduced to hydrogen or remain stable, depending upon whether E is below or above curved. At high potentials iron will be oxidised to Fe3+and = 0-76V); water will be stable FQO, will dissolve to form Fe3+ (EFe3+/Fez+ or will be oxidised to oxygen, depending upon whether E is below or above curve m,respectively. Passivation According to Fig. 1.15 (top) all the Fe will be converted to FQO,, whilst the rust originally present will be unaffected. According to Fig. 1.15 (bottom) the rust will be unaffected, whilst the iron surface exposed to the solution through pores in the rust will be passivated by a protective film of Fe,O, Water will be stable except at high potentials where it will be oxidised to Oz
.
.
1:68
CORROSION IN AQUEOUS SOLUTIONS
Thus the tendency for an electrochemical reaction at a metal/solution interface to proceed in a given direction may be defined in terms of the relative values of the actual electrode potential E (experimentallydetermined and expressed with reference to the S.H.E.) and the reversible or equilibrium potential E, (calculated from E e and the activities of the species involved in the equilibrium). When E > E, the reaction can only proceed in the direction of oxidation. When E < E, the reaction can only proceed in the direction of reduction. When E = E, the reaction is at equilibrium. This can be summarised by the relationship
. . .(1.23) ( E - E,)Z 2 0 in which Z, the reaction current, is regarded as positive in the case of oxidation and negative in the case of reduction. The parameter (E - E,), which may be positive, negative or zero, is termed the overpotential or ufinity, and gives the tendency for the reaction to proceed in the direction of oxidation or reduction or to be at equilibrium, respectively. However, the precise magnitude of I will depend upon kinetic factors, which will be considered subsequently. Advantages and Limitations of Diagrams
Although the zones of corrosion, immunity and passivity are clearly of fundamental importance in corrosion science it must be emphasised again that they have serious limitations in the solution of practical problems, and can lead to unfortunate misconceptions unless they are interpreted with caution. Nevertheless, Pourbaix and his co-workers, and others, have shown that these diagrams used in conjunction with E-i curves for the systems under consideration can provide diagrams that are of direct practical use to the corrosion engineer, It is therefore relevant to consider the advantages and limitations of the equilibrium potential-pH diagrams. The M-H,O diagrams present the equilibria at various pHs and potentials between the metal, metal ions and solid oxides and hydroxides for systems in which the only reactants are metal, water, and hydrogen and hydroxyl ions; a situation that is extremely unlikely to prevail in real solutions that usually contain a variety of electrolytes and non-electrolytes. Thus a solution of pH 1 may be prepared from either hydrochloric, sulphuric, nitric or perchloric acids, and in each case a different anion will be introduced into the solution with the consequent possibility of the formation of species other than those predicted in the M-H20system. In general, anions that form soluble complexes will tend to extend the zones of corrosion, whereas anions that form insoluble compounds will tend to extend the zone of passivity. However, provided the relevant thermodynamic data are available, the effect of these anions can be incorporated into the diagram, and diagrams of the type M-H,O-Xare available in Cebelcor reports and in the published literature. The effect of anions on the zones of corrosion and passivation can be exemplified by a comparison of the Pb-H20 and Pb-H,O-SO:equilibrium diagrams (see Section 4.3, Figs. 4.13 and 4.14) and it can be seen that in the presence of SO:- the corrosion zone corresponding with stability of
1:69
CORROSION IN AQUEOUS SOLUTIONS
Pb2+is completely replaced by PbSO, so that passivation is possible in the acid region owing to the thermodynamic stability of PbS0,. Similar considerations apply to the potential E which can be varied by means of an auxiliary electrode and an external source of e.m.f., or by varying the redox potential of the solution, and in the case of the latter a given redox potential may be achieved by using different oxidants. Thus at pH 7 and E = -0-44V,iron will be in the zone of corrosion (aFe2+= l), and the potential could be raised into the passive region by either dissolved oxygen, potassium chromate or potassium perchlorate. However, the effects produced will depend upon a variety of factors, and whereas passivation can be achieved if chromate is present in sufficient concentration, it may cause pitting at lower concentrations. Perchlorate will tend to cause pitting, and dissolved oxygen can result in localised attack and will passivate iron only if it is brought rapidly and simultaneously to all parts of the metal surface. A further serious limitation is that diagrams evaluated from thermodynamic data at 25OC have little relevance in high-temperature aqueous corrosion, but it is now possible to construct that are applicable at elevated temperatures from data obtained at 25" C (see Section 2.1). Pourbaix6 has studied the behaviour of iron in city water (Brussels) at Table 1.12 State environments and iron in BNSWIS water; see also Fig. 1.16 (after Pourbaix6 , Experiment
a
b C
C'
d
e
0
Sample number 1 2 3 4 5 6 7
8 9 10 11 12 13 14 15 16 17 14' 15' 16' 17' 18 19 20 21 22 23 24 25
EH
Solution
H 2 0 distilled NaCl
1 g/l 1 g/l NaHSO, 1 g/l 1 g/l NaOH K2 CrO, 1 g/l K2Cr04 NaCl 1 g/l KMn0, 0.2 g/l KMn04 1 g/l H20 2 0.3 g/l H20 2 3.0 g/l Brussels city water NaOH 40 g/l degassed city water-iron-copper city water-iron-zinc city water-iron-magnesium city water-iron-platinum city water-iron-copper city water-iron-zinc city water-iron-magnesium city water-iron-platinum NaHCO, 0.1 M Pole NaHCO, 0 - 1 M Pole NaHC0, 0.1 M Pole NaHCO, 0 - 1 M Pole NaHC0, 0.1 M Pole NaHC0, 0.1 M Pole + NaHCO, 0.1 M Pole NaHC03 0.1 M Pole +
General corrosion; a local corrosion;
+
+
0
absence of corrosion.
pH
(v)
8.1 6.9 2.3 6.4 11.2
-0.486 -0.445 -0.351 -0.372 +0.026 +0.235 -0.200 -0.460 +0.900 -0.200 +0.720 -0-450 -0.810
8 .5
8.6 6.7 7.1 5.7 3.4 1-0
13.7 7.5 7.5 7-5 7-5 7.8 7.7 8.7 8-4 8-4 8-4 8.4 8-4 8-4 8.4 8.4
-0.445
-0.690 -0-910 -0.444 -0.385 -0.690 -0.495 -0 860 -0.350
-
-0.885
Stateof metu/* e e e
e 0 0
Q
e 0 0 0 0
e e 0 0 0
e 0 0
e 0
e 0
+1-380
0
-0-500
e
+lo550 -1-OOO
0
f1.550
0
0
GUS
1:70
CORROSION IN AQUEOUS SOLUTIONS
I
I
I
2
4
6
I 8
I
I
10
12
5
PH
Fig. 1.16 Potential-pH diagram for the Fe-H20 system in which results obtained for the behaviour of iron in Brussels water have been inserted (see Table 1.12) (after Pourbaix6)
various pHs and potentials (Table 1.12); the study of the former was effected by adding acids and alkalis and the latter by applying an external e.m.f., by coupling the iron to either more positive (Cu, Pt) or more negative metals (e.g. Zn, Mg), or by adding oxidants (e.g. K2Cr0,, KMnO,, H,O,). The corrosion rate of the iron has been determined and results have been inserted in the potential-pH diagram for Fe-H,O (Fig. 1.16), and it can be seen that in this particular water there is good agreement between the predictions of the diagram and the corrosion behaviour of the iron. However, it does not follow that this correlation would necessarily apply to all fresh waters or to sea-water. Zones of Corrosion
The Zn-H,O (Fig. 1.17) diagram lo shows that extensive corrosion zones exist at both Iow and high pH values (compare the very restricted corrosion zone in the Fe-H,O diagram at high pHs); similar zones in the region of low and high pH are obtained with other amphoteric metals such as aluminium, lead and tin. The diagram for Zn-H,O predicts with some accuracy the behaviour of the metal in practice, where it has been established that zinc corrodes rapidly outside the range pH 6-12-5 but is passive within
CORROSION IN AQUEOUS SOLUTIONS
1:71
I!
0
Zn (OH12
-> II: -1
Zn
PH
Fig. 1.17 Simplified potential-pH diagram for the Zn/H20 system (after Delahay. Pourbaix and Van Rysselberghe'O)
it. On the other hand, iron in sodium hydroxide at 25°C at potentials and pHs corresponding with the triangle of corrosion on the right-hand side of Fig. 1.15 [FeO,H-(aq.) stable] shows little evidence of attack, although the presence of this zone does explain the phenomenon of caustic cracking. However, it should be noted that caustic cracking normally occurs in alkaline waters at elevated temperatures and that temperature will have a marked effect on both the thermodynamics and kinetics of the reaction (see Section 2.1). Finally, it should be noted that although the arbitrary activity of g ion/l represents a very low concentration of metal ions it could be significant in certain circumstances, e.g. lead at that concentration would render potable water toxic. It should also be noted that if the equilibrium is continuously disturbed, e.g. by a flowing solution, significant amounts of metal g ion/l. will corrode even at an equilibrium activity of Zone of Immunity
The region of immunity [Fig. 1.15 (bottom)] illustrates how corrosion may be controlled by lowering the potential of the metal, and this zone provides the thermodynamic explanation of the important practical method of cathodic protection (Section 11.1). In the case of iron in near-neutral solutions the potential E = -0.62 V for immunity corresponds approximately with the practical criterion adopted for cathodically protecting the metal in most environments, i.e. -0.52 to -0.62V (vs. S.H.E.). It should be observed, however, that the diagram provides no information on the rate of charge transfer (the current) required to depress the potential into the region of immunity, which is the same ( < -0.62 V) at all values of pH below 9 - 8 . Consideration of curve L for the H2/H,0 equilibrium shows that as the pH
1:72
CORROSION IN AQUEOUS SOLUTIONS
decreases the thermodynamic tendency of water to become reduced to hydrogen increases, and although theoretically cathodic protection is feasible in the acid region it would be economically impracticable, since hydrogen evolution at a very high rate would occur at the potentials required to achieve immunity. Thus cathodic protection is normally confined to near-neutral solutions. In the case of the AI-H20 diagram system (Section 4.1, Fig. 4.41) immunity in the near-neutral region can be achieved only at potentials e - 1 * 82 V, which cannot be attained in practice owing to the hydrogen evolution reaction, which is the thermodynamicallypreferred process. However, owing to the presence of a surface oxide film the potential of aluminium in practical environments is far more positive than the reversible potential, e.g. the corrosion potential in sea-water is -0.55 V compared to E&+,AI= -1 -7 V, and cathodic protection may be achieved in practice by making the potential 100 mV more negative than the corrosion potential. This is because aluminium in neutral chloride-containing environments corrodes by pitting, and the criterion of cathodic protection is thus the critical pitting potential (see Sections 1.5 and 1.6) and not the zone of immunity of the potential-pH diagram, and similar considerations apply to the cathodic protection of stainless steels. The Al-H20 diagram does show, however, the danger that may arise due to an increase in pH when the metal is cathodically protected in near-neutral solutions; indeed, the possibility of alkaline corrosion has seriously limited the use of cathodic protection for aluminium structures. Zone of Passivity
Although thermodynamics can predict the region of pH and potential in which solid oxides, hydroxides and other compounds are stable, it can provide no other information; thus on the basis of these considerations alone a metal in the passive region should be completely converted to a solid compound by reacting with water with a consequent loss of properties. Implicit in the concept of passivity is the assumption that the solid compound forms a kinetic barrier between the reactants so that further interaction becomes very slow. Whether this occurs in practice will depend on the position of formation of the oxide (an oxide produced by the sequence M M2'(aq.) + M20,.H20 is likely to precipitate away from the metal surface owing to the mobility of the Mz" (aq.) ion and to be non-protective, and an oxide produced directly, e.g. M + MxO, is likely to form on the metal surface and to be protective), the adhesion of the oxide to the metal, the solubility of the oxide, its cohesion, crystal form, etc. Thus iron in a neutral chloride solution maintained in the region of passivation by dissolved oxygen will corrode owing to the fact that the hydrated oxide Fe20,.H20 precipitates away from the metal surface and is therefore non-protective. Similarly, metals such as magnesium, aluminium and zinc, which according to the relevant potential-pH diagram are all passive in near-neutral solutions at elevated potentials, can be used as sacrificial anodes in sea-water, since the presence of the chloride ion precludes passivation; in fact in this particular application it is essential to ensure that the metal does not passivate, and in the case of aluminium and zinc, additions of mercury may be used to prevent +
1:73
CORROSION IN AQUEOUS SOLUTIONS
the formation of a protective film thus facilitating uniform corrosion (see Section 10.2). A number of metals and alloys can be passivated in the acid region at elevated potentials, although this phenomenon is not evident from the pH-potential diagrams, which give the impression that the metal will corrode. Thus iron is passivated in fuming nitric acid, and aluminium in nitric acid at concentrations > = 70%; iron, nickel and cobalt can be passivated in sulphuric acid by raising the potential by applying an external source of e.m.f. The reason for this behaviour is that although the passive zone is based on the thermodynamic stability of solid compounds it is possible for these compounds to exist as metastable phases outside the regions defined by thermodynamic data. Under these circumstances the rate of corrosion of the metal will be controlled by the rate of transport of metal cations through the film and by the dissolution of the oxide in the solution (see Section 1.4). The fact that oxides can exist as metastable phases is illustrated by the Ni-H20 diagram (Fig. 1.18) in which the curves for the various oxides of nickel have been extrapolated into the acid region of Niz+stability, and this diagram emphasises the fact that nickel can be passivated outside the region of thermodynamic stability of the oxides". The converse situation occurs when complexants are present, and this can be exemplified by the Cu-H,O-NH, system (Fig. 1.19) in which the zones
I
"
-0.4 -0.6
-
NiO
-
1
0
-.
Ni I
1
I
2
I
3
1
4
1
5
I
6
1
7
I
8
I I 1 9 1 0 1 1
I
I
1213
I
PH Fig. 1.18 Modified potential-pH diagram for the Ni-H20 system; the curves showing the stability of the nickel oxides have been extrapolated into the acid region to indicate the formation of metastable oxides (after De Gromoboy and Shreir")
CORROSION IN AQUEOUS SOLUTIONS
Fig. 1.19 Potential-pH diagram for copper in solutions containing Cu2+ and (NH,),S04 (after Mattson ") with superimposedtimes to fracture Tf of direct-loaded o-brasswires held at various potentials in the solution of pH 7.2; the specimen without external polarisation had Tf= 3; h (after Hoar and Booker 1 3 )
of passivation due to the stability of Cu,O and CuO are confined to a narrow region of pH, whereas in the Cu-H,O water system passivation can be achieved at all pHs > = 6.8 (Section 4.2, Fig. 4.10). Importance of Potential-pH Di8gr8mS in Corrosion Science
Since the remainder of this section will be confined to a consideration of rates of corrosion reactions, it is appropriate to conclude this review of potential-pH diagrams with an assessment of their significance in corrosion science. In this connection, it is relevant to consider the compositiontemperature equilibrium phase diagrams of alloy systems, which provide the foundation for the study of the structure of metals and alloys, although it is recognised that these diagrams have serious limitations owing to the fact that many commercial alloys are not in a state of equilibrium. This can be exemplified by the Fe-C equilibrium diagram, in which ferrite (a-Fe) and
1:75
CORROSION IN AQUEOUS SOLUTIONS
graphite are the phases at equilibrium; however, a variety of metastable phases such as cementite and martensite can be formed by suitable heat treatments. The Fe-graphite equilibrium diagram predicts that a liquid alloy containing 3% C when cooled to ambient temperatures will consist of ferrite and graphite, but if the alloy is cooled rapidly graphite formation is suppressed, resulting in a hard, brittle white iron composed of cementite and ferrite. Subsequent annealing at 870°C will cause the system to tend to the equilibrium state, and the cementite will decompose slowly with the formation of the stable phase graphite. There are numerous examples that could be quoted to show that commercial alloys contain metastable phases that do not conform with the predictions of the relevant equilibrium diagram, but the usefulness of these diagrams is not disputed. However, in the case of steels the kinetics of the isothermal transformation of austenite can be presented in the form of transformation-time-temperature (TTT) diagrams which show the phases that form at different temperatures after a given time (the axes are transformation temperature vs. the logarithm of the time of transformation). These TTT diagrams (see Section 20.4) may be regarded as being analogous to the E-log I diagrams, since the former provide information on the isothermal rate of mass transfer, whilst the latter provide information on both charge transfer and mass transfer. Thus the potential-pH diagrams and the E-I diagrams may be regarded as complementary in the study of corrosion phenomena and in the solution of corrosion problems. A survey of the literature (see pages 1.114 to 1.117) shows that numerous workers in the field of corrosion have used potential-pH diagrams in order to throw more light on the mechanism of a corrosion process. As an example, some consideration will be given to the stress corrosion of a-brass, which also serves to illustrate diagrams of the type M-H,O-X, where in this particular case X i s NH, . Mattsson” constructed an equilibrium potential-pH diagram for copper in solutions of (NH,),SO, at various pHs (Fig. 1.19), and also studied experimentally the rate of cracking of brass in dilute solutions containing NH,, Cu2+ ions and SO:- ions at various pH values of from 2 to 11. It was found that intergranular cracking occurred mainly in the range pH 6-3-7-7, and was most rapid in the range pH 7 1-7 -3. The diagram shows that, thermodynamically, Cu,O formation becomes increasingly easy up to pH 7 . 3 , and although previous workers in this field had observed that a black film of corrosion product (subsequently identified as Cu,O) accompanied cracking, it was regarded as of little significance. Mattsson suggested that the cuprous oxide stimulated cracking, and this was confirmed by the subsequent work of Hoar and Bookerl3 who studied the time to failure (T,) of stressed brass wires at various pHs and potentials (E,) corresponding with significant zones in the potential-pH diagram. They found that rapid fracture occurred in the pH range 7 1-7 - 3, and that at fracture the potential of the brass had risen from 0.15 to 0.25 V, corresponding with the anodic formation of Cu,O from Cu. At potentials below -0-05 V, at which Cu,O cannot form, cracking was arrested. It is not appropriate here to consider the precise r6le of Cu,O in the cracking process, but this example does serve to illustrate the usefulness of potentialpH diagrams in providing information that can assist in establishing the mechanism of corrosion
-
1:76
CORROSION IN AQUEOUS SOLUTIONS
Electrochemical Mechanism of Corrosion The rate (or kinetics) and form of a corrosion reaction will be affected by a variety of factors associated with the metal and the metal surface (which can range from a planar outer surface to the surface within pits or fine cracks), and the environment. Thus heterogeneities in a metal (see Section 1.3) may have a marked effect on the kinetics of a reaction without affecting the thermodynamics of the system; there is no reason to believe that a perfect single crystal of pure zinc completely free from lattic defects (a hypothetical concept) would not corrode when immersed in hydrochloric acid, but it would probably corrode at a significantly slower rate than polycrystalline pure zinc, although there is no thermodynamic difference between these two forms of zinc. Furthermore, although heavy metal impurities in zinc will affect the rate of reaction they cannot alter the final position of equilibrium. The essential features of the electrochemical mechanism of corrosion were outlined at the beginning of the section, and it is now necessary to consider the factors that control the rate of corrosion of a single metal in more detail. However, before doing so it is helpful to examine the charge transfer processes that occur at the two separable electrodes of a well-defined electrochemical cell in order to show that since the two half reactions constituting the overall reaction are interdependent,their rates and extents will be equal.
Electrochemical and Electrolytic Cells More positive
Fig. 1.20 Cell consisting of two reversible Ag+/Agelectrodes ( A g in AgNO3 solution). The rate and direction of charge transfer is indicated by the length and arrow-head as follows: g a h of electrons by Ag+ + e Ag-b;loss of electrons by Ag Ag+ + e t . (a) Both electrodes at equilibrium and (b)electrodes polarised by an external source of e.m.f.; the position of the electrodes in the vertical direction indicates the potential change. (V,high-impedance voltmeter; A , ammeter; R , variable resistance) +
+
CORROSION IN AQUEOUS SOLUTIONS
1:77
An electrochemical cell is a device by means of which the enthalpy (or heat content) of a spontaneous chemical reaction is converted into electrical energy; conversely, an electrolytic cell is a device in which electrical energy is used to bring about a chemical change with a consequent increase in the enthalpy of the system. Both types of cells are characterised by the fact that during their operation charge transfer takes place at one electrode in a direction that leads to the oxidation of either the electrode or of a species in solution, whilst the converse process of reduction occurs at the other electrode. For simplicity a cell consisting of two identical electrodes of silver immersed in silver nitrate solution will be considered first (Fig. 1.2Oa), Le. Ag, /AgNO, /Ag,, . On open circuit each electrode will be at equilibrium, and the rate of transfer of silver ions from the metal lattice to the solution and from the solution to the metal lattice will be equal, i.e. the electrodes will be in a state of dynamic equilibrium. The rate of charge transfer, which may be regarded as either the rate of transfer of silver cations (positive charge) in one direction, or the transfer of electrons (negative charge) in the opposite direction, in an electrochemical reaction is the current I, so that for the equilibrium at electrode I
. . .(1.24) In this equation Io,,is th%equilibrium exchange current, and the arrow convention adopted is that I,, represents the rate of cathodic reduction
Ag+(aq.) and
+ e+Ag(l)
. . .(1.25)
I’,represent the rate of anodic oxidation Ag(1) 4 Ag+(aq.)
+e
. . .(1.26)
If the areas of the electrodes are assumed to be 1 cm’, and taking the equilibrium exchange current density io for the Ag+/Ag equilibrium to be lo-’ A then Io will be lo-’ A, which is a very high rate of charge transfer. A similar situation will prevail at electrode 11, and rates of exchange of silver ions and the potential will be the same as for electrode I. It is apparent from this that since the rates of the cathodic and anodic processes at each electrode are equal, there will be no net transfer of charge; in fact, with this particular cell, consisting of two identical electrodes in the same electrolyte solution, a similar situation would prevail even if the electrodes were short-circuited, since there is no tendency for a spontaneous reaction to occur, i.e. the system is at equilibrium and AG = 0. Consider now the transfer of electrons from electrode I1 to electrode I by means of an external source of e.m.f. and a variable resistance (Fig. 1.20b). Prior to this transfer the electrodes are both at equilibrium, and the equilibrium potentials of the metal/solution interfaces will therefore be the same, i.e. E, = E,, = E,, where E, is the reversible or equilibrium potential. When transfer of electrons at a slow rate is made to take place by means of the external e.m.f., the equilibrium is disturbed andfhe rates of the charge transfer processes become unequal. At electrode I, I**,, > I,,,, and there is
1:78
CORROSION IN AQUEOUS SOLUTIONS
+
c
now a net cathodic reaction (equation 1.25). At electrode 11, IAg,II > IAg,,I, and there is now a net anodic reaction (equation 1.26). The rate of transfer of electrons in the external circuit I,, which is the rate actually measured by the ammeter, is the difference between rates of the dominant or forward reaction and the subsidiary or reverse reaction at each electrode, and it follows that
By definition, electrode I1 at which oxidation is the predominant reaction is the anode, whereas electrode I at which reduction is the predominant reaction is the cathode. It is apparent that the removal of electrons from Ag,, will result in the potential of its interface becoming more positive, whilst the concomitant supply of electrons to the interface of Ag, will make its potential become more negative than the equilibrium potential: Ep,c
Er
c Ep,a
. . .(1.28)
where Ep,cand Ep,aare the polarised potentials of the cathode and anode, respectively. If the resistance in the external circuit is decreased sufficiently so that Ep3c
Er
ab.II, and for the equilibrium Ag' e Ag the e.m.f. of the cell will be
+ *
If aAS,* = 1-0 and in which must be positive since ab,l > ah,,, = 0-1, then Er,m,l = 0-059V and electrode I will be the cathode [E, = (0.79 O-O06)V]and electrode I1 the anode [E, = (0.079 - O.O6O)V]. If the cell operates spontaneously, charge transfer takes place until aAg.!= aAg,,, when AG becomes zero, i.e. the system is at equilibrium.
+
CORROSION IN AQUEOUS SOLUTIONS
1:79
V
-r
-
‘Cathode
Anode’
‘Ail 11
Fig. 1.21 Concentration cell in which uAg+,lI< uAg+,rso that charge transfer occurs spon-
taneously and proceeds until the activities are equal (Ej is the liquid junction potential at the sintered glass plug that is used to minimise mixing of the two solutions)
Rates and Extents 1 Cs-’. The The current l i s the rate of charge transfer, and a rate of 1 A charge on the electron is 1-6020 x C, and since 1 mol of an element contains 6-023 5 x loz3atoms (Avogadro’s number) the cathodic reduction x 1.602 0 x of a univalent metal ion M + will require 6.023 5 x = 96 494 C. This statement is essentially Faraday’s law and 1 faraday = 96 494 C = 96 500 C. In the more general case of the anodic oxidation or cathodic reduction of a species by change transfer, zF coulombs will be required for 1 mol where
1:80
CORROSION IN AQUEOUS SOLUTIONS
z is the number of electrons required to carry out one act of the electron transfer process. Thus the rate of charge transfer of an electrochemical reaction is given by
r
k, = ZF
(molss-')
or
. . .(1.32) .(1.33)
where M is the molar mass (kgmol-') of the species involved. In these examples of cells the areas of the electrodes have been disregarded, but in electrode kinetics it is the rate per unit area of the electrode that is of significance. The rate per unit area is the current density i, and
. . .(1.34)
i = Is-'
where S is the area, and i therefore has units of A cm-', A m-*, mA cmP2, etc. Thus equation 1.33 can be expressed in terms of a rate per unit area .(1.35)
where i is in A cm-2. (Note that although the metre is the recommended S.I. unit, the centimetre is so widely used that it will be retained in certain units in this book.) The extent of an electrochemical reaction is the quantity of charge Q (coulombs) transferred in a given time t, and
. . .(1.36) Q = It where t is the time (s). Since 1 mol of charge transfer requires ZF coulombs It
Q=-
ZF
orQ=-
it
ZF
(molcm-2)
itM ItM alternatively, Q = - (kg) , or Q = - (kg crn-') ZF ZF
. . .(1.37) . . .(1.38)
It is apparent (Fig. 1.21) that at potentials removed from the equilibrium potential (see equation 1.30) the rate of charge transfer of (a) silver cations from the metal to the solution (anodic reaction), (6) silver aquo cations from the solution to the metal (cathodic reaction) and (c) electrons through the metallic circuit from anode to cathode, are equal, so that any one may be used to evaluate the rates of the others. The rate is most conveniently determined from the rate of transfer of electrons in the metallic circuit (the current I) by means of an ammeter, and if I is maintained constant it can also be used to evaluate the extent. A more precise method of determining the quantity of charge transferred is the coulometer, in which the extent of a single well-defined reaction is determined accurately, e.g. by the quantity of metal electrodeposited, by the volume of gas evolved, etc. The reaction Ag+(aq.) + e = Ag is utilised in the silver coulometer, and provides one of the most accurate methods of determining the extent of charge transfer.
CORROSION IN AQUEOUS SOLUTIONS
1:81
Partial Cathodic and Anodic Reactions
The reversible cells described are characterised by the fact that the same charge transfer process Ag+(aq.)+ Ag(1) takes place at each electrode. However, this type of cell is not typical, and the fact that the exchange processes usually differ can be exemplified by considering an electrolytic cell consisting of two platinum electrodes immersed in a solution of deoxygenated Na,SO,, in which water is decomposed into hydrogen and oxygen
HzO -,H, + +02 by the interdependent reactions 2H,O 2e = H, 20H(cathode) and H,O = +O, 2H' + 2e (anode)
+
+
+
. . .(1.39) . . .(1.40)
In contrast to the cell consisting of two reversible Ag+/Ag electrodes, charge transfer cannot occur until the e.m.f. of the equivalent reversible cell is exceeded (> 1-23V)*, and in view of the high overpotential for oxygen evolution the rate will not be significant until the e.m.f. is more than about 1 -6 V. If the deoxygenated Na,SO, solution is replaced by an oxygenated sodium chloride solution, the followingadditional reactions are also possible . . .(1.41) 0, 2H,O 2e = 20H- (cathode)
+
+
+
(anode) . . .(1.42) 2C1- + C1, 2e It is not appropriate here to consider the kinetics of the various electrode reactions, which in the case of the oxygenated NaCl solution will depend upon the potentials of the electrodes, the pH of the solution, activity of chloride ions, etc. The significant points to note are that (a) an anode or cathode can support more than one electrode process and (b)the sum of the rates of the partial cathodic reactions must equal the sum of the rates of the partial anodic reactions. Since there are four exchange processes (equations 1.39-1.42) there will be eight partial reactions, but if the reverse reactions are regarded as occurring at an insignificant rate then This leads to the fundamental concept that irrespective of the number of electrode processes or whether they occur on one or more than one electrode surface CI, = EIa (cs-') . . .(1.44) i.e. during charge transfer the sum of the partial cathodic currents must equal the sum of the partial anodic current. From Faraday's law .( 1.45)
which means that the sum of the oafhodic processes must equal the sum of the anodic processes with respect to both the rate at any instant of time and the extent after any period of time. This law applies equally to these cells, *The value of 1-23 V follows from Table 1.8 and Fig. 1.15 (bottom).
1:82
CORROSION IN AQUEOUS SOLUTIONS
the electrodes are separable, and to those corrosion cells in which the electrodes cannot be distinguished physically (see Table 1.Al).
Electrode Area end Cumnt Density
It is now necessary to consider equation 1.45 in terms of the rates per unit area of the electrode surfaces, which may be equal or unequal depending on circumstances. When the area of the cathode equals the areas of the anode, equation 1.44 is applicable, and I can be replaced by i, the current density: Ci, = Cia
( C cm-2 s-')
and
. . .(1.46) .(1.47)
which may be regarded as the criterion for the uniform corrosion of a single metal. However, if the area of the anode is smaller than that of the cathode, then i, > i, and Cia > Ci,
. . .(1.48)
whereas if the cathode area is smaller, i, 2 i, and Ci, > Cia
.. .(1.49)
and it follows that the criterion for localised attack (or pitting) will be .(l.SO)
and that the greater the magnitude of this ratio the more intense will be the attack. Equation 1.SO is referred to as the pitting ratio. The current density i requires a knowledge of both the current Z and the area of the electrode S, and the latter is seldom equal to the geometrical or superficial area. In fact, it is possible to distinguish the following types of areas in relation to electrode processes that take place on metal surfaces: geometrical, true, active and eflective. In uniform corrosion the supe@kial or geometrical area of the metal is used to evaluate both the anodic and cathodic current density, although it might appear to be more logical to take half of that area. However, surfaces are seldom smooth and the true surface area may be twice to three times that of the geometrical area (a cleaved crystal face or an electropolished single crystal would have a true surface area that approximates to its superficial area). It follows, therefore, that the true current density is smaller than the superficial current density, but whether the area used for calculating i, and i, is taken as either equal to or half of the superficial area, is unimportant compared to the fact that they are equal. Furthermore, the number of metal atoms at the metal surface that are active in the processes of metal dissolution or cathodic reduction are usually far less than the total number available. Thus Hoar and Notmani6 have shown during the anodic dissolution of nickel that the current density at any given potential is increased by a
CORROSION IN AQUEOUS SOLUTIONS
1:83
factor of 10 by cold working the annealed nickel. Similar considerations apply to the hydrogen evolution reaction, in which the coverage of the atoms on a metal surface with adsorbed hydrogen atoms can range from very small to about 100V0, depending on the mechanism of the reaction (see Section 20.1). Thus the active surface area (the number of atomic sites that participate in the electrode process) may be appreciably less than the total number of sites available. Finally, it is important to point out that although in localised corrosion the anodic and cathodic areas are physically distinguishable, it does not follow that the total geometrical areas available are actually involved in the charge transfer process. Thus in the corrosion of two dissimilar metals in contact (bimetallic corrosion) the metal of more positive potential (the predominantly cathodic area of the bimetallic couple) may have a very much larger area than that of the predominantly anodic metal, but only the area adjacent to the anode may be effective as a cathode. In fact in a solution of high resistivity the effective areas of both metals will not extend appreciably from the interface of contact. Thus the effective areas of the anodic and cathodic sites may be much smaller than their geometrical areas. It follows from equation 1.45 that the corrosion rate of a metal can be evaluated from the rate of the cathodic process, since the two are faradaically equivalent; thus either the rate of hydrogen evolution or of oxygen reduction may be used to determine the corrosion rate, providing no other cathodic process occurs. If the anodic and cathodic sites are physically separable the rate of transfer of charge (the current) from one to the other can also be used, as, for example, in evaluating the effects produced by coupling two dissimilar metals. There are a number of examples quoted in the literature where this has been achieved, and reference should be made to the early work of Evans” who determined the current and the rate of anodic dissolution in a number of systems in which the anodes and cathodes were physically separable. More recently, Fontana and Greenel*measured the current between a pit in stainless steel and the surrounding metal; the pit was allowed to form, and cut out from the surrounding metal (the cathode), its edge was insulated and it was then replaced in the hole with a suitable connection for measuring the current flow between the pit and the surrounding metal. These workers showed that under certain conditions i, was about a thousand times i, . When the anodic and cathodic sites are inseparable the corrosion current cannot be determined directly by an ammeter, but it can be evaluated electrochemically by the linear polarisation technique (see Sections 19.1- 19.3).
Electrochemical Cells and Corrosion Cells One of the most well-known electrochemical cells that is used for the conversion of chemical energy into electrical energy is the Daniel1 cell Zn IZnSO,( aq. 1I CuSO,( as.) I Cu . . .(1.51) in which the spontaneous reaction . . .(1.52) Cu2+(aq.) + Zn -, Zn2+(aq.) Cu takes place. The half reactions that constitute the overall reaction are
+
1:84
CORROSION IN AQUEOUS SOLUTIONS
Zn
+
Zn2+(aq.)
Cuz+(aq.)
+ 2e
+ 2e -+
Cu
(anodic reaction) (cathodic reaction)
and electrons are transferred from the zinc to the copper through the metallic circuit. During the operation of the cell (or during the direct interaction of zinc metal and cupric ions in a beaker) the zinc is oxidised to ZnZ+and corrodes, and the Daniell cell has been widely used to illustrate the electrochemical mechanism of corrosion. This analogy between the Daniell cell and a corrosion cell is perhaps unfortunate, since it tends to create the impression that corrosion occurs only when two dissimilar metals are placed in contact and that the electrodes are always physically separable. Furthermore, although reduction of Cuz+(aq.)does occur in certain corrosion reactions it is of less importance than reduction of H,O+ ions or dissolved oxygen.
-
Electronic condudion I,
e
OS
Fig. 1.22 Spontaneous corrosion of zinc in acid illustrated by the reversible cell Zn I Zn2+ I H, O + , H, 1 Pt. The individual potentials of the electrodes are determined by a reference electrode (Ref) and a Luggin capillary to minimise the IR drop in the solution
CORROSION IN AQUEOUS SOLUTIONS
1:85
For these reasons a somewhat different approach will be adopted here, and an attempt will be made to show how a corrosion reaction may be represented by a well-defined reversible electrochemicalcell, although again there are a number of difficulties. Consider the corrosion of metallic zinc in a reducing acid Zn
+ 2H30+ Zn2+(aq.) + H, + H,O +
. . .(1.53)
which occurs spontaneously when zinc is immersed in hydrochloric acid; the dissolution of the zinc is usually quite uniform so that there is no means by which anodic and cathodic sites can be identified physically. The half reactions involved are Zn + Zn2+
+ 2e
(anodic reaction)
2H++ 2e --t H2 (cathodic reaction)
. . .(1.54) . . .(1.55)
and it can be seen although the anodic reaction is the same as that in the Daniel1 cell the cathodic reaction is different. It will be assumed that when the zinc corrodes, randomly dispersed atoms on the surface form the anodic and cathodic sites, and that equations 1.54 and 1.55 can proceed with charge transfer through the zinc. Since the zinc corrodes uniformly the total anodic and cathodic areas must be equal to one another, and this electrochemical reaction, in which the anodic and cathodic sites are inseparable, could be represented by the reversible cell Zn I Znz+I H,O+, H,I Pt
. . .(1.56)
consisting of a reversible Znz+/Znelectrode and a reversible hydrogen electrode (Fig. 1.22). There will of course be a liquid junction (indicated by the line) and a corresponding liquid junction potential, but the latter will be disregarded for the purpose of the present discussion. This cell clearly does not represent what actually occurs during the corrosion of zinc, and an obvious objection that can be raised is that during corrosion the hydrogen evolution reaction (equation 1.55) occurs in a zinc surface and not on one of platinised platinum. Nevertheless, a reversible cell of this type does serve as a convenient starting point. By means of a resistance in the circuit the spontaneous corrosion reaction can be made to proceed at a predetermined rate, and the rate can be measured by means of an ammeter A . At the same time the potentials of the individual electrodes can be measured by means of a suitable reference electrode, a Luggin capillary and high-impedance voltmeters VI and V,. At equilibrium there is no net transfer of charge (I, = I, = 0), and the e.m.f. of the cell is a maximum and equals the difference between the reversible potentials of the two electrodes
. . (1.57) = Er,c - E r , a where Er,&!is the reversible e.m.f. of the cell, and Er,cand Erma are the reversible potential of the cathode and an anode, respectively. The driving force of the reaction is the free energy change A G which is related to the reversible or equilibrium e.m.f. of the cell by the relationship Er,cell
AG = -*E,dI
. . .(1.58)
1:86
CORROSION IN AQUEOUS SOLUTlONS
and, as emphasised in Section 20.2, both AG and E , , , , are thermodynamic quantities that provide a means of evaluating the equilibrium constant K , and hence the activities of the reactants and products when the reaction comes to equilibrium (Table 1.9). If now the resistance in the external circuit is decreased slightly the reaction will proceed at a finite rate, and the electrodes constituting the cell will become mutually polarised and displaced from their equilibrium values, Le. the polarised potential of the anode (Zn2+/Zn) will become more positive, whilst that of the cathode (2H+/H2) will become more negative (Fig. 1.23). The displacement of the potential of an electrode from its reversible value is the overpotentid v, and q = Ep
- E,
. . .(1.59)
where E,, is the polarised potential and E, the reversible or equilibrium potential. Since Ep,c< Er,c(more negative) vc
= Ep,c
.
- Er,c < 0
*
.(1.60)
and the cathode overpotential vc is always negative, although Ep,cmay be positive or negative depending on the sign of E, and the magnitude of vc. (If the potential of a Cu*+/Cu electrode, where E, = 0.34V, is polarised cathodically to 0.32V, then 7, = -0-02OV; if the same procedure is
lu 1
I
!I n!
I
I
I
I
I
I
J
0
I' Current,
Fig. 1.23
I
I
I Irnax
I
E-[curves for the corrosion of zinc (see Fig. 1.22) showing the relationship between E,, E,, and r) for the cathodic and anodic half reactions
1:87
CORROSION IN AQUEOUS SOLUTIONS
adopted with a Zn2+/Zn electrode where E, = -0*76V, Ep will be -0-78V, but qc will still be -0.020V.) Conversely, Epma > Er,a(more positive), and since qa = E p , a
. ..(1.61)
- Er.c > 0
the anode overpotential is always positive. It should be noted that whereas E is always relative to a specified reference electrode this will not apply to the overpotential (see equation 1.59). As the rate of charge transfer is increased by decreasing the resistance Re in the circuit, the magnitudes of qc and qa increase thus decreasing the It follows from magnitude of the polarised e.m.f. 'of the cell Ep,m,,. Fig. 1.23 that for any given rate of charge transfer I
.
= Er,ccll - ( q a + q c + Z R o l n , ) . .(1.62) where qa and qc are the magnitudes of the overpotentials (the negative sign is the elecfor qc must be omitted) corresponding to the rate I , and RWIn. trolytic resistance of the solution. Ep,cell
Since
Ep.ce~= IRe
. .(1.63) which shows that for any given value of Re the rate of the process I increases with
(a) Increase in the magnitude of the reversible e.m.f. of the cell. (b) Decrease in the magnitudes of the anode and cathode potentials. (c) Decrease in the electrolytic resistivity of the solution. Thus, irrespective of Er,ce,l,a thermodynamic parameter, the rate will be controlled by the irreversibility of the reaction, which is reflected in the magnitudes of the anode and cathode overpotentials. If the two electrodes are short-circuited Re 0, and IR, 0, and E,,,, will attain its minimum value. If the conductivity is very high and E,,,,, is small enough to be disregarded it follows from equation 1.62 that +
Er,ce~= q a
+ t7c + ZRsoIn.
. . .(1.64) . .(1.65)
and
Equations 1.62-1.65 apply when the anodes and cathodes are separable so that the rate of transfer of charge can be measured by means of an ammeter in the metallic circuit. If Rsoln. is significant, then EP,, > Ep,a,and E,,,,,, > 0; if RsOln. is very small Ep,c Ep,aand Ep,cel, 0, but q, will not necessarily be equal to qa. It is now appropriate to apply the above considerations of the operation of a well-defined electrochemical cell to the uniform corrosion of a metal in a solution of high conductivity, and under these circumstances both IR, and ZRSoln. may be regarded as negligible. Thus Ep,cell will tend to zero, and Ep,cwill tend to be equal to Ep,a(within 1-2 mV)
-
..
EP,c= Ep,== Ecorr.
. . .(1.66)
1:88
CORROSION IN AQUEOUS SOLUTIONS
where E,,,,. is the corrosion potential, and from equation 1.64 Er.ce11= ~c
. . .(1.67)
+ 'la
The above considerations show that the rate of a corrosion reaction is depenand the kinetic paradent on both the thermodynamic parameter ErvW,, meters qaand qC. It is also apparent that (a)the potential actually measured when corrosion reaction occurs on a metal surface is mixed, compromise or corrosion potential whose magnitude depends on and on the Ep,c-I and Ep,a-Zrelationships, and (b) direct measurement of I, is not possible when the electrodes are inseparable. Overpoientiels
"9
The various types of overpotentials are dealt with in more detail in Section 20.1 but it is appropriate here to outline the significant factors in relation to their importance in controlling the rate of corrosion reaction. Activation overpotential vA For any given electrode process under specified conditions, charge transfer at a finite rate will involve an activation overpotential qA, which provides the activation energy required for the reactant to surmount the energy barrier that exists between the energy states of the reactant and product. Some reactions are kinetically easy (e.g. Ag+(aq.) e + Ag) and thus require only a small activation overpotential, whilst others (e.g. H30++ e -,f H, on metals such as Hg, Pb and Zn) are kinetically difficult and high activation overpotentials are required. Most electrode processes involve more than one step; one of them is usually more sluggish than the others and is thus rate determining, and the activation energy is required, therefore, to maintain the rate of the rate-determining step (r.d.s.), since the other steps may be regarded as being at equilibrium. The activation energy Et is given by
+
. . .(1.68) E' = f l r ) ~ where E f is in joules per mole and z is the number of electrons involved in one act of the rate-determining step. The activation overpotential, and hence the activation energy, varies exponentially with the rate of charge transfer per unit area of electrode surface, as defined by the well-known Tafel equation . . .(1.69) r), = a + blogi where i is the current density, and a and b are the Tafel constants which vary with the nature of the electrode process and with the nature of the solution. Thus qA will be linearly related to log i at overpotentials greater than O.OlOV, and the position and slope of the curve will be dependent on the magnitudes of a and b, which are in turn dependent on the equilibrium exchange current density io, the transfer coefficient a and the number of electrons z involved in one act of the rate-determining step. The Tafel equation for a cathodic process can be expressed (see Section 20. l) in the form v ~=-lni,--lnr, ,RT~ CYZF
RT CYZF
.
.( 1.70)
1:89
CORROSION IN AQUEOUS SOLUTIONS
and since 2.303 RT/F In x = 0.059 log x at 25°C VA.c
0.059 - 0.059 log i, at 25°C = -log io acz
-
. . .(1.71)
acz
where qA.cis the activation overpotential of the cathodic process. Similarly, the activation overpotential of an anodic process is given by qA.a
=
0-059
log ia at 25°C --%Jlog io + 0.059 a&
. . .(1.72)
It is evident from these expressions that since in the Tafel region i (the current density actually determined) must be greater than io (the equilibrium exchange current density), the signs of the overpotentials will conform to equations 1.60 and 1.61, Le. qA,cwill be negative and qA,+will be positive. Furthermore, the smaller the magnitude of io the greater the magnitude of qA and the lower the rate of the electrode process at any given polarised
I I 1 Log ‘0
Log
I
Fig. 1.24 Tafel lines for a single exchange process. The following should be noted: (u)linear E-log i curves are obtained only at overpotentials greater than 0-052 V (at less than 0.052 V E vs. i is linear); (b) the extrapolated anodic and cathodic E-log i curves intersect at io the equilibrium exchangecurrent density; and (c) iaand icthe anodic and cathodic current densities c
+
+
c
actually measured at the differences between i and i , and i and i , respectively
1:90
CORROSION IN AQUEOUS SOLUTIONS
potential E,. Thus the equilibrium exchange current density io is the most significant parameter in controlling the rate of a corrosion process in which one (or both) of the electrode processes involve an appreciable activation energy. Figure 1.24 shows the cathodic and anodic Tafel lines for a single exchange process at an electrode, in which i, and i, are the anodic and cathodic current densities actually measured. Transport (diffusion and concentration) overpotentials qT Previous considerations have been confined to the kinetics of charge transfer but the rate of an electrode reaction will also depend on mass transfer, Le. the rate at which the reactant is transported to the surface of the electrode and the rate at which the product is transported away from the electrode. Transport through the solution to and from the metal surface occurs by diffusion, ionic migration (transport of electrical charge through the solution) and convection, and of these diffusion through the thin static layer of solution adjacent to the metal surface, the diffusion layer 6, is usually of the greatest significance. However, this is not always the case in practical systems, particularly where dissolved oxygen is the cathodic reactant, and in certain circumstances the rate of diffusion through the bulk solution to the metal/solution interface may be rate determining. The limiting current density (the maximum possible rate/unit area under the conditions prevailing) for a cathodic process is given by
. .( 1.73) where i, is the limiting current density (Acm-*), z is the number of electrons required for one step of the electrode process involving 1 mol of the cathode reactant, D is the diffusion coefficient, c is the concentration of the reactant (mol dm-3), 6 is the thickness of the diffusion layer (cm), and n, the transport number of the cation; the term (1 - n + ) can be neglected if ions other than the species involved in the electrode process are responsible for ionic migration. The relation between transport overpotential and current density for a cathodic reaction is given by RT vI,=-ln
ZF
[ [) I--
0.059
=-
z
[
log 1 -
ti
at 25°C
. . .(1.74)
and it is evident that the smaller iL the greater the magnitude of the overpotential due to transport. Unlike activation overpotential, transport overpotential is not controlled by the kinetics of charge transfer, and the magnitude of qT will be the same for any cation (providing z, Di and c are the same) and any metal surface. Thus the rate-controlling parameter in transport overpotential is iL,and it will be seen that any factor in equation 1.73 that causes iLto increase will result in an increase in the corrosion rate, providing the latter is solely determined by the kinetics of the cathodic process. Figure 1.254 shows the relationship between q and log i when the rate is controlled solely by transport, and Fig. 1.25b shows the relationship when both transport and activated charge transfer are involved. It should be noted that whereas in transport overpotential z is the number of electrons involved in one act of the reaction, in activation overpotential z is the number of electrons involved in one act of the rate-determining step.
1:91
CORROSION IN AQUEOUS SOLUTIONS
Log i
Log i
(a)
(6)
Fig. 1.25 1) vs. log i curves for a cathodic reaction ((I) when the rate is solely controlled by transport and (b) when both transport and activated charge transfer are rate determining. (Derivations of the relationships are provided in Section 9.1)
Resistance overpotential q R Since in corrosion the resistance of the metallic path for charge transfer is negligible, resistance overpotential qR is determined by factors associated with the solution or with the metal surface. Thus resistance overpotential may be defined as
.
. .(1.75) = l(Rsoh. + Rf) where &,". is the electrical resistance of the solution, which is dependent on the electrical resistivity (a2 cm) of the solution and the geometry of the corroding system, and R, is the resistance produced by films or coatings formed on or applied to the surface of the sites. Thus, in addition to the resistivity of the solution, any insulating film deposited either at the cathodic or anodic sites that restricts or completely blocks contact between the metal and the solution will increase the resistance overpotential, although the resistivity of the solution is unaffected. This applies particularly to the deposition of CaCO, [and Mg(OH), J at the cathodic sites during corrosion in hard waters due to the increase in pH produced by the cathodic process, and since the anodic and cathodic sites are usually close together the calcareous scale will also block the anodic sites, and thus decrease the corrosion rate. Similar considerations also apply to the dielectric films formed on the metal surface during anodising, and, for example, in the case of the valve metals (Al, Ti, Ta, Nb, etc.) IR drops of hundreds of volts may be produced by the anodic oxide film formed on the metal surfaces. Paint films applied to a metal surface also exert resistance control (see Section 14.3). A11 these types of polarisation will be present to a greater or lesser extent in most corrosion reactions, but if one is more significant than the others it will control the rate of the reaction. This leads to a classification of corrosion reactions according to whether the cathodic or anodic reaction is rate VR
1:92
CORROSlON 1N AQUEOUS SOLUTlONS
determining (cathodic control or anodic control), which can be made even more specific by including such terms as 'activation', 'transport' and 'resistance'. Thus the slow corrosion of zinc in solutions of reducing acids is controlled by the high activation energy required for the hydrogen evolution reaction (cathodic activation control), whereas the rapid corrosion of the metal in concentrated sodium hydroxide is controlled by transport of OHand ZnO; to and away from the metal/solution interface, respectively (anodic transport control).
Graphical Methods of Expressing Corrosion Rates The graphical method of showing how the corrosion rate is dependent on the extent of the polarisation of the anodic and cathodic reactions constituting the corrosion reaction was due originally to Evans who used " J ~
/'
\
iewtw= ( i,d-ia)
.'.
Cathodic reduction
Limiting ditturia current density, iL
10'
102
103 Current density, ,uAcrn-'
Fig. 1.26 E vs. log i curves for the corrosion of a metal in a reducing acid in which there are two exchange processes (c.f. Fig. 1.24) involving oxidation of M 4 M 2 +are reduction of H + 4 H 2 . Note that (a) the reverse reactions for exchange process are negligible at potentials removed from E,, (b) the potential actually measured is the corrosion potential Emm., which is mixed potential, and (c) the E vs. iaPpl, curves (where ism,. is the applied current density) when extrapolated intersect at Ecom.
CORROSION IN AQUEOUS SOLUTIONS
1:93
the co-ordinates E and Z to illustrate how the electrochemical mechanism of corrosion could be applied to a variety of corroding systems. In these ‘Evans’ diagrams, both the cathodic and anodic partial reactions constituting the overall corrosion reaction are presented as linear E-Z curves that converge and intersect at a point, which defines the corrosion potential E,, and the corrosion current Imm..Figure 1.26 shows the E log i curves for the two half-reactions involved in the corrosion of a metal in an acid. Comparison should be made with Fig. 1.24 for a single exchange process, and it should be note that at significantly high overpotentials the reverse reaction for each half-reaction may be neglected. A typical Evans diagrams for the corrosion of a single metal is illustrated in Fig. 1.260 (compare with Fig. 1.23 for two separable electrodes), and it can be seen that the E,-Z and E,-Z curves are drawn as straight lines that intersect at a point that defines E,,,, and Zcom. (it is assumed that the resistance for the solution is negligible). E,,, can of course be determined by means of a reference electrode, but since the anodic and cathodic sites are inseparable direct determination of Z,,,,. by means of an ammeter is not
Current. I (a1
Fig. 1.27 Evans diagrams illustrating (a)cathodic control, (b)anodic control, (c) mixed control, (d)resistance control, (e) how a reaction with a higher thermodynamictendency (E,. =,,) may result in a smaller corrosion rate than one with a lower thermodynamic tendency and (f)how E,,,,. gives no indication of the corrosion rate
1:94
CORROSION IN AQUEOUS SOLUTIONS
possible, and indirect methods must beused (e.g. weight loss and the application of Faraday’s law). and Er,, can be calculated from the stanThe equilibrium potentials Er,, dard electrode potentials of the H+/H2 and M / M z + equilibria taking into account the pH and a,+ ;although the pH may be determined an arbitrary value must be used for the activity of metal ions, and aMr+= 1 is not unreasonable when the metal is corroding actively, since it is the activity in the diffusion layer rather than that in the bulk solution that is significant. From these data it is possible to construct an Evans diagram for the corrosion of a single metal in an acid solution, and a similar approach may be adopted when dissolved O2or another oxidant is the cathode reactant. Figures 1 . 2 7 ~to d show how the Evans diagram can be used to illustrate how the rate may be controlled by either the polarisation of one or both of the partial reactions (cathodic, anodic or mixed control) constituting corrosion reaction, or by the resistivity of the solution or films on the metal surface (resistance control). Figures 1.27e andfillustrate how kinetic factors may be more significant than the thermodynamic tendency (Er,,J and how E,,,,. provides no information on the corrosion rate. The Evans diagram has been used for illustrating various types of corrosion phenomena ranging from the uniform corrosion of a single metal to the enhanced corrosion of one metal when it is coupled to another (bimetallic corrosion), and since the diagram can include only the predominant cathodic and anodic reactions all others are regarded as negligible. Thus if zinc is coupled to iron and the couple is immersed in an oxygenated neutral solution there are at least four possible exchange processes (eight half-reactions), but for the purpose of the Evans diagram only the reduction of dissolved oxygen on the iron surface and the oxidation of Zn + Zn2+need to be considered. This tends to create the erroneous impression that each metal sustains only one electrode reaction, whereas in reality the more anodic metal may support a cathodic reaction, although it is predominantly anodic, and the converse applies to the cathodic metal. In the Evans diagram the curves show the E,-I relationship, whereas it is evident from previous consideration that Ep and are functions of the current density i. In the case of a single metal I, = I,, and since S, = Sa, i, = i,. However, this is not possible when the anodic and cathodic areas are not equal, and Fig. 1.28 shows how bimetallic corrosion of two dissimilar metals can be represented by an Evans-type diagram. It can be seen that although the curves intersect at which must be determined by placing the reference electrode at some distance from the couple, the magnitudes of i, and i, are different; it is also evident that the more dangerous bimetallic situation is when S, is large and Sa is small (see Section 1.7). Over the years the original Evans diagrams have been modified by various workers’’ who have replaced the linear E-I curves by curves that provide a more fundamental representation of the electrode kinetics of the anodic and cathodic processes constituting a corrosion reaction (see Fig. 1.26). This has been possible partly by the application of electrochemical theory and partly by the development of newer experimental techniques. Thus the cathodic curve is plotted so that it shows whether activation-controlled charge transfer (equation 1.70) or mass transfer (equation 1.74) is rate determining. In addition, the potentiostat (see Section 20.2) has provided
1:95
CORROSION IN AQUEOUS SOLUTIONS
I I
I I
I
I I
I
I
I
‘c
I
I
I
Q
iU
‘C
Current density, i
(b)
(0)
Fig. 1.28 Evans diagram illustratinga corrosion process (e.g. a bimetallic couple) in which the area of the cathode is not equal to that of the anode. (a) S, > Sa so that i, < ia and (b)Sa > S, so that ia < i,
a powerful tool for studying the detailed shape of the anodic curve of metals that show an active-passive transition, which has meant that the linear anodic E-I curve used originally has been replaced by the characteristic discontinuous potentiostatic curve. Nevertheless, all these modifications are based on the original concepts of U.R. Evans whose ‘Evans diagrams’ provided a major step forward in our understanding of the electrochemical mechanism of corrosion. In conclusion it is appropriate to mention that whereas in the Evans diagrams both the anodic and cathodic currents are drawn on the same side of the E axis (Le. both positive) many workers (particularly Pourbaix and his co-workers) adopt the approach originally devised by Wagner and Traud2*, in which the cathodic curve is taken as negative and drawn on the left-hand side of the E axis whilst the converse applies to the anodic curve (Fig. 1.29).
E (more positive)
Reduction
t
Oxidation
Reduction
Oxidation
Fig. 1.29 Wagner-Traud method of representing (u) a single reversible reaction and (b)a corrosion reaction (note that E,,,. is the potential when i, = ia)
1:96
CORROSION IN AQUEOUS SOLUTIONS
Cathodic Reactions in Corrosion General Considerations
It follows from the electrochemical mechanism of corrosion that the rates of the anodic and cathodic reactions are interdependent, and that either or both may control the rate of the corrosion reaction. It is also evident from thermodynamic considerations (Tables 1.9 and 1.10) that for a species in solution to act as an electron acceptor its redox potential must be more positive than that of the M ' + / M equilibrium or of any other equilibrium involving an oxidised form of the metal. The hydrogen evolution reaction (h.e.r,) and the oxygen reduction reaction (equations 1.11 and 1.12) are the two most important cathodic processes in the corrosion of metals, and this is due to the fact that hydrogen ions and water molecules are invariably present in aqueous solution, and since most aqueous solutions are in contact with the atmosphere, dissolved oxygen molecules will normally be present. In the complete absence of oxygen, or any other oxidising species, the h.e.r. will be the only cathodic process possible, and if the anodic reaction is only slightly polarised the rate will be determined by the kinetics of the h.e.r. on the particular metal under consideration (cathodic control). However, when dissolved oxygen is present both cathodic reactions will be possible, and the rate of the corrosion reaction will depend upon a variety of factors such as the reversible potential of the metal/metal ion system, the pH of the solution, the concentration of oxygen, the kinetics of the h.e.r. and the oxygen reduction reaction on the metal under consideration, temperature, etc. In general, the contribution made by the h.e.r, will increase in significance with decrease in pH, but this too will depend upon the nature of the metal and metal oxide. Thus metals like zinc and aluminium, whose oxides are amphoteric, are thermodynamically unstable in alkaline solutions (see Fig. 1.17) and will react with water at high pHs with consequent hydrogen evolution and formation of metal anions. In this connection it should be noted that in neutral or alkaline solutions the activity of H,O+ is too low for it to participate in the h.e.r., and under these circumstances the water molecule will act as the electron acceptor . . .(1.76) H,O+ + e -+ 4H2 + H 2 0 (acid solutions) H,O + e + t H 2 + OH- (neutral and alkaline solutions) . . .(1.77) and for the oxygen reduction reaction to, + 2H,O+ 2e 3H20 (acid solutions)
+ 40, + H,O + 2e -,20H-
(neutral and alkaline solutions)
. . .(1.78) . . .(1.79)
It should also be noted that both reactions will result in an increase in pH in the diffusion layer.
The Hydrogen Evolution Reaction (H.E.R.) 1 9 ~ 2 0 Although the h.e.r. involves transport of H 3 0 + ions (or H,O molecules) to the metal surface by diffusion and migration, the activation energy for
CORROSION IN AQUEOUS SOLUTIONS
1:97
charge transfer is usually of the greater significance, and a corrosion reaction in which the h.e.r. is the cathodic process is frequently controlled by the activation overpotential of the latter. If it is assumed that the transfer coefficient 01 = 0.5, and taking z = 1, equation 1.70 becomes qA," = 0-12 log io
. . .(1.80)
- 0- 12 log i,
which is identical with the original Tafel equationz, since io is a constant for a given metal and for given conditions of the solution (see Chapter 21.1, Table 21.12, for values of io). Thus for activation-controlled transfer the significant parameter is the equilibrium exchange current density io,and the smaller io the smaller ic at a given overpotential. For a corrosion reaction in which both the anodic and cathodic reactions are under activation control, then
NO; > Ac > SO:- > ClO;
for aluminium:
NO; > CrO; > Ac > benzoate > SO$
where Ac signifies the acetate anion. The effect of pH appears to be controversial. Some workers find a slight increase in Eb with increase in pH in the acid region, whilst others report that there is practically no change. In the alkaline region, however, Eb becomes significantly more positive with increase in pH owing to the passivating ability of the OH- ion. An increase in temperature significantly decreases Eb, an observation that is frequently neglected with unfortunate consequences. An interesting example was observed recently in which Fe-18Cr-8Ni stainless-steel steamheated pans used for the manufacture of synthetic cream containing a small concentration of sodium chloride were found to pit after 3-4 weeks. The cream was manufactured at 70°C, but the pan was heated with superheated steam, and on removal of the cream by an outlet at the bottom of the pan the residue of cream on the sides of the pan was subjected to temperatures well above 7OoC, with consequent pitting as a result of the small amount of salt present in the cream. The obvious solution to this problem was to use an Fe-18Cr-lONi-3Mo stainless steel, which is more resistant to pitting attack.
Induction Perid for pitting
It is apparent that the critical pitting potential for a given alloy depends on the concentration of chloride ions, on the concentration of inhibiting anions in the solution and on the temperature of the solution. Unfortunately, the situation is complicated further by the fact that there is an induction period for the onset of pitting, which means that the pitting propensity
LOCALISED CORROSION
1:179
of an alloy cannot be assessed precisely on the basis of potentiostatic determination of short duration. The induction time T will decrease with increase in potential and with increase in chloride ion concentration, and in connection with the latter Stolica3' obtained the following relationship: For pure iron
7
= 21 - 4 cc,- min-'
For Fe-5 * 6Cr
7
= 1.54 (cc,- -0*02)min-'
T
= 2 - 3 (ccI- -0*069)min-'
-
For Fe-1 1 6Cr
. . .(1.140) . . .(1.141) . . .(1.142)
Mutually Protective Effect
Pits seldom form in close proximity to one another and it would appear that the area of passivated metal, which acts as the cathode for the local cell, is protected by the anodic dissolution of metal within the pit -a phenomenon that is referred to as the mutually protective eflect (see Section 1.5). Protection Potential
It has been pointed out in Section 1.3 that although the equilibrium potential-pH diagrams are based solely on thermodynamic data it is possible to construct practical potential-pH diagrams from experimentally determined E - i curves. Figure 1.56a shows the potentiokinetic E - i curves for Armco iron in chloride-free solutions of different pHs obtained by Pourbaix3*,in which general corrosion occurs below the passivation potential P and above the region of immunity, and it can be seen that it is possible to construct a practical potential-pH diagram from these curves showing the zones of immunity, general corrosion and passivity that will prevail under various conditions of pH and potential. However, if the same procedure is adopted with solutions of different pH containing mol dm-3 of chloride ion, a sudden increase in current occurs when the potential is raised to a value r, the breakdown potential Eb, at which pits are initiated and give rise to an anodic current (Fig. 1.56b). If after attaining Eb the potential is now lowered, the curve is not retraced (electrochemical hysteresis) and will intersect the i axis (i = 0) at a potential p at which neither anodic oxidation nor cathodic reduction can occur, i.e. pitting is arrested. The potential p is referred to as the protection potential Ep against pitting, since at and below E,, the metal, will not pit and the whole surface will remain passive. EDis always more negative than Eb,and whereas pitting will occur on a pitfree surface above &, it will occur only in the range of potentials between E, and Ebif the surface is already pitted, i.e. between E, and Eb prior pits will continue to propagate, but initiation of new ones will not be possible. Pourbaix, on the basis of the breakdown potential Eb and the protection potential E,, distinguishes between the following states of a metal surface, which have been incorporated in the potential-pH diagram shown in Fig. 1.566: 1. Perfect passivation. The potential-pH region between the passivation potential E,, and the protection potential E,, in which pits are not
1: 180
LOCALISED CORROSION
initiated nor do they propagate if already present owing to passivation. 2. Imperfect passivation. The potential-pH region between E, and Eb, in which pits already present can continue to propagate. 3. Pitting region. The potential-pH region above Eb, at which pits can both initiate and propagate. The electrochemical hysteresis method just described is now becoming widely used to characterise the pitting propensity of alloys, and it has been used by Verink, and Pourbaix3’ to study the behaviour of a range of Fe-Cr Cr), Cu-1ONi and Cu-1ONi-1Fe alloys in solutions of alloys (0-5-24-9Vo different pH and chloride ion concentrations. However, Wilde396has shown that E, is not a unique parameter since it varies in magnitude with the amount of localised attack produced during anode polarisation. This is discussed more fully in Chapter 19 when considering testing for crevice corrosion and pitting.
pH-5
pH=7
pH.9
pH.11
0
2
4
6
8 PH
101214
0
2
4
6
8 PH
1012
pH:13 (a)
pH.7
pH.9
pH.11
pH.13
4
(b)
Fig. 1.56(u) E - i curves and experimental potential-pH diagram for A m c o iron in chloridefree solutions of different pHs (A is the unpolarised potential and P the passivation potential) and (b) E - i curves and experimental potential-pH diagram for Armco iron in solutions of different pHs containing mol dm-3 of chloride ion (r is the rupture potential and p the protection potential). (After Pourbaix38*39)
LOCALISED CORROSION
1: 181
Mechanism
In view of the fact that there are two opposing views on the mechanism of passivity it is not surprising that a similar situation prevails concerning the mechanism of breakdown of passivity. The solid film theory of passivity and breakdown of passivity is dealt with in some detail in Section 1.5, so that it is appropriate here to discuss briefly the views based on the adsorption theory. Uhlig and G i l m a ~ ~ and ’ ~ , Kolotyrkin” explain the breakdown of passivity in terms of competitive adsorption between chloride ions in solution and the adsorbed monolayer of oxygen on the surface of the metal. According to Uhlig20v34 although the metal has a greater affinity for oxygen than chloride ions, adsorption of the latter will be favoured by an increase in the potential until a value is reached where the adsorbed oxygen at specific sites is replaced by chloride ions, which catalyse anodic dissolution. Thus the induction period required for pitting, the decrease in the breakdown potential with increase in chloride ion concentration and the increase when passivating anions are also present in solution, are all explained in terms of competitive adsorption between chloride ions (and the passivating oxyanions) and the adsorbed oxygen on the metal surface. Vermilyea4’has adopted a thermodynamic approach to pitting, and considers that the critical pitting potential is the potential at which the metal salt of the aggressive ion (e.g. AlCl,) is in equilibrium with metal oxide (e.g. A120,). On the basis of this theory the critical pitting potential should decrease by 0.059 V per decade increase in chloride ion concentration. Vermilyea’s theory successfully predicts the values of the critical potentials for AI, Mg, Fe and Ni, but in the case of Zr, Ti and Ta there are large discrepancies. An extensive review of pitting corrosion has been compiled by SzklarskaSmialowski4’, and the reader is recommended to consult this publication for further details. Control of Pitting
Since stagnant conditions favour pitting attack it follows that it will be stimulated by the presence of crevices, deposits and by stagnant volumes of solutions, and these should therefore be avoided. In fact, many of the precautions enumerated to avoid crevice corrosion apply equally to pitting, However, the best procedure is to use an alloy that is resistant to pitting under the environmental conditions prevailing, and reference should be made to the recent paper by Streicher” who has examined the resistance to pitting and crevice corrosion of a range of metals including stainless steels, Inconels, Hastelloys and titanium. Pitting of Carbon Steels
At the beginning of this section a simple explanation was provided of the localised attack that can occur when steel with a discontinuous coating of
1:182
LOCALISED CORROSION 402
402
102
102
102
NeutraL aerated NaCl solution
Fig. 1.57 Electrochemicalreactions that occur when a pit is initiated at sulphide inclusion in a carbon steel (after ran glen^^)
magnetic is exposed to an oxygenated water (Fig. 1.46). However, the situation is far more complex than that described, and is characterised by the formation of a hemispherical membrane of corrosion products (a tubercle, cap or mound) that inhibits diffusion of dissolved oxygen to the metal beneath, thus resulting in an occluded cell, Wranglen4j considers that sulphide inclusions are responsible for the initiation of attack in both carbon steels and stainless steels, and on this basis he has provided a detailed exposition of the pitting of a carbon steel at an inclusion of MnS when the steel is immersed in an oxygenated chloride solution (Fig. 1.57). The reactions of significance are given in the diagram, but certain features of the mechanism are of interest since they illustrate the complexity of the process.
Pit interior Within the pit the primary anodic reaction is Fe Fe2++ 2e which is followed by hydrolysis and the generation of H+ Fe2+ + H 2 0 -+FeOH++ H+ -+
. , .(1.143)
. , .(1.144)
The decrease in pH results in dissolution of some of the MnS MnS
+ 2H+
+
H2S
+ MnZ+
. . .(1.145)
thus providing S2- and HS-that stimulate attack by decreasing the activation overpotential for the dissolution of Fe (and Ni). The electrons released are partly accepted by dissolved oxygen at the surface millscale and partly by the H + , with the consequent formation of H,.gas. The concentration of chloride ions within the pit will increase owing to migration, and this too will stimulate dissolution. Pit mouth A membrane of magnetite (Fe,04) and rust (FeOOH) is formed, which prevents the intermingling of the acid anolyte and alkaline catholyte, by the following steps:
1: 183
LOCALISED CORROSION
Oxidation of FeOH+ and Fez+ by dissolved oxygen occurs
+ io, + 2H+ 2FeZ++ to, + 2H+
2FeOH2+ + HzO
+ HzO
. . .(1.146) . . .(1.147)
+ H,O+Fe(OH): + H+ Fe3++ H 2 0 FeOH2+ + H+
. . .(1.148) . . .(1.149)
2FeOH'
+
+
2Fe3+
followed by hydrolysis of the reaction products FeOH2+
+
and the precipitation of magnetite and rust
+
2FeOHZ+ Fez+ Fe(OH),+
+ 2H20
+ OH-
+
+
Fe,O,
FeOOH
+ 6H+
+H20
. . .(1.150) . . .(1.151)
Outside the pit Reduction of dissolved oxygen 0 2
+ 2H20 + 4e
-+
40H-
. . .(1.152)
and reduction of rust to magnetite 3FeOOH + e+ Fe,O,
+ H,O + OH-
. . .(1.153)
This area will be passivated by the increase in pH due to the cathodically produced OH- ions, and partially cathodically protected by the electrons liberated by the anodic processes within the pit. The tubercle thus results in an occluded cell with the consequent acidification of the anodic sites. Wranglen considers that in view of the fact that crystals of FeC1, -4H,O are sometimes observed at the bottom of a pit the solution within the pit is a saturated solution of that salt, and that this will correspond with an equilibrium pH of about 3 * 5. It is also of interest to note that Wranglen considers that the decrease in the corrosion rate of steel in the atmosphere and the pitting rate in acid and neutral solution brought about by small alloying additions of copper is due to the formation of Cu,S, which reduces the activity of the HS- and S2ions to a very low value so that they do not catalyse anodic dissolution, and a similar mechanism was put forward by Fyfe etal.& to explain the corrosion resistance of copper-containing steels when exposed to industrial atmospheres. pitting of Aluminium
The pitting of aluminium in chloride-containing waters follows a similar mechanism to that of steels (Fig. I .58), and again the characteristic feature of the process is the formation of acid within the occluded cell45.The passivating film of A120, surrounding the pit acts as the cathode, but its effectiveness in reducing dissolved oxygen .is significantly enhanced if copper is either deposited on the surface or enters the lattice of the A1203, and it is well known that the pitting of aluminium occurs rapidly when the water contains a trace of copper ions (see Section 4.1). Similar considerations apply to intermetallic phases such as FeA1, and CuAI,, which can increase the kinetics of oxygen reduction.
1: 184
LOCALISED CORROSION
C
Concentrated aci
Fig. 1.58 Pit on aluminium showing how the rate of pitting may be facilitated by an intermetallic phase (A1,Fe) or by a deposit of copper (after wrangle^^^^)
Pitting of Copper Previous considerations of pitting have been largely confined to metals and alloys that have a strong tendency to passivate, but since the pitting of copper has a number of unusual features it is appropriate to consider it in some detail. Reference to the potential-pH diagram for the Cu-H,O (Section 4.2) system shows that in neutral solutions at the potentials encountered in oxygenated waters the stable form of copper is Cu,O, and the corrosion resistance of copper thus depends upon whether or not the Cu,O forms a protective film. Copper and its alloys in certain fresh waters give rise to a form of localised attack that is referred to as nodular pitting in which the attacked areas are covered by small mounds or nodules composed of corrosion products and of CaC03precipitated from the water. This is a serious problem in view of the extensive use of copper pipes and tanks for water supplies, and in aggressive water these may perforate in a relatively short time. that form on Figure 1.59 shows the type of pit and corrosion copper pipes used for hard (or moderately hard) well waters, and this type of attack is most prevalent when the pipe is used for conveying cold water. The pit interior is almost invariably covered with solid Cu,C1,, and across the mouth of the pit there is a thin membrane of Cu,O containing one or more holes; this membrane is supported on the underside by a more substantial layer of coarsely crystalline Cu,O formed by hydrolysis of Cu,CI,. Above the Cu,O membrane there is a roughly hemispherical mound of CaCO, containing insoluble copper salts, mainly basic carbonate and chloride. According to Campbell4'*" it is possible to distinguish two types of pits and those described above are most prevalent in waters used in the USA,Belgium, Holland and the UK. Another type of pit occurs in certain soft water areas (mainly in Sweden and Germany), but only when the temperature is above 60°C;these pits are of a smaller cross-section than those obtained in hard waters, and contain a very hard crystalline Cu,O
1: 185
LOCALISED CORROSION
Basic clrpru salts and calcium carbonate
Cuprous oxide membrane Crystalline cuprous oxide Cuprous chloride
Fig. 1.59 Pit formed on a copper surface (protected by a film of Cu20) in a hard water (after ~ucey~~)
that may be capped by small black or greenish-black mounds of Cu,O and basic copper sulphate, but often no mound of corrosion product is produced. Subsequent considerations will be confined to the type of pit shown in Fig. 1.59. Campbell49”’during an investigation of pitting of copper tubes in the UK showed that pitting only occurred in deep well supplies of very low organic content, whereas river or lake waters that contain organic impurities did not give rise to pitting. He suggested that these non-aggressive waters contained an organic substance that facilitated the formation of a protective layer of Cu,O, whereas in its absence the Cu,O formed a loose, coarsely crystalline, non-protective deposit. The inhibitor has not been identified, but it has been shown to be a negatively charged colloidal substance with acidic properties, which can be detected by a whitish-blue fluorescence when the water is exposed to ultra-violet radiation; it has been suggested that it is possibly an unsaturated delta lactone. Campbell also showed that a high proportion of copper pipes that failed in the aggressive cold waters contained a carbon deposit that was formed by the breakdown of the drawing lubricant residue during bright annealing. A glassy thin film of Cu,O Can also form during bright annealing, and initially it was considered that both the carbon and the glassy Cu,O could act as a cathode, and thus stimulate attack at discontinuities in the film and lead to pitting. Devroey and Depommier5’in Belgium, and Campbell have found, however, in further investigations that glassy cuprous oxide scales formed in tubes during annealing do not cause pitting corrosion but become converted in service to a dull protective oxide. The glassy cuprous oxide found in tubes that had failed in service were shown to have been formed by the action of the water and to be associated with the presence of very thin carbon films produced in the tubes during manufacture. The compositions of waters that give rise to pitting have been the subject of numerous investigations, and studies by Obrecht et a/.31of the pitting of copper tubes in the USA have shown that it is invariably a cold water phenomenon that occurs with hard well waters. These waters contained over 5 p.p.m. of dissolved C 0 2 . 10-40 p.p.m. being typical, a pH in the range
1: 186
LOCALISED CORROSION
7.0-7-8, 10-12p.p.m. of dissolved 0,and an SO:-/Cl- ratio of 3-4:l. Lucey5,has analysed about 120 waters in the UK, whose behaviour in relation to the pitting of copper is well established, and has constructed a nomogram that provides a means of predicting pitting propensity from an analysis of the water. It would appear that an increase in the SO:- or Na’ ion concentration or an increase in the concentration of dissolved oxygen increases the pitting propensity, whereas the converse applies to the C1- or NO; ion concentration and pH. Thus whereas the pitting of stainless steels is favoured by a high C1-: SO:- ratio (see Fig. 1.55) the pitting of copper is favoured by a high SO:-: C1- ratio. A high pH reduces the pitting propensity, and many surface waters contain ‘humic acid’ that enables more calcium bicarbonate to be held in solution than the equilibrium value, thus resulting in a high and stable pH. Mechanism
The mechanism of pitting is highly complex, and reference should be made to the original papers for further details. However, it is of interest to consider certain views on this subject, since some of them introduce new concepts. May” was the first to stress the important r61e of Cu,Cl, within the pits on the mechanism, and he considered that it acted as a screen that prevented dissolved oxygen gaining access to the bottom of the pit thus preventing the formation of a protective Cu,O film; the low solubility of Cu,Cl, also maintained the activity of copper ions at a low value and thus facilitated anodic dissolution of the copper. Pourbaix 12,55 has constructed equilibrium potential-pH diagrams for the Cu-Cl-H, 0 system and for the Cu-C1-C0,-SO, -H,O system using concentrations of ions that correspond to those actually present in Brussels water. This work has shown that at the bottom of a pit the phases Cu, Cu,O and Cu,C1, are at equilibrium in a solution containing 246 p.p.m. of C1- and 270 p.p.m. of Cu2+only at a potential of 270 mV (vs. S.H.E.) and pH3-5. The potential of the metal outside the pit in a water of pH 7-8 depends on the concentration of oxygen, the overpotential for its reduction on the passive surface film, etc. but it is normally about 300mV, which is only 30-70mV higher than the potential within the pits. Under these circumstances the pits do not grow appreciably. However, if a carbon film is present on the surface of the copper, a differential aeration cell will be set up, and the potential of the surface will increase this in turn will increase the potential of the surface of the interior of the pit above the equilibrium value of 270 mV. Under these circumstances the equilibrium will be disturbed and the copper within the pit will corrode rapidly forming Cu2+.This mechanism thus involves many of the conventional features of pitting, Le. differential aeration, a large cathode: anode surface area ratio and the development of acidity within the pit by hydrolysis of Cu,C12 that prevents a protective film of Cu,O from forming. LuceyMexamined a number of examples of pitting of copper pipes and tanks from hard water districts, and found that there was no more calcium carbonate scale deposited around the pits than on other parts of the metal surface. There was, however, a large amount of CaCO, in the mound
1: 187
LOCALISED CORROSION
immediately above the pit, and this suggested that the reduction of dissolved oxygen takes place immediately above the pit and not on the surrounding surface. He also showed that the Cu,O membrane could act as thin bipolar electrode, the upper surface acting as a cathode and the under surface as an anode. Thus the Cu,Cl, produced within the pit is anodically oxidised to Cu2+,and this ion can then attack the copper within the pits by the disproportionation reaction c u 2 +c u
+
2cu+
. . .(1.154)
The principal cathodic reaction on the upper surface of the membrane is the reduction of Cu2+ that is formed by the reaction of Cu+ with dissolved oxygen in the water; these Cu+ ions are provided partly from the diffusion through the pores in the oxide membrane from within the pit and partly from those produced by cathodic reduction (equation 1.154). Lucey’s theory thus rejects the conventional large cathode:small anode relationship that is invoked to explain localised attack, and this concept of an electronicallyconducting membrane has also been used by Evanss6to explain localised attack on steel due to a discontinuous film of magnetite.
Selective Leaching or De-alloying In certain alloys and under certain environmental conditions selective removal of one metal (the most electrochemically active) can occur resulting in either localised attack, with the consequent possibilityof perforation (plug type), or in a more uniform attack (layer type) that results in a weakening of the strength of the component. Although the selective removal of metals such as Al, Fe, Co, Ni and Cr from their alloys is known, the most prevalent form of de-alloying is the selective removal of zinc from the brasses-a phenomenon that is known as dezinciJcation. Plan
Corrosion Droducts Seclnon (magnified)
Section
(d1
Fig. 1.60 Dezincification and impingement attack of copper-alloy tubes. (0) Uniform layer dezincificationof a brass, (b)banded dezincification of a brass, (c) plug-typedezincificationand ( d ) impingement attack
1:188
LOCALISED CORROSION
In this connection it is of interest to refer to the parting of Ag-Au alloys in nitric acid in which the silver can be completely removed from the alloy providing the ratio of Ag:Au is greater than 2-5:l; at somewhat lower ratios separation is incomplete and at low ratios no separation is possible and the alloy is unattacked by the acid. Similarly, dezincification does not occur when the zinc in a brass is less than 15%, and red brass (15% Zn) has only a slight tendency to dezincify, whereas the a-brasses (Cu-3OZn) are highly susceptible and the @-brasses(Cu-40Zn) even more susceptible; low alloying additions of A1 and Mn in brasses appear to have very little effect in reducing dezincification. Dezincification is readily apparent, since the yellow colour of the brass is replaced by the characteristic red of copper, which may take the form of small plugs or of layers that in some cases can extend over the whole of the surface (Fig. 1.60). In plug-type dezincification a mechanically weak, porous residue of copper is produced, which may remain in situ or become removed by the pressure of water, leading to a perforation. In the layer type the transformation of the alloy into a mechanically weak layer of copper results in loss of strength, and failure may occur by splitting when the metal is subjected to water pressure or to external stress. Dezincification can occur over a wide range of pH, although the pH appears to affect the form of attack. Thus the layer type is favoured when the environment is acid and the brass has a high zinc content, whilst the plug type is more prevalent when the environment is neutral, slightly acid or alkaline and the zinc content is relatively low. In both cases dezincification is favoured by stagnant conditions (compare erosion-corrosion that becomes prevalent at high velocities), by the presence of chloride ions and by the formation of porous scales and deposits that lead to stagnant crevice conditions. Dezincification of a-brass can be readily prevented by suitable alloying additions, and this was achieved first by adding 1% Sn. However, elements such as As, Sb and P are more effective, and alloying additions of 0~02-0-06% As are widely used for this purpose. Unfortunately, no alloying element has been found that prevents the dezincification of the two-phase ap-brasses, which are more susceptible than the a-brasses, and their use must be avoided under environmental conditions that are conducive to dezincification. Mechanism
Two theories have been proposed to explain dezincification, but since both have considerable support the precise mechanism remains unresolved. One theory proposes that the zinc is selectively leached from the alloy leaving a porous residue of metallic copper in situ (cf.parting of Ag-Au alloys), whilst the other proposes that the whole of brass dissolves and that the copper immediately redeposits at sites close to where the brass was dissolved. There is considerable metallographic and electrochemical evidence in support of each theory and it is of interest to note that two of the most authoritative works on corrosion appear to support opposite views Uhlig” favours the selective dissolution of zinc theory, whereas Fontana and Greene22favour the dissolution-precipitation theory.
1: 189
LOCALISED CORROSION
Metallographic studies by Polushkin and Shuldener57 revealed that the copper residue in a dezincified a-brass contained twins and residual grain boundaries that resembled those present in the parent metal, which appears to support the dissolution of zinc theory. However, other workers claim that twins found in the copper residue are similar to those found in electrolytically deposited copper. H ~ r t o nused ~ ~ time-lapse colour photomicrography to observe the changes that took place during the dezincification of an a-brass in 3% NaCl at 50°C. The brass was cold worked and annealed to produce a coarse-grained structure with a number of annealing twins, but none of these features was reproduced in the copper, indicating that the whole alloy dissolved. Any selective dissolution mechanism must explain how the zinc within the alloy diffuses to the surface at which the reaction takes place. Pickering and Wagner 59 used X-ray and electron diffraction to study the selective dissolution of copper from Cu-Au alloys, and subsequently a similar approach was used by Pickering@' to study the selective dissolution of zinc from Eand y-brasses (zinc-rich alloys) during anodic polarisation in a variety of electrolyte solutions. Pickering found that the partially dissolved alloys gave rise to new intermediate and terminal phases having a higher copper concentration than the original alloys. It was concluded that volume diffusion of zinc could occur via the vacancies (mono- and di-) created at the surface by anodic dissolution; the diffusion coefficient for di-vacancies in copper at 25°C was calculated to be 1 . 3 x 10-'*cm2 s-'. Lucey6' concludes from his electrochemical studies that dezincification involves anodic dissolution of both copper and zinc followed by the cathodic deposition of copper, and on this basis he has explained why arsenic is capable of inhibiting dezincification of a-brass but not of &brass. When dezincification occurs in service the brass dissolves anodically and this reaction is electrochemically balanced by the reduction of dissolved oxygen present in the water at the surface of the brass. Both the copper and zinc constituents of the brass dissolve, but the copper is not stable in solution at the potential of dezincifying brass and is rapidly reduced back to metallic copper. Once the attack becomes established, therefore, two cathodic sites exist-the first at the surface of the metal, at which dissolved oxygen is reduced, and a second situated close to the advancing front of the anodic attack where the copper ions produced during the anodic reaction are reduced to form the porous mass of copper which is characteristic of dezincification. The second cathodic reaction can only be sufficient to balance electrochemically the anodic dissolution of the copper of the brass, and without the support of the reduction of oxygen on the outer face (which balances dissolution of the zinc) the attack cannot continue. The potentials of film-free a-brass and &brass in solutions comparable to those existing inside the alloy at the advancing front of attack were found to be -0-38V and - 0 . 5 6 V (vs. S.H.E.), respectively. It was also established, taking into account the activities of copper ions in equilibrium with the sparingly soluble corrosion product Cu,Cl,, that whereas Cu2+ ions can be reduced to copper at -0- 16 V the reduction of Cu ions is possible only at potentials more negative than -0.41 V. Thus whereas the P-phase of an a@-brasscan reduce both Cu2+and Cu ions, the a-brass can reduce only the Cu2+ion. Lucey points out that although arsenic can prevent dezincification of an +
+
1:190
LOCALISED CORROSION
a-brass it cannot prevent pitting, during which CuzC1, is formed and is subsequently hydrolysed or oxidised to the secondary corrosion product Cu,O. Thus during dezincification the advancing front of attack results in the formation of Cu,Cl,, and the Cu' ions can be readily reduced to metallic copper by the very negative /3-phase, but not by the cy-phase. The fact that cy-brasses dezincify is explained by the formation of Cu2+by the disproportionation of Cu+ ions (formed from Cu,CI,) 2cu+
* cu2++ c u
. . .(1.155)
which are readily reduced directly to copper by the a-phase. Under these circumstances dezincification will proceed as long as Cu2+ions are present at the advancing front. If, however, metallic arsenic is deposited at the advancing front in preference to copper metal, the Cu2+ions are reduced to Cu+ according to the following cycle:
in which arsenic metal is regenerated. Thus dezincification of a-brass is prevented by the presence of a very small quantity of arsenic, and slow hydrolysis of Cu,Cl, to Cu,O takes place.
Erosion-Corrosion The effect of movement of the solution or of the metal on the rate and form of corrosion is complex, and on the basis of previous considerations (see also Sections 1.4 and 2.1) the situation can be summarised as follows: 1. Increase in velocity may increase the rate by bringing the cathode reactant more rapidly to the surface of the metal thus decreasing cathodic polarisation, and by removing metal ions thus decreasing anodic polarisation. 2. Increase in velocity may decrease the rate by bringing the cathode reactant to the surface at a rate that exceeds imit., thus causing passivation. 3. Decrease in velocity will favour all forms of localised attack in which an occluded cell is involved in the mechanism, and will also favour selectivedissolution of alloys that are susceptibleto this form of attack.
However, movement at appreciable rates can result in another form of attack that is brought about by the conjoint action of erosion and corrosion; hence the term erosion-corrosion that includes all forms of accelerated attack in which protective films, and even the metal surface itself, are removed by the abrasive action of movement of a fluid (gas or liquid) at high velocity. In general, the higher the velocity the more abrasive the solution. Erosion-corrosion in the widest sense of the term will include impingement attack, cavitation damage and fretting corrosion, but since the latter two are dealt with in separate sections (see Sections 8.7 and 8.8) they will not be considered here. The most significant effect of erosion-corrosion is the constant removal
1: 191
LOCALISED CORROSION
of protective films (which may range from thick visible films of corrosion products to the thin invisible passivating films) from the metal's surface, thus resulting in localised attack at the areas at which the film is removed. This can be caused by movement at high velocities, and will be particularly prone to occur if the solution contains solid particles (e.g. insoluble salts, sand and silt) that have an abrasive action. Impingement attack is a form of erosioncorrosion in which a turbulent stream of water containing entangled air bubbles and solid particles hits a metal surface, disrupts the protective film and thus results in pitting. In addition to the mechanical damage of the protective film, velocity or movement will also bring the cathode reactant more rapidIy to the metal surface thus decreasing cathode polarization. The ability of a metal or alloy to withstand erosion-corrosion depends upon the nature of the environment which can range from a natural water to a concentrated acid, so that it is difficult to make sweeping generalisations and each system must be considered individually. In most systems the rate of attack increases with velocity, and metals and alloys that have an acceptable corrosion rate in static solutions may corrode rapidly when the solution (or the metal) is moving at a high velocity. Table 1.22**shows that up to about 120 cm s-' the effect is small, but at about 820 cm s-' there is a rapid increase in the corrosion rate. However, there are exceptions to this rule, and Fontana and GreeneU point out that whereas the corrosion rate of aluminium in fuming nitric acid at 42OC increases with velocity, the converse applies to type 347 stainless steel owing to the different corrosion mechanisms involved. Thus aluminium is protected by films of aluminium nitrate and aluminium oxide and since the former is removed at intermediate velocities (up to 100 cm s-') and the latter at higher velocities, the corrosion Table 1.22 Corrosion of metals by sea-watermoving at different velocities (grn-*d-'
X
IO2)*
Velocity
Metd 30.5 cm s-' (1 ft s-')t
Carbon steel Cast iron Silicon bronze Admiralty brass Hydraulic bronze G bronze Aluminium bronze (10% Al) Aluminium brass Cu-1 ONi-0- 8Fe Cu-3ONi-0.05 Fe Cu-3ONi-0-5Fe Monel Stainless steel (316) Hastelloy C Titanium
3.4 4.5 0.1 0.2 0.4 0.7
122 cm s-l (4 ft s-'H 7.2
-
0.2 2.0 0.I 0.2
823 cm s-' (27 ft s-')$ 25.4 27.0 34.3 17.0
33.9
28.0
0.5
23.6
0.2 0.5 0.2 c0.1 co.1
10.5
0.1 C0.1 0
99.99070 A1 aluminium with the magnesium alloys AZ31B (3% Al, 1070 Zn, 0.4% Mn) and AZ3IA (3% Al, 1% Zn, 0.4% Mn, 0.15% Ca) in sodium chloride and sea-water. Here the higher electronic resistance of the natural oxide film on super-purity aluminium appears to limit the current flow and the local rise of pH at the cathode, thereby preventing normal aggressive cathodic corrosion. Akimovs0was one of the earliest investigators to report that the pitting of aluminium and aluminium alloys in sea-water could be prevented by coupling to zinc. He also drew attention to the occurrence of a tenacious
BIMETALLIC CORROSION
1 :233
black film on the aluminium. Since then additional observations have been made on the beneficial effects of coupling zinc to various aluminium alloys in chloride solutions. For instance, zinc will often stop the stress-corrosion cracking of susceptible AI-Mg alIoys. The protective action of zinc is, however, somewhat unreliable and strongly dependent on the method of surface preparation of the aluminium alloy*'. Experiments carried out by Keir, van Rooyen and Pryor have clarified the variable behaviour of the Al-Zn couple in chloride solutions. Using highpurity aluminium and zinc electrodes of equal size coupled together in sodium chloride solution, it was found that zinc is initially anodic to aluminium but that within one day the polarity of the couple reverses and remains as such subsequently (Fig. 1.70). This reversal in polarity appears to be due to the accumulation of Zn2+in solution. Accordingly, with decrease in distance between the electrodes, and in solution-volume: electrode-area ratio, the polarity reversal occurs much more rapidly. The accumulation of Zn2+ in solution depresses the potential of the aluminium from an initial value of about -0.5V to a final open-circuit value of about -1.OV (vs. S.H.E.). The corrosion rates of both the aluminium and the zinc electrodesare greater than in the absence of bimetallic contact, but the corrosion of the aluminium is changed from the characteristic pitting, usually observed in nearly neutral chlorides, to a desirable mild uniform attack. The polarity reversal is not
Fig. 1.70 Polarity reversal of the AI-Zn couple in 1 a 0 N sodium chloride at 25°C. Curve a aluminium and zinc electrodes 150 mm apart; 16 ml of solution per square ccntimetre of electrode; curve b aluminium and zinc electrodes 20 mm apart, 10 ml of solution per square centimetre of electrode
1 :234
BIMETALLIC CORROSION
detectable by potential measurements alone, since the potential of the couple remains constant, to within a few millivolts at a value of around -0.83 V. When the aluminium is cathodic, polarisation curves indicate that the galvanic corrosion is under cathodic control; after the polarity reversal, the much weaker galvanic corrosion appears to be under anodic control. which It remains to be determined whether the previous experimentsa0*81, have been interpreted as confirmingthe cathodic protection of aluminium by zinc, can be truly interpreted in this fashion or whether they are due to the accumulation of Znz+in the electrolyte. Under laboratory conditions, and under some practical conditions in stagnant solutions or in recirculating systems, the latter explanation is quite likely. Two dissimilar metals, such as iron and aluminium, may cause aggravated corrosion effects even if they are not in electrical contact. This subject is, however, outside the scope of this section, and has been treated in detail e l ~ e w h e r e * Heavy ~ * ~ ~ .metal ions, such as copper ions, are particularly liable to produce galvanic effects by redeposition on a less noble metal; the phenomenon is discussed in Sections 4.1, 4.2 and 9.3.
Protective Measures Under Conditions of Total Immersion Protective measures against bimetallic corrosion should ideally start before the particular installation or equipment is built lS. Reference should be made to tables showing compatibility of metals, alloys and non-metallic materials (see Table 1.25) and to the literature. However, it must be emphasised that the environment obviously plays a most important r61e in bimetallic corrosion, and that there are a number of situations in which apparently incompatible materials in contact can be used without adverse effects. Assuming that some incompatible materials of construction must be used, much can be done in the way of initial design to minimise future problems. Under conditions of total immersion in high-conductivity electrolytes containing dissolved oxygen, the catchment area principle is fundamental in minimising galvanic corrosion problems. Reduction of the area of the more noble metal together with use of the maximum area of the less noble metal gives a combination of small galvanic current and minimum intensity of attack on the less noble metal. This approach becomes less effective as the conductivity of the electrolyte decreases. In simple equipment, galvanic corrosion can be eliminated by complete electrical insulation. Bushings, washers and pipe fittings of nylon or Teflon have become quite popular for this purpose despite their obvious limitations in mechanical properties (see Section 9.5). Particular examples of the use of this type of fitting include insulating domestic aluminium or galvanised-iron hot-water heaters and tanks from black iron or copper plumbing. Electrical insulation often tends to be ineffective in complicated equipment on account of the numerous other electronically conducting paths that may exista4.It is always well to check supposedly electrically isolated metal components with an ohmmeter in order to confirm that the desired electrical isolation has, in fact, been achieved.
BIMETALLIC CORROSION
1 :235
Increasing the electrolytic resistance of the solution path is a possible method of reducing the galvanic corrosion rate. Little significant practical benefit accompanies this approach in high-conductivity electrolytes since severe galvanic corrosion can exist at locations which are several metres distant from the actual bimetallic joints4. This approach, accordingly, has merit mainly in low-conductivity electrolytes such as certain supply waters, and in the case of atmospheric galvanic corrosion. Deaeration has occasionally been used as a means of controlling bimetallic corrosion under conditions of total immersion, and this method of control can be used successfully, if physical conditions permit, provided that the less noble metal is not sufficientlyelectrochemicallyactive to permit rapid evolution of hydrogen at the more noble metal, as is observed, for instance, in many bimetallic couples involving magnesium anodes. Metallic coatings have been widely and successfully used as a means of alleviating many bimetallic corrosion problems both under conditions of total immersion and in corrosive atmospheres. If, for instance, aluminium and steel must be jointed together in sea-water, the galvanic corrosion can be largely eliminated by aluminising the steel either by hot dipping or by flame spraying, as is more popular in Europe. Both zinc and cadmium are also fairly compatible with aluminium and so the steel may be protected with thin coatings of these metals without incurring the risk of aggravated galvanic corrosion; cadmium plating has even been applied to stainless steel for this purpose. The use of dissimilar metallic coatings eliminates bimetallic corrosion only if the coating is initially free from voids and remains so in service, a circumstance seldom realised in practice. Metallic coatings on the steel that contain or develop voids still reduce the galvanic corrosion rate (because of the smaller area of steel exposed) provided that they are anodic to the substrate. Cathodic protection with a sacrificial anode that is less noble than either member of the couple is frequently used to reduce the severity of bimetallic corrosion, particularly that resulting from the use of bronze propellers in steel ship hulls. Paint coatings also receive extensive practical use for protecting against galvanic corrosion in atmospheres and under conditions of total immersion. The best practice, where feasible, calls for complete painting of both members of the bimetallic couple. If only one member of the couple can be painted, the cathodic metal should receive this treatment; since paint coatings are seldom free from holidays, painting the cathode will reduce the total cathodic area and hence the galvanic corrosion rate. Painting the anodic metal alone represents bad practice under conditions of total immersion in a high-conductivity electrolyte, because the original cathode area is undiminished, and corrosion will then take place at holidays or damaged areas in the coating on the anodic metal at a high intensitys5.For optimum protection against galvanic corrosion, repainting should be carried out on a regular schedule since the protection afforded by most paints can be rather limited in duration. The use of soluble inhibitors as a means of controlling bimetallic corrosion presents many technical problems. Apart from the fact that this method is limited in applicability to recirculating systems, efficient anodic inhibitors, such as chromates, are frequently quite specific in their action and so certain bimetallic couples, such as the AI-Cu couple in chloride solutionsa, are
1 :236
BIMETALLIC CORROSION
extremely difficult to control by a single anodic inhibitor. Accordingly, other methods of treating bimetallic corrosion, as already described, are often preferred, with inhibition being relegated to special applications such as automotive cooling systems where these previous methods are either not feasible or economical. Inhibition of automotive cooling systems is generally achieved by highly complex mixtures of inhibitors, often involving combinations of borax, nitrates and organic adsorption inhibitors, and even then complete success is not always attained (see Section 17.2). Various inhibitors were compared in tests by Brunoro et a1.86. Advice on the reduction of bimetallic corrosion at welded and brazed joints can be found in Reference'". The use of replaceable wastage pieces to take up the bimetallic corrosion in various systems is proposed in References84and85. In Corrosive Atmospheres
Bimetallic corrosion in atmospheres is confined to the area of the less noble metal in the vicinity of the bimetallic joint, owing to the high electrolytic resistance of the condensed electrolyte film. Electrolytic resistance considerations limit the effective anodic and cathodic areas to approximately equal size and therefore prevent alleviation of atmospheric galvanic corrosion through strict application of the catchment area principle. With this exception, many of the methods already described for protecting against bimetallic corrosion under conditions of total immersion may be similarly used for preventing atmospheric galvanic corrosion. These include selection of compatible metals, metallic coatings and painting. It is, however, more common to use the principle of increasing the resistance of the solution path for preventing galvanic corrosion. Since the solution-path resistance is already high, the additional means that are required to increase resistance further are simple and generally inexpensive. In many cases, taping the immediate joint area with mastic tapes, with or without chromate impregnation, will suffice, provided that the whole of the bimetallic contact is covered to a distance of about 25 mm from the junction and on either side. Vulcanising a rubber or Neoprene ball around small joints has also been used very satisfactorily. While these methods protect a joint against atmospheric galvanic corrosion, it can hardly be overemphasisedthat they are not applicable to protecting against bimetallic corrosion where the same joint is totally immersed in a high-conductivity electrolyte.
Some Beneficial Effects of Galvanic Coupling So far this section has been primarily concerned with the harmful aspects of bimetallic corrosion, in which the less noble member of the couple is subjected to attack of unusual severity. It is, however, implicit that bimetallic corrosion can be beneficial in that it will usually reduce or prevent corrosion of the more noble metal. Refer to Sections 11.2 and 11.4 for further details. Another very beneficial aspect of bimetallic corrosion is power generation from chemical cells, but this subject is outside the scope of this section.
BIMETALLIC CORROSION
1:237
The principles of bimetallic corrosion have, in addition, been used in an elegant fashion for the development of highly corrosion-resistant alloys. Draley and Ruther" observed that commercial-purity aluminium (1 100 alloy) was subject to catastrophic intergranular corrosion in distilled water above 200°C with the corrosion rate increasing very rapidly with temperature. In most cases enhanced attack occurred at grain boundaries and around second-phase stringers. Draley and Ruther showed that the rapid intergranular disintegration of the aluminium was associated with the entry of hydrogen into the metal from cathodic sites. They proposed that, if alternate cathodic sites of lower hydrogen overpotential could be provided, the hydrogen would have a much better chance of being evolved as harmless bubbles instead of entering the aluminium and causing intergranular disintegration. This hypothesis was confirmed by adding 5 p.p.m. of Ni2+to the distilled water. The small amounts of nickel plating-out on the aluminium surface were sufficient to protect against catastrophic corrosion of 1100 alloy at 275"C, although the 1100 was subject to slow and uniform attack. Subsequently, 1100 alloy specimens were electroless nickel-plated and found to resist catastrophic corrosion for 80 days at 315°C. Finally Draley and Ruther alloyed small amounts of nickel (0.5% or more) with commercially pure aluminium and obtained consistent protection against catastrophic corrosion at temperatures of up to 350°C. The nickel, being largely insoluble in aluminium, exists primarily as NiAl, constituent which evidently possesses a low hydrogen overpotential and protects against catastrophic corrosion. Its action is augmented by the simultaneous presence of iron in the alloy, the improved corrosion resistance probably being due to the presence of an Al-Ni-Fe constituents9. A commercial aluminium alloy designated as 8001 (1@lo Ni, 0 -6% Fe) now exists for high-temperature water service. It has long been known that alloys such as austenitic stainless steel and metals such as titanium, while exhibiting passive behaviour in mildly or strongly oxidising solutions, often suffer active corrosion at a high rate in were among the first to point out the reducing acids. Tomashov et aLWs9' possibility of improving the corrosion resistance of stainless steel, chromium and titanium by increasing the stability of the passive state with small alloying additions of noble metals such as silver, palladium and platinum. This work was extended for titanium in reducing-type acids by Stern and Wissenberg9*who primarily investigated the effect of platinum and palladium. The principle by which noble alloying additions are effective in improving the corrosion resistance of titanium is illustrated in Fig. 1.71 which is taken from the work of Stern and Wissenberg. In a reducing acid where hydrogen is evolved from a titanium surface, the exchange current is relatively small. It may be increased and the cathodic Tafel slope decreased by providing local noble-metal cathodes. The intercept of the cathodic and the anodic polarisation curves is shifted in the more noble direction and, if the shift is large enough to raise the mixed potential into the passive potential region in Fig. 1.71, essentially passive behaviour in reducing acids such as boiling HCI can result (see Section 5.4). Similar improvements in the corrosion resistance of chromium in sulphuric and hydrochloric acids have been found by Greene, Bishop and Stern93to accompany alloying with small amounts of rhodium, palladium or osmium. These noble-metal alloying
1 :238
BIMETALLIC CORROSION
Current
Fig. 1.71 How alloying with a noble metal produces a passive mixed potential and a marked reduction in corrosion rate (after Stern and W i ~ s e n b e r g ~ ~ )
additions have the further advantage of not significantly reducing the normally excellent corrosion resistance of chromium in oxidising acids such as nitric acid, whereas elements such as platinum, iridium and ruthenium confer excellent corrosion resistance on chromium in non-oxidising acids, but increase its corrosion rate in oxidising acids. It is evident from the foregoing that the principles of bimetallic corrosion are being applied in a progressively more widespread and successful fashion to the development of alloys of maximum corrosion resistance.
Distribution of Bimetallic Corrosion in Real Systems The influence of electrolyte conductivity on the distribution of bimetallic corrosion has already been described in qualitative terms earlier in the section (see Fig. 1.69). For a bimetal couple of small size, one could expect an approximately even distribution of corrosion in an electrolyte of high conductivity, such as sea-water. In real systems, such as heat-exchangers,% steam condensers, pumps, pipework and off-shore rigs, the almost inevitable presence of a mix of different metals leads to the development of galvanic corrosion which is unevenly distributed, even in electrolytes of high conductivity, because of large system dimensions. Furthermore, the effective cathode to anode area ratio in real mixed-metal systems will differ from the geometrical area ratio. Indeed, there could well be doubt as to which metals will be cathodic and which anodic in complex systems. Thus laboratoryderived data relating to simple bimetal couples made up from small electrode samples cannot be used to give an accurate indication of even the maximum corrosion rates of the more negative metals in a real systemg. Maximum
BIMETALLIC CORROSION
1:239
corrosion rates of the more negative metals in a real systemg. Maximum corrosion rates calculated on the basis of geometrical area ratio can be typically two or three times too low. The quantification of the probable extent and magnitude of bimetallic corrosion for a new system at the design stage, for it is at this stage that remedial actions, such as the provision of wastage pieces of thicker material, is difficult to achieve by means of a corrosion can most readily be made34*35, evaluation using a-scaled-down model of the full-scale prototype design. Such a procedure is time consuming and also presents grave problems associated with necessary scaling of the conductivity of the test electrolyte by dilution which can itself have an effect upon the anodic and cathodic react i o n ~ An ~ ~alternative . to setting up scaled-down tests is to mathematically model the corrosive processes for the full-sized system in order to predict the distribution of bimetallic corrosion, using cathodic and anodic polarisation curves relating to fresh or filmed metal, as desired. Such modelling also, of course, enables the prediction of the extent of cathodic protection on the more positive metals in the system. As the corrosion rate, inclusive of local-cell corrosion, of a metal is related to electrode potential, usually by means of the Tafel equation and, of course, Faraday’s second law of electrolysis, a necessary precursor to corrosion rate calculation is the assessment of electrode potential distribution on each metal in a system. In the absence of significant concentration variations in the electrolyte,%a condition certainly satisfied in most practical sea-water systems, the exact prediction of electrode potential distribution at a given time involves the solution of the Laplace equation for the electrostatic potential (P)in the electrolyte at the position given by the three spatial coordinates (x, y, 2). a2P a2P a2P +-+-=o a x 2 a y 2 az2 The solution of the Laplace equation is not trivial even for relatively simple geometries and analytical solutions are usually not possible. Series solutions have been obtained for simple geometries assuming linear polarisation kinetics97-101. More complex electrode kinetics and/or geometries have been dealt with by various numerical methods of solution such as finite difference10z,103, finite element 104*10s and bdundary element.41* The numerical approaches to the solution of the Laplace equation usually demand access to minicomputers with fast processing capabilities. Numerical methods of this sort are essential when the electrolyte is unconfined, as for an off-shore rig or a submarine hull. However, where the electrolyte is confined, as within essentially cylindrical equipment such as pipework and heat-exchangers, or for restricted electrolyte depths, a simpler modelling procedure may be adopted in the case of electrolytes of good conductivity, such as ~ e a - w a t e r ~ ~This ’ ~ ’ . simpler procedure enables computation to be carried out on small, desk-top microcomputers. For electrolytesof low resistivity, it can be shown that the electrode potential distribution within cylindrical equipment is often very closely approximated to by neglecting the radial potential variation Le. by assuming current flows only axially34935~1m.108. Astley” has demonstrated that sea-water systems with diameters of up to at least 500mm can be examined making a ‘unidirectional current flow’ assumption.
1:240
BIMETALLIC CORROSION \
cothodic tubeplote
steel hull
copper- olloy heoder \
'31 *.
cothodc tubes
I
1 I
I
\
I
\
I
\
I I
I
I I I 1
I I I
tf
uncoupled cmwm rate
1 Dirt once
Fig. 1.72 Schematic representation of calculated bimetallic corrosion distribution; ( 0 ) connected header, (b)electrically-insulatedheater
The differential equation to be solved for a cylindrical system assuming unidirectional current flow conditions apply is34'35 2E -d= dx2
2pi
r
where E is the electrode potential at distance, x, along the system, p is the electrolyte resistivity, i is the local surface current density at distance, x, and r is the system radius. The solution of this equation depends upon the potential dependence of i, Le. the form of the cathodic or anodic polarisation curve for each metal in the system. Analytical solutions have been derived for linear polarisation kinetics 34, Io'* 109-111 , f or Tafel conditions 349"', for Butler-Volmer conditions9#3s*lo', llo* l1I, 113* I L 4 and for combinations of linear polarisation kinetics with a potential-independent current den~ity'~. Astley has used unidirectional current flow analysis to assess bimetallic corrosion and cathodic protection distribution within a number of real seawater Thus, design-stage analysis has been made of the bimetallic corrosion distribution within a marine heat-exchanger system having a cathodic metal tube-bundle, two conical bronze headers and two seven-metre long, 350 mm dia cupronickel feed-pipes each connected to a steel hull by lengths of plastic ~ i p i n g . ~ ~ Mathematical ."~ modelling for this system revealed the likely magnitude of corrosion rates within the headers and feed-pipes, both before and after electrical insulation of the headers
BIMETALLIC CORROSION
1 :241
at their flanges, and also quantified any possible decrease in bimetallic corrosion that could arise due to electrical connection of the steel hull to the feed-pipes. The schematic corrosion rate distributions are shown in Figure 1.72. Design decisions with respect to header insuiation and the employment of wastage pieces were thereby facilitated. Quantitative confirmation of the predicted corrosion pattern in this system was subsequently obtained in pilot-scale trials. The development of mathematical modelling techniques is proving to be a significant advance in the assessment of the bimetallic corrosion hazard in real systems.
M.J. PRYOR D.J . ASTLEY REFERENCES 1. Evans, U. R., The Corrosion and Oxidation of Metals, Arnold, London (1961) 2. Gatty, 0.and Spooner, E. C., Electrode Potential Behaviour of Corroding Metals in Aqueous So/utions, Oxford University Press, London (1938) 3. Mansfeld, F., Corrosion, 29, 403 (1973) 4. Baboian, R., Paper 58, ‘Corrosion 85’, N.A.C.E., Houston (1985) 5. Bauer, 0. and Vogel, O., Mitt. MatPrifAmt. Inst. Metallforsch. Berl., 36, 114 (1918) 6. Latimer, W.M., Oxidation States ofthe Elements and Their Potentials in Aqueous Solutions. Prentice-Hall, New York, 294 (1950) 7. Mansfeld, F.,Corrosion, 30, 343 (1974) 8. Scholes, I. R., Astley, D. J. and Rowlands, J. C., Sixth European Congress on Metallic Corrosion, SCI. London, 161 (1977) 9. Astley. D.J. and Rowlands. J. C., Br. Corr. J., 20. 90 (1985) 10. Linder, M. and Mattson, E., Seventh Scandinavian Corrosion Congress, Norway, 19 (1975) 11. Davis. G.0..Kolts, J. and Sridhar, N.. Corrosion, 42, 329 (1986) 12. Hack. H.P. and Scully, J. R., Corrosion, 42, 79 (1986) 13. Lee, H.Y.,Son, U. T. and Kim, S. J., J. Korean Inst. Met., 20, 31 (1982) 14. Kuron, D.,Kilian, R. and Grafen, H., 2. Werkstofltech, 11, 382 (1980) 15. Guides to Practice in Corrosion Control, No 14;‘Bimetallic Corrosion’, Dept. of Industry (1982) 16. Reboul, M. C., Corrosion, 35,423 (1979) 17. Shalaby, L. A., Corrosion Science, 11, 767 (1971) 18. Promisel, N. E. and Mustin, G. S., Corrosion, 7, 339 (1951) 19. Evans, U.R. and Rance, V. E., Corrosion and Its Prevention at Bimetallic Contacts, H.M.S.O., London (1958) 20. ‘Commentary on corrosion at bimetallic contacts and its alleviation’, British Standards Institute, PD6484 (1979),confirmed August (1984) 21. Godard. H.P., Corrosion, 7, 93 (1951) 22. Lauer, G. and Mansfeld, F., Corrosion, 26, 504 (1970) 23. Cotton, J. B. and Downing, B. P., Trans. Inst. Mar. Eng., 69, 311 (1957) 24. Johnson, K. E. and Abbott, J. S., Br. Corr. J., 9, 171 (1974) 25. Southwell, C. R., Bultman. J. D. and Alexander, A. L., Materials Performance, 15, 9 (1976) 26. Danek, G. J., ‘The effect of seawater velocity on the corrosion behaviour of metals’, Naval Engineers Journal, No. 763, (1966) 27. Pryor, M.J. and Keir, D. S., J. Electrochem. SOC., 104, 269 (1957) 28. Dyess. J. B. and Miley, H. A.. Trans. Amer. Inst. Min. (Metall.) Engrs.. 133, 239 (1939) 29. Pryor, M. J. and Keir, D.S.. J. Electrochem. Soc., 102. 605 (1955) 30. Pryor, M. J., Nature, Lond., 178, 1245 (1956) 31. Mayne, J. E. O., Menter, J. W. and Pryor, M. J., J. Chem. SOC., 1831 (1949) 32. Evans, U. R., J. Chem. Soc., 478 (1930) 33. Pryor, M.J. and Keir, D. S., J. Electrochem. SOC., 105, 629 (1958)
I :242
BIMETALLIC CORROSION
34. Astley. D. J., Corrosion Science, 23, 801 (1983) 35. Astley, D.J.. Gulvunic Corrosion E d . H. P. Hack) ASTM STP 978 pp. 53-78 (1988) 36. Whitman. W.G.and Russell, R. P.,Industr. Engng. Chem., 16.276 (1924) 37. Mansfeld, F., Corrosion, 27, 436 (197ij 38. Yau, Y. H.and Streicher, M. A., Corrosion, 43, 366 (1987) 39. Uhlig, H.H.,Corrosion and Corrosion Control, Wiley, New York (1971) 40. Eldridge, G. G.and Mears, R. B., Industr. Engng. Chem., 37, 736 (1945) 41. Bardal, E., Johnsen, R. and Per Olav Gartland, Corrosion, 40,628 (1984) 42. Mansfeld, F., Corrosion Science, 15, 183 (1975) 43. Mansfeld, F., Corrosion Science, 15, 239 (1975) 44. Mansfeld, F., and Kenkel, J. V., Corrosion, 33, 376 (1977) 45. Lennox, T. J., Peterson, M. H., Smith, J. A. and Groover, R. E., Materials Performance, 13, 31 (1974) 46. Robson. D. N. C., Section 2.3 of ‘Corrosion and Marine Growth on Offshore Structures’, 69, J. Wiley & Sons, (1984) 47. Wei, M. W.. Corrosion, 23, 261 (1967) 48. Venczel, J. and Wranglen. G.. Corrosion Science, 7, 461 (1%7) 49. Wranglen, G.and Inam Khokar. M.,Corrosion Science, 9,439 (1%9) 50. Monticelli, C., Brunoro. G., Trabanelli, G. and Frignani, A., Werkst. Korros., 38, 83 (1987) 51. Marek, M., ‘Corrosion of Dental Materials’, Encyclopedia of Muterials Science and Engineering, Vol. 2, Pergamon Press, 896 (1986) 52. Trzaskoma, P. P., Corrosion, 42, 609 (1986) 53. Belluci, F. D.,Martino, A. and Liberti, C., J. Appl. Electrochemistry, 16, 15 (1986) 54. Rosenfel’d, N., Proceedings of the First International Congress on Metallic Corrosion, London, I%], Butterworths, London, 243 (1962) 55. ‘Determination of bimetallic corrosion in outdoor exposure tests’, BS 6682 (1986) 56. Kucera, V. and Mattson, E., Atmospheric Corrosion of Bimetallic Structures, ex Atmospheric Corrosion, 561, J. Wiley and Sons, (1982) 57. Pelensky, M. A., Jaworski, J. J. and Gallaccio, A., ASTM STP 646,58 (1978) 58. Baboian, R., ASTM STP 646, 17 (1978) 59. ‘Protection of Electrical Power Equipment against Climatic Conditions’, BS C P 1014 (1963) 60. Godard. H. P., Muterials Protection. 2, 40 (1963) 61. Latimer, K. G.,2nd Inter. Congress Metal Corrosion, 780 New York City, (1%) 62. Escalante, E. and Gerhold, W. F., ASTM ‘Field and Laboratory Studies’, 81 (1976) 63. Schick, G. and Mitchell, D. A., ASTM ‘Field and Laboratory Studies’, 69 (1976) 64. Vrable, J. B., Materials Performonce, 21, 51 (1982) 65. Toy, S. M., English, W. D. and Crane, W. E., Corrosion, 24, 418 (1968) 66. Jones, D.A. and Wilde, B. E., Corrosion, 33,46 (1977) 67, Schikorr, G.,Trans. Electrochem. Soc., 76, 247 (1939) 68, . Hoxeng,-R. B. and Prutton, C. F., Corrosion, 5, 330 (1949) 69. Kenworthy, L. and Smith, M. D., J. Inst. Met., 70, 463 (1944) 70. Schuldener, H.L. and Lehrman, L., J. American Wat. Wks. Assoc., 49, 1432 (1957) 71. Cohen, M., Thomas, W. R. and Sereda, P. J., Cunad. J. Technol.. 29, 435 (1951) 72. Schuldener, H.L. and Lehrman, L.. Corrosion, 14, 54% (1958) 73. Caplan, D.and Sereda. P. J.. Cunud. J. Technol., 31, 172 (1953) 74. Gilbert, P. T.,J. Electrochem. Soc., 99, 18 (1952) 75. Glass, G.K. and Ashworth. V.. Corrosion Science, 25, 971 (1985) 76. Gabe, D.R. and El Hassan. A.M., Br. Corr. J., 21, 185 (1986) 77. Zanker, L. and Yahalom. J.. Corrosion Science, 9, 157 (1969) 78. Gouda, V. K., Shalaby, L. A. and Abdul Azim, A. A., Br. Corr. J., 8, 81 (1973) 79. Bothwell, M. R., J. Electrochem. SOC., fod, 1014, 1019 (1959) 80. Akimov,-G., Korros. Metaflsch., 6 , 84 (1930) 81. Edeleanu, C. and Evans, U. R., Trans. Faraday Soc., 47, 1121 (1951) 82. Bird, C. E. and Evans, U. R., Corr. Techno!., 3, 279 (1956) 83. See for instance Evans, U. R., The Corrosion and Oxidution of Metals, Arnold, London, 205-6(1960) 84. Gilbert, P. T., ‘Considerations arising from the use of dissimilar metals in seawater piping systems’, 5th International Congress on Marine Corrosion and Fouling, Barcelona (1980) 85. Rowe, L. C., Automotive Engineering, 82, 40 (1974)
BIMETALLIC CORROSION
1 :243
86. Brunoro, G.. Zucchi, F. and Zucchini, M . , Mater. Chem.. 5 , 135 (1980) 87. Jarman, R. A. and Shreir, L. L., Welding and Metal Fabrication, 444 (1987) 88. Draley, J. E. and Ruther, W. E., Corrosion, 12,48Ot (1956);J. Electrochem. Soc., 104, 329 (1957) 89. Phillips, H.W. L., J. Znst. Met., 69, 275 (1943) 90. Tornashov. N. D. and Chernova, G. P., C,R. Acad. Sci. U.R.S.S.,89, 121 (1953) 91. Tomashov. N. D.,Altovsky, R. M. and Arakelov, A. G., C. R. Acad. Sci. U.R.S.S.,121, 885 (1958) 92. Stern, M. and Wissenberg, H., J. Electrochem. Soc., 106,755, 759 (1959) 93. Greene, N. D., Bishop, C. R. and Stern, M.. J. Efectrochem. SOC.,108, 836 (1961) 94. Gehring, G. A., Kuester, C. K. and Maurer, J. R., Paper 80, ‘Corrosion SO’,N.A.C.E., Houston (1980) 95. Agar, J. N. and Hoar, T. P., Discuss. Faraday Soc., 1, 158 (1947) 96. Newman, J.. Electrochemical Systems, Prentice-Hall (1973) 97. Waber, J. T.and Ruth, J. M., Los Alamos Lab. Microfich LA-1993 (1956) 98. McCafferty. E., Corrosion Science, 16, 183 (1976) 99. McCafferty, E., J. Electrochem. Soc., 124, 1869 (1977) 100. Melville, P. H.,J. Electrochem. SOC.. 126,2081 (1979) 101. Melville, P.H., J. Electrochem. SOC.. 127, 864 (1980) 102. Doig, P. and Flewitt. P. E. J., J. Electrochem. Soc.,126, 2057 (1979) 103. M u m , R. S. and Clark, J. H., Paper 74, ‘Corrosion 83’, N.A.C.E., Houston (1983) 104. Helle, H.P. E., Beck, G. H. M.and Ligtelijn, J. Th., Corrosion, 37, 522 (1981) 105. Forrest, A. W., Fu, J. W. and Bicicchi. R. T., Paper 150, ‘Corrosion 80’, N.A.C.E., Houston (1980) 106. Danson. D. J. and Warne, M. A.. Paper 21 1, ‘Corrosion 83’,N.A.C.E.. Houston (1983) 107. Frurnkin, A. N., Zh. j7z. Khim., 23. 1477 (1949) 108. de Levie, R. in Advances in Electrochemistry and Electrochemical Engineering, Vol. 6, Ed. Delahay, P., Znterscience (1967) 109. Sato, S. and Yamauchi, S., Sumitomo, Light Metal Techn. Rep., 17, 24 (1976) 110. Chizmadzhev, Yu. A., Markin, V. S., Tarasevich, M. R. and Chirkov, Yu. G., Macrokinetics of Processes in Porous Media, Nauka, Moscow (1971) 1 1 1. Reingeverts, M.D., Parputs, I. V. and Sukhotin, A. M., Soviet Electrochemistry, 16,35 (1 980) 112. Mueller, W. A.. J. Electrochem. Soc.. 110,698 (1%3) 113. Posey, F. A.. J. Electrochem. Soc., 111. 1173 (1964) 114. Alkire, R. and Mirarefi, A. A., J. Electrochem. SOC., 120, 1507 (1973) 115. Astley, D. J., ‘Prediction of Galvanic Corrosion in Marine Heat-exchangers’, Institute of Metals Conference, Bristol (1986)
1.8 Lattice Defects in Metal Oxides
When a metal oxide is in contact with one of its components (metal or oxygen), the condition for thermodynamic equilibrium cannot, in general, be satisfied unless the crystal is non-stoichiometric, Le. unless it contains an excess of one of the two components. The reason for this is that although energy must be expended in incorporating the excess component, the entropy of the system increases extremely rapidly at first, and then more slowly as the non-stoichiometry increases. Thus the equilibrium condition, namely that the free energy of the system is a minimum, is satisfied only for some finite degree of non-stoichiometry. The thermodynamic functions are shown schematically as functions of a,the degree of non-stoichiometry, in Fig. 1.72. We note at this stage that it is sufficient to discuss the equilibrium between an oxide and oxygen gas because the other case (equilibrium between the oxide and the metal) is then covered by putting the oxygen pressure equal to the dissociation pressure of the oxide.
Fig. 1.72 Thermodynamic functions G, H and TS as functions of a, the degree of nonstoichiometry. G is the free enthalpy, H the enthalpy and S entropy
1:244
LATTICE DEFECTS IN METAL OXIDES
1 :245
There are two basic questions which can be decided only by experiments. First, we must know whether the metal or the oxygen is present in excess, and second, we must know how the excess component is incorporated in the oxide lattice. In connection with the latter question we have to remember that a non-stoichiometric crystal remains electrically neutral (except in narrow regions near the surfaces), so that if the excess component is present in the crystal as ions, lattice defects with charges of opposite sign must necessarily be present also (see Figs. 1.77 and 1.78). The most important defect structures will be discussed in this section. The presence in an oxide of an excess of one component provides a mechanism for the transport of material. This transport mechanism, which is vital in understanding the formation of a continuous oxide film on a metal, is also discussed in this section. An important feature here is that an excess of one component may provide a transport mechanism, not for itself, but for the other component.
p-type Oxides Cu,O contains excess oxygen which is taken up in such a way as to build up new layers of the oxide. Thus, the excess oxygen is present as 0'- ions on their normal lattice sites. To form oxygen ions, electrons are needed, and to build up new layers of the crystal, Cu+ ions are needed. Both species, electrons and Cu', ions, are supplied from the interior of the crystal. Thus excess oxygen is incorporated in C u 2 0 by forming vacant cation sites (Cu' O), and vacant electron levels (en) in the crystal. The vacant electron levels are called positive holes, and in Cu,O they may be pictured as Cu2+ ions at normal cation sites in the lattice. The cation vacancies are negative charges in the Cu,O lattice, while the positive holes are positive charges. Consequently, they attract each other, and at low temperatures may stick together. We shall assume that the temperature is high enough for this trapping of positive holes by cation vacancies to be ignored. The chemical reaction giving non-stoichiometric Cu,O may then be written
io2
-+
Cu20
+ 2(Cu+O) + 2 ( e 0 )
. . .(1.157)
four cation vacancies and four positive holes being produced by every oxygen molecule absorbed. The non-stoichiometric oxide conducts electricity with the movement of positive holes, and because of this the conductivity is said to be p-type (positive carriers). If n is the concentration of defects (cation vacancies or positive holes) at equilibrium, then, applying the law of mass action to equation 1.157 = pl/SKl/4
. . .(1.158)
wherep is the oxygen pressure, and K the equilibrium constant for the reaction. The formula for K can be found only from statistical mechanics, and a simple calculation is instructive. The important quantity which we need to estimate is the entropy change when one molecule of oxygen is absorbed, with the formation of four defects in the lattice. This entropy change may be divided into two parts. The first is the entropy change, AS,,when one molecule is absorbed, with the formation of defects at specified lattice
1 :246
LATTICE DEFECTS IN METAL OXIDES
points; the second is the entropy change, AS,, when we allow for the fact that the defects may be formed anywhere in the crystal. AS,is independent of the existing defect concentration, and, in addition, is almost certainly negative because it involves the loss of translational and rotational degrees of freedom of the oxygen molecule. AS, is positive and depends on the existing defect concentration. To calculate AS, we proceed as follows. Let there be N cation sites in the crystal and n defects. The number of arrangements, P,of the cation vacancies on the lattice sites is
. . .(1.159)
P = M / n !( N - n ) !
which contributes an entropy S, = kdnP, where k is Boltzmann’s constant. Taking logarithms in equation 1.159, using Stirling’s approximation (Inw! = xlnx - x as x - *
m)
and putting n / N = CY we find S, = - ~ k ([1 - a)ln( 1 - a)
. . .(1.160)
+ alna]
We note that S, is positive and goes through a maximum as a increases. If the positive holes were localised on the cations, they would give an entropy contribution S, exactly equal to S,. The positive holes have, however, considerable mobility (see below), and are perhaps best treated as an ideal gas consisting of particles of effective mass m. In this case’ S, = ~ & 4 + i n ( @ P ~ )
+ $ - lna + 1x121
. . .(1.161)
where 4 = 2?rmkT/hZ,h being Planck’s constant. 0 is the volume of the oxide per metal ion, and the term Nkaln 2 takes account of electron spin. We note that Sp is positive but shows no maximum. Adding equations 1.160 and 1.161, and differentiating with respect to a, we obtain AS,. Remembering that CY p o , then n, > no, and the oxidation proceeds with cation vacancies and positive holes being created at the Cu,O/O, interface and moving inwards to be destroyed at the Cu/ Cu,O interface. A similar situation exists whenever a p-type oxide is formed, for example with Ni/NiO. As an example of a different type of oxide, we may consider ZnO. This oxide evolves oxygen and forms cations in interstitial positions (Zn’ 0 )or (Znz+0),and free electrons (eo). If the interstitial zinc ions are only singly charged, the reaction describing the non-stoichiometry may be written
*
ZnO
-+
to, + ( z n + 0)+ (eo)
Using the simple mass action formula, the equations corresponding to 1.169 and 1.170 are
1 :256
CONTINUOUS OXIDE FILMS
. . .(1.171) with po equal to the dissociation pressure of ZnO. If p , > po then n, < no, and the oxidation proceeds with interstitial cations and free electrons moving outwards from the Zn/ZnO interface. A similar situation exists whenever the oxidation product is an n-type oxide. The system Al/Al,O, is another example.
Fig. 1.75 Movement of defects in the oxidation of copper
Wagner's Theory of the Parabolic Law For definiteness, the oxidation of copper to copper(1) oxide may be considered. Our picture of the process is that cation vacancies and positive holes formed at the Cu,O/O, interface by equation, 1.166 are transported to the Cu/Cu,O interface where they are destroyed by copper dissolving in the non-stoichiometricoxide. We require an expression for the rate of oxidation. We denote by x the distance from the metal surface, and by n,(x) and n p ( x ) the concentrations of cation vancancies and positive holes in the oxide. Let u, and up be their mobilities, and 0,and Dp their diffusion coefficients. Let F(x) be the electrostatic field in the oxide. J,, the flux of cation vacancies (number crossing unit area per second), will be expressed by .( 1.172)
and J,, the flux of positive holes, by dnP + npvpF Jp = -Dp -
dx
. . .(1.173)
In equation 1.172 the first term on the right-hand side is the flux due to a concentration gradient, the second is that due to the electric field. It should be noted that cation vacancies are negatively charged carriers of electricity. The terms in equation 1.173 have similar meanings; positive holes are positively charged carriers. If the system were at equilibrium we would have J, = Jp = 0 and the field F would be negligible except in narrow regions near the two interfaces, but when oxidation is proceeding this is not generally the case.
CONTINUOUS OXIDE FILMS
1 :257
In a steady state of oxidation Jp = J, = J, say. Eliminating F from equations l. 172 and l. 173 and using Einstein’s relation (see below) u/D = e/kT, where k is Boltzmann’s constant and T the absolute temperature, J = - ( k T / e ) [n,u,n,v,/(n,u,
d + npu,)] ln(n,n,) . . .(1.174) dx
The negative sign means that cation vacancies and positive holes move inwards, i.e. in the negative direction of x. For Cu,O, the positive holes are much more mobile than the cation vacancies, and we can assume that n, up >> n, 0,. The oxidation flux is then d J = - ( k T / e ) n u - In( n,n,)
,dx
. . .(1.175)
This equation can be obtained in another way which may be more instructive. Assume that the slow step in the oxidation is the transport of cation vacancies. The positive holes may then be considered to take up their equilibrium distribution, defined by Boltzmann’s equation np = n,,(O)exp( -eV/kT) . . .(1.176) Here np is the concentration of positive holes at any point in the oxide where the electrical potential is V , and n,(O) is their concentration at the Cu/Cu,O interface where V = 0. Differentiating equation 1.176,we obtain an equation for the electric field, namely dV d F = --- ( k T / e ) . . ., (1.177) Inn,
dx
dx
Substituting this in equation 1.172 and using Einstein’s relation between the mobility and the diffusion coefficient of cation vacancies, we obtain equation 1.175 for the oxidation flux J,. This derivation shows that equation 1.175 is valid if all processes involving positive holes (and therefore electrons) are so fast that the oxidation flux can be carried without significantly disturbing their equilibrium distribution. We note at this stage that to derive Einstein’s relation it is only necessary to compare equation 1.176 with the equation obtained by integrating equation 1.173 for J, = 0. The integration gives n, = n, (0)exp ( -up WD,) and this establishes the result up/Dp= e/kT. The corresponding result for cation vacancies is derived similarly. Returning now to equation 1.175,we cannot in general proceed without knowing how n, and n, vary through the oxide. However, if the oxide layer is thick enough, the situation is simple, for we can assume that electrical neutrality is preserved in the oxide except in narrow regions near the two interfaces where there are space charges. Thus we have n, = np = n say, except in the space-charge regions which, however, we can neglect if the oxide layer is thick enough. Then equation 1.175 gives dn
J = -2 ( k T / e )u, -
dx
. .(1.178)
However, J must be independent of x, so this equation can be integrated to give
1:258
CONTINUOUS OXIDE FILMS
JX = -2(kT/e)u,(n,
- no)
. . .(1.179)
where Xis the thickness of the oxide layer, and nx and no are the concentrations of cation vacancies or positive holes at the two interfaces. These concentrations are of course given by equations 1.169 and 1.170. Our picture of the transport process in these thick oxide layers is that there is a uniform concentration gradient of defects (cation vacancies and positive holes) across the layer. But it is important to notice that the oxidation flux is exactly twice that to be expected if diffusion alone were responsible for the transport of cation vacancies. The reason for this is, of course, that the more mobile positive holes set up an electric field which assists the transport of the slower-moving cation vacancies. If Q is the volume of the oxide per metal atom, the rate of growth, dX/dt, is equal to I J l n . Thus from equation 1.179 we derive the parabolic law dX/dt = k2/X k, = W,Q(n, - no)
. .(1.180)
This formula for k2can be cast into another form by using equations 1.169 and 1.170. We note first that in these latter equations K’14is the concentration of defects in Cu,O at 1 atm pressure of oxygen, so that (K”4DcQ)is the self-diffusion coefficient of Cu’ in CuzOat this oxygen pressure. Call this self-diffusion coefficient 09,then
kz = 2 D ! ( p F -PA’*) . . .(1.181) This equation shows that if px >> p o ,k, depends upon the oxygen pressure and the temperature in the same way as the self-diffusion coefficient of Cu’ in Cu,O. Regarding the temperature dependence, self-diffusion coefficients in solids depend exponentially on the temperature (see Section 1.8), i.e.
Dz = Aexp( -Q/kT) and Q is the actiuation energy for self-diffusion. Thus for the temperature dependence of k, we have
k2 = Bexp( -Q/kT) and the activation energy for the oxidation reaction should be the same as that for self-diffusion of cations in the oxide. The above account of the oxidation of Cu to Cu,O is a simplified version of the more general theory developed by Wagner’.’. Cu,O is a p-type oxide. As an example of a system where an n-type oxide is formed, we shall consider the oxidation of Zn to ZnO. Here Zn dissolves in the oxide at the Zn/ZnO interface to give interstitial cations and free electrons. These defects cross to the ZnO/O, interface and react with oxygen to build up new layers of the oxide. The slow step in the oxidation is the transport of interstitial cations, and if these are singly charged we still have equation 1.179 for the flux, except that u, is the mobility of interstitial cations and n, and no are given by equation 1.171. We note further that J is positive in this case because defects move outwards. The expression for the parabolic rate constant corresponding to equation 1.181 is
k, = ~ D : ( P ; ’‘ ~pX1’4)
1:259
CONTINUOUS OXIDE FILMS
and if px >> po, the oxidation rate should be independent of the oxygen pressure. This pressure independence should be noted since the self-diffusion coefficient of Zn in ZnO does depend on the oxygen pressure. Further, the temperature dependence of k, is not simply that of the self-diffusion coefficient, since p,, , the dissociation pressure of ZnO, is also temperature dependent. Because of this the activation energy for the oxidation reaction is less than that for self-diffusion by half the heat of formation of ZnO.
Thin Oxide Films Wagner’s theory of the parabolic law involves the following assumptions: 1. The oxide layer is compact and adherent. 2. The slow step is the transport of material through the oxide layer. 3. The layer is so thick that the space-charge regions at the two interfaces are unimportant and the oxide can be regarded as electrically neutral.
In the early stages of oxidation when the oxide layer is thin, it is clear that assumption (3) must be invalid. The limiting simple case when the layer is so thin that space charges can be neglected because they are small compared with the surface charges has been considered by Mott All other assumptions are the same as in Wagner’s theory. In the oxidations of Cu, for example, we assume that electronic equilibrium is established in the system Cu/Cu,O/O,. This sets up an electrical potential difference across the oxide layer because electrons are transferred from the metal to form oxygen ions adsorbed on the outer surface of the oxide. If the surface charge formed in this way is large compared with the space charge in the oxide, the electric field is uniform and equal to V / X where Y is the potential drop across the oxide. In a thin oxide layer this field may be very large. For example, a potential difference of 1 V gives a field of 10SV/cmin a layer 100 nm thick. Now the flux of cation vacancies (and hence the oxidation flux) is given by equation l. 172, and if F is as large as this, a significant oxidation rate is to be expected even at ordinary temperatures where the diffusion coefficient is very small. For the system Cu/Cu,O the theory gives a cubic law of growth (dX/dt = k , / X 2 ) at ordinary temperatures, and a parabolic law at high temperatures. The parabolic rate constant is, however, entirely different from that in Wagner’s theory. When the oxidation product is an n-type oxide like ZnO or AlzO,, the law of growth is parabolic both at ordinary temperatures and at high temperatures. The two rate constants are different, and both differ from that in Wagner’s theory. For further details, the original papers’“ should be consulted.
’.
Very Thin Oxide Layers Many metals oxidise rapidly at first when exposed to oxygen at sufficiently low temperatures, but after a few minutes, when a very thin oxide layer has been formed, the reaction virtually ceases. Oxide layers formed in this way are about 5 nm thick. Aluminium and chromium are well-known examples, showing this sort of behaviour at room temperature. A theory of the effect has been proposed by M ~ t t ~ . ~ .
1:260
CONTINUOUS OXIDE FILMS
For definiteness consider the system Al/Al,O,. It is assumed as before that electronic equilibrium is established so that there is a field in the oxide associated with the presence of oxygen ions adsorbed on the outer surface. In a very thin layer this field will be enormous (about 106V/cm if X = lOnm), and we cannot assume, as we did in equation 1.172, that the contribution which this field makes to the flux of cations is simply proportional to the field. An investigation of the transport process in such strong fields (see below) shows that the flux increases exponentially with the field, and because of this cations are transported through the oxide much more rapidly than would be expected on the basis of equation 1.172. It seems unlikely therefore that cation transport can be the slow step in the reaction, and the rate will be controlled instead by a surface reaction. For the system All Al,O,, the slow step is probably that by which A13+ions enter interstitial positions in the oxide at the Al/AI,03 interface. The rate of this process is also influenced by the strong electric field. The potential energy diagram for an ion leaving the metal and entering the oxide is shown in Fig. 1.76. P represents an ion in the metal surface and I,,I,, are interstitial positions in the
I
,
c a----
Fig. 1.76 Potentiaienergy of an interstitial ion near the metaVoxide interface
oxide. The height of the barrier between P and I, is Q when the field is zero. The field lowers this barrier by an amount zeaF where ze is the charge on the ion and a the distance between P and the top of the barrier. The probability per second that an ion jumps from P to I , in the absence of a field is u exp ( - Q / k T ) , where u is the vibration frequency of an ion at P.In a field F, this probability is increased by the factor exp (zeaF/kT).Hence if 9 is the rate at which ions can enter the oxide when there is no field, the rate in the field is* G?exp (zeaF/kT)
If this process determines the oxidation rate, then with F = V / X , the law of growth is * An argument similar to this establishes the exponential relation between the cation flux through the oxide and the field F.
1:261
CONTINUOUS OXIDE FILMS
1
dX/dt = Q4 exp ( M X ) X = zeaV/kT
. . .(1.182)
If V = 1 V , a = 0.25nm, a n d z = 3, X = 30nm at 300K, so that for afilm l n m thick, the field increases the rate of growth by a factor of about The term in the growth law due to the field, namely exp (X/X), is large only when X i s small. Because of this a thin oxide film can form even at low temperatures where 4,the ordinary rate of entry of ions into the oxide, is negligible. As the film thickens, the factor exp (WX)decreases rapidly, and the rate of growth soon falls to such a low value that, for practical purposes, oxidation has ended.
Effects of Alloying An important aspect of any theory of the oxidation of a pure metal is that it enables us to see how the protective power of the oxide layer can be altered by the introduction of alloying constituents into the metal. According to Wagner’s theory, the parabolic rate constant for the system Ni/NiO for example depends upon the concentration of cation vacancies in the oxide in equilibrium with oxygen gas. If this concentration can be reduced, the oxidation rate is reduced. Now this can be done if cations of lower valency than NiZ+can be got into the oxide (Fig. 1.77). Suppose, for example, that a little Li is added to the Ni. Each Li+ ion which replaces Ni2+is a negative
0’-
NiZf
0’-
Ni3+
O2-
Ni2+
Ni2+
0’-
NI’+
O2-
Ni”
O2-
0 2 -
0
0’-
Ni3+
02-
Ni’+
Ni2+
0’-
Ni’+
02-
Ni2+
0‘-
0’-
LI
0 2-
Ni.\+
NiZ+
0’-
Ni’+
0’-
Ni2+
0’-
0 2 -
Li
0 2 -
Ni’+
0’-
Ni’+
Ni2+
0’-
Ni2+
0 2 -
Ni”
0’-
0’-
Niz+
O2-
Ni3+
0’-
Ni2+
;
Cr’ +
0 2 -
NiZ+
O2-
Ni’+
0’-
I
0 2 -
0
0’-
Ni3+
0 2 -
Ni2+
O 2-
Cr’+
02-
0
+
NiO
I
v Ni”
NiO+Li,O +
I
1
NiO+Cr,O,
N”+ I 0’-
1
I
Fig. 1.77 Effects of Li,O and Cr,03 on the defect structure of NiO
1 :262
CONTINUOUS OXIDE FILMS
charge in the NiO lattice. To preserve electrical neutrality, one positive hole (eo) must be created for each Li+ ion introduced. But the product n(Ni2+0 )n ( e 0 ) is fixed by the reaction governing the non-stoichiometry of NiO. Hence n(Ni2+0 )falls and the oxidation rate is reduced. By a similar argument, an alloying constituent of higher valency than AiZ+(Cr3+for example) which enters the oxide layer in place of Ni2+ increases the oxidation rate. When the oxidation product is an n-type oxide like ZnO, the conditions are reversed (Fig. 1.78). If a monovalent ion like Li+ enters the oxide layer in place of Zn2+ one free electron (eo) is destroyed. But the product n(Zn+ O)n(eo) is fixed by the reaction governing the non-stoichiometry of ZnO. Hence n(Zn+ 0), the concentration of interstitial Zn+ ions, increases, and the oxidation rate, which depends upon the concentration of these ions in the oxide in equilibrium with metallic Zn, increases. This simple account of the effect of alloying constituents is valid only if the second metal shares in the oxide formation by dissolving freely in the oxide of the basis metal. Further, the second metal should be present in the oxide layer in such low concentrations that it can be regarded as an impurity in the oxide of the basis metal. If the alloying constituent is insoluble in the oxide of the original metal, or if a new phase, for example a spinel, is formed, the discussion fails. The spinel NiCr2O4is in fact formed in the oxidation of Ni-Cr alloys when the Cr content is high enough, and the oxidation rate is decreased, not increased as we would expect from the simple discussion above. I
1
Zn2+
02-
Zn2+
02-
Zn2+
02Zn
e
02-
Zn2+
Zn2
O2-
+
zn2+ 02-
0 2 -
e
0'-
-
Zn
O2-
Zn2+ 0 2 -
02-
Znz+
Zn2+
02-
02-
Zn2+
Zn2+
0 2 -
Znz+
O2
O2 -
Zn2+
O2-
Zn2+
Cr'
0 2 -
Zn2+
02-
Zn
O 2-
-
+
j
Zn
ZnO
ZnO + Li,O
+
e
e
j
+
ZnZ+ Zn
Zn2+
02-
+
-
+
I
+
Zn2+
O2-
Zn2+
O2
Zn2+
02-
Zn2+
02-
ZnZ+
O2
O2-
Zn2+
02-
zn2+ j
-
ZnO+Cr,O, l
e
Zn2+
-
Fig. 1.78 Effects of Li,O and Cr,03 on the defect structure of ZnO
1 :263
CONTINUOUS OXIDE FILMS
To examine the situation with alloys in a little more detail, the Cu-Ni alloys will first be considered. Here the mutual solubility of the two oxides NiO and CU20 can probably be neglected, and these are the only two possible oxidation products. Assume for simpkity that the alloy is thermodynamically ideal, and let x,, and xNibe the mole fractions in the alloy. Consider the reactions
+ Ni + +O,
2cu
+ 0 2
-P
cu,o NiO
-+
whereby Cu,O and NiO are formed by oxidation of the alloy. The equilibrium conditions are X&Jl”Z
= p’”(cu,o)}
. . .(1.183)
xNipl/’ = p’”(Ni0)
where p(Cu,O) and p(Ni0) are the dissociation pressures of the two oxides, a n d p is the effective oxygen pressure at the alloy surface. The two relations of equation 1.183 are illustrated in Fig. 1.79 by plotting p against xNi. Note that p(Ni0) < p(Cu,O) and xNi x,, = 1. The two curves intersect at one value of xNi,and this defines the alloy composition for which C u 2 0 and NiO can co-exist on the surface. If the Ni content is higher than this critical amount, only NiO is stable; for lower Ni contents only C u 2 0 is stable. If therefore the diffusion coefficients of Cu and Ni in the alloy were very large so that the composition of the alloy in the surface region did not change during oxidation, the situation would be simple, and the oxidation product would be either Cu,O or NiO except at one critical alloy composition where
+
xNi
I--cu,o--;-
Ni 0
Fig. 1.79 Surface oxides on Cu-Ni alloys
I U I
1 :264
ONTINUOUS OXIDE FILMS
Fig. 1.80 Surface oxides on Fe-Cr alloys
both oxides would be formed. But the diffusion coefficients in the alloy are not usually large enough for this simple analysis to apply, and if we start with a bulk alloy composition in the region where we would expect only NiO to be formed then, as oxidation proceeds, the surface region of the alloy is depleted of Ni. Thus Cu diffuses inwards and Ni has to be supplied from the interior of the alloy. The composition of the alloy at the surface necessary to maintain the supply of Ni into the oxide layer may be such that the Ni content is below the critical concentration for which only NiO is formed. If this is so, both Cu,O and NiO are formed as oxidation products; if not, we still get only NiO. The effect of finite diffusion rates in the alloy is therefore that the critical bulk alloy composition for the exclusive formation of NiO is pushed to a higher Ni content than we would expect from elementary consideration. Similarly, the minimum Cu content for the exclusive formation of Cu,O is pushed higher. There is therefore a range of compositions of the bulk alloy in which both Cu,O and NiO are formed together. This analysis shows that if the oxides of the two components of a binary alloy are mutually insoluble, and if one of the components has a much greater affinity for oxygen than the other, then the oxide of the baser metal will be formed exclusively even though it is present in the alloy in only a small amount. It seems that the importance of beryllium as an alloying constituent can be explained in this way. It has a high affinity for oxygen [p(BeO) = 10-30atmat 1000°C] and also forms a highly protective oxide layer. The
CONTINUOUS OXIDE FILMS
1 :265
oxidation resistance of metals more stable to oxygen than Be, but which normally oxidise faster, should therefore be improved by the addition of Be, provided that the oxide of the basis metal is not soluble in BeO. The addition of Be to Cu is an example. Another example which can be argued on the same lines is that of Cr-Fe alloys. This is more complicated, and for simplicity we may assume that only FeO could be formed as the oxidation product of Fe. In addition, Cr203 and the spinel FeCr,O, can be formed. We expect the dissociation pressures to be in the order p(Fe0) > p(FeCr,O,) > p(Cr203)and Fig. 1.80 may be constructed showing two critical Cr contents of the alloy. Below the first only FeO is formed, between the first and the second only FeCr,O,, and above the second only Cr203.The existence of finite diffusion rates in the alloy will, of course, smear out these divisions. There is, however, the possibility that by adding Cr to Fe, either the spinel FeCr204or the single oxide Cr203 is formed exclusively as the oxidation product. A Cr,O, layer is certainly protective. A spinel layer will be protective if the diffusion coefficient for Fe2+(or Fe3+)in the spinel is lower than that in the oxides of iron. We note that the protective layer (either Cr203or FeCr,O,) is formed next to the alloy. Beyond this there will almost certainly be another layer composed mainly of the oxides of iron. This portion is without influence on the protective properties of Cr.
Experimental Techniques Oxidation is followed by measuring the gain in weight of the specimen with time. An electrostatic field applied across the growing oxide enhances or reduces the oxidation rate according to the polarity of the field, and the charge on the moving species. The movement, or lack of it, of an inert marker placed on the metal prior to oxidation indicates whether the oxide grows by metal moving outwards, or oxygen moving inwards (see Section 1.10). Techniques of modem surface science (Auger Electron Spectroscopy (AES), Secondary Ion Mass Spectrometry (SIMS), X-ray Photoelectron Spectroscopy ( X P S ) , Ion Scattering Spectroscopy (ISS), for example) are used to determine the composition, and the thickness of tarnish films. Three examples must suffice. Firstly, ion scattering has been used' to analyse airformed films on Fe-Cr alloys. Incident ,"Ne or 3Heions with energies in the range 1.3 keV scattered at 90°,when energy-analysed, have peaks for each element in the surface of the film, and since the incident beam sputters the surface, a depth profile is also obtained. As expected from the discussion above, at the oxide/air interface the Cr/Fe ratio is low, as is the metal/ oxygen ratio, and the Cr/Fe ratio increases going into the oxide. But an unexpected finding is that the latter ratio peaks a short distance into the oxide. No explanation of this has been given. A second example of the use of surface-sensitive analytical techniques is the investigation* using AES, and argon-ion sputtering, of the composition, and thickness of the films formed on Ni in air in various relative humidities. The findings of this work will be mentioned below. Angularly resolved XPS, unlike depth profiling by sputtering, is non-destructive. Photoelectrons from the metal, and from its different oxides, are identified by their chemical shifts. Those originating
1:266
CONTINUOUS OXIDE FILMS
from the metal are attenuated by the oxide film, and the current I ( X , e), measured at an angle 0 to the normal for a film of thickness X is
I(x,
e) a exp( - x~xoxcose)
where A, is the electron mean free path in the oxide. Oxide thickness up to around 20nm depending on the system, can be measured in this way, and in addition, by varying 8, the uniformity of the film can be checked. Measurements using different XPS lines (when they are present) enables the kinetics of multi-layer film growth to be determined. XPS studies of the oxidation of Nb at 300 K have been published '. Nb,O, grows on a thin layer ( - .5nm) of NbO, (x < 1) to a limiting thickness of about 6nm in several days, with the growth law dX/dt = 2Qgsinh ( W X )
which is (1.182) modifided to include the back-reaction.
Atmospheres other than Pure Oxygen Atmospheres containing H,O vapour, or SO, are technically very important, but much more fundamental work is needed, and there is space here for only elementary considerations. vapour mixtures. Possible oxidation Consider Ni exposed to 02/H20 products are NiO and Ni (OH), , but the large molar volume of Ni (OH), , (24 cm3compared with that of Ni, 6.6 cm3)means that the hydroxide is not likely to form as a continuous film. From thermodynamic data, Ni (OH), is the stable species in pure water vapour, and in all O,/H,O vapour mixtures in which 0,is present in measurable quantities, and certainly if the partial pressure of 0, is greater than the dissociation pressure of NiO. But the actual reaction product is determined by kinetics, not by thermodynamics, and because the mechanism of hydroxide formation is more complex than oxide formation, Ni (OH), is only expected to form in the later stages of the oxidation at the NiO/gas interface. As it does so, cation vacancies are formed in the oxide according to
HzO(g)
+ fO,(g)
-+
Ni(OH),
+ (Ni2+O)+ 2 ( e 0 )
which lowers the degree of protection afforded by the NiO film'. Similar considerations apply to O,/SO, mixtures. Taking Cu as an example, thermodynamic data show that in the presence of SOz, the sulphate CuSO, is the stable species even when the partial pressure of 0, is as low as the dissociation pressure of Cu,O. Even so, for kinetic reasons, Cu,O should form first, and be converted to CuSO, at the Cu,O/gas interface in the later stages of the reaction. Because the volume of the sulphate per metal atom is so much larger than that of the oxide it replaces, the sulphate is unlikely to be continuous. Furthermore, its growth creates defects in the CuzO film:
SO,(g) + Oz(g) -,CuSO, which lowers its protective power.
+ (Cu+O) + ( e o )
1 :267
CONTINUOUS OXIDE FILMS
The situation with Ni in 0 2 / S 0 2 mixtures is different. When the partial pressure of O2 is as low as the dissociation pressure of NiO, the sulphide NiS, not the sulphate, is the stable species. Consequently, in the presence of NiO, which for kinetic reasons is expected to form in preference to the that the sulphate, NiS forms at the Ni/NiO interface. It is observedLo*” sulphide forms as a thin layer between the metal, and the oxide, and also grows into the oxide as a network providing an easy path for the transport of Ni to the solidlgas interface. No quantitative theory of oxidation leading to this film morphology, which is observed” also with Co, exists. Film growth is initially linear indicating that a surface reaction controls the rate, but in the later stages of the reaction, film growth obeys a parabolic law because the transport of Ni in the sulphide network controls the rate.
T.B. GRIMLEY
REFERENCES 1. Wagner, C., Z. Phys. Chem. E., 21, 25 (1933) 2. Wagner, C., Z . Phys. Chem. B., 32, 47 (1936) 3. Mott, N. F., J. Chim. Phys., 44, 172 (1947) 4. Cabrera, N. and Mott, N. F., Rep. Progr. Phys., 12, 163 (1948-49) 5. Grimley, T. B. and Trapnell, B. M. W., Proc. Roy. SOC. A , , 234, 405 (1956) 6. Hauffe, K. and Schottky, W.. Halbleiterprobleme, 5, 203 (1960) 7. Frankenthal, R. P. and Malm, D. L., J. Electrochem. SOC.123, 186 (1976) 8. Kulpa. S. H. and Frankenthal, R. P . . J. Electrochem. SOC. 124, 1588 (1977) 9. Grundner, M. and Halbritter, J., Surface Sci. 136, 144 (1984) 10. Luthra. K. L. and Worrell, W. L., Metall. Trans. 94, 1055 (1978) 11. Luthra, K. L. and Worrell, W. L., Metall. Trans. 1OA 621 (1979) 12. Jacobson, N. S. and Worrell, W. L., J. Electrochem. SOC.131, 1182 (1984)
BIBLIOGRAPHY Kofstad, P., High-Temperature Oxidation of Metals, John Wiley. New York (1966). Lawless, K. R., “The Oxidation of Metals”, Rep. Prog. Phys. 37, 231 (1974) Birks, N. and Meier, G. H., Introduction to High Temperature Oxidation of Metals, Edward Arnold, London (1983) Kubaschewski, 0. and Hopkins, B. E., Oxidation ofMetals andAlloys, Butterworths, London (1962)
Hauffe, K.,Oxydation von Metallen und Metallegiemngen, Springer, Berlin (1956) Garner, W. E. (Ed.), Chemistry of the Solid State, Butterworths, London (1955) Btnard, J. (Ed.), L’Oxydation des Mkteaux, Vol. 1, Gauthier-Villars et Cie, Paris (1962)
1.10 Discontinuous Oxide Films
The Applicability of Rate Laws Section 1.9 showed that as long as an oxide layer remains adherent and continuous it can be expected to increase in thickness in conformity with one of a number of possible rate laws. This qualification of continuity is most important; the direct access of oxidant to the metal by way of pores and cracks inevitably means an increase in oxidation rate, and often in a manner in which the lower rate is not regained. In common with other phase change reactions the volume of the solid phase alters during the course of oxidation; it is the manner in which this change is accommodated which frequently determines whether the oxide will develop discontinuities. It is found, for example, that oxidation behaviour depends not only on time and temperature but also on specimen geometry, oxide strength and plasticity or even on specific environmental interactions such as volatilisation or dissolution. The models derived for continuous oxide layers remain valuable when porous oxides are formed; they provide a frame of reference against which deviations may be examined and give a basis for understanding the factors governing the location of new oxide. In many cases, however, the experimentally derived rate 'laws' no longer have a unique interpretation. For example, the linear rate law relating the thickness of oxide, x , to the time, t x = k/t
. . .(1.184)
can describe a situation in which an oxide regularly fails when it reaches a critical thickness', or one in which the oxide volatilises' at a uniform rate. Similarly, the logarithmic rate law x = k,log( 1
+ kbt)
. . .(1.185)
has been shown to describe many differing situations, including rate control by chemisorption3, by oxide nucleation4, or by cavity production5. In certain circumstances even the parabolic rate law may be observed under conditions in which the oxide is porous and permeated by the oxidising environment6. In these cases it has been shown that it is diffusion of one or other of the reactants through the fluid phase which is rate controlling. More usually however the porous oxide is thought to grow on the surface of a lower oxide which is itself growing at a parabolic rate. The overall rate of growth is then said to be paralinear'**and may be described by the sum of linear and parabolic relationships (see equations 1.197 and 1.198). 1 :268
1 :269
DISCONTINUOUS OXIDE FILMS
It will be seen that although each system must be described on its merits there are general principles governing mass transport which can be used as a guide.
The Volume Change on Oxidation of a Metal The formation of discontinuities, particularly the grosser forms of pores and cracks, in an oxide layer is often attributable to the mass flows and volume changes occasioned by oxidation. As can be seen from Table 1.27 it is usual Table 1.27 Metals which form porous oxides in dry oxygen Linear oxidation
Paralinear oxidation
Type 1 Curve (Fig. 1.89)
Type 2 Curve (Fig. I.89)
Oxidelmetal volume ratio
Metal
MetaI
(0
(r)
Mg
0.81 0.64
Ca Nb (400')
2.49
Oxiakltnetal volume ratio
Ce
La Mo W
U Nb (450") Ta Ti Zr Hf Th
1.22
1.1 3.24 3-35 1 -94
2-49 2.54 1.73 1.56 1.60
1.35
Metal
~~~~~~
Oxide drift
:
Markers remain on surf ace
A
Zone of metal consumption
,fl/
Zone of oxide formation
Metal consumption
Oxide for mation
Fig. 1.81 Oxidation of flat surfaces. (a) When cations diffuse the initially formed oxide drifts towards the metal; (b)when anions diffuse the oxide drifts in the opposite direction
1 :270
DISCONTINUOUS OXIDE FILMS
for the volume of an oxide to differ greatly from that of the equivalent amount of metal. It is in fact remarkable that such a phase change can occur in so many cases of oxidation without oxide disruption. Moreover, oxidation reactions are rarely topotactic but take place by transport of one of the reactants to a plane of reaction. It is thus not only the net volume change which must be considered, but also the local volume changes due to oxide creation in some zones and metal consumption in othersg. Fortunately the oxidation of many metals takes place by the diffusion of the metal cation". This flux is outwards through the oxide layer, and the work of adhesion" enables the loss of metal to be compensated for by a drift of the oxide towards the metal (Fig. 1.81). Thus the stresses set up in the maintenance of oxide/metal contact are compressive and, as such, can be more readily withstood by most oxides. Nevertheless, it is these general movements of the oxide scale which are ultimately responsible for discontinuities in the majority of cases and it is appropriate to discuss transportinduced flows before proceeding any further.
Mass Transport in Growing Oxide Layers Oxide Drift
In the very early stages of oxidation the oxide layer is discontinuous; both kinetic and electron micro~cope'~-'~ studies have shown that oxidation commences by the lateral extension of discrete oxide nuclei. It is only once these interlace that the direction of mass transport becomes of importance. In the majority of cases the metal then diffuses across the oxide layer in the form of cations and electrons (cationic diffusion), or as with the heavy metal oxides, oxygen may diffuse as 0'-ions with a flow of electrons in the reverse direction (anionic diffusion). The number of metals oxidising by both cationic and anionic diffusion is believed to be small, since a favourable energy of activation for one ion generally means an unfavourable value for the other lo. A consequence of single-ion diffusion is that the mass movement must be compensated for by an opposing drift (relative to a fixed point deep in the metal) of the existing oxide layer if oxidation is not to be stifled by lack of one of the reactants. The effect may be illustrated by reference to a metal surface of infinite extent (Fig. 1.81). When cations are the diffusing species (Fig. 1.81a), metal is consumed either by solution in the oxide as interstitial cations and electrons M=M++e-
. . .(1.186)
or by reaction with metal ion vacancies in the oxide
M+M+O
+
o =null
. .(1.187)
where as is conventional the right-hand side of the equation represents the excess or deficiency of ions in the stoichiometric oxide, and M + O and o represent the vacant sites in the ionic lattice and positive holes in the full band, respectively. Since oxide is formed at the gadoxide interface, oxidemetal contact is only maintained by translation of the initial oxide through
DISCONTINUOUS OXIDE FILMS
1 :271
the volume occupied by that part of the metal which has been oxidised. In systems in which anionic diffusion prevails (Fig. 1.81b), metal is consumed by direct reaction to form the diffusing oxygen species
kf+ MO
+ 0 2 - 0+ 2e-
. . .(1.188)
The oxygen vacancies then diffuse to the gas interface where they are annihilated by reaction with adsorbed oxygen. The important point, however, is that metal is consumed and oxide formed in the same reaction zone. The oxide drift has thus only to accommodate the net volume difference between the metal and its equivalent amount of oxide. In theory this net volume change could represent an increase or a decrease in the volume of the system, but in practice all metal oxides in which anionic diffusion predominates have a lower metal density than that of the original metal. There is thus a net expansion and the oxide drift is away from the metal. Oxide movements are determined by the positioning of ‘inert’ markers on the surface of the oxide’”’’. At various intervals of time their position can be observed relative to, say, the centreline of the metal as seen in metallographic cross-section. In the case of cation diffusion the metal-interfacemarker distance remains constant and the marker moves towards the centreline; when the anion diffuses, the marker moves away from both the metal-oxide interface and the centreline of the metal. In the more usual observation the position of the marker is determined relative to the oxide/ gas interface. It can be appreciated from Fig. 1.81 that when anions diffuse the marker remains on the surface, but when cations move the marker translates at a rate equivalent to the total amount of new oxide formed. Bruckman l9 recently has re-emphasised the care that is necessary in the interpretation of marker movements in the oxidation of lower to higher oxides.
Oxidation of Non-planar Surfaces Oxide movements on plane surfaces, such as those just described, do not create stress; stress will arise however, when the oxide movement is constrained by the presence of a corner, or when the metal is curved, so that there is a progressive strain on the lateral dimensions of the oxide. Since oxides are brittle the appearance of tensional stress can be expected to lead to brittle failure; examples are given in Figs. 1.82 and 1.83. When convex surfaces oxidise by cationic diffusion compressive stresses occur within the oxide along planes parallel to the metal surfaces and reach a maximum at the metal/oxide interface. Frequently oxides are able to flow plastically under compression at moderate temperatures and the stress is relieved. Hales”, for example, has proved this to occur by the density of dislocations present in a nickel oxide layer. Sometimes however, relief is abrupt, giving rise to oxide distortions of the type illustrated in Fig. 1.82. An example of oxide buckling observed during the oxidation at 500°C of a Cu-1ONi alloy is shown in Fig. 1.84; features of this kind are not unusual and have been followed in the oxidation of pure iron using a hot stage and oxygen blanket in the scanning electron microscopez1. The compressive forces may be contained by adhesion and cohesion in the oxide-metal system; experiments designed to reveal the presence and
1 :272
DISCONTINUOUS OXIDE FILMS
(a)
Compressive failure
Metat
Direction of oxide drift
( b)
Void growth by vacancy condensation
Fig. 1.82 Oxidation of a convex surface by cation diffusion; the compressive stress in the initially formed oxide may lead t o (a) failure by buckling or to (b) void precipitation
Metal
__ Direction of oxide drift
Oxide failure under tension
Fig. 1.83 Oxidation of a convex surface by anion diffusion; the outward translation of the oxide gives tensile cracking in the initially formed oxide
DISCONTINUOUS OXIDE FILMS
1:273
magnitude of such stresses in continuous films will be described later. Anionic diffusion in the oxidation of a convex surface creates a situation which is the reverse of that just described. The oxide is in tension along planes parallel to the surface and fracture may be expected to occur readily in perpendicular directions and starting from the gadmetal interface. Although very thin films may have resistance to fracture22,thick films frequently acquire the morphology shown in Fig. 1.83. Concave surfaces are of industrial importance, in relation to the internal surface of bores, holes and pipes, but are not found on typical solid testpieces and have received much less discussion. The stress patterns will tend to be the opposite of those found on convex surfaces; for example, an oxide growing by cation diffusion should be in tension at the metal interface. Bruce and H a n ~ o c khave ~ ~ discussed the oxidation of curved surfaces and show how the time to adhesive failure of the oxide can be predicted if its mechanical properties are known.
Fig. 1.84 Surface of a Cu-1ONi alloy after oxidation in oxygen at 500OC. showing blistering, probably associated with CuO formation over voids at the metal/oxide interface (courtesy Central Electricity Research Laboratories)
The potential influence of shape on the correct design of laboratory testpieces has been discussed in detail by RomanskiU. Samples of iron in the form of discs, cylinders, plates or parallelepipeds, and of a wide range of areas, were sulphidisedunder controlled conditions. The parabolic rate constant could be expressed in terms of the area A of the samples by
k* = A / ( a + bA)
. . .(1.189)
at any given temperature. In this expression the constant a depended on the geometry of the specimen and hence on scale rigidity, and b on specimen purity, and thus probably on scale plasticity. The form of this equation was
1 :274
DISCONTINUOUS OXIDE FILMS
confirmed with the systems Ni-S, Cu-S and Cu-0. Romanski concluded that plate or disc specimens with an area of 30-40cm2 were required to obtain rates approaching the maximum for infinite surfaces. Oxide Drift and Pore Formation
Clearly geometric constraints have an important influence on the formation of oxide layers. There are, however, many reported example^^^-^^ of the development of porosity at the oxide-metal interface on plane surfaces well away from corners. This phenomenon is generally ascribed to the suitability of the interface as a sink for lattice vacancies generated during oxidation; the point was lucidly discussed by VermilyeaZ9.Voids of this type may range in size from a few nanometres, as described by BoggsZ6for the oxidation of tin, through the crypt-like cavities in the beautiful micrographs of Howes3', to the complete separation witnessed in the oxidation of iron by Pheil" and described in a paper which also recorded the first use of inert markers. Birchenall has also given interesting observations on the behaviour of large cavities32.Such porosity is more commonly associated with cation diffusion although Pemsler 33 has suggested that condensation of oxygen vacancies at the metal-oxide interface may be the cause of the voids observed during the oxidation of zirconium. The work of Cohen and his group at Ottawa concerning the oxidation of pure iron^^^"^ and more recently, of nickel" has made a notable contribution to this field. In particular, they have shown how the dislocation network of the underlying metal is effective in removing vacancies from the metal-oxide interface and thus preventing void formation. The effect is less marked in the case of nickel where the more important effect of cold work in the metal is to yield an oxide structure which is consistent with high diffusion rates. It seems likely that the major difference between the oxidation of these two metals lies in the plasticity of their oxide layers. Bruce and Hancock4' have detected repetitive oxide cracking during the oxidation of iron by use of a vibrating testpiece, whereas oxide cracking did not occur when nickel testpieces were used. This is in good accord with the behaviour of dislocations in nickel oxide during oxidation; Hales" has shown that these become optimised to permit maximum creep rates and moreover that the preferred orientation and columnar nature of the oxide also derives from this selection principle. The work is related very nicely to the analysis by Harris and Masters" of the work of metal/oxide adhesion. When voids do form the first formed oxide may lose contact with the retreating metal surface; the way is then open for the metal consumption zone9 to be filled with a secondary growth which is usually fine grained and porous". Often the inner layer appears to have formed from the inception of oxidation since the inner-outer layer interface retains the shape and dimensions of the original metal surface"-". More often, however, the balance between inner and outer layer foFmation alters from point to point, with the inner layer being favoured at corners and edges. On alloys the inner layer may consist of particles of the more noble metal oxide which, as in the case of the copper-nickel alloys45,provide a spaceframe supporting the outer layer. Duplex layers of this nature are sometimes seen on pure metals, the classic example being the case of nickel. Sartell and Li46have shown
DISCONTINUOUS OXIDE FILMS
1 :275
from diffraction measurements that the two layers on nickel differ in their lattice constants, the outer being in compression, They envisaged the outer oxide to grow by anionic diffusion and the inner by cation movement. It now seems likely that the inner layer grows in oxygen formed by decomposition of the outer layer; the case has been well argued by Bruckman’ and supported by the work of Douglass”. What is not so clear is how the outer layer on pure metals continues to grow since it demands the movement of both metal and oxygen across the inner layer when they already have sufficient chemical potential to react. The problem does not arise when alloys oxidise since the inner layer can be traversed at potentials below those at which the species react. Mrowec and Webber43have made numerous observations on systems of this kind and neatly summarise their findings in a discussion of a paper by Kofstad and Hed47on cobalt-chromium alloys. There are a number of factors which will trigger void formation during oxidation: Cohen and his c o - ~ o r k e r s ~have ~ - ~ ’shown that annealing of the metal (to remove dislocations) will do so, whilst and Wulf, Carter and Wallworka have shown that the presence of a continuous film of a more noble alloying element is just as effective. Tuck et aL4’ consider that the extra adhesion of the oxide on iron when this metal is heated at 950°C in oxygen containing water vapour is due to the beneficial effect of hydrogen on oxide plasticity. There are many similar observations in the literature.
The Influence of Voids on Oxidation Kinetics Voids as Diffusion Barriers
It was Evans’.’” who first suggested that cavities could act as diffusion barriers and derived the logarithmic rate law from the progressive nucleation of voids. Boggs”, and several ~ o - ~ o r k eproved r s ~ ~the~ Evans ~ ~ mechanism to apply to the oxidation of tin over the temperature range 150-220°C in a series of papers illustrating the combined use of electron microscopy and kinetic measurements. The logarithmic rate law also describes the oxidation of a number of other metals in the thin-film range, e.g. Mg’2, Cu53and NiS4,but there is as yet no evidence that the Evans mechanism applies to these cases. Possibly pores have not been sought with the same rigour as that used by Boggs. However, D o u g l a ~ sused ~ ~ a rearrangement of a method first used by Tylecote” to show that void condensation causes a three-fold reduction in the parabolic rate constant for a dilute chromium-nickel alloy at 800°C; he compared the rates of oxidation of two cylinders, one of them being pIugged at either end to exclude the atmosphere from the interior. Voids did not form at the metal-oxide interface on the plugged cylinder since the interior then acted as a sink for diffusing vacancies. Probably the most comprehensive measurements of the effect of voids on rates are those of C ~ h e n ~and ’ ~ his ~ school. They have published data on the oxidation of pure irons for a wide temperature range and for oxygen pressures ranging from 1 3 x N/m2 to 100 kN/mZ. The interactions between void formation and oxygen uptake are complex but only at presN/m2 do voids have no effect. Some of their results sures below 1 * 3 x are summarised in Fig. 1.85; over the pressure range 1 3 x lo-’ N/m2 to
-
1:276
DISCONTINUOUS OXIDE FILMS 0.8
0.6
-
1.3 50 c .-
Time (h)
O
m c
z
2 0.5
E
0.4
I
0.3
0.2 0.1
U Time ( h j
Fig. 1.85 Oxidation of high-purity iron in oxygen at differing pressures. All figures on curves are in N m-’. At 1-3 x to 1 e 3 x N/m* torr the rate is controlled by the impact of molecular oxygen; at 1 - 3 x lo-’ N/m2torr the initial rate of oxidation is sufficiently hi h to give void precipitation and the rate decreases with pressure increasing to 1 - 3 N/m ; at pressures greater than this the crack-heal mechanism becomes operative and the rate again increases with pressure (after Hussey and Cohen 38)
8
1 3 N/m’ the rate decreases as vacancies start arriving at the interface more rapidly than they can be assimilated. Hussey and C ~ h e n ~have ’ ? ~evidence ~ that the voids form in the Fe,O, layer before the outer Fe,O, is nucleated. However, this rapidly takes place once the supply of iron is stifled by the voids, oxide blisters and cracks, and oxygen ingress causes the rate to accelerate again. The curves indicate that blistering is a repetitive process and this is confirmed by micrographs from the hot-stage scanning electron microscope (Fig. 1.86). It is a salutary fact that the material showing the most protective behaviour (Fig. 1.85) has an oxide which is already partly separated from the metaI.
DISCONTINUOUS OXIDE FILMS
1:277
Fig. 1.86 'Stills' from a scanning electron microscope study by time-lapse photography of iron oxidation showing the results of the crack-heal mechanism. Left, 1 m m r 1 pm; right, 1 mm 'I0.5 pm (courtesy Central Electricity Research Laboratories)
Notwithstanding the large amount of work on pure iron and binary alloys, it remains difficult to translate the results to commercially useful steels. It is believed, on the one hand, that effusion of carbon monoxide can cause non-healing fissures in the scale", and on the other, that silicon creates selfhealing layers at the metal interface". The Importance of Voids in 'Short-circuit Diffusion
Several have suggested that in some systems voids, far from acting as diffusion barriers, may actually assist transport by permitting a dissociation-recombination mechanism. The presence of elements which could give rise to carrier molecules, e.g. carbon or hydrogen@* ',' and thus to the behaviour illustrated in Fig. 1.87, would particularly favour this mechanism. The oxidant side of the pore functions as a sink for vacancies diffusing from the oxide/gas interface by a reaction which yields gas of sufficiently high chemical potential to oxidise the metal side of the pore. The vacancies created by this reaction then travel to the metal/oxide interface where they are accommodated by plastic flow, or they may form additional voids by the mechanisms already discussed. The reaction sequence at the various interfaces (Fig. 1.87b) for the oxidation of iron (prior to the formation of Fe203)would be
20' = Fe304+ Fef 0 + 8e . . .(1.190) at B . . .(1.191) Fe304 3Fef + 0 + 8e + 4CO = 4C0, at C . . .(1.192) 4C0, = Fe304 3Fef + 0 + 8e + 4CO at D 3Fef + 0 + 8e + 3Fem,,,, = null . . .(1.193) Notice that oxide is utilised by the reaction at interface B at the same rate as it is formed at A, so that the void effectively moves through the growing oxide with the distanceAB remaining constant. It may be recalled that a truly at A
+
+
1 :278
DISCONTINUOUS OXIDE FILMS
inert marker placed at the point B in the continuous oxide would be expected to remain fixed with reference to the point D (the metal-oxide interface). The distinction is important since voids involved in the mechanism in this way remain in a string across the oxide and can thus co-operate to move the oxidant towards the metal at a high ratew. What is not so easy to explain is why, when this mechanism is thought to apply, the oxide/metal and metal/ oxide interfaces remain flat; the implication is that the abnormally high flux carried by the porous regions redistributes, at the gadoxide interface, to the benefit of the whole surface. G i b b has ~ ~ shown ~ that as long as surface diffusion is fast this mechanism gives rise to an altered parabolic rate law. Smeltzer%,however, has argued that short-circuited routes may be progressively lost as the oxide increases in thickness; thus there is a transition from a short-circuit diffusion process at short times to the usual parabolic dependence at long times.
1
Fe+ot
Oxide
Metal
I01
A
IF 2 ;
Oxide
+ CIF4; D
Metal [ 6)
Fig. 1.87 Voids at an oxide/metal interface. They may grow by (a) condensation of vacancies from the metal as well as the local oxide; they impede transport as shown in Fig. 1.85. However, voids which become filled with a carrier gas (b) may act as short-circuit paths by the reactions at A , B , C and D,given in the text (after Birks and Rickerts8). Alternatively (c) gaseous by-products of oxidation may maintain fissures in the oxide (after Boggs and K a ~ h i k ~ ~ )
All voids described so far have been formed at, or released from, the metal-oxide interface. Birchena16’ has discussed the formation and growth of voids within the oxide scale by condensation of vacancies. For this to occur anions must be removed in some way and he has suggested creep and slip in the anion lattice as possible mechanisms. More recently, Coxm has used a sensitive porosimeter as well as special metallographic techniques to study strings of voids formed in zirconium oxide as a result of recrystallisation-the work is discussed in more detail below. Although Birchenal’s niodel predicted greater deviations from the parabolic rate law than were in fact found, the phenomenon of pore growth within the scale seems to be real and it is as well to keep the implications for anion movement in mind.
DISCONTINUOUS OXIDE FILMS
1 :279
Oxide Cracking and Void Cavities As the area of individual cavities at the metal-oxide interface increases there is an increased probability that the oxide will crack and permit oxygen access to a large area of unprotected metal. Such cracking is then reflected in the overall kinetics and the rate curve takes the form shown by the curves in Fig. 1.85 for pressures greater than 1* 3 kN/mZ; it is known to occur with c ~ p p e r ~iron3’ . ~ ~ and , iron-chromium al10ys’~~’’over certain temperature ranges. The oxide formed has become known as a Pfeil-type porous oxide3’. When failure occurs rather infrequently the rate curve appears as a succession of parabolas, each having as origin the critical values of time and thickness (t, and x,) at which the previous film failed. The net rate of oxidation is thus close to that given by the linear rate law with the value of the constant k, equal to x,/?,, or
. . .(1.194) k, = (x,/k,)f In some cases the number of oxide layers can be related directly to the number of breaks in the curve and there is then no doubt that the acceleration derives from repetitive stress-induced oxide cracking. Growth Laws of Oxidation Oxidation at a Linear Rate
Metals which oxidise at a linear rate can follow two types of oxidation curve; there are those (Type 1) for which the transition to the linear rate is far more abrupt and irreversible than that associated with Pfeil-type behaviour. The oxidation curve (Fig. 1.88) shows that in the region of the point A , i.e. at ‘breakaway’, the rate alters from a value which is initially very small to one which is both large and which no longer decreases with time, e.g. Cathcart et al.” have shown that sodium will oxidise protectively in pure oxygen, forming a film with a limiting thickness of about l5Onm at 2 5 ° C whereas in air breakaway occurs and a crust many centimetres thick may be f ~ r m e d ’ ~The . situation is similar with magnesium; at 575OC, and prior to breakaway, the protective oxide is transparent although after this event the rate may be so large that the metal takes fire’. Not all rate transitions are as large in magnitude; that of steel in the C02 coolant of nuclear reactors , this, because of the inaccessileads to a linear rate of about 25 ~ m / y ’ ~and bility of the steel, has necessitated temperature reductions in operating reactors, consequently with a considerable loss in power output. The second type of behaviour (Fig. 1.89) is much closer to that which one might predict from the regular cracking of successive oxide layers, i.e. the rate decreases to a constant value. Often the oxide-metal volume ratio (Table 1.27) is much greater than unity, and oxidation occurs by oxygen transport in the continuous oxide; in some examples the data can be fitted by the paralinear rate law, which is considered later. Destructive oxidation of this type is shown by many metals such as molybdenum, tungsten and tantalum which would otherwise have excellent properties for use at high temperatures.
1 :280
DISCONTINUOUS OXIDE FILMS
Time (h) Fig. 1.88 Early stages of oxidation of magnesium at 525"C, but at a lower pressure of 13 kN/m2 than the example in Fig. 1.89 illustrating breakaway
Fig. 1.89 Oxidation of magnesium at 500°C illustrating the increase in rate to the constant value (Type 1) (after Gregg and Jepson') and the oxidation of tungsten at 7W°C (after Webb, Norton and Wagner") illustrating the decrease in rate to the constant value (Type 2 )
Probably the only feature common to the mechanism of oxidation of the two groups is that, because of crack or pore formation in the continuous oxide, the rate of transport of oxygen in a molecular form has increased to the point where a phase-boundary reaction has assumed rate control. In
DISCONTINUOUS OXIDE FILMS
1:281
accord with this interpretation there is frequently a marked reduction in the activation energy for oxidation when non-protective oxidation commences. The mechanism and the theories of linear oxidation must be discussed with reference to specific examples. This will shortly be done, but it will be helpful to return first of all to the theories of Pilling and Bedworth. The Pillling-Bedworth Theory
Pilling and Bedworth6*made the earliest attempt to classify and interpret the oxidation behaviour of metals, and since they were working before any marker experiments had been carried out, they assumed that the predominating transport was always of oxygen to the metal. They consequently argued that the oxide layer would be continuous if its volume was greater than that of the metal consumed by its production. Conversely, if the oxide was less in volume than that of the equivalent mass of metal, the layer would be cellular and porous so that gaseous oxygen would reach the metal surface. As experimental evidence has accumulated it has become clear that this simple rule does not fit the facts; and understandably so since the original premise concerning oxygen transport is not generally true. In particular it is now known that all the metals with oxide: metal volume ratios of less than unity, e.g. Na,O: Na = 0-5572,MgO: Mg = 0-815’and CaO: Ca = 0-6475, form oxides which remain thin and continuous indefinitely at temperatures below critical values t, which are characteristic of the metals and which are often close to their melting points (for Na, t, = 48°C; for Mg, t, = 550OC). As Table 1.25 shows, the oxide ratios of metals which do exhibit linear oxidation of one form or another cover the whole range of possible values so that the rule gives no guidance at all in this respect. In some circumstances the rule is found to be in accord with the sign of the stress set up in the metal during oxidation so that some authors have felt that some residual anion diffusion must occur, notwithstanding the relative values of self-diffusion coefficients. This point will be returned to later. The Mechanism of Breakaway Corrosion
Over the years, breakaway has become very strongly associated with stressinduced oxide cracking, especially following the work of Pilling and Bedworth, but such a proposition is unwarranted as a generalisation and is difficult to prove for any specific case. The only general feature of breakaway is the very fine state of subdivision of the porous This is usually beyond the resolving power of even the scanning electron micro~cope~~. The coarse cracks which are seen in many optical micrographs are almost certainly secondary features associated with the proximity of the reaction zone to the metal-oxide interface and the subsequent drift of the oxide scale. The factors suggesting that the development of fine-grained porosity in oxides is not merely a stress-assisted reticulation type of failure are (a) the long induction periods, when stresses should be relieved rather than developed; (b)the major r6le of gas phase impurities (examples are water in the case of and beryllium78,hydrocarbons with magnesium5’and CO
1 :282
DISCONTINUOUS OXIDE FILMS
with steel in CO,"); (c) the promotion of breakaway of steel" in hightemperature water or CO,", and of zirconiuma0by layers of platinum; and (d) the curious fact that metals which happily formed a protective oxide when they first contacted oxygen refused to do so when the process was repeated. Much of the difficultyin demonstrating the mechanism of breakaway in a particular case arises from the thinness of the reaction zone and its location at the metal-oxide interface. Workers must consider (a) whether the oxide is cracked or merely recrystallised"; (b) whether the oxide now results from direct molecular reaction, or whether a barrier layer remainsa; (c) whether the inception of a side reaction (e.g. 2CO -+ CO, + C)" caused failure; or ( d ) whether a new transport process, chemical transport or volatilisationa3, has become possible. In developing these mechanisms both arguments and experimental technique require considerable sophistication. As a few examples one may cite the use of density and specific surface-area measurements as r ~ u t i n e ' ~of; porosimetry by a variety of methodsa; of optical electron microscopy" and X-ray diffractions6 at reaction temperature; of tracera, electric field *' and stress measurements. Excellent metallographic sectioning is taken for granted in this field of research. As has been intimated certain breakaway reactions are of great technological importance and a correspondingly large amount of research has been carried out on these, but as yet no consensus has obtained for the mechanism of linear oxidation of Type 1 (Fig. 1.89) for any one of the metals. The papers of COX^*^' on the oxidation of zirconium and its alloys are, however, well worth study; the work included the development of a mercury porosimeter sensitive to pores of about 10 nm in size and the investigation of electron transport in the pre-breakaway oxide layer. Cox concludes that it is electron transport which is rate controlling in the early stages and that breakaway is the recrystallisation of the oxide, induced by a tensional stress, which creates continuous porosity" by void condensation at grain boundaries. Intermetallic particles, which in zirconium alloys are associated with easy routes for electron transport in the pre-breakaway film, appear to be located at the base of pores after breakaway. The porosimetry measurements indicate that the pores have a diameter of less than 10 nm and this correlates with the size of pore-like features seen in replicas by transmission electron microscopy. The model received additional support from hot-stage X-ray diffractions6which showed recrystallisation to a columnar structure to be concurrent with the rate transition. Bradhurst and Heuer8' prefer to turn the mechanism around, arguing that recrystallisation occurs as a result of cracking in the traditional way; they have confirmed that there is stress relief in the oxide-metal system at the time of breakaway and have evidence from the hot stage microscope of crack-like features moving across the surface. The viewpoints of both .schools are summarised in letters to the Journal of Nuclear The discovery by Fiegna and WeisgerberaOthat noble metals are able to catalyse the breakaway corrosion of zirconium has not been built into either of the main theories. Antill et af." have also found it difficult to explain their similar observation for the oxidation of steel by CO,. Reactor grade CO, contains both water and CO as impurities; CO is also produced by the reaction
DISCONTINUOUS OXIDE FILMS
+
1 :283
+
3Fe 4C0, = Fe,O, 4CO . . .(1.195) and Antill's thesis is that it is the disproportionation of the carbon monoxide (the Boudouard reaction)
+
. . .(l.l%) 2 c o = co, c which causes fragmentation of the protective oxide. However, in this system the action of the noble metal appears to be associated with the water impurity. Water as an impurity is known to promote the breakaway corrosion of a number of metals; in addition to iron in CO, the effect has been reported for magnesium (hydrocarbons have more effect on the oxidation of this metal), beryllium, zirconium and sodium. In the latter case water is known to convert the oxide to deliquescent NaOH but acceleration of beryllium oxidation probably results from hydride formation and mechanical damage to the oxide. The Mechanisms of Paralinear Oxidation
Some metals oxidise at a rate which decreases, rather than increases (Type 2 in Fig. 1.89). Cerium behaves in this fashion at temperatures between 40°C and 130°C, and Loriers7*' has suggested that the curve derives from the competition between the two oxides Ce,O, and CeO,. It was proposed that the inner layer Ce,O, was continuous and grew under diffusion control but transformed at a constant rate to an outer layer of CeO,. That is, if we writey and z as the thickness of the inner and outer layers respectively, then
0
Time
Fig. 1.90 Kinetic interpertation of paralinear oxidation. Curves u and b correspond to the growth of the inner compact layer and the outer porous layer, respectively;curve c represents the total weight and is the algebraic sum of curves a and b. Note that as oxidation proceeds, y tends to a limiting value ymm,(curve u) and the overall rate of oxidation tends to a constant value fb
1:284
DISCONTINUOUS OXIDE FILMS
dy/dt = (a/y) - b
. . .(1.197)
and &/dt
=fa
. . .(1.198)
wherefis the ratio of the oxygen content per gramme atom in the outer layer to that in the inner layer, a is one half the parabolic rate constant for diffusion in the inner layer, andfb is the linear rate constant. The integrated forms of equations 1.197 and 1.198 are illustrated in Fig. 1.90 together with that for the total weight gain w which is given by the sum of y and z . It is evident that as oxidation proceeds the rate of thickening of the inner layer progressively decreases and its thickness tends to a limiting value, i.e.
. . .(1.199) = a/b so that the overall rate of oxidation tends to the constant value Ymax.
dw/dt = j3
. . .(1.200)
The particular interest in this form of oxidation stems from the fact that the important group of metals Nb, Ta, Mo and W show similar behaviour although it is only with tungsten that the kinetics are strictly paralinear. The model was first applied to the oxidation of tungsten by Webb, Norton and Wagnerss for the temperature range 700-1OOO”C. They were able to demonstrate the presence of a barrier film (probably a metastable modification of one of the intermediate oxide phases of tungsten such as W4O1,) between the metal and the outer layer of porous tungstic oxide. Subsequently Jepson and AylmoreS9measured the amount of porous oxide formed on tungsten oxidised at temperatures in the range 750-800°C by krypton sorption and by metallographic methods. They discovered that the porous oxide did not form at a constant rate as required by the paralinear model; instead its rate of formation decreased with time with kinetics similar to those of weight gain vs. time. They concluded that the weight of combined oxygen in the barrier layer must be very small. Kellet and Rodgersw have since thrown even more doubt on the concept of a barrier layer by their finding that the black oxide ‘barrier layer’ and the yellow porous oxide have the same chemical composition.
Oxidationin the Presence of Subscales
The aforementioned inconsistencies between the paralinear model and actual observations point to the possibility that there is a different mechanism altogether. The common feature of these metals, and their distinction from cerium, is their facility for dissolvingoxygen. The relationship between this process and an oxidation rate which changes from parabolic to a linear value was first established by Wallwork and Jenkins” from work on the oxidation of titanium. These authors were able to determine the oxygen distribution in the metal phase by microhardness traverses across metallographic sections; comparison of the results with the oxidation kinetics showed that the rate became linear when the metal surface reached oxygen
DISCONTINUOUS OXIDE FILMS
1:285
saturation, at a composition of about Ti,O. It was thought that the porous layer of TiO, was formed by exfoliation of metal layers but remained in good contact since (a) when the oxygen demand of the metal fell, cations diffused into, and discoloured, the oxide and (b)if the oxidising atmosphere was removed, the oxide redissolved. The general thesis of this work was supported by the work of Osthagen and K o f ~ t a dwith ~ ~zirconium in oxygen at 800"C, who found that the rate became linear at a surface composition of Zr,O, and by that of Smeltzer et al. The former authors related the rate change to the poor influence of the suboxide (which is believed to be volatile) on the interface adhesion. Pemsler%, in an important series of papers, has developed the theme of oxidation of prior-formed zirconium-oxygen alloys but, curiously, was unable to reproduce the linear portion of the oxidation curve. He found the rate to remain parabolic at all temperatures up to 1300°Cand with a rational rate constant which was independent of the degree of oxygen saturation. Pemsler developed a sensitive technique for the determination of the oxygen profile in the metal and showed ordered alloys to exist at the compositions ZrO,, where x takes, in turn, the values 0.16, 0.21, 0.26 and 0.32. Kofstad9' had earlier suggested that the initial protective period in the oxidation of niobium (this metal belongs to the group which shows Type 1 behaviour) is to be interpreted in terms of the dissolution of oxygen and the formation of a suboxide. Breakaway then corresponds to the nucleation of the pentoxide Nb205 on the surface. Cathcart% and his co-workers observed blister-like cracks in the oxide at breakaway, which is to be expected since NbzO, grows by anion diffusion and has an oxide: metal volume ratio of 2.49:l. Later work has confirmed this and it thus seems possible that there is no barrier layer, merely an interaction with the underlying metal9', at least for certain conditions of temperature and oxygen pressure. Theories of the oxidation of tantalum in the presence of suboxide have been developed by Stringer9*.By means of single-crystal studies he has been able to show that a rate anisotropy stems from the orientation of the suboxide which is precipitated in the form of thin plates. Their influence on the oxidation rate is least when they lie parallel to the metal interface, since the stresses set up by their oxidation to the pentoxide are most easily accommodated. By contrast, when the plates are at 45" to the surface, complex stresses are established which create characteristic chevron markings and cracks in the oxide. The cracks in this case follow lines of pores generated by oxidation of the plates. This behaviour is also found with niobium, but surprisingly, these pores are not formed when Ta-Nb alloys are oxidisedw, and the rate anisotropy disappears. However, the rate remains linear; it seems that this is another case in which molecular oxygen travels by submicroscopic routes.
',.
The Role of Metal Dissolution or Volatilisation in the Formation of Porous Oxides As we have seen, a consequence of the formation of porous oxide is that the rate-controlling step reverts to that of a phase boundary reaction and
1:286
DISCONTINUOUS OXIDE FILMS
therefore becomes independent of the oxide thickness. When these circumstances are such that the porous oxide becomes unusually thick, or if the oxidising medium is unusualIy dense, a form of the parabolic rate law may be re-established. In this case the relevant diffusion coefficient is that of the transported component in the fluid phase permeating the oxide layer. Perhaps the best known system in which the apparently paradoxical association of porous oxide formation with parabolic kinetics is observed is that in which iron or mild steel reacts with water or alkaline solutions at temperatures within the region of 300°C. At this temperature water acts as an oxidising agent, even in the absence of electrochemical coupling, and very hard compact layers of oxide are formed. This oxide scale is found to consist of two layers, both magnetite, and marker experiments have been used to establish the important fact that the inner layer occupies exactly the volume of the metal consumed by oxidation'". It follows from these observations that the inner layer of oxide supports counter diffusing and nearly equivalent fluxes of iron and oxygen and, by inference from the observed parabolic rate law, one of these is rate controlling. Potter'" considered the inner layer to be continuous and to grow by oxygen ion diffusion; however the rate of oxygen transport during corrosion is 10' times greater than would be expected to occur by diffusion in a continuous oxide film, and it is difficult to account for the equivalence of the ion fluxes over a wide range of temperature and solution composition'02. The fact that the outer layer was porous was known to Potter and Mann, but the possibility that the inner layer was porous was first discussed by Field et aZ.103who adduced evidence that the oxide had only 90% of its bulk theoretical density. It was subsequently shown that the inner layer consists of individual crystallites of 0-1-0-2pm in szie@ '' and that the porosity between them was interconnected'. The knowledge that molecular water penetrated to the reaction zone close to the metal surface enabled Castle and Masterson6 to construct a model in which the dissolution of the metal matrix represented a competing reaction with the growth of oxide nuclei. When dissolution is sufficiently fast the growing oxide nuclei can be undermined before a continuous stable film of oxide is formed. Since dissolution will also expose new sites for oxide nucleation the process can be repeated indefinitely so long as the resolved iron is removed from the vicinity of the metal surface. It is this efflux of iron in soluble form from the metal surface which becomes the rate-controlling transport reaction during the oxidation of steel in high-temperature water. The model is useful since it correctly predicts the dependence of the reaction rate on the solution pH63LoS. This model, or variants'06 of it, is also able to explain the behaviour of aluminium in high-temperature water'", of steels in molten salts'08and of nonferrous metals and alloys in low-temperature aqueous solutions '09, "O. There is also evidence that the transport of metals in volatile form, across similarly porous oxide, may be an important feature of oxidation in steamE3and in special circumstances where the vapour pressure of the metal is high". The more important cases of oxide volatilisation occur in the platinum metals'" and with the refractory metals' at high temperatures. In these systems, unlike the aforementioned, it is the higher valence oxide which is the more volatile so that at sufficiently high temperature the metal may be oxide free. Gulbransen' has shown that the rate of oxidation is then con-
DISCONTINUOUS OXIDE FILMS
1:287
stant and agrees well with that derived from the kinetic theory of gases and the relevant thermochemical data.
Stresses in Oxide Layers Much of the earlier part of this chapter has dealt with the observable effects of stress in an oxide layer. Oxide buckling or even failure, recrystallisation, the promotion of columnar grain growth or whiskers, and, above all else, the creation of non-protective oxide, have all been attributed to stress. It has been shown (page 1.270) how mass transport in oxides leads to oxide drift and thus, on finite and non-planar surfaces, to a change of shape with consequent generation of stress. Other mechanisms of stress generation are: epitaxial strain, Le. the development of oxide layers with an orientation which permits some correspondence between the lattice parameter of the metal sublattice in the oxide and that of the metal; oxide formation in cracks or grain boundaries; the formation of higher oxides; oxygen solution in the metal, with or without subscale formation; and shrinkage of the metal by assimilation of vacancies from the metal oxide interface. The experimental evidence relating to these forms of stress generation has been reviewed by Stringer"' in an article which draws together work from wide-ranging fields. In order to understand the more catastrophic effects of oxide stress its magnitude must be determined in oxide layers which are still continuous. Several methods have been evolved to permit this measurement including low-energy electron diffraction (L.E.E.D.)Il3 for the very early stages of oxide growth, X-ray diffraction-line br~adening"~ and techniques in which the strain developed by the oxide layer when grown under conditions which permit stress relief is measured by bending or extension of the metal substrate 112,1 IS, 116. This latter class encompasses a variety of geometrically designed metal testpieces. The method has become known as the flexure or Stoney "'method and is worthy of special comment particularly since it is the method most widely adopted for measurements on thick film. The Flexure Method
The method makes use of the tendency of a metal foil to bend when it is oxidised unilaterally, Le. on one side only, and has been developed from the method used by Stoney to measure stresses in electrodeposits, as long ago as 1909. There are important problems in the translation of the technique to oxidation studies. Firstly, it is difficult to completely arrest oxidation on the 'inert' side of the testpiece. The problems of diffusion across a barrier layer of electrodeposited or vapour-deposited metal for a long time restricted its use to low temperatures, e.g. Ta at 35O-55O0C"*, Nb at 425°C"s and Cu at 200-400"C, but recently Pawel and Cathcart have reported the use of AI-3Au evaporated layers which, on uranium alloys, will enable temperatures of up to 800°C to be used'''. Secondly, there is an interpretive difficulty since although there are two stress systems, acting at right angles in the plane of the oxide, a strip (or even a disc) generally bends in a plane perpendicular to only one of them"'.
1 :288
DISCONTINUOUS OXIDE FILMS
Because of the experimental difficulty other workers have circumvented the problem in various ways. In one of the first demonstrations of stress, Evans 12' examined the flexure of detached oxides at room temperature whilst soon after this Dankov and Churaev'U examined the stresses present at reaction temperature by oxidation of layers of the metals evaporated onto mica substrates. Their technique has been criticised2' on the grounds that the major stresses would refer to oxide formation in the pores of the evaporated layer -and also that the thin films oxidised may unduly reflect epitaxial stresses. Engell and Wever'*' oxidised both sides of a spiral of iron at 700°C and relied on the different lengths of the inward- and outward-facing surfaces of the helix to provide a differential force. This technique would be unduly influenced by rate differences on either side of the helix-especially as derived from the curvature itself 'I2. Bradhurst and Heuer 124 also oxidised both sides of the testpiece and obtained their results from the change in curvature at room temperature when the oxide was removed from one face. The method was used to determine stresses in the oxide on Zircalloy 2 and zirconium at 500-700'C and is successful only if the oxide does not spa11 on cooling and if the correction for differential thermal expansion can be experimentally determined. Bradhurst and Heuer were successful and showed that stresses steadily increased until relieved on the Zircaltoy 2 at the point of breakaway. Yet another technique has been utilised by Appleby and T y l e c ~ t e who ' ~ ~ protected one side of disc-shaped specimens by reducing atmospheres while the upper side was oxidised at temperatures up to 950°C. The interpretive difficulty has been discussed in detail by Pawel and by Morton'" in papers which appeared almost simultaneously. Both authors use arguments to show that the simple formula of bending-beam theory utilised by Stoney
Et2 6rd
a=-
.. .(1.201)
where E is Young's modulus, t is the thickness of metal, r the radius of curvature, d the oxide thickness and u the stress, is suitable only as a first approximation. Since there is an isotropic growth stress in the plane of the oxide it is necessary to consider the two principal stresses ax and uy given by
. . .(1.202) for biaxial plane stress in a plate, and uy =
+4 1 - v2
. . .(1.203)
where r, and are the strains, x is the long axis and y the short axis of the testpiece, respectively, and v is the Poisson's ratio. If the testpiece is free to bend in both directions under these stresses then E, = ~y and
Et2 . . .(1.204) ( 1 - v)6rd More usually however the initiation of bending on a plane perpendicular to y (say) and through x will render the testpiece sufficiently rigid to preclude ax =
DISCONTINUOUS OXIDE FILMS
1 :289
bending in the orthogonal plane through y . As Morton points out this mode of deformation must inevitably occur in the helical testpieces used in the spiral contractometer. An appropriate formula is then ry = 0 and a, =
Et2 (1
- v2)6rd
. .(1.205)
although Timoshenko's "'early analysis shows that it is this relationship which reverts to the Stoney formula for suitably shaped specimens. The rigidity of the y axis prevents the development of spherical surfaces for all but very small displacements. Morton suggests that the limit is reached when the displacement is equal to the metal thickness. This condition was satisfied in the high-temperature studies of Appleby and Tylecote lZ5 and spherical doming of the disc specimen occurred. When the oxide is not very thin compared with the metal both the moduli for oxide and metal must be considered. Stringer 'I2, in his excellent review of stress generation and relief in oxide layers, quotes a corrected formula, originally due to Brenner and Senderoff lZ8
. .(1.206) (which omits the Poisson's ratio correction) and quotes data for magnesium which show that the simple formula would be in error by more than 50% when the oxide layer had a thickness of 10% of that of the metal. It is perhaps apposite to remind the reader that these corrections are pertinent to films which are very thick compared with those formed at breakaway in the Type 1 oxidation. Magnesium undergoes breakaway when the oxide has maintained a stable thickness of about 5 x lO-'m for several hundred hours. Either on the Pilling-Bedworth model or the oxide drift argument the film would contain a biaxial stress of the order of only about 2 MN/m2 tension or compression even if there has been no stress relief by plastic flow. The oxide should be well able to support this. In fact it is occasionally possible to evaporate magnesium leaving a near-perfect box of transparent oxide which shows no sign whatsoever of having been in strained conditions -yet eventually such magnesium will go into breakaway. As reliable stress measurements become extended to greater temperatures the extent to which growth stresses are dissipated by plastic flow of the oxide becomes more apparent -the values of around 700 MN/m2 measured for the oxidation of niobium '1**129 and ZircaIloy 2'" must be contrasted with the value of 70MN/m2 (no breakaway) found for the oxidation of zirconium at 700-900"C'" and 869 kN/m* for copper at this same temperaturelZ. The stress measured for the formation of cuprous oxide at 900°C was only 371 kN/mZ('Z5).The very low stress calculated for the oxidation of copper at 700°C is, however, surprising since the crack-heal mechanism of oxidation is known to be operative at slightly lower temperatures, and also in view of the direct relation between stress in the oxide scale and the formation of whiskersL3'. They are, however, in reasonable accord with the X-ray strain values measured by Homma and Issike for the oxidation of copper at 500 and 600°C and the very high values are undoubtedly associated with oxygen solution and oxide wedging in the metal.
1 :290
DISCONTINUOUS OXIDE FILMS
Stress Relief by Metal Creep
The fact that uniaxially oxidised metals bend suggests that thin foils should stretch when biaxially oxidised. This has been frequently observed and moreover thin-walled tubing may both decrease in diameter ‘32 and increase in under the compressive loading of oxidation. Such a loading on the metal may seriously reduce its effective tensile ~trength’~’ without relieving much of the compressive stress in the oxide, and introduces a dependence of creep behaviour on the oxygen potential of an environment’36.Although outside the scope of this section, the paper on anodic oxidation of loaded aluminium wires by Leach and Ne~feld’~’ gives an indication of the probable depth of this field, in which little work has yet been carried out.
J.E. CASTLE REFERENCES Gregg, S. J. and Jepson, W. B., J. Znsr. Met., 87, 187 (1958-59) Gulbransen. E. A., Corrosion, 26, 1 (1970) Landsberg, P. T., J. Chem. Phys., 23, 1079 (1955) Vernon, W., Akeroyd, E. and Stroud. E., J. Znst. Met., 65, 301 (1939) Evans, U. R., Trans. Electrochem. SOC.,91, 547 (1947) Castle, J. E. and Masterson, H. G.. Corros. Sci., 6, 93 (1966) Lorriers, J., C.R. Acad, Sci., Park., 229, 547 (1949) Lorriers, J., C.R. Acad. Sci., Paris, 231, 522 (1950) Bruckman, A , , Corros. Sci., 7 , 51 (1967) Harrop. P. J., J. Mats. Sci., 3. 206 (1968) Masters, B. C. and Harris, J. E., Proc. Roy. SOC.,292A,240 (1966) Boggs. W. E.,Kachik, R. H. and Pellissier, G. E., J. Electrochem. SOC., 112,539 (1965) Taylor, M. E., Holmes, E. and Boden, P. J., Corros. Sci., 9,683 (1969) Pignocco. A. J. and Pellissier. G. E., J. Electrochem. SOC.,112, 1188 (1965) Uhlig, H. H., Corros. Sci., 7,325 (1967) Sacks, K.,Mefallurgia. 54. 11 (1956) Mrowec, S. and Weber, T., Acta Met., 8, 819 (1960) Jorgenson. P. J., J. Chem. Phys., 31, 874 (1962) Bruckman, A. and Simkovich, G., Corrcs. Sci., 12, 595 (1972) 20. Hales, R., Corros. Sei., 12, 555 (1972) 21. Castle, J. E. and Hurit, M. R., to be published in Corros. Sci. 22. Pashley, D. W., Advances Phys., 5, 173 (1956) 23. Bruce, D. and Hancock, P., J. Iron and Steel Znsr., 208, 1021 (1970) 24. Romanski, J., Corros. Sci.. 8, 67 (2 papers) (1968) 25. Tylecote, R. F., J. Iron Steel Znst., 195, 380 (1960) 26. Boggs. W.E., Kachick, R. H. and Pellisier, G. E., J, Electrochem. SOC., 1086 (1961) 27. Boggs, W. E., Trozzo, P. S. and Pellisier, G. E.,ibid., 13 (1961) 28. Caplan, D. and Cohen, M., Corrosion, 15, 141t (1969) 29. Vermilyea, D. A., Acta Mef., 5 , 492 (1957) 30. Howes, V. R.. Corros. Sci., 10. 99 (1970) 31. Pfeil, L. B., J. Iron and Steel Inst., 119, 501 (1929) 32. Juenker, D. W.. Meussener. R. A. and Birchenall, C. E., Corrosion, 14, 57 (1958) 33. Pemsler, J., J. Electrochem. SOC.,112,477 (1965) 34. Caplan, D. and Cohen, M., Corros. Sci.. 6, 521 (1966) 35. Caplan, D. and Cohen, M., Corros. Sci., 7, 725 (1967) 36. Caplan, D., Graham, M. J. and Cohen, M.. Corros. Sci., 10, 1 (1970) 37. Hussey, R. J. and Cohen, M. J., Corros. Sci., 11, 699 (1971) 38. Hussey, R. J. and Cohen, M. J., Corros. Sci., 11. 713 (1971) 39. Caplan, D . , Sproule, G . I. and Hussey, R. J., Corros. Sci., 10,9 (1971) 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.
DISCONTINUOUS OXIDE FILMS
1:291
40. Caplan, D., Graham, M. J. and Cohen, M., J. Electrochem. Soc., 119, 1205 (1972) 41. Bruce, D. and Hancock, P., J. Inst. Met., 97. 140 (1969) 42. Douglass, D. L., Corros. Sci., 8, 665 (1%8) 43. Mrowec, S. and Weber, T., J. Electrochem. SOC., 117, 1531 (1970) 44. Antill, J. E., Peakall, K. A. and Warburton, J. B., Corros. Sci.. 8, 689 (1%8) 45. Whittle, D. P. and Wood, G. G., Corros. Sci., 8, 295 (1968) 46. Sartell. J. A. and Li, C. H..J. Inst. Met., 90, 92 (1961) 47. Kofstad, P. N. and Hed, A. Z . , J. Electrochem. Soc., 116, 1542 (1969) 48. Wulf, G. L., Carter, J. J. and Wallwork, G. R., Corros. Sci., 9, 689 (1969) 49. Tuck, C. W., Odgers, M. and Sacks, K., Corros. Sci., 9, 271 (1969) 50. Evans, U. R., The Corrosion and Oxidation of Metals, Arnold, London (1961) 51. Boggs, W. E., J. Electrochem. SOC., 108, 124 (1961) 52. Castle, J. E., Gregg, S. J. and Jepson, W. B., J. Electrochem. SOC., 109, 1018 (1962) 53. Uhlig, H. H., Acta Met., 4, 541 (1956) 54. Uhlig, H. H., Pickett, J. and MacNairn, J., Acta Met., 7, 111 (1959) 55. Tylecote, R. F. and Michett, T. E., J. Iron and Steel Inst., 1%, 445 (1960) 56. Boggs, W. E. and Kachik, R. H., J. Electrochem. Soc.,116, 424 (1%9) 57. Wood, G. C., Richardson, J. A., Hobby, M. C. and Bouttead, J., Corros. Sci., 9, 655 ( 1969) 58. Birks, N. and Rickert, H., J. Inst. Met., 91, 30 (1962) 59. Wood, G. C., Wright, J. G. and Fergusson, J. M., Corros. Sci., 5 , 645 (1965) 60. Fuji, C. T. and Meussner, R . A., J. Electrochem. Soc., 114,435 (1967) 61. Mrowec, S., Corros. Sci., 7, 563 (1967) 62. Antill, J. E., Peakall, K. A. and Warburton, J. B., Corros. Sci., 8. 689 (1968) 63. Birks, N., British Corros. J . , 3, 56 (1968) 64. Cox, B., J. NUC.Mats., 27, 1 (1968) 65. Gibbs, G. B., Corrosion Sci., 7, 165 (1%7) 66. Smeltzer, W. W., Haering, R. R. and Kirkaldy, J. S., Acta. Met., 9, 880 (1961) 67. Birched. C. E., J. Electrochem. Soc.. 103, 619 (1956) 68. Pilling, N. B. and Bedworth, R. E.,J. Inst. Met., 29, 529 (1923) 69. Tylecote, R. F., J. Inst. Met., 81, 681 (1953) 70. Caplan, D. and Cohen, M.,J. Metals, Trans. Amer. Inst. Min. (Metall.) Engrs., 203,336 (1955) 71. Mortimer, D. and Post, M. L., Corros. Sci., 8, 499 (1968) 72. Cathcart, J. V., Hall, C. C. and Smith, G. P., Acta. Met., 5 , 249 (1957) 73. Howland, W. H. and Epstein, L. F., Ind. Eng. Chem.. 49, 1931 (1957) 74. Antill, J. E., Campbell, C. B., Goodison, D., Jepson, W. B. and Stevens, C. G., Proc. 3rd Conf. Peaceful Uses of Atomic Energy, 9, 523 (1964) 75. Gregg, S. J. and Jepson, W. B., J. Chem. SOC.,712 (1960) 76. Aylmore, D. W., Gregg, J. J. and Jepson, W. B., J. Electrochem. Soc., 106, 1010 (1959) 77. Castle, J. E. and Wood, C. G., Scanning Electron Microscopy (Ed. 0 . Johari), I.I.T. Research Inst.. Chicago, 39 (1%8) 78. Aylmore, D. W., Gregg, S. J. and Jepson, W. B., J. NUC.Mats., 3, 190 (1961) 79. Castle, J. E. and Mann, G. M. W.,Corros. Sci., 6 (1966) 80. Fiegna, A. and Weisgerber, P., J. Electrochem. Soc., 115, 369 (1968) 81. Cox, B., J. Nuc. Mots.. 41, 96 (1971) 82. Dupre, B. and Shreiff, R., J. NUC.Mats., 42, 260 (1972) 83. Surman, P. L. and Castle, J. E., Corros. Sci., 9, 771 (1969) 84. Fern, F. H. and Antill, J. E., Corros. Sci., 10, 649 (1970) 85. Bradhurst, D. H. and Heuer, P. M., J. NUC.Mats., 41, 101 (1971) 86. Roy, C. and David, G.. J. NUC.Mats., 37. 71 (1970) 87. Cox, B., J. NUC.Mats., 31, 48 (1969) 88. Webb, W. W., Norton, J. T. and Wagner, C., J. Electrochem. Soc., 103, 107 (1956) 89. Jepson, W. B, and Aylmore, D. W., J. Electrochem. SOC., 108, 942 (1%1) 90. Kellett, E. A. and Rodgers, S. E.,J. Electrochem. SOC.,110, 503 (1965) 91. Wallwork. G. R. and Jenkins, N. E., J. Electrochem. Soc.. 106. 10 (1959) 92. Osthagen, K. and Kofstad, P., J. Electrochem. SOC., 109, 204 (1962) 93. Smeltzer, W. W., Can. Met. Quart., 11, 41 (1962) 94. Pemsler. J., J. Electrochem. Soc.. 111, 383 (1964) 95. Kofstad, P., Proc. 1st Int. Cong. Met. Corrosion, London, 1961, Butterworths, London, 181 (1962)
1 :292
DISCONTINUOUS OXIDE FILMS
Cathcart, J. V., Campbell, J. J . and Smith, G. P., J. Electrochem. Soc., 105,442 (1958) Weirich, L. J. and Larsen, W. L., J. Electrochem. Soc., 119,465 (1972) Stringer, J., J. Less Common Metals, 12, 301 (1%7) Dooley, R. B. and Stringer, J., J. Less Common Metals, 24, 139 (1971) 100. Potter, E. C. and Mann, G. M. W., Proc. 1st Int. Cong. Met. Corm., London, 1961, Butterworths, London, 417 (1962) 101. Potter, E. C., Mitt. V.G.B.. 76. 19 (1962) 102. Castle, J. E. and Surman, P. L.. J. Phys. Chem., 71,4255 (1967) 103. Field, L. M., Stanley, R. C., Adams, A.M. and Holmes, D. R., Proc. 2ndInt. Cong. Met. Corrosion, N.A.C.E., New York (1963) 104. Harrison, P. L., Holmes, D. R. and Teore, P., V.G.B. Conference on Feed-water Treatment, Essen (1%5) 105. Potter, E. C. and Mann, G. M. W.,B. Corrosion J., 1,26 (1965) 106. Bignold, G . J., Garnsey, R. and Mann, G. M. W., Corros. Sci., 12,325 (1972) 107. Castle, J . E., unpublished work 108. Holmes. D. R.. Discussion of Paper by A. U. Seybolt, 4th Int. Cong. on Met. Corros., Amsterdam, 560 (1969) 109. Vermilyea, D. A. and Vedder, W.,Trans. Farad, Soc., 66, 2644 (1970) 110. Green, J. A. S., Mengelberg, H. D. and Yolhen, H. T., J. Electrochem. SOC., 117, 433 ( 1970) 111. Betteridge, W.and Rhys, D. W.,1st Int. Cong. Met. Corros., Butterworths, London, 186 (1962) 112. Stringer, J., Corros. Sci., 10, 513 (1970) 113. Sickafus, E. N. and Bonzel, H. P., Recent Progress in Surface Science I V , A.P., New York and London (1971) 114. Swank, T. F. and Lawless, K. R., Advances in X-ray Analysis, 10, Plenum Press, New York, 234 (1966) 115. Pawel, R. E., Cathcart, J. V. and Campbell, J. J . , J. Electrochem. SOC., 110,551 (1963) 116. Jaenicke, W., Leistikow, S. and Stadder, J., J. Electrochem. SOC., 111, 1031 (1964) 117. Stoney. G. C . , Proc. Roy. SOC., Lond., A82, 172 (1909) 118. Pawel, R. E. and Campbell, J. J., Acta Met., 14, 1827 (1%6) 119. Pawel, R. E. and Cathcart, J. V., J. Electrochem. Soc., 118, 1776 (1971) 120. Pawel, R. E., J. Electrochem. SOC., 116, 1144 (1969) 121. Evans, U. R., Inst. Met. Symp. on Stresses in Metals, 219 (1947) 122. Dankov, D. D. and Churaev, P. V., Dokl. Akad. Nuuk. SSSR. 73, 1221 (1950) 123. Engell, H. and Wever, F., Acta Metall., 5 , 695 (1957) 124. Bradhurst, D. H. and Heuer, P. M., J. NUC.Mats., 37, 35 (1970) 125. Appleby, W. K. and Tylecote, R. F., Corros. Sci., 10, 325 (1970) 126. Morton, V. M., Corros. Sci., 9,261 (1969) 127. Timoshenko, S.. Mech. Engrs.. 45, 259 (1923) 128. Brenner, A. and Senderoff, S., J. Res. Nrrtn. Bur. Stand.. 42, 105 (1949) 129. Weirich, L. J. and Larsen, W. L., J. Electrochem. SOC., 119, 465 (1972) 130. Sartell, J . A., Stoke
tt.2
Y
-
IC)
cy
4
+o-8
w r vl
II)
+O*L
u 0
0 v) N
-.-
0.0
ydrogen evolution
L
0
E
-0.4
0
a
4
z
-0.8
c
U
w
-1.2
0
2
6 8 10 pH ( measured a t 2 5 O C )
4
12
14
Fig. 2.8 Potential-pH diagram calculated for Fe-H20 system at 250°C. The pH scale refers to the solution measured at 25OC and then raised to 250°C (after Ashworth55)
The work of Porter et aLQ has shown that for copper in phosphoric acid the interfacial temperature was the main factor, and furthermore this was the case for positive or negative heat flux. Activation energies were determined for this system; they indicated that concentration polarisation was the rate-determining process, and by adjustment of the diffusion coefficient and viscosity for the temperature at the interface and the application of dimensional group analysis it was found that: 2
iL
= 4 3 * 3 ( c s- C B ) Du-"
1
where C, = surface concentration of ions, C, = bulk concentration of ions, D = diffusion coefficient and u = kinematic viscosity. This equation provides a means of predicting iL (A/m2), from which an approximation of i,,, can be made from physical measurements alone. Boiling Heat Transfer
This represents a special case of high-level turbulence at a surface by the formation of steam and the possibility of the concentration of ions as water evaporates into the steam bubble^^^.^^. For those metals and alloys in a particular environment that allow diffusion-controlled corrosion processes, rates will be very high except in the case where dissolved gases such as oxygen are the main cathodic reactant. Under these circumstances gases will be expelled into the steam and are not available for reaction. However, under conditions of sub-cooled forced circulation, when cool solution is continually approaching the hot metal surface, the dissolved oxygen
EFFECT OF CONCENTRATION, VELOCITY AND TEMPERATURE
2 :25
appears to be effective” and cathodic processes are stimulated. When the activation step is rate controlling the boiling temperature represents a maximum in the rate. Only small changes (of about 10°C) are possible for metal temperatures to exceed the boiling point because of film boiling, when steam effectively covers the whole surface and corrosion rates become negligible. There is a danger of metal damage by a rapid rise in temperature (Fig. 2.9) when the ‘cooling’ action of evaporation at the surface is prevented, a situation that is obviously to be avoided in the design. When activation processes are in control the small temperature rise allowable above the equilibrium boiling point may increase rates of dissolution by two orders of magnitude ”. Butler and I s o have ~ ~ suggested ~ that variation in corrosion rate can be influenced by surface roughness, which allows a large number of nuclei for steam bubble formation. In these circumstances they have suggested that concentration of ions in solution next to the surface will be greater, and their observations on corrosion damage indicate that the steam bubbles may provide crevices or at least enhanced conditions for dissolution at the triple interface (solution/metal/steam).
Thermogalvanic Corrosion It is impossible to design heat exchangers where all surfaces are isothermal and in many cases such differences are required by the design. For instance, a steam cooler may have a de-superheating zone, a condensing zone and a liquid cooling zone on the same metal tube, but at different positions along its length (Fig. 2.10). The question arises as to whether such temperature differences on the same metal surface in contact with the same electrolyte
1000 ‘Burn out’of
IO
Fig. 2.9 Typical boiling heat-transfer characteristics; At is the temperature differencebetween the solution and the metal surface
2 :26
EFFECT OF CONCENTRATION, VELOCITY AND TEMPERATURE
solution (on the cooling side), can have sufficient electrode potential differences to give rise to a galvanic cell, i.e. a thermal galvanic cell. Electrode potentials change with temperature, but as shown previously, temperature changes may also affect the kinetics of dissolution, especially activationcontrolled processes. The main r61e of thermogalvanic cells is in polarising existing electrode processes, which, depending on other aspects of the environment, may accelerate or decelerate corrosion.
Condensing of vapour
_--_----__-------------
-
Temperature of zone
t
Coolent inlet
J.
Outlet
Fig. 2.10 Temperature distribution in a typical heat exchanger
Origin and Magnitude of Thermogalvanic Potentials
The e.m.f. of a thermogalvanic cell is the result of four main effects": ( a ) electrode temperature, (b) thermal liquid junction potential, ( c ) metallic thermocouple and ( d ) thermal diffusion gradient or Soret. The driving force of a thermogalvanic corrosion cell is therefore the e.m.f. attributable to these four effects, but modified by anodic and cathodic polarisation of the metal electrodes as a result of local action corrosion processes. In practical systems, ( c ) and ( d ) are often very small especially on the same metal surface when solution flow occurs by convection or forced circulation. In neutral solutions, (b) may be small but is somewhat larger in acid solution. On this basis several workers have determined effect ( a ) as a
EFFECT OF CONCENTRATION, VELOCITY AND TEMPERATURE
2 :27
guide to the subsequent behaviour of a thermogalvanic cell. The main usefulness of such a calculation is to decide whether a hot anode or a hot cathode is produced. In many corroding systems, a large cathode area to anode area is detrimental, because of the many situations where corrosion is controlled by diffusion of reactants to the cathode. Such a situation exists at the entrance to a heat exchanger producing a hot zone, and if this is anodic to the larger area of cooler metal then a thermogalvanic cell is set up, having a potentially enhanced corrosion rate s2,s3. It should be noted that the simple Nernst equation cannot be used since the standard electrode potential E e is markedly temperature dependent. By means of irreversible thermodynamics4' equations have been computed to calculate these potentials and are in good agreement with experimentally determined results. In general, temperature coefficients of electrode potential are in the range f 0.1 to 2 mV/"C and in many practical systems temperature differences rarely exceed 75"C, so that the driving force for thermogalvanic corrosion is small and would be subject mainly to resistance control. However, in many instances the temperature change also decreases polarisation (see Fig. 2.11) so that if the resistance of the solution is not high severe attack can ensue. Because of the resistance effect attack is confined to a small area of largest temperature gradient, leading to deep notches at the edge of the heated zone, i.e. the dangerous situation of a small anode and large cathode.
Active-passive Transitions Whilst temperature coefficients suggest modest potential differences, these calculations do not take into account the large potential changes that can occur when thermal effects allow transition from active to passive states.
4
2-
-0.01
Hot metal surface
0 .-
-
Ecorr.
1
at 9 4 O C I
1
0-1
-1.0
Current dcntitylA/m2 )
I
1
#- *
Thermogalwnic currents
Fig. 2.1 1 Influence of temperature on the anodic polarisation of copper in aerated 3% NaCl solutions1.E,,,. is the corrosion potential of the hot metal when not in contact with the cold metal
2 :28
EFFECT OF CONCENTRATION, VELOCITY AND TEMPERATURE
Potentials become more positive (a hot cathode) as a result of thickening of the passive film. Such changes have been observed4’ (Fig.2.12), and moreover, in the presence of aggressive anions, when the thermal effects allow the change to passivity, then overall general corrosion is changed to deep pitting on the electrode surface. Thermogalvanic coupling would enhance this effect.
- 0.3 -
potential for filmed electrodes
hot anode
-0.8 -
film-free electrodes
10
20
30
b0
50
60
Time(min1
Fig. 2.12 Potential-timecurves on mild steel in sodium borate/hydrochloric acid buffer solutions, pH 7.60, oxygen-saturated solution (after Ashworth”)
These observations show that kinetic factors can outweigh thermodynamic effects and the situation of the mutual polarisation of two electrodes in a corrosion cell leads to either negative or positive temperature coefficients. Prediction of thermogalvanic action from consideration of the anode process alone can therefore be misleading. Thermogalvanic corrosion rates may be low in most circumstances but they are persistent for long periods, existing as long as temperature differences exist, i.e. the operating period of plant. They represent a dormant situation that can accelerate corrosion if the environment changes, e.g. high conductivity, and increase in aggressive ion concentration.
P.J. BODEN REFERENCES 1. Pourbaix, M., Electrochim. Acta, 12,184 (1967) and Ness, P., Electrochim. Acta, 12, 161 (1967) 2 . Pourbaix, M..Corrosion, 25, 267 (1967) 3. Mattson, E., Electrochim. Acto, 3, 279 (1961) 4. Horvath, J . and Hackl, L., Corr. Sci., 5, 528-538 (1965)
EFFECT OF CONCENTRATION, VELOCITY AND TEMPERATURE
2 :29
5 . Greene, N. D., J. Electrochem. Soc., 107,457 (1960)and France, W. D. and Greene, N. D., Corrosion, 24, 403 (1968) 6. Cron, C. J., Payer, J. H. and Staehle, R. W., Corrosion, 27, 1 (1971) 7. Bockris, J. OM., Drazic, D. and Despic, A. R., Electrochim. Acta, 4. 325 (1961) 8. Bonhoeffer, K. F. and Heusler, K. E., Z. Phys. Chem., N.F., 8 , 390 (1956) 9. Kabanov, B. N. and Leikis, D. I., Dokl. Akad Nauk. SSSR, 58, I 685 (1947) 10. Florianovich, G. M., Sokolova, L. A. and Kolotyrkin, Ya. M., Electrochim.Acta, 12,897
(1967) 11. Oakes. G. and West, J. M . . Brit. Corr. J.. 4, 66 (1969) 12. Hines, J. G.,Electrochim. Acta, 10, 225 (1965) 13. Florianovich, G. M.,Kolotyrkin, Ya. M. and Kononova, M.D., Proceedings of the 4th International Congress on Metallic Corrosion, Amsterdam, N.A.C.E. (1969) 14. Ammar, I., Darwish, S. and Etman, M., Electrochim. Acta, 12,485 (1967) 15. Brasher, D.M. and Mercer, A. D., Brit. Corr. J., 3,120(1968) 16. Brasher, D.M., Reichenberg, D. and Mercer, A. D., Brit. Corr. J., 3,144 (1968) 17. Brasher, D.M., Brit. Corr. J., 4, 122 (1969) 18. Gouda, V. K., Khedr, M. G. A. and Am Sham El Din, Corr. Sci., 7 , 221 (1967) 19. Legault, R. A., Mori, S. and Leckie, H. P., Corrosion, 26 (1970) 20. Ibi, N., Electrochim. Acta, 1. 117 (1959) 21. Levich, V. G.,Physicochemical Hydrodynamics, Prentice-Hall Inc. (1962) 22. King. C. V.. Surface Chemistty of Metah and Semiconductors, editor H.C. Gatos, John Wiley, New York, 357 (1959) 23. Zembura. Z.,Corr. Sci., 8, 703 (1968) 24. Ross,T.K. and Hitchen, B. P. L., Corros. Sci., 1, 65 (1961) 25. Ross, T. K., Wood, G. C. and Mahmud, I., J. Electrochem. Soc., 113, 334 (1966) 26. Van Shaw, P.,Reiss, L. P. and Hanratty, T. J., Am. Inst. Chem. Engrs. J . , 9,362 (1963) 27. Ross, T.K. and Wragg, A. A., Electrochim. Acta, 10, 1 093 (1965) 28. Marangozis, J., Corrosion, 24, 255 (1968) 29. Mahato, B. K., Steward, F. R. and Shemilt, L. W., Corr. Sci., 10, 737 (1968) 30. Conway, B. E., Beatty, E. M. and DeMaine, P. A. D.. Electrochime Acta, 7 , 39 (1962) 31. Speller. F., Corrosion, McGraw-Hill, London (1951) 32. Cowan, R. L. and Staehle, R. W., J. Electrochem. Soc.. 118, 557 (1971) 33. Murgulescu. I. G. and Radovici, O., Proceedings ofthe 1st Cong. On Metallic Corrosion, Butterworths, London, 109 (1961) 34. Griess, J. C., Corrosion, 24, 97 (1968) 35. Finley, T.C. and Myers, J. R., Corrosion, 26, 544 (1970) 36. Okomoto, G.0.and Kobayashi, H., Z. Electrochem., 62,755 (1958) 37. Pugh, M., Warner, D. Gabe, Corr. Sci., 7 , 807 (1967) 38. Robinson, F. P.A. and Golante, L., Proc. 2nd Int. Cong. on Metallic Corrosion, N.A.C.E., New York (1963) 39. Kucera, V., Novak, P., Franz, F. and Koritta, J., Korrozija ZaEita Titana, G.N.T.I.M.L., Moscow (1964) 40. Pourbaix, M.,Atlas of ElectmchemicalEquilibriain Aqueous Solutions, Pergamon Press, Oxford (1966) 41. Criss, C. M. and Cobble, J. W., J. Am. Chem. SOC., 86, 5 390 (1964) 42. de Bethune, A. J., Licht, T. S. and Swendeman, N., J. Electrochem. SOC.,106,616 (1959) 43. Townsend, H. E.,Corr. Sci., 10, 343 (1970) 44. Ashworth, V. and Boden, P. J., Corr. Sci., 10,709 (1970) 45. Brook, P.A., Corr. Sci., 11, 389 (1971) 46. Brook, P. A., Corr. Sci., 12 (1972) 47. Ross, T.K., Brit. Corr. J., 2, 131 (1967)* 48. Porter, D. T., Donimirska, M. and Wall, R ., Corr. Sci., 8. 833 (1968) 49. Butler, G. and Ison, H.C. K.. Roc. 1st Int. Cong. on Metallic Corrosion, Butterworths, London (1%1) 50. Freeborn, J. and Lewis, D., J. Mech. Eng. SOC.,4, 46 (1962) 51. Boden. P. J., Corr. Sci., 11, 353 (1971) 52. Breckon, G.and Gilbert, P. T., Proc. 1st Int. Cong. on Metallic Corrosion, Butterworths, London (1961) 53. Bem, R. S. and Campbell, H. S., ibid. 54. Ashworth, V. and Boden, P. J., J . Electrochem. SOC., 119,6 (1972) 55. Ashworth, V. and Boden, P. J., Corr. Sci., 14,209 (1974)
2 :30
EFFECT OF CONCENTRATION, VELOCITY A N D TEMPERATURE
Chilton, T. H. and Colburn, A. P., Ind. Eng. Chem., 26, 1183 (1934) Poulson, B., Corr. Sci., 23, 391 (1983) Chin, D. T., Tsang, C. H.. J. Electrochem. Soc., 125, 1461 (1978) Martin, H., ‘Advances in Heat Transfer’, 13, 1, Academic Press New York (1977) 60. Tagg. D. J., Pattrick, M.A., Wragg, A. A., Trans. I. Chem. E., 57, 176 (1979) 61. Heitz, E., WerkstofJe und Korrosion, 15, 63 (1964) 62. Poluboyartseva, L. A. etul., J. Appl. Chem., 36, 1210 (1963) 63. Oldfield, J. and Todd, P.. Desalinution. 31, 365 (1979) 56. 57. 58. 59.
* Detailed review presented at symposium on ‘Corrosion under Heat Transfer in Liquid Media’, reported in Br. Corr. J., 2 (1967)
2.2 The Atmosphere
Metals are more frequently exposed to the atmosphere than to any other corrosive environment. Atmospheric corrosion is also the oldest corrosion problem known to mankind, yet even today it is not fully understood. The principal reason for this paradox lies in the complexity of the variables which determine the kinetics of the corrosion reactions. Thus, corrosion rates vary from place to place, from hour to hour and from season to season. Equally important, this complexity makes meaningful results from laboratory experiments very difficult to obtain. However, the object of this section is to outline the principles which govern atmospheric corrosion, and the emphasis is placed on metals whose atmospheric corrosion is of economic importance. These include iron and steel, zinc, copper, lead, aluminium and chromium.
Classification of Atmospheric Corrosion Atmospheric corrosion can be conveniently classified as follows: ( a ) Dry oxidation.
(b) Damp corrosion. ( c ) Wet corrosion. Dry Oxidation
This takes place in the atmosphere with all metals that have a negative free energy of oxide formation. Gold does not oxidise and this property is utilised in the coating of electronic components where even the thinnest layers of corrosion product cannot be tolerated. For metals forming non-porous oxides (alkali metals are an exception) the films rapidly reach a limiting thickness since ion diffusion through the oxide lattice is extremely slow at ambient temperatures, and at the limiting thickness, the oxideofilmson metals are invisible. For example, those on iron are typically 30A thick. For certain metals and alloys these films are so fault-free or rapidly self-healing that they confer remarkable protection on the substrate, e.g. stainless steel, titanium and chromium. 2:31
2:32
THE ATMOSPHERE
The tarnishing of copper and silver in dry air containing traces of hydrogen sulphide (Table 2.6) is another example of film growth by lattice diffusion at ambient temperatures. In these cases defects in the sulphide lattice enable the films to grow to visible thicknesses with the consequent formation of tarnish films which are aesthetically objectionable and may have a significant effect on the behaviour of the metals in particular applications, e.g. electrical contacts. Table 2.6 Typical concentration of atmospheric impurities Impurity
Sulphur dioxide**’ Sulphur trioxide Hydrogen sulphide Ammonia3 Chloride3 (air sampled) Chloride3 (rainfall sampled) Smoke particles I
Typicu! concentrutions@g/m
Industrial region: winter 350. summer 100 Rural region: winter 100. summer 40 Approximately 1% of the sulphur dioxide content Industrial region: 1-5-90 Urban region: 0-5-1.7 Rural region: 0-15-0-45 Industrial region: 4.8 Rural region: 2. I Industrial inland: winter 8.2, summer 2 - 7 Rural coastal: annual average 5 . 4 Industrial inland: winter 7 - 9 , summer 5.3 Rural coastal: winter 57, summer 18 (these values in mg/l) Industrial region: winter 250, summer 100 Rural region: winter 60,summer 15
There are two methods that arc commonly used for estimating sulphur dioxide: h a d peroxide ‘candle’method. The weight gain. caused by lead sulphate formation as sulphur dioxide reacts with a specified surface area of kad peroxide paste. is measured. ( b ) Hydrogen psroxide titrimetric method. A known volume ofair is pumped through a weak hydrogen peroxide solution in which the sulphur dioxide is o x i d i d IO sulphuric acid. The acid content is estimated by titration. In Ref. I the second method was used, the air first being filtered IO yield an estimate of paniculate matter. (0)
However, in this section emphasis is placed upon damp and wet atmospheric corrosion which are characterised by the presence of a thin, invisible film of electrolyte solution on the metal surface (damp type) or by visible deposits of dew, rain, sea-spray, etc. (wet type). In these categories may be placed the rusting of iron and steel (both types involved), ‘white rusting’ of zinc (wet type) and the formation of patinae on copper and its alloys (both types). The corrosion products may be soluble or insoluble. If insoluble, they usually reduce the rate of corrosion by isolating the substrate from the corrosive environment. Less commonly, they may stimulate corrosion by offering little physical protection while retaining moisture in contact with the metal surface for longer periods. Soluble corrosion products may increase corrosion rates in two ways. Firstly, they may increase the conductivity of the electrolyte solution and thereby decrease ‘internal resistance’ of the corrosion cells. Secondly, they may act hygroscopically to form solutions at humidities at and above that in equilibrium with the saturated solution (Table 2.7). The ‘fogging’of nickel in SO,-containing atmospheres, due to the formation of hygroscopic nickel sulphate, exemplifies this type of behaviour. However, whether the corrosion products are soluble or insoluble, protective or non-protective, the
2:33
THE ATMOSPHERE Relative humidities of air in equilibrium with saturated salt solutions at 2OoC4-'
Table 2.7
Salt in solution
r.h.
Salt in solution
r.h.
(TO)
C U S O -~5 H l 0
98 98 93 92 92 90 89 86 81
KZ
Na, SO, Na2C0, .10H,O FeSO, .7H,O ZnSO, .7H, 0 3CdS04 .8H20 KCI (NH4)zSOd
(%lo)
NaCl C U C I '2Hz0 ~ Feel2 NiCI, K2CO3 *2H2O MgCI, * 6Hz 0 CaCI, *6H,O ZnClz*xH20 NH4Cl
76 68 56 54 44 34 32 10
80
corrosive atmosphere experienced by the substrate (often referred to as the 'micro-environment') is modified from the macro-environment experienced by a bare substrate. For this reason, corrosion rates are rarely constant for extended periods of atmospheric exposure.
Composition of the Atmosphere Nominal Composition
The composition given in Table 2.8 is global and, for most components, is reasonably constant for all locations, but the water vapour content will obviously vary according to the climatic region, season of the year, time of the day, etc. However, only oxygen, carbon dioxide and water vapour need to be considered in the context of atmospheric corrosion. Carbon dioxide was once thought essential for the rusting of ferrous metals (viz. the carbonic acid theory of rusting) but is now considered of relatively minor However, basic zinc carbonate is frequently found in the corrosion products of zinc and small amounts of siderite (FeCO,) are found in ferrous rusts. Table 2.8
Constituents Air Nitrogen Oxygen
Argon Water vapour Carbon dioxide
Approximate constitution" of the atmosphere at 10°C and 100 kN/m2 (excluding impurities) g/m'
1172 879 269 15 8 0-5
Weight ('70) 100
75 23
1.26 0.70 0.04
Constituents
rng/m3
Neon
14
Krypton Helium Xenon Hydrogen
p.p.rn. by weight 12
4
3
0.8 0-5 0-05
0.7 0.4 0-04
Water vapour is essential to the formation of an electrolyte solution which will support the electrochemical corrosion reactions, and its concentration in the atmosphere is usually expressed in terms of the relative humidity (r.h.).
2:34
THE ATMOSPHERE
This is defined as the percentage ratio of the water vapour pressure in the atmosphere compared to that which would saturate the atmosphere at the same temperature. Alternatively, the difference in temperature between the ambient atmosphere and that to which it would have to be cooled before moisture condensed from it, is also used as a measure of moisture content. This difference in temperature is called the dew point depression. The actual temperature at which condensation takes place is known as the dew point. The relative humidity is then expressed as: r.h. =
Saturated vapour pressure of H,O at the dew point x 100%. Saturated vapour pressure of HzOat ambient temp.
Oxygen from the atmosphere, dissolved in the electrolyte solution provides the cathode reactant in the corrosion process. Since the electrolytesolution is in the form of thin films or droplets, diffusion of oxygen from the atmosphere/electrolyte solution interface to the solution/metal interface is rapid. Moreover, convection currents within these thin films of solution may play a part in further decreasing concentration polarisation of this cathodic process". Oxygen may also oxidise soluble corrosion products to less soluble ones which form more or less protective barriers to further corrosion, e.g. the oxidation of ferrous species to the less soluble ferric forms in the rusting of iron and steel.
Atmospheric Contaminants In a sense this subdivision of the composition of the atmosphere is arbitrary since some of the so-called contaminants are derived partly or wholly from natural sources. However, in that their concentrations vary appreciably within very narrow geographical limits, they may be distinguished from the contents of Table 2.8 (with the possible exception of water vapour). Table 2.6 lists those contaminants which are important from a corrosion standpoint. Excluded are contaminants found only in very specific locations, e.g. in the vicinity of a chemical works. The concentrations given are intended only to indicate general levels in the usual classification of environments and not to define a particular environment. Sulphur oxides These (SO,is the most frequently encountered oxide) are powerful stimulators of atmospheric corrosion, and for steel and particularly zinc the correlation between the level of SO, pollution and corrosion rates is However, in severe marine environments, notably in the case of zinc, the chloride contamination may have a higher correlation coefficient than SO,. The SOzin the atmosphere is derived from two sources. Firstly, from the aerial oxidation of H,S produced naturally (see later) and secondly from the combustion of sulphur-containing fuels. In industrialised countries the second source predominates, but on a global scale only about one-fifth of the total sulphur pollution is derived from human activity. In 1969, the total sulphur emission, expressed in terms of SO,, from burnt fuel in the UK was 6.06 x lo6 tons. In densely populated countries sulphur pollution levels are very much related to the domestic heating cycle, and in the UK maximum
THE ATMOSPHERE
2:35
pollution levels are reached in JanuaryiFebruary and the minimum usually occurs in August I . This cyclic pattern is closely reflected by corrosion rate variations Is* 16, corrosion being heaviest in the winter months despite lower average temperatures. A more detailed consideration of the r81e of SO, as a corrosion stimulator will be given later. Hydrogen sulphide This is produced by the putrefaction of organic sulphur compounds or by the action of sulphate-reducing bacteria in anaerobic conditions (e.g. in polluted river estuaries). It is fairly rapidly oxidised to SO2 and concentrations are considerably lower than those of SO: (Table 2.6). Nevertheless it is responsible for the tarnishing of copper and silver at normal atmospheric concentrations. Nitrogen compounds These also arise from both natural and synthetic sources. Thus ammonia is formed in the atmosphere during electrical storms, but increases in the ammonium ion concentration in rainfall over Europe in recent years are attributed to increased use of artificial fertilisers. Ammonium compounds in solution may increase the wettability of a metal” and the action of ammonia and its compounds in causing ‘season cracking’, a type of stress-corrosion cracking of cold-worked brass, is well documented. Saline particles These are of two main types. The first is ammonium sulphate formed in heavily industrialised areas where appreciable concentrations of ammonia and SO, or of HISO, aerosol co-exist. It is a strong stimulator of the initiation of corrosion, being hygroscopic and acidic. The second is marine salt, mainly sodium chloride but quite appreciable quantities of potassium, magnesium and calcium ions are analysed in rainfall3. Chlorides are also produced in industrial areas and for the UK the fall-off in concentration of marine salt with distance from the sea is partially masked by chloride produced by the industrial regions in the centre of the country’. Chlorides are also hygroscopic and the chloride ion is highly aggressive to some metals, e.g. stainless steel. Other airborne particles These are also divisible into two groups. Firstly, the inert non-absorbent particles, usually siliceous, which can only affect corrosion by facilitating differential aeration processes at points of contact. Secondly, absorbent particles such as charcoal and soot are intrinsically inert but have surfaces or infrastructures that adsorb SO,, and by either coadsorption of water vapour or condensation of water within the structure, catalyse the formation of a corrosive acid electrolyte solution. ‘Dirt’ with soot assists the formation of patinae on copper and its alloys by retaining soluble corrosion products long enough for them to be converted to protective, insoluble basic salts.
Other Atmospheric Variables Temperature This may be more or less of an important factor, depending on the metal considered. For example, while zinc is characterised by a very low positive temperature coefficient of corrosion ratel’, steel has a high
2:36
THE ATMOSPHERE
positive The rate of drying of electrolyte solution from the metal surface, directly into the atmosphere or through layers of corrosion product, is strongly temperature dependent. In these contexts the metal surface temperature is probably more important than ambient temperature although the latter obviously strongly influences the former. However, many other factors will affect the metal temperature, including the thermal capacity of the metal structure, its orientation with respect to the sun, the intensity of sunlight, the reflectivity of the metal surface or its corrosion products, wind velocity and direction, the thermal insulating properties of insoluble corrosion products, and so on. The prevailing wind direction is also an important factor in relation to increases in corrosion rates to be expected from the proximity of large industrial plants producing appreciable concentrations of potentially corrosive pollutants.
Electrolyte Solution Formation Wetness of a metal surface The time of wetness of the metal surface is an exceedingly complex, composite variable. It determines the duration of the electrochemical corrosion process. Firstly it involves a consideration of all the means by which an electrolyte solution can form in contact with the metal surface. Secondly, the conditions under which this solution is stable with respect to the ambient atmosphere must be considered, and finally the rate of evaporation of the solution when atmospheric conditions change to make its existence unstable. Attempts have been made to measure directly the time of wetness”, but these have tended to use metals forming non-bulky corrosion products (see Section 20.1). The literature is very sparse on the rBle of insoluble corrosion products in extending the time of wetness, but considerable differences in moisture desorption rates are found for rusted steels of slightly differing alloy content, e.g. mild steel and Cor-Ten. Critical relative humidity The primary value of the critical relative humidity denotes that humidity below which no corrosion of the metal in question takes place. However, it is important to know whether this refers to a clean metal surface or one covered with corrosion products. In the latter case a secondary critical humidity is usually found at which the rate of corrosion increases markedly*. This is attributed to the hygroscopic nature of the corrosion product (see later). In the case of iron and steel it appears that there may even be a tertiary critical humidity”. Thus at about 60% r.h. rusting commences at a very slow rate (primary value)’’; at 75-80% r.h. there is a sharp increase in corrosion rate probably attributable to capillary condensation of moisture within the rust8”’. At 90% r.h. there is a further increase in rusting rate” corresponding to the vapour pressure of saturated ferrous sulphate solution’, ferrous sulphate being identifiable in rust as crystalline agglomerates 16. The primary critical r.h. for uncorroded metal surfaces seems to be virtually the same for all metals, but the secondary values vary quite widely. Moisture precipitation Apart from wetting by sea-spray, moisture may either be deposited on a surface by rainfall or dew formation. For a known ambient humidity the dew point can be calculated, using the expression given previously, from standard tables giving the saturated vapour pressure of
THE ATMOSPHERE
2:31
% r.h.at ambient temperatureIOC)
Fig. 2.13 Dew point depression below ambient temperature as a function of the relative humidity of the ambient atmosphere over a range of temperature
water at various temperatures (e.g. Handbook of the Chemical Rubber Company). However, Fig. 2.13 sets out these relationships graphically and from a knowledge of the ambient relative humidity and ambient temperature, the dew point depression may be read off. Since gaseous pollution, particularly of SO,, tends to be concentrated near ground level, dew can be considerably more acid than rain which forms at higher altitudes. Moreover, dew can, unlike rain, wet completely sheltered surfaces. Thin sheets of metal which closely follow changes in ambient temperature are more likely to have dew formed on them than more massive sections of higher thermal capacity which will cool more slowly. ‘White rusting’ of galvanised sheeting is usually attributable to dew formation in poorly ventilated conditions. Rainfall, besides wetting the metal surface, can be beneficial in leaching otherwise deleterious soluble species and this can result in marked decreases in corrosion A recent survey of rainfall analyses for Europe has shown that, with the exception of the UK, the acidity and sulphate content of rainfall markedly increased in the period 1956 to 1966, pH values having fallen by 0.05 to 0.10 units per ann22.The exception of the UK may be due to anti-pollution measures introduced in this period. However, even in the UK a pH of 4 is not uncommon for rainfall in industrial areas. The significance of electrolyte solution pH will be discussed in the context of corrosion mechanisms. The remaining cases of electrolyte formation are those in which it exists in equilibrium with air at a relative humidity below 100%. Capillary condensation The vapour pressure above a concave meniscus of water is less than that in equilibrium with a plane water surface. It is therefore possible for moisture to condense in narrow capillaries from an atmosphere of less than 100% r.h.
2:38
THE ATMOSPHERE
The relative lowering of the saturated vapour pressure of water is described by the Thomson equation:
P = poe -2aM/dR Tr where p and p o are the saturated vapour pressures above a concave meniscus of radius r, and a plane surface, respectively; u is the surface tension of the liquid at an absolute temperature T,d its density and M its molecular weight; and R is the gas constant. Thus, as the value of r decreases (r can be approximately equated to the radius of the capillary concerned) so the relative humidity at which condensation takes place within the capillary also decreases (Table 2.9). Table 2.9 Capillary radii for condensation at given humidities
Capillart radius ( A )
Relative humidity for condensation
360
98 90 80 70
94 41
30 21
15
60 50
This concept may be invoked to account for electrolyte formation in microcracks in a metal surface or in the re-entrant angle formed by a dust particle and the metal surface. More importantly, it can also explain electrolyte formation in the pores of corrosion product and hence the secondary critical humidity discussed earlier. Ferric oxide gel is known to exhibit capillary condensation characteristicz3 and pore sizes deduced from measurements of its adsorptive capacityz3are of the right order of magnitude to explain a secondary critical relative humidity r70% for rusted steel”. Chemical condensation This occurs when soluble corrosion products or atmospheric contaminants are present on the metal surface. When the humidity exceeds that in equilibrium with a saturated solution of the soluble species, a solution, initially saturated, is formed until equilibrium is established with the ambient humidity. The contaminants have already been detailed and of the corrosion products, obviously sulphates, chlorides and carbonates are most important in this context. However, in some cases there is a lack of reliable data on the vapour pressure exerted by saturated solutions of likely corrosion products. The useful data was summarised in Table 2.7. In practice, however, the soluble components are often contained in a matrix of insoluble product and formation of electrolyte by both capillary and chemical condensation may occur in the same humidity range. Adsorbed electrolyte layers In this case the water molecules are bound to the metal surface by Van der Waals’ forces. It is estimated that at 55% r.h. the film on polished iron is about 15 molecular layers thick, increasing to 90 molecular layers at just below 100V0r.h.~.Such films are capable of
2:39
THE ATMOSPHERE
supporting electrochemicalcorrosion processes and these have been studied. As the humidity is reduced below 100% and the moisture layers become thinner, polarisation of the cathodic and particularly the anodic process rapidly becomes enormous and corrosion virtually ceases below about 60% r.h.,.
The Rale of Sulphur Dioxide in Atmospheric Corrosion Sulphur dioxide plays such an important r6le in the corrosion of metals in the atmospheres of industrialised countries that detailed consideration of its action seems justified. For all metals SO, appears to be selectively adsorbed from the atmosphere, less so for aluminium than for other metals, and for rusty steel it is almost quantitatively adsorbed even from dry air at OOC”. Under humid conditions sulphuric acid is formed, the oxidation of SO, to SO, being catalysed by metals and by metallic oxides. For some non-ferrous metals (copper, lead, nickel) the attack by sulphuric acid is probably direct with the formation of sulphates. Lead sulphate is barely soluble and gives good protection. Nickel and copper sulphates are deliquescent but are gradually converted (if not leached away) into insoluble basic sulphate^'^, e.g. Cu(Cu(OH),),SO,, and the metals are thus protected after a period of active corrosion. For zinc and cadmium the sulphur acids probably act by dissolution of the protective basic carbonate film Is. This reforms, consuming metal in the process, redissolves, and so on. Zinc and cadmium sulphates are formed in polluted winter conditions whereas in the purer atmospheres of the summer the corrosion products include considerable amounts of oxide and basic carbonate ’’. Thus for non-ferrous metals, SO, is consumed in the corrosion reactions whereas in the rusting of iron and steel it is believed”.” that ferrous sulphate is hydrolysed to form oxides and that the sulphuric acid is regenerated. Sulphur dioxide thus acts as a catalyst such that one SO:- ion can catalyse the dissolution of more than 100 atoms of iron before it is removed by leaching, spalling of rust or the formation of basic sulphatez4. These reactions can be summarised as follows:
+ + + + +
SO, 0, $ FeSO, Fe 4FeS0, + 0, 6H,O 4FeOOH 4H,S04 4H,S04 4Fe 0, S 4FeS0, 4H20,et seq.
+ +
Rosenfel’d’’.ZSconsiders that SO, can act as a depolariser of the cathodic process. However, this effect has only been demonstrated with much higher levels of SO2(0.5%) than are found in the atmosphere (Table 2.4) and the importance of this action of SO2has yet to be proved for practical environments. However, SO, is 1300 times more soluble than 0,in water” and therefore its concentration in solution may be considerably greater than would be expected from partial pressure considerations. This high solubility would make it a more effective cathode reactant than dissolved oxygen even though its concentration in the atmosphere is comparatively small.
2:40
THE ATMOSPHERE
Electrochemistry of Atmospheric Corrosion This has already been touched upon in several of the previous paragraphs. Russian workers have extensively examined the electrochemistry of corrosion under thin moisture films and the reader is referred to the work of Rosenfel'd, Tomashov, Klark and co-workers for fuller details4,1 1 . 2 5 * 2 6 . It has been found that the corrosion rate reaches a maximum when the moisture film is around 150 pm thick. The cathodic process in atmospheric corrosion is often stated to be oxygen reduction, and indeed in many cases the evidence is that this is i.e.
O2
+ 2 H 2 0 + 4e- S 40H'
Kaesche' considers that proton reduction may also play a r61e in polluted environments where the pH of the electrolyte is likely to be low. This would be particularly likely in the case of iron if the Schikorr mechanism, involving the presence of sulphuric acid, did in fact operate. However, Russian work4." has shown that oxygen depolarisation is many times more efficient in thin moisture films than in bulk solutions and therefore proton reduction may not be important in affecting corrosion rates. In the rusting of iron and steel, Evans29considers that the anodic reaction of F e e Fe2+
+ 2e-
is balanced by the cathodic reduction of ferric rust to magnetite under wet conditions when access of oxygen is limited: 4Fe,03
+ Fez+ + 2e- * 3Fe304
As the rust dries and is permeated by oxygen, magnetite is reoxidised to rust
with a net gain of 0.5Fe20,: 3Fe,04
+ 0.7502 + 4-5Fe2O3
There is considerable evidence that under certain conditions this mechanism may be operative.
Effect of Corrosion Products on Corrosion Rates The change in corrosion rate with time varies markedly for different metals due to the differing degrees of protection conferred by the corrosion products. Lead, aluminium and copper corrode initially but eventually form completely protective films4 . Nickel in urban atmospheres does not form a completely protective film, the corrosion/time curve being nearly parab o k 4 . The corrosion rate of zinc appears to become linear after an initial period of decreasing corrosion rate4. The behaviour of steel depends very much on the alloying elements present for any given environment. Thus the decrease in corrosion rate with time for mild steel is very much slower than for a low-alloy steel. This can be attributed to the much more compact nature of the rust formed on the latter type of steel and this is clearly illustrated in Figs. 2.14(a) and ( b ) .
2:41
THE ATMOSPHERE
Fig. 2.14 Surface textures of rust on ( a ) mild steel and ( b ) Cor-Ten steel exposed in an industrial atmosphere for 2.5 years [ ( a ) x 1260 and ( b ) x 1 3201
Weather conditions at the time of initial exposure of zinc and steel have a large influence on the protective nature of the initial corrosion products 15. This can still be detected some months after initial exposure. Finally, rust on steel contains a proportion of ferrous sulphate which increases with increase in SO2pollution of the atmosphere. The effect of this on corrosion rate is so strong that mild steel transferred from an industrial atmosphere to a rural one corrodes for some months as though it was still exposed to the industrial environment Conclusion
In a section of this brevity it is i’mpossible to cover all aspects of the ‘atmosphere’. There are therefore gaps concerning such topics as corrosion at sub-zero temperatures, effect of surface orientation and inclination on corrosion rates and the influence of organic vapours on metallic corrosion. Neither has it been possible to describe the intensive efforts now being made to monitor continuously atmospheric variables such as ‘time of wetness’with a view to predicting the long-term corrosion behaviour of metals in a particular area without resorting to long-term trials.
D. FYFE
REFERENCES 1. Investigation of Air Pollution, April 1968 to March 1969-National Survey, Smoke and
Sulphur Dioxide, Min. Tech., Warren Spring Laboratory, Stevenage, Herts. 2. Smith, A. F. etal., J. Appl. Chem., 11, 317 (1961) 3. Stevenson, C. M., Q. J . Roy. Meteorol. Soc., 94, 56 (1968)
2:42
THE ATMOSPHERE
4. Tomashov. N. D.,Theory of Corrosion and Protection of Metals, Macmillan, New York (1966) 5 . Schikorr. G., Werk. Korr., 18, 514 (1967) 6. Young, J. F., J. Appl. Chem., 17, 241 (1%7) 7. O’Brien, F. E.M., J. Sci. Instrum., 25, 73 (1948) 8. Vernon, W.H.J., Trans. Far. SOC., 31, 1 668 (1935) 9. Kaesche. H.. Werk. Korr., 15, 379 (1964)and B.I.S.I.T.S. No. 5 271 10. Meetham, A. R.. Atmospheric Pollution: its Origins and Prevention, Pergamon, London (1956)and Stem, A. C., (Ed.). Air Pollution. Academic Press, New York. 2nd edn. (1968) 11. Rosenfel’d, I. L., Proc. 1st Int. Corros. Cong. (London), Butterworths (1962) 12. Hudson, J. C.and Stanners, J. F., J. Appi. Chem., 3, 86 (1953) 13. Chandler. K. A. and Kilcullen, M. B., Br. Corros. J., 3, 87 (1968) 14. Haynie, F. H.and Upham. J. B., Mats. Prot., 9 No. 8 , 35 (1970) 15. Schikorr, G.,Werk. Korr.. 15,457 (1964)and B.I.S.I.T.S. No. 3 947 16. Schwarz, H.,ibid., 16, 93 (1%5) and B.I.S.I.T.S. No. 4269 17. Ross, T.K. and Callaghan, B. G., Nature, 211, 25 (1966) 18. Sereda, P.J., Ind. Engng. Chem., 53, 157 (1960) 19. Stanners, J. F., Br. Corros. J.. 5. 117 (1970) 20. Skorchelletti, V. V. and Tukachinsky. S. E., J . Appl. Chem. (USSR), 28. 615 (1955) 21. Barton,K. andBartonova,Z., Werk. Korr.,21No. 2,85(1970)andB.I.S.I.T.S.N0. 8349 22. Persson, G.,Acidity and Concentrationof Sulphate in Precipitation Over Europe, Report, Swedish National Nature Conservancy Office, December (1968) 23. Broad, D.W.and Foster, A. G.. J. Chem. SOC.,446 (1946) 24. Schikorr, G., Werk. Korr., 14, 69 (1963) 25. Rosenfei’d, I. L.and Zhigalova, K . , Corrosion of Metals and Alloys (Ed. by C . Booker), Metallurgizdat, Moscow 26. Klark, G. B., etal., ibid. 27. Brokskii, A. I.. Zhur. Fir. Khim., 30, 676 (1956) 28. Roikh, I. L., ibid., 32, 1137 (1958) 29. Evans, U.R., Trans. Inst. Met. Fin., 37, 1 (1960)
2.3 Natural Waters
Introduction Metals immersed or partly immersed in water tend to corrode because of their thermodynamic instability. Natural waters contain dissolved solids and gases and sometimes colloidal or suspended matter; all these may affect the corrosive properties of the water in relation to the metals with which it is in contact. The effect may be either one of stimulation or one of suppression, and it may affect either the cathodic or the anodic reaction; more rarely there may be a general blanketing effect. Some metals form a natural protective film in water and the corrosiveness of the water to these metals depends on whether or not the dissolved materials it contains assist in the maintenance of a self-healing film. The metals most commonly used for water systems are iron and steel. These metals often have some sort of applied protective coating; galvanised steel, for example, relies on a thin layer of zinc, which is anodic to the steel except at high temperatures. Many systems, however, contain a wide variety of other metals and the effect of various water constituents on these must be considered. The more usual are copper, brasses, bronzes, lead, aluminium, stainless steel and solder. The passage of a natural water through a pipe may modify the composition of the water and hence its corrosive properties. Consumption of constituents which in the circumstances may be corrosion inducing-e.g. oxygen or carbon dioxide, may reduce the water's corrosive properties. Dissolution of a metal into water may, on the other hand, make it more corrosive. An example of this is the attack of some waters on copper and the subsequent increased pitting corrosion of less noble metals such as iron, galvanised steel and aluminium. It has been suggested that this enhanced pitting is caused by the redeposition of minute quantities of copper on the less noble metal thus setting up numerous bimetallic corrosion cells I . Failure of the metal can be the most important effect of a corrosive water, but other effects may arise from small concentrations of metallic ion produced by corrosion. A natural water passed through a lead pipe may contain a toxic concentration of that metal; with copper there is a greater tolerance from the toxicity point of view but staining of fabrics and sanitary fittings may be objectionable. With iron, similarly, discoloration of the water may be unpleasant and may cause damage to materials being processed. 2:43
2:44
NATURAL WATERS
Constituents or Impurities of Water The concentrations of various substances in water in dissolved, colloidal or suspended form are relatively low but vary considerably; for example, a hardness of 300-400p.p.m. (as CaCO,) is sometimes tolerated in public supplies, whereas dissolved iron to the extent of 1 mg/litre would be unacceptable. In treated water for high-pressure boilers or where radiation effects are important, as in some nuclear projects, impurities are measured in very small units (e.g. pg/litre or p.p. lo9), but for most purposes it is convenient to express results in mg/litre. In water analysis, determinations (except occasionally for dissolved gases) are made on a weight/volume basis but some analysts still express results in terms of parts per million (p.p.m.). The differencebetween mgllitre and p.p.m. is small and for practical purposes the two units are interchangeable. For some calculations, the use of milliequivalents per litre or equivalents per million (e.p.m.) has advantages but has not found much application. Hardness, whatever the constituent salts, is usually expressed as p.p.m. CaCO, (see Table 2. IO). Table 2.10 Units of measurement and of hardness Units of Measurement - Conversion Factors Milligrams per litre (mg/P) = parts per million (p.p.m.) Part per 100 OOO = 10 mg/litre Grains per Imperial gallon = 14-25 mg/litre. Grains per US gallon = 17.1 mg/litre.
Hardness Units Parts per million (mg/P) as CaCO, . Degree French = parts per 100 OOO as CaCO, (= 10 mg/litre CaCO,). Degree Clark, English or British = grains per Imperial gallon or p.p. 70 OOO as CaC0, (= 14.25 mg/litre CaC03). Degree German = parts per 100 OOO as CaCO (= 17.8 mg/litre CaCO,).
Water analysis for drinking-water supplies is concerned mainly with pollution and bacteriological tests. For industrial supplies a mineral analysis is of more interest. Table 2.11 includes a typical selection and gives some indication of the wide range that can be found. The important constituents can be classified as follows: 1. Dissolved gases (oxygen, nitrogen, carbon dioxide, ammonia, sulphurous gases). 2. Mineral constituents, including hardness salts, sodium salts (chloride, sulphate, nitrate, bicarbonate, etc.), salts of heavy metals, and silica. 3. Organic matter, including that of both animal and vegetable origin, oil, trade waste (including agricultural) constituents and synthetic detergents. 4. Microbiological forms, including various types of algae and slimeforming bacteria.
2:45
NATURAL WATERS
Table 2.11 Typical water analyses (results in mg/litre) Slightly Hard hard Very Moderately Slightly Moderately borehole borehole soft soft hard hard water water (chalk containing underriver river lake surface water water water water forma- sodium tion) bicarbonate
: [
"$:
pH value Alkalinity to methyl orange (CaCO,) Total hardness (CaCO, 1 Calcium hardness (CaCO, 1 Sulphate (SO,) Chloride (Cl) Silica (SiOz) Dissolved solids
6.3
6.8
7.4
7-5
7-1
8.3
7.1
2
38
90
180
250
278
470
10
53
120
230
340
70
559
85 39 24 3
210 50 21 4 332
298 17 4 7
40 109 94 12 620
45 1 463 149 6 1670
5 6 5
trace 33
36 20 11 0.3 88
185
400
Dissolved Gases
Of the dissolved gases occurring in water, oxygen occupies a special position as it stimulates the corrosion reaction. Carbon dioxide is scarcely less important; this constituent must, however, be considered in relation to other constituents, especially calcium hardness. Nitrogen is present with oxygen although the ratio is not the same as in air. It has little importance in connection with corrosion, but can be a nuisance if changes in physical conditions bring about its release from solution. Other gases which are occasionally present usually arise from pollution. Ammonia, which in various forms may be present in waste waters, attacks copper and copper alloys; its presence in estuarine waters is one of the main causes of condenser-tube corrosion. Hydrogen sulphide and sulphur dioxide are also usually the result of pollution; sometimes they are produced by the interaction of two contaminants, but sometimes bacterial action may be contributory. Both gases may initiate or accelerate corrosion of most metals. The significance of small concentrations of these and other impurities in high-pressure steam-boiler feed water is discussed in Section 17.4. Oxygen Dissolved oxygen is probably the most significant constituent affecting corrosion, its importance lying in the fact that it is the most important cathodic depolariser in neutral solutions. Other depolarisers also occur, but as oxygen is an almost universal constituent of natural waters its importance will readily be understood. In surface waters, the oxygen concentration approximates to saturation, but in the presence of green algae supersaturation may occur. Underground waters are more variable in oxygen content and some waters containing ferrous bicarbonate are oxygen-free. Contact with air, however, usually gives
2:46
NATURAL WATERS
rise to an oxygen concentration similar to the figures in Table 2.12, which are for distilled water. The solubility is slightly less in the presence of dissolved solids, but this effect is not very significant in natural waters containing less than 1 OOOp.p.m. dissolved solids. Table 2.12
Temperature
("C) 0 5 10 I5 20 25
Solubility of oxygen in distilled water
Oxygen content of air-saturated water ~
~~
~~
mg/litre
ml/litre
14.16 12.37 10,92 9.76 8.84 8.11
9.90 8.65 7-64 6.83 6.18 5-67
For some applications, notably feed-water treatment for high-pressure boilers, removal of oxygen is essential. For most industrial purposes, however, de-aeration is not applicable, since the water used is in continuous contact with air, from which it would rapidly take up more oxygen. Attention must therefore be given to creating conditions under which oxygen will stifle rather than stimulate corrosion. It has been shown that pure distilled water is least corrosive when fully aerated and that some inhibitors function better in the presence of oxygen'. In these cases oxygen acts as a passivator of the anodic areas of the corrosion cells. These facts do not, however, modify the foregoing statements on the significance of oxygen in waters as used in practice. A major difficulty in applying corrosion inhibitors is that the oxygen content of the water may not be the same at all points. For example, in a thin layer of water between a flake of scale (or almost any other foreign body) and the metal on which it is lying the oxygen can be depleted; the difference in oxygen content between the body of water and the stagnant water will then set up a corrosion current which is difficult to suppress. Rather similar conditions occur at the water line of a vessel containing water with air above it. Even if the water is conditioned to prevent corrosion under submerged conditions, the protection may not extend to the water line, especially if the water has a high dissolved-solids content. Fluctuation in water level extends the area of localised attack. Carbon dioxide and calcium carbonate The effect of carbon dioxide is closely linked with the bicarbonate content. Normal carbonates are rarely found in natural waters but sodium bicarbonate is found in some underground supplies. Calcium bicarbonate is the most important, but magnesium bicarbonate may be present in smaller quantities; in general, it may be regarded as having properties similar to those of the calcium compound except that on decomposition by heat it deposits magnesium hydroxide whereas calcium bicarbonate precipitates the carbonate. Calcium bicarbonate requires excess carbon dioxide in solution to stabilise it; the necessary concentration depends on the other constituents of the water and the temperature.
2:47
NATURAL WATERS
The concentrations of carbon dioxide in water can be classified as follows: 1. The amount required to produce carbonate. 2. The amount required to convert carbonate to bicarbonate. 3. The amount required to keep the calcium bicarbonate in solution. 4. Any excess over that accounted for in types 1, 2 and 3.
With insufficient carbon dioxide of type 3 (and none of type 4) the water will be supersaturated with calcium carbonate and a slight increase in pH (at the local cathodes) will tend to cause its precipitation. If the deposit is continuous and adherent the metal surface may become isolated from the water and hence protected from corrosion. If type 4 carbon dioxide is present there can be no deposition of calcium carbonate and old deposits will be dissolved; there cannot therefore be any protection by calcium carbonate scale. The mathematical relationship between carbon dioxide, calcium bicarbonate and calcium carbonate has been studied by several workers, including Langelier’?‘. The simpler form of his equation is pH, = pCa
+ pAlk + (pK2 - pK,) at constant temperature
where pH, = saturation pH value, pCa = negative logarithm of the calcium concentration expressed as p.p.m. CaCO,, pAlk = negative logarithm of the alkalinity to methyl orange expressed in p.p.m. of equivalent CaCO,, pK, = ionisation constant of HCO; ,
[‘ ~ ~ ~ ~ ~ 1 ] ] and
pK, = solubility product of CaCO, . This simple formula does not apply to pH values over 9.0, and high salinities affect its accuracy. The term (pK2 - pK,) is a function of temperature and ionic strength (dissolved solids). In an analysis of a given water at a constant temperature much useful information can be obtained from the equation. The saturation index of a water (S.I.) is defined as: S.I. = pH - pH,
where pH is the actual pH of the water. If the saturation index is positive the water will be supersaturated with calcium carbonate whereas if it is negative the water will be aggressive to calcium carbonate. Graphical forms of the expression are of most practical value and that devised by Powell, Bacon and Lill’ is shown in Fig. 2.15. These authors have also developed the method so that it can be applied to provide a water which has a constant saturation index over a fair temperature range; this is mainly of interest to operators of industrial cooling systems6. The corresponding formula for the magnesium hydroxide equilibrium is: pH,(Mg) = 4 (PMg - PKdhl,,) is incorporated in the Langelier diagram4.
+ PKW
2:48
NATURAL WATERS
PARTS PER MILLION
Fig. 2.15
Langelier saturation index chart (after the American Chemical Society).
A distinction must be made between a thick layer of deposit -whether of calcium carbonate or of other material-and a protective layer. The ideal protection in fact consists of layers of negligible thickness which do not impede water or heat flow and which are self-healing. This is difficult to
2:49
NATURAL WATERS
achieve with natural waters. A water which is exactly in equilibrium in respect of calcium carbonate is normally corrosive to steel (unless it contains natural inhibitors of other types) because it has no power to form a calcium carbonate deposit. Supersaturated waters on the other hand, unless suitably treated, will form a substantial scale, but whether this inhibits corrosion or not depends on adherence to the metal and porosity. The degree of protection afforded by calcium carbonate has been studied by McCauley The carbon dioxide content (Le. of types 3 and 4) can be ascertained from the pH of the water and its alkalinity by a formula devised by Tillmans’
’.
pH = log
c
1
alkalinity x 0.203 x lo7 free CO,
where alkalinity (expressed as CaCO,) and free CO, are in p.p.m. The actual figure is, however, of value only in relation to calcium carbonate content and for calculation of alkali additions for pH corrections. A graphical form is included in the Langelier diagram4. The significance of carbon dioxide in corrosion is also discussed in some detail by Simmonds’. Mineral Constituents
Hardness salts The hardness figures for natural supplies are very varied but most natural supplies in the U.K. fall into well-defined groups. The most important of these are: 1. Upland waters of low hardness, as supplied to most towns in Scotland,
Wales and the North of England. 2. Hard underground waters, mainly in the East and South of England,
mostly from chalk, sandstone or limestone strata. A few supplies are intermediate in composition. Many of them are derived from river sources and vary according to season. The usual classification of water by hardness (Thresh, Beale and Suckling) is as follows: 4 0 p.p.m. 50-100 p.p.m. 100-150 p.p.m. 150-250 p.p.m. 250-350 p.p.m.
CaCO, CaCO, CaCO, CaCO, CaCO, >350p.p.m. CaCO,
soft moderately soft slightly hard moderately hard hard very hard.
The corrosive properties of natural waters are governed by many factors and cannot be related to hardness alone, but the following trends are apparent: 1. Soft upland waters are aggressive to most metals, their behaviour depending to some extent on pH values as discussed on p. 2.53. They
are inevitably unsaturated with respect to calcium carbonate and it is not usually practicable to modify the carbonate equilibrium to make them non-aggressive.
2:50
NATURAL WATERS
2. Very hard waters are usually not very aggressive provided that they are supersaturated with calcium carbonate. Underground waters with a low pH value and high carbon dioxide content are, however, aggressive unless corrective treatment is applied. 3. Waters of intermediate hardness frequently contain fair amounts of other constituents and there is often a tendency for the scale to be loosely attached, permitting corrosion to occur irregularly underneath. In most waters the bicarbonate content is less than the hardness, but a few natural waters are known where the reverse is the case. These waters have been partially softened by the zeolite process which occurs underground, and then contain sodium bicarbonate which, together with the high concentration of chloride and other minerals, may accelerate attack.
Dissolved mineral salts The principal ions found in water are calcium, magnesium, sodium, bicarbonate, sulphate, chloride and nitrate. A few parts per million of iron or manganese may sometimes be present and there may be traces of potassium salts, whose behaviour is very similar to that of sodium salts. From the corrosion point of view the small quantities of other acid radicals present, e.g. nitrite, phosphate, iodide, bromide and fluoride, have little significance. Larger concentrations of some of these ions, notably nitrite and phosphate, may act as corrosion inhibitors, but the small quantities present in natural waters will have little effect. Some of the minor constituents have other beneficial or harmful effects, e.g. there is an optimum concentration of fluoride for control of dental caries and very low iodide or high nitrate concentrations are objectionable on medical grounds. Chlorides have probably received the most study in relation to their effect on corrosion. Like other ions, they increase the electrical conductivity of the water so that the flow of corrosion currents will be facilitated. They also reduce the effectiveness of natural protective films, which may be permeable to small ions; the effect of chloride on stainless steel is an extreme example but a similar effect is noted to a lesser degree with other metals. Turner" has observed that the meringue dezincification of duplex brasses is affected by the chloride/bicarbonate hardness ratio. Nitrate is very similar in its effects to chloride but is usually present in much smaller concentrations. Sulphate in general appears to behave very similarly; Hatch and Rice have shown that small concentrations in distilled water increase corrosion more than similar concentrations of chloride ' I . In practice, high-sulphate waters may attack concrete, and the performance of some inhibitors appears to be adversely affected by the presence of sulphate. Sulphates have also a special rale in bacterial corrosion under anaerobic conditions. Both sulphates and nitrates are acceptable in low-pressure boiler feed water as they are believed to be of value in controlling caustic cracking. Conventional combinations Salts of strong acids and alkalis are, of course, almost completely ionised in dilute solutions. For some purposes, however, it is convenient to regard the ions as being in combination, and various systems of 'conventional combinations' have been developed. In Britain, the system most used takes the metals and acid radicals in the order shown in Table 2.13 (after Thresh, Beale and Suckling). For example, if the amount
2:51
NATURAL WATERS
Table 2.13 Ferrous carbonate Calcium carbonate Calcium sulphate Calcium chloride$ Calcium nitrate$ Magnesium carbonate Magnesium sulphate Magnesium chloride# Magnesium nitrate$ Sodium carbonate Sodium sulphate Sodium chloride$ Sodium nitrate$ Sodium silicate5
Conventional combinations*+ 56 Fe 6 40Ca E 40Ca a 40CaE
40Ca 0
60 C 0 3 = I16 FeC03
60 C 0 3 = 100 CaCO, % SO4 71 CI 124 NO3
= 136 CaSO,
= IIICaC12 = 164 Ca(N03h
24 Mg I 60 C 0 3 ZE 84 MgCO, 24 Mg = 96 CO, = I 2 0 MgSO, 24 Mg E 71 CI 95 MgCI, 24Mg = 124 NO3 i 148 Mg(N03), 46 Na = 60 C 0 3 = 106 Na2C03 46Na E 96 SO, c 142 Na2C04 = 117NaCl 4 6 N a = 71 CI 46 Na H 124 NO3 = 170 NaNO, 46 Na E I36 Si,Os zz 182 Na2Si205
If potassium is present in significant quantities and is determined it is usually inserted in the conventional combination table after magnesium and before sodium (78 K 138 KzCOj I 174 KzS04 I 149 KCI m 202 KNOj). t The figuresincorporated in the table are ‘equivalent to CaCOf or double the chemical equivalent weights. I f it is desired to express analyticalfigures as milliiquivaknts per litre (e.p.m.), the conantrations in mg/litre (p.p.m.) must be divided by half the table figure 8.0 e.p.m.). If this procedure is (e.$. 150 p.p.m. CaCOj 3 e.p.m.; 96 p.p.m. Mg adopted for both anions and cations, the totals of each should be identical. t Sometimeschloridesand nitrates are t a k a in reversed order throughout but this is rarely of much significance. $ The sodium silicates premt are variable in composition: the formula given above must be regarded as a ‘rough average’. f
-
of bicarbonate ion is more than the sum of the equivalents of ferrous iron and calcium the presence of magnesium bicarbonate, and possibly sodium bicarbonate, is postulated. If, however, the bicarbonate content is less than the calcium equivalent the water is assumed to contain calcium sulphate. The significanceof conventional combinations arises largely in two classes of supply: ( a ) those in which the method indicates sodium bicarbonate to be present and (6) those similarly found to contain magnesium chloride or calcium chloride. Waters containing sodium bicarbonate are derived from underground sources where zeolitic materials are present. They occur in various parts of Britain but the best-known group comprises those in central London and other parts of the Thames valley where the water, originating in the chalk, is modified by passage through zeolitic rock. Although some softening occurs, the hardness may still be appreciable; the waters also have high chloride concentrations and usually a fair amount of carbon dioxide. (A typical analysis is given in Table 2,ll.) They are among the most difficult corrosive waters to deal with unless the bicarbonates and excess carbon dioxide are removed. Alkali additions are rarely effective and conditions are far from ideal for the effective use of inhibitors. The importance of magnesium chloride has probably been exaggerated. There is little doubt that it can act as a catalyst in corrosion reactions by hydrolysing to form hydrochloric acid, being then regenerated by reaction between ferrous chloride and magnesium hydroxide. There is, however, little evidence that this reaction takes place in cold- or hot-water systems, and it is probably confined to steam boilers where it might be a cause of corrosive attack underneath scale deposits; it does not constitute a problem in a properly conditioned boiler water.
2:52
NATURAL WATERS
Brief reference has already been made to iron- and manganese-bearing waters. The small amount of deposit formed from these waters is not likely to have much effect on corrosion although there is always a possibility that attack will occur under sludge deposits. Most iron-bearing waters contain substantial amounts of carbon dioxide which may be troublesome. Manganese-bearing waters may be of a similar type but they sometimes contain complex organic compounds of manganese for which special treatment may be needed. Manganese deposits have been associated with type I1 copper pitting corrosion. Another mineral constituent of water is silica, present both as a colloidal suspension and dissolved in the form of silicates. The concentration varies very widely and, as silicates are sometimes applied as corrosion inhibitors, it might be thought that the silica content would affect the corrosive properties of a water. In general, the effect appears to be trivial; the fact that silicate inhibitors are used in waters with a high initial silica content suggests that the form in which silica is present is important. Organic Matter
The types of organic matter in supplies are very diverse and may be present in suspension, or in colloidal or true solution. It is largely decaying vegetable matter but there are many other possible sources including run-off from fields and domestic and industrial wastes. An increasingvolume of literature is appearing on organic pollution but the significance of this in relation to corrosion receives little, if any, attention. A comprehensive account of all the possible constituents is beyond the scope of this section but it may be useful to consider the effects of some types of organic matter. In the first place, there is the masking effect of deposits which may result from suspended matter thrown down on to hot surfaces or at areas where velocity is reduced. They may also form from material coming out of true or colloidal solution. A partially covered surface is always liable to attack. In an aerated water, the distribution of oxygen will be uneven so that corrosion currents will be set up by the cells so produced, corrosion normally occuring at the points where the oxygen content is lower. In waters free from oxygen, other ‘differentials’ may result in corrosion cells. Another aspect, especially in systems where de-aerated water might be used, is that deposits may lead to over-heating and failures of a different kind, e.g. bursting of boiler tubes. Among contaminants one of the most objectionable is oil, especially in systems where water is strongly heated. A relatively small amount of oil on a heating surface can produce very rapid failures. An indirect effect of oil, or other contaminants which form films on the water surface, is that the film isolates the water from air so that in polluted water anaerobic conditions may develop with the encouragement of objectionable bacterial activity. Some of the worst corrosive effects in soft waters are attributed to a rather wide group of organic acids abstracted from peat and mosses, sometimes called peaty acids. Such waters have low pH values and are often discoloured. They affect ferrous metals appreciably and also attack lead and
NATURAL WATERS
2:53
copper. Corrosion control, either of steel or of copper, is rarely achieved solely by pH correction of such waters. Organic Growths
Natural waters may contain organic growths of various kinds, including algae and slime-forming bacteria, which may have a direct or indirect effect on corrosion. The effect may be of two main types: ( a ) the masking effect of living or dead organisms which is little different from that of other materials, and ( b ) the effect of alterations in composition brought about by the organisms. Algae, for example, may remove carbon dioxide and produce oxygen, while other organisms often consume oxygen. Under anaerobic conditions the sulphate-reducing bacteria produce sulphide, and hence hydrogen sulphide, from dissolved sulphate, with disastrous effects. This mechanism is often responsible for external attack on cast iron mains in waterlogged soil, but is not unknown in hot-water systems where the temperature in stagnant branches may promote the development of such organisms (see Sections 2.5 and 2.6). The voluminous deposits associated with iron bacteria, although objectionable in other ways, rarely have much effect on corrosion as they form only over a long period and the alteration in water composition is negligible.
pH of Water Reference has previously been made to pH in connection with calcium carbonate, but it has also a more general significance. The pH of natural waters is, in fact, rarely outside the fairly narrow range of 4.5 to 8.5. High values, at which corrosion of steel may be suppressed, and low values, at which gaseous hydrogen evolution occurs, are not often found in natural waters. According to weight-loss measurements, steel corrodes at approximately the same rate throughout the range of pH found in natural waters. The form which the corrosion takes is, however, affected by the pH. At values between 7 * 5 and 9.0 there is a tendency for the corrosion products to adhere in a hard crusty deposit 12. Sometimes there are separate ‘tubercles’, but these are more usually joined up to form a more or less continuous layer. Attack under the deposit is, however, usually irregular. At lower pH values, adherent corrosion products are not so evident although a very hard form of deposit is sometimes seen in pipes which have been in service for some years. Loss of head due to scaling of a pipe is more commonly found in the higher pH range; at lower pH values, ‘red water’ complaints, arising from corrosion products in suspension, are more common. Inhibitors may reduce the amount of corrosion, but if inhibition is not complete the type of attack is unaltered. For this reason, it is difficult to prevent corrosion in the tuberculating range as a small amount of attack produces an adherent corrosion product which puts a barrier between the inhibitor and the metal. Cast iron behaves in a manner similar to steel at alkaline pH values but at low pH values it is subject to graphitisation.
2:54
NATURAL WATERS
Copper is affected to a marked extent by pH value. In acid waters, slight corrosion occurs and the small amount of copper in solution causes green staining of fabrics and sanitary ware. In addition redeposition of copper on aluminium or galvanised surfaces sets up corrosion cells resulting in pitting of these metals. In most waters the critical pH value is about 7.0 but in soft water containing organic acids it may be higher. The 'pitting' corrosion of copper is independent of the general nature of the water but occurs only when ( a ) certain carbonaceous or oxide films occur on the metal surface and ( b ) when the water does not contain a natural organic inhibitor* which is, in fact, present in many supplies'3*'4(see also Sections 1.6 and4.2). Lead is affected by carbonate content, pH value and mineral constituents. With soft waters the simplest method of control is usually to increase the pH value by adding alkali. Zinc coatings on steel (galvanised) are attacked in the same way as iron, but usually more slowly. Very alkaline waters are usually aggressive to zinc and will often remove galvanised coatings; the corrosion products consist of basic zinc carbonate or other basic compounds and may take the form of a thick creamy deposit or hard abrasive particles.
Rates of Flow The effect of deposits has been referred to in relation to organic matter. Oxygen depletion can, of course, also occur under other types of deposit. Water velocity plays some part here, as with a good flow deposition is less likely. Apart from this effect, increased velocity usually increases corrosion rates by removing corrosion products which otherwise might stifle the anodic reaction and, by providing more oxygen, may stimulate the cathodic reaction".
Temperature The effect of temperature is complex. At very high temperatures such effects as reversal of polarities, as in the Zn/Fe couples of a galvanised surface, may be produced, and, where there are temperature gradients, corrosion cells may be set up. The more general effects may, however, be summarised as follows: ( a ) the velocity of corrosion reactions is greater at increased temperatures, ( b ) temperature changes may affect solubility of corrosion products or shift the position of such equilibria as that existing between calcium carbonate and carbon dioxide, ( c ) gases are less soluble at increased temperature, an effect which is, however, partly offset by greater diffusion rates and ( d ) modification of pH value. This last effect is bound up with the previous two and is mainly of importance in affecting the form and distribution of corrosion products. The overall effect is that corrosion is usually more rapid at higher temperatures, the corrosion products being often more objectionable in nature. There are, however, exceptions to this generalisation and the increased rate * The chemical nature of the inhibitor is not known
NATURAL WATERS
2:55
of reaction at high temperature can sometimes be put to good advantage when corrective measures are being applied.
Assessing the Corrosivity of Natural Waters from their Chemical Analysis Although the Langelier index is probably the most frequently quoted measure of a water’s corrosivity, it is at best a not very reliable guide. All that the index can do, and all that its author claimed for itI6, is to provide an indication of a water’s thermodynamic tendency to precipitate calcium carbonate. It cannot indicate if sufficient material will be deposited to completely cover all exposed metal surfaces; consequently a very soft water can have a strongly positive index but still be corrosive. Similarly the index cannot take into account if the precipitate will be in the appropriate physical form, i.e. a semi-amorphous ‘egg-shell’ like deposit that spreads uniformly over all the exposed surfaces rather than forming isolated crystals at a limited number of nucleation sites. The egg-shell type of deposit has been shown to be associated with the presence of organic material which affects the growth mechanism of the calcium carbonate crystals17. Where a substantial and stable deposit is produced on a metal surface, this is an effective anticorrosion barrier and forms the basis of a chemical treatment to protect water pipes’*. However, the conditions requir.ed for such a process are not likely to arise with any natural waters. As well as the conventional chemical parameters generally useful in gauging a water’s corrosivity e.g. pH, chloride, sulphate etc., various ratios of ions have been found to be significant for particular problems. Thus an increase in the corrosion rate of iron occurs when the ch1oride:carbonate ratio exceeds 3: 1 l9 and attack of the dezincification prone brasses arises when the chloride to carbonate hardness ratio exceeds 1:3”. More recently the aggressivity of the chloride ion to galvanically coupled lead or tin-lead solders has been found to be suppressed when the su1phate:chloride ratio exceeds 2: 1’I. The most spectacular example of this approach is that involving six parameters, pH, chloride, sulphate, nitrate, sodium ion and dissolved oxygen, that have to be taken into account when calculating the propensity of waters to support type I copper pitting”. However such examples, which require a computer program to carry them out conveniently and provide semi-quantitative answers, are unfortunately rare. More usually only very limited correlations can be made between water composition and corrosivity, and even where no multiple ion effects are involved, the response to a change in one parameter may be difficult to model mathematicallye.g. the corrosion of iron which passes through a maximum between pH 7.5-8 in some natural waters”. Considerations such as these can lead to unexpected problems where waters are mixed, either at a treatment works or in a tidal zone within a distribution network into which two sources are fed separately. Within the author’s experience, problems of an erosion attack on copper pipe have occurred at fittings, especially where the ends of the copper tube have been belled out to meet the requirements of the bye-laws for underground pipe24, with mixtures of waters that were satisfactory when supplied separately.
2:56
NATURAL WATERS
Calculations to determine if the mixing of the supplies had produced an increase in free COz, the most likely explanation of this effect, proved negative. More recently, attempts have been made to correlate mathematically the chemical composition of natural waters and their aggressivity to iron by direct measurements on corrosion couponszSor pipe samples removed from distribution systems26.This work has been of limited success, either producing a mathematical best fit only for the particular data set examined or very general trends. The particular interest to the water supply industry of the corrosivity of natural waters to cast iron has led to the development of a simple corrosion rig for the direct measurement of corrosion rates”. The results obtained using this rig has suggested an aggressivity classification of waters by source type i.e. Source type hard aerated borehole lowland river derived soft upland
Corrosivity to iron (mdd) typically 5 typically 15 typically 40
All these rates are, of course, quite low and the problem with corrosion of cast iron water mains is its effect on water quality rather than deterioration of the asset. The corrosion rig has been used to study the effect of inhibitors e.g. silicate and phosphate commonly used to overcome problems with iron. This has revealed that these ‘inhibitors’ hardly affect the long-term corrosion rate, indeed in certain circumstances they may actually increase it. They produce their effect by stabilising the corrosion product developed, thereby preventing the water quality deterioration which is the real complaint ”. The above catalogue of difficulties, in relating the aggressivity of natural waters to their chemical composition, arises precisely because of the low corrosion rates that are usually found with most metals. Under such circumstances, water composition is only one of many factors that determine the rate of attack. The other factors include flow regime, temperature and the conditions under which the initial corrosion product is laid down. The best summary of the behaviour of metals commonly used in natural waters is still that produced by Campbell for the Society of Water Treatment and Examination”. Recent Developments
EC Directive relating to the quality of water intended for human consumption The significanceof this directivez9,from the corrosion point of view, is that for the first time legally enforceable limits for the concentrations of toxic metals in drinking water have been defined. This has greatly increased the importance of contamination as a consequence of corrosion, as opposed to simple mechanical failure, and has required a reassessment of the suitability of various metals and alloys traditionally used in the supply of water for domestic purposes.
NATURAL WATERS
2:57
Chief among these has been the use of lead, mainly for service pipes but also for header tanks in parts of Scotland, and in tin-lead solders for capillary joints of copper tube. For both materials corrosion rates are low in the range of drinking waters supplied. However, the limit for lead set in the drinking water standard is so very low, 0.05 mg P-’ in a running water sample, that it can be readily exceeded, especially in large plumbing systems where the water can have significant residence times. Because of these considerations, no new lead pipe is being installed. Also, as a result of extensive research3’, the contamination from pipe already in place is being controlled by reducing the solubility of the lead corrosion product in the water concerned. For soft waters (carbonate hardness of < 50mg CaCO, t-’) this is readily achievable by increasing the pH to 8 - 8 ~ 5 ~Thus ‘ . the pH values of 6.3 and 6 . 8 , given above in Table 2.11 as typical of soft water, are no longer typical for treated supplies. For waters with higher carbonate hardness values, where contamination problems can arise because of the formation the soluble lead carbonate ion pair comp l e ~ ’ ~raising , the pH is much less effective. For these waters orthophosphate additions are made which converts the corrosion product to a lead phosphate complex with a sufficiently low solubility. Because of the galvanic interaction with the copper, lead contamination from tin-lead soldered joints, or lead-copper pipe junctions, cannot be controlled by reducing the solubility of the corrosion product layer. Galvanic coupling does not simply increase the rate of corrosion, it also increases the corrosivity of the water by converting chloride, which otherwise acts as a corrosion inhibitor for lead, into an aggressive ion2’. Although unacceptable contamination only arises where soldered joints are made badly, problems have occurred in practice, especially in large buildings with long water residence times such as hospitals and schools where particularly vulnerable members of society are exposed. Given that practically equivalent non-lead solders are already available, it has now become policy not to use tin-lead alloys in contact with potable water and the appropriate British Standard has been amended a ~ c o r d i n g l y ~ ~ . A British Standards ‘draft for development’ has been developed 34 which defines a test procedure to determine the potential of metals to contaminate drinking water in contravention of the requirements of the EC Directive. Although primarily meant for new materials, traditional plumbing alloys will also have to be shown to be satisfactory.
Organics There is an increasing tendency to treat drinking waters to remove organic material. This is to minimise the formation of haloforms, produced when the water is chlorinated, which have health implications”. Organics are known to affect certain corrosion processes, e.g. type I copper pitting and the formation of protective corrosion product layers. However, the outcome of this development is difficult to predict as not all the organic material present is removed.
2:58
NATURAL WATERS
NiZf8teS
There has been an increasing level of nitrate contamination of borehole supplies in the east of England, because of the use of agricultural fertilisers since the Second World War36. Nitrates are known to exacerbate certain corrosion processes e.g. at soldered joints; however the maximum value allowed for this ion by the EC drinking water directive (50mg NO, 4-’) should limit its significance. C. W. DRANE
R. J. OLIPHANT REFERENCES 1. 2. 3. 4.
5. 6. 7. 8. 9. 10. 11. 12.
13. 14. 15.
Kenworthy, L., J. Inst. Met., 69, 67 (1943) Uhlig, H. H.,Triadis, D. N. and Stern, M., J. Electrochem. SOC., 102, (1955) Langelier. W. F.. J. Amer. Wat. Wks. Ass., 28, 1500 (1936) Langelier. W. F., J . Amer. Wut. Wks. Ass., 38, 169 (1946) Powell, S. T., Bacon, H. W. and Lill, J. R., Industr. Engng. Chem., 37, 842 (1945) Powell, S. T.. Bacon, H. W. and Lill. S . R., Industr. Engng. Chem., 40, 435 (1948) McCauley, R. F. and Abdullah, M. O., J. Amer. Wut. Wks. Ass., 50, 1419 (1958) Tillmans, J., Die Chemische Untersuchung von Wusser und Abwasser, Verlage von Willhelm Knapp, Saale (1936) Simmonds, M. A., ‘Carbon Dioxide in Domestic Water Supplies’, Proc. SOC.Water Treutment and Examination, 12,4, 197 (1963) and 13, 1, 40 (1964) Turner, M. E. D., ibid., 10, 2, 162 (1961) and 14, 2, 81 (1965) Hatch, G. B. ‘and Rice, Owen, J . Amer. War. Wks. Ass., 51. 719 (1959) Hatch, G. B. and Rice, Owen, Industr. Engng. Chem., 37, 710 (1945) Gilbert, P. T., ‘Dissolution by Fresh Waters of Copper from Copper Pipes’, Proc. Soc. Water Treatment and Examination. 15, 3, 165 (1966) Lucey, V. F.. Br. Corros. J., 2, 175 (1967) Eliassen, R., Perada, C., Romeo, A. S. and Skrinde, R. T., J. Anrer. War. Wks. Ass., 48,
1005 (1956) 16. Langelier, W. F.. Amer. War. Wks. Ass.. 38, 169 (1949) 17. Campbel, H. S., Turner, M. E. D., Jour Inst. Wut. Eng. & Sci., 37, 1 , 55 (1983) 18. Hasson, D., Karman, M., 5th. International Conference on the Internal and External Pro-
tection of Pipes, Innsbruck, Austria, conference sponsored by BHRA, Cranfield, England (October 1983) 19. Lawson, T. E., ‘Corrosion by Domestic Water’, Bull 59, Ill. State Water Survey, Urbana (1975) 20. 21. 22. 23. 24. 25. 26. 27.
28. 29.
30. 31. 32.
Turner, M. E. D., Jour SOC. War. Treat. & Exurn., 10, 2, 162 (1961) Oliphant, R. J., Water Research Centre Report ER125 E (November 1983) Lucey, V. F.. BNFMRA Research Report No A.1838 (December 1972) Larson, T. E., Skold, R. V., Corros.. 14, 6, 43 (1958) Guide to the application and interpretation of the model water byelaws (1986 Edition), Ellis Harwood Limited, Publishers, Byelaw 52, 120. Singley, J. E., J. Amer. Wut. Wks. Ass., 73. 579 (1981) Oliphant, R. J., Assoc. Wut. Offices Jour, 23, 3, 29, (1987) Williams, S. M., Ainsworth, R. G., and Elvidge, A. F., ‘A method of assessing the corrosivity ofwater towards iron’, Source document 3, Water Mains Rehabilitation Manual, Water Research Centre/Water Authorities Association (1986) Campbell, H. S., J. SOC. Wur Treut (e Exam., 20, 1, 11 (1971) Council Directive relating to the quality of water intended for human consumption, Official Journal of the European communities No. L 229, 11 (August 1980) Seminar ‘Lead in Drinking Water’, Lorch Foundation, Lane End, High Wycornbe, Organised by the Water Research Centre, (March 1981) Gregory, R. ibid. Paper 16 Hunt, T. E and Jackson, P. J., ibid. Paper 9
NATURAL WATERS
2:59
33. BS 219:1977 Specification for soft solders (as amended 1987) 34. BS DD 201 (1991) 35. Fawell, J. K.,Fielding, M. and Ridgway, J. W., J. Institute of Water and Environmental Management 1, 1, 61 (1987) 36. Beresford, S. A. A.. International Journal of Epidemiology, 14, 1, 51 (1985)
BIBLIOGRAPHY Holden, W. S. (Ed.), Water Treatment and Examination, Churchill, London (1970) Tarzwell. C. M., Proc. 29th International Water Conference, Engineers’ Society of Western Pennsylvania, 1 (1968) Corrosion of Iron and Steel by Industrial Waters and its Prevention, Special Report No. 41, I.S.I., London (1949) Manual of British Water Engineering Practice, Vols. I to 111. Published for the Institution of Water Engineers, Heffer, Cambridge, 4th edn. (1969) Manual on Industrial Water and Industrial Waste Water, American Society for Testing Materials, Philadelphia (1962) Water Quality and Treatment, American Water Works Association, New York, 3rd edn. (1971) Water Quality Criteria, Resources Agency of California, State Water Quality Control Board, 2nd edn. (1963) Water Quality Criteria, American Society for Testing and Materials, S.T.P. No. 416 (1967) Water-1968, Chemical Engineering Progress, Symposium Series, American Institute of Chemical Engineers (1968)
2.4 Sea Water
Sea water is the only electrolyte containing a relatively high concentration of salts that occurs commonly in nature, covering as it does over two-thirds of the earth’s surface. It is both the most familiar and one of the most severe of natural corrosive agents.
Chemical Composition Ocean sea water is roughly equivalent in strength to a 3f To w/v solution of sodium chloride, but it has a much more complex composition, embodying a number of major constituents, and traces at least of almost all naturally occurring elements. For convenience, however, the concentration of salts in any sample of sea water is expressed in terms of the chloride content, either as chlorinity or as salinity. Both these units are again subject to arbitrary definition and do not conform simply to the chemical composition. Chlorinity When a sample of sea water is titrated with silver nitrate, bromides and iodides, as well as chlorides are precipitated. In calculating the chlorinity (Cl), the entire halogen content is taken as chloride, and chlorinity is defined as the weight in grams of silver required for precipitation of total halogen content per kilogram of sea water, multiplied by 0.328 533. (Chlorinity is always expressed as parts per thousand, using the symbol %o-) Salinity This term is intended to denote the total proportion of dissolved salts in sea water. As it is inconvenient to determine directly, it is normally derived from the chlorinity, defined and determined as above, using the empirical relationship:
Salinity = 1 * 80655 x chlorinity Like chlorinity, it is expressed in parts per thousand. Constancy of composition The validity of these arbitrary conversions depends on the constancy of the ratios of the various dissolved salts. It is a remarkable and important fact that, except where there is gross dilution or contamination, the relative proportions of the major constituents of sea water are practically constant all over the world. 2:60
2:61
SEA WATER
Table 2.14’ gives the composition of sea water of 19 parts per thousand chlorinity. Table 2.14 Major constituents of sea water (parts per thousand) (Chlorinity = 19%, density at 20°C = 1 a 0 2 4 3 ) Chloride (CI-) Sulphate (SO:-) Bicarbonate (HCO,-) Bromide (Br-) Fluoride (F-) Boric acid (H3 BO,) Sodium (Na+) Magnesium (Mg2+) Calcium (Ca2+) Potasium (K+ Strontium (Sr )
2,
18.979 9 2.648 6 0.139 7 0.064 6 0.001 3 0.026 0 10.556 1 1.272 0 0.400 1 0.380 0 0.013 3
(Data courtesy Prentice-Hall, lnc., USA.)
Variations of salinity In the major oceans the salinity of sea water does not vary widely, lying in general between 33 and 37 parts per thousand, a figure of 35 parts per thousand, equivalent to 19.4 parts per thousand chlorinity is commonly taken as the average for ‘open-sea’ water. Local conditions may modify this profoundly in special areas. In the Arctic and Antarctic, and where there is dilution by large rivers, the salinity may be considerably less, and it may vary greatly according to season. Salinity is well below normal in the Baltic, and may fall nearly to zero at the head of the Gulf of Bothnia. In enclosed seas like the Mediterranean, Black Sea and Red Sea, on the other hand, where there is rapid evaporation, salinity may reach 40 parts per thousand. The total salt content of the inland Dead Sea is 260 g/kg compared to 37 g/kg for the Atlantic Ocean.
Minor constituents Sea water contains a multitude of organic and inorganic molecules some of which form metallic complexes which even in trace amounts can significantly affect the corrosion mechanism. Trace metallic complexes also play an important r6le in determining the physiology of biological organisms whose presence in sea water can exert considerable control over corrosion reactions. The presence of such complexing agents in sea water could explain the difficulty of simulating the natural product for corrosion research investigations in the laboratory. (See Section 20.1 for compositions of artificial sea waters.) Variability of Seawater Vertical sections through seawater showing the distribution of temperature, salinity, and oxygen for the Pacific Ocean and Western Atlantic Ocean are shown in Figures 21.3 and 21.4. The global variability of natural seawater and its effects on corrosion have been reviewed=, in particular with respect to seasonal variation of temperature, salinity, oxygen and pH in the Pacific surface water. Data is also given on
2: 62
SEA WATER
the depth profiles for temperature, salinity, oxygen and pH at various sites around North America. Similar information on temperature, dissolved oxygen, salinity and density has been published for the Northern North Atlantic", and in more detail for the seas around the British IslesB; the latter also includes hydrographic data on the contents of dissolved metals (Zn, Ni, Cu, Cd, Hg, Mn) and nutrient cations (phosphate, nitrate). Data from extensive trials investigating the effect of depth on corrosion of materials in the Pacific have now been published29.
Physical Properties Density The density of sea water is, of course, related to its salinity (or chlorinity). If po is the density of sea water at 0" C in g/ml, u, is defined as (Po - 1)l OOO and the relationship between density and chlorinity is given by the equation a, = -0.069
+ 1 *470 8 C1-0*001 570 C12 + 0.OOO 039 8 C13
Since, however, density is affected to a considerable degree by temperature, and since its accurate measurement demands special apparatus and great care, it is not a reliable measure of the 'strength' of sea water.
Electrical Conductivity This is often a convenient and accurate measurement of salinity or chlorinity. Here, too, there is considerable variation with temperature, so that simultaneous observation of temperature is essential. Figure 2. 163shows the relationship between conductivity and chlorinity at various temperatures. Temperature The surface temperature of sea water ranges between about -2" C and 35"C, while the temperature of a shallow surface layer may run even higher. A general picture of the variation with geographical location is given by Table 2.154. Seasonal variations are associated not only directly with the elevation of the sun, but also with changes of surface currents depending on the prevailing winds. The annual variation is generally quite small in the tropics and greatest in the temperate zones, where it may amount to about 10"
c.
In general, water at great depths in the oceans is not subject to temperature fluctuations and even in the tropics seldom exceeds 10" C. The 'freezing point' of sea water, defined as the temperature at which ice crystals begin to form, is -2" C.
Dissolved Gases Dissolved oxygen is a very important factor in the corrosion of metals immersed in sea water. Because of its biological significance, a vast amount
2:63
7
O
C
2 2
O
C
10°C
L L 5OC
/
ooc
err-
Chlorinity ('/..)
Fig. 2.16 Relationship between conductivity and chlorinity of sea water (after Prentice-Hall, Inc., USA). Note that S
=
n-'
Table 2.15
Average surface temperature of the oceans between parallels of latitude ("C)
North latitude
Atlantic ocean
Indian ocean
Pacific ocean
70'-60' 6Oo-5O0 50"-40' 40"-30" 30"-20" 20"-10" 10"- ' 0
5.60 8.66 13.16 20* 40 24.16 25.81 26.66
-
-
26.14 27.23 27.88
5.74 9.99 18.62 23.38 26.42 27.20
South latitude
Atlantic ocean
Indian ocean
Pacific ocean
70'-60" 6Oo-5O0 50"-40" 40"-30" 30"-20" 20D-10" 10'- 0"
-1.30 1.76 8.68 16.90 21.20 23.16 25.18
-1.50 1.63 8.67 17.00 22.53 25.85 27.41
-1.30 5.00
11.16 16.98 21.53 25.11 26.01
of information about its variation in ocean masses has been collected, but insufficientdetail is available about the coastal and harbour waters which are of most importance in the corrosion of fixed structures. Sea water of normal salinity, in equilibrium with the atmosphere, has the following oxygen contents (compare Table 2.14): Temperature ("0 DLrrolvedoxygen(ml/l)
-2
0
8-52
8.08
5 7.16
10 15 20 30 6-44 5-86 5-38 5-42
The dissolved oxygen content of surface oceanic water is mainly determined by its biological history; it always tends, by solution from the air, towards
2:64
SEA WATER
saturation values. Estuarial water may be grossly deficient in oxygen; this results in the rapid multiplication of anaerobic bacteria, and in extreme cases the rate of corrosion may be controlled by the bacteria rather than by dissolved oxygen (Section 2.6). It has been suggested’ that the oxygen content of the deep water in the Atlantic ocean is high due to the southward flow of the cold oxygensaturated water through the funnel of the north Atlantic, but in the Pacific ocean the oxygen content decreases with depth due to negligible water flow through the Bering Strait.
PH
-
Natural sea water is well buffered and normally lies between 8 1 and 8 3 but may fall to 7.0 in stagnant basins with the formation of hydrogen sulphide produced by anaerobic bacteria. For the solubility in seawater of oxygen, nitrogen and carbon dioxide at various temperatures and chlorinities refer to Tables 21.21 and 21.22. The freezing point, temperature of maximum density, osmotic pressure and specific heat for seawater of various salinities are given in Table 21.23.
Potentials of Metals in Sea Water An important factor in the corrosion of a metal in sea water is its electrical potential. This is of course especially the case when two or more electrically connected metals are immersed in a single system. The ‘open-circuit’ potential of most metals in sea water is not a constant and varies with the oxygen content, water velocity, temperature and metallurgical and surface condition of the metal. In static air-free sea water the potential of iron or steel reaches a steadystate value of -0.75 V (us. S.C.E.,E = 0.246 V) which should be compared with the more noble potential of -0.61 V observed under conditions of high velocity and aeration (Table 2.16). This potential of -0.75 V for iron in sea water is important in the practice of cathodic protection. The values in Table 2.16 show how the potentials obtained under service conditions differ from the standard electrode potentials which are frequently calculated from thermodynamic data. Thus aluminium, which is normally coated with an oxide film, has a more noble value than the equilibrium potential E&+,N = -1-66V vs. S.H.E. and similar considerations apply to ‘passive’ stainless steel (see Chapter 21). Although Table 2,16shows which metal of a couple will be the anode and will thus corrode more rapidly, little information regarding the corrosion current, and hence the corrosion rate, can be obtained from the e.m.f. of the cell. The kinetics of the corrosion reaction will be determined by the rates of the electrode processes and the corrosion rates of the anode of the couple will depend on the rate of reduction of hydrogen ions or dissolved oxygen at the cathode metal (Section 1.4).
2:65
SEA WATER
Table 2.16 Potentials of metals in aerated moving sea water (Potentials are negative to the S.C.E.,E = 0.246 V)
Metal
Potential
Reference from which figures taken
(VI Magnesium Zinc Aluminium Cadmium Steel Lead Solder (50/50) Tin Naval brass Copper Aluminium brass Gun metal Cupro-nickel %/lo Cupro-nickel 80/20 Cupro-nickel 70/30 Nickel Silver Titanium Stainless steel 18/8 (passive) Stainless steel 18/8 (active)
1-5 1 e03 0.79 0.7 0.61
5 6 6
5 6
0.5 0.45 0-42
0.08
5 5 5 5 5 5 5 5 5 5 5 6 7 6
0.53
6
0.30 0.28 0.27 0.26 0.26 0.25 0-25 0.14
0.13 0.10
Table 2.17 Effect of exposure period on corrosion rate of mild steel, copper and aluminium
Exposure time (months) 1 2 3 6
12 24 48
Average corrosion rate for period (mm/y)
Steel
Copper
0-33
-
0.25 0.19 0.15 0.13 0.11 0.11
Aluminium -
0.034
0.004 3
0.019 0.018
0.002 1 0.0017
Corrosion Rates Ferrous Metals
Ferrous metals, of which steel is technically the most important, have a remarkably steady rate of corrosion when fully immersed in sea water. The corrosion of mild steel is very rapid initially but falls off gradually over several months to a fairIy steady rate (Table 2.17). In extended exposure periods of,up to 16 years in tropical sea water, Southwell and Alexander' obtained an average corrosion rate for steel of 0*18mm/y in the first year, falling off to a constant rate after 4 years at 0.025 mm/y. They also quote pitting rates as 1 mm/y in the first year falling
2:66
SEA WATER
off dramatically over the second to fourth years and ultimately continuing at a rate comparable with the average rate of penetration giving an average rate for exposure for 16 years of 0.08 mm/y. However, the pitting rate is generally quoted as several orders of magnitude greater than the average rate of penetration, with values of 0-25 to 0.4 mm/y9. A comprehensive table of corrosion rates in sea water has been compiled by LaQue". This appears to show no obvious dependence of corrosion rates on the geographical location of the testing site, and few of the rates depart widely from an average of 0 - 11 mm/y. It is suggested that a figure of 0.13 mm/y may be taken as a reasonable estimate of the expected rate of corrosion of steel or iron continuously immersed in sea water under natural conditions, in any part of the world. The theory has been advanced that the rapid growth of marine fouling in the tropics may provide a protective shield which counteracts the effect of the greater activity of the hotter water, and LaQue" has pointed out that in flowing sea water, when no fouling organisms became attached to small fully immersed specimens, corrosion of steel at 11 C proceeded at 0-18 mm/y compared with 0-36mm/y at 21" C. This increase corresponds with what would be expected from chemical kinetics, where the rate of reaction is approximately doubled for a rise of IO" C. It is significant that most of the data from which a remarkable uniformity of attack is deduced are derived from small isolated panels. This is the most convenient form of specimen for measurements of corrosion rates by loss of weight; but it eliminates the important effect of galvanic currents passing between remote parts of a large structure. It is believed that the experience of civil engineers and other users would not support the conclusion suggested by panel tests that corrosion is no faster in tropical than in temperate waters. Ambler and Bain" found that isolated panels exposed in half-tide conditions are normally more rapidly corroded than those fully immersed, a factor of 2 to 4 being not unusual, but in commercial ports the presence of oil contamination may greatly reduce half-tide corrosion by filming the metal surface. Humble l 3 investigated the corrosion of coupled and uncoupled steel plates distributed in a vertical line extending above high-water and below low-water levels and gives a diagram showing the corrosion profile of steel piling in sea water, based on five years' exposure at Kure Beach. This shows two maxima, one in the 'splash zone' above high-tide level, and the other just below low-tide level. In the tidal zone, between these, there is a minimum corrosion rate. The explanation of this pattern is that the well-aerated areas in the tidal zone become strongly cathodic while the metal just below water becomes anodic. This distribution is in striking contrast to the results quoted by Ambler and Bain". It is generally agreed that steel composition within the range practical for ship plate has little influence on the corrosion rate in sea waterI4-". Owing to the laborious task of obtaining corrosion rates from gravimetric measurements, data for the effect of exposure time on corrosion rates have been very limited. However, with the more recent use of polarisation resistance measurements it would appear that in the absence of macro-biofouling
2:67
SEA WATER
settlement the depth of penetration rate for mild steel in the Channel, North Sea and North Atlantic varies with time according to the relationship: d = 0.126
where d is the average depth of corrosion penetration (mm) occurring in time t (years). This expression agrees fairly well with experience gained from the wreck of the Holland I where the average corrosion of the steel hull had occurred to a depth of 6 mm in 70 years3'. The presence of biofouling settlement may, however, considerably reduce the depth of corrosion, as for instance in the case of artifacts jettisoned by Captain Cook on the Coral Reef in 1770, where coral formation had almost completely protected cast and wrought iron for 200 years3'. The presence of shell fouling affects the corrosion of steel structures in the intertidal zone where it has been found that the rust formed consists of irregular layers or iron oxides and lime, the latter accounting for up to 15% by weight of the corrosion product32. The corrosion rate of mild steel in UK waters for the full immersion and intertidal zone is typically 0.08 mm/y compared with 0.1 to 0.25 mm/y in the splash zone according to the strength of wave action. Above the splash zone corrosion diminishes rapidly to 0.05-0.1 mm/y3*. Non-ferrous Metals
Many of the common non-ferrous metals corrode relatively slowly in still or slowly-moving sea water. Typical figures are given in Table 2.18. The effect of exposure time on the corrosion of copper and aluminium is illustrated in TabIe 2.17. The results quoted by Southwell, Hummer and Table 2.18 Corrosion rates of non-ferrous metals and alloys in sea water Corrosion rate
Material
(mm/y)
Copper (a) full immersion (b) half-tide Brass (Cu-10 to 35 Zn) Aluminium brass (Cu-22Zn-2Al) Admiralty brass (Cu-29Zn-1Sn) Gun-metal (Cu-1OSn-2Zn) Phosphor bronze Aluminium bronze (95Cu-4A1) Copper-nickel-iron (Cu-5Ni- 1 Fe) Cupro-nickel (70Cu-30Ni) Nickel Monel Aluminium (99.8Vo) (9SVO) (5 Mg)
Lead Zinc
0.003 8 0.002 5
0.004 5 0.002 0 0.004 6 0.002 5 0.002 5 0.003 8 0-003 8 0.001 3 0.002 5 0.002 5 0-OOO38 O*OOO 76
O-OOO 30 0.001 0 0.001 8
2:68
SEA WATER
Alexander "-I9 for corrosion rates of copper and aluminium in tropical waters compared with those obtained around the British Isles suggests that the corrosion rate increases by a factor of two for every 10' C rise in temperature. The effect of alloying additions on the marine corrosion properties of nonferrous metals can be very significant, and for copper-based alloys has been comprehensively reviewed by Bradley". For comprehensive reviews of published marine corrosion data refer to references 29 and 33. Crevice Corrosion
Stainless steel, and aluminium and its alloys, derive their excellent corrosion resistance from the self-repairing protective oxide film which renders them passive. Repair of the film depends on access of oxygen, and in crevices this is often inadequate, with the result that the metal in a crevice becomes 'active'. As the fully exposed areas, usually relatively large, act as a cathode, rapid and sometimes disastrous corrosion may result. The use of stainless steel under sea water needs the greatest care to avoid this trouble, and as a rule one of the resistant copper alloys is the better choice;. the danger from built-in crevices may be foreseen and avoided by careful design, but crevice corrosion also occurs behind marine fouling organisms or other deposits which it may not be possible to prevent. (See Section 1.6.) In recent years the mechanism of crevice has been mathematically modelled and a more thorough understanding of the corrosion processes has been e ~ o l v e d ' ~ -From ~ ~ . such mathematical modelling it is feasible to predict critical crevice dimensions to avoid crevice corrosion determined with relatively simple electrochemical measurements on any particular stainless steel. Effect of Depth
Little scientific examination of the deterioration of materials at depth has been undertaken except that by the US Naval Civil Engineering Laboratory and Naval Research, Laboratory. The results of this work were reported by Reinhart in 1966 and more recently the work has been reviewed by Kirk2'. Typical corrosion data for a selection of metals exposed in the Pacific Ocean at several sites and for different times are shown in Table 2.19 and are compared with results obtained in surface waters at Wrightsville Beach by International Nickel Inc. The general indication of the results in this table is that the corrosion rates of nonferrous metals increase with depth in spite of lower temperatures and lower oxygen concentrations than at the surface. It was noted in the paper by Kirk2' that the results at depth were typical of the variation of performance of these materials experienced on numerous occasions in surface sea water. A notable exception was for aluminium alloys of the 5000 (AI-Mg) and 6000 (Al-Mg-Si) series which had good resistance to corrosion
2:69
SEA WATER
Table 2.19 Effect of depth on the corrosion rate of some metals and alloys Rote of metal penetration at various exposure depths (mm/y)
Material
Om 704m
1600m ~~~~
Zinc Mild steel Aluminium alloy 5052* ‘G‘ bronze? CU-1ONi Cu-3ONi Stainless steel (type 410) Incoloy 825 Stainless steel (type 316) Monel400
Exposure conditions Temperature (“C) Oxygen concentration (p.p.m.) AI-4Mg allay.
t
0.015 0.058 0-127 0.043
-
~~~
0.018 0.023
-
0.008
0.008 0.005 0.008 0.020 0.005 0.023
0.018 0.015
1.270 Slight 0 0.035
1.270 Slight 0-025 >0.035
-
5-30 5-10
7.2 0.6
1700m
2-5 1.8
0.091
0.020 >0.576 0-018 0-015
0.025 1.270 0 0 0-038 2-3 2.8
2050m
Form of uttack
~~
General General Pitting 0.008 General 0.015 General 0-030 General 1.270 Pitting Slight Pitting Slight Pitting >0.092 Pitting 0.150
0.058
-
2.7 1.7
Admiralty Gunmetal (Cu-IOSn-2Zn. BS 1400 GI).
in shallow waters, but were found to suffer very severe crevice corrosion in deep sea water. Interpreting these data into practical terms it would seem that there is little justification for expecting lower corrosion rates with increased depth in spite of changes in the sea-water chemistry, and corrosion rates would be expected to show the normal variations encountered in surface waters. Although in the deep-ocean corrosion tests the oxygen concentrations were considerably lower than at the surface, it is perhaps not too surprising that this does not lead to reduction in corrosion since water movement brings a fresh suppIy of oxygen to the corroding metal surface. In desalination investigations it has been shown that even one part per million of dissolved oxygen in sea water can sustain a corrosion reaction on some materialszz. In terms of applications for deep-water engineering it is probably the deterioration of materials by the combined action of mechanical loading and corrosion such as stress corrosion and corrosion fatigue which is of major concern. Drisco and Brouillett23 have examined a number of protective coatings on mild steel and compared their performance at 2 - 1 km with that at shallow depth. They concluded that with thick coatings of over 0 . 3 mm there was negligible difference, but with thinner coatings there was some loss of protection at holidays under deep immersion conditions. An exception was with soft coatings such as asphalt and coal tar which performed better at depth owing to their susceptibilityto damage by certain marine fouling organisms such as barnacles at the surface, whereas such species were not encountered at depth. See reference 29 for further information.
2: 70
SEA WATER
Effect of Weter Speed
Hardly any quantitative results on the effect of movement on corrosion of steel are available. Water movement can markedly affect the corrosion process in controlling the rate of transport of reactants to the corrosion site, and the removal of the corrosion reaction products. A curve is given by LaQue'' which indicates that the corrosion rates are approximately as follows: Water speed (m/s)
Corrosion rate (mm/y)
0 0.13
1.5 0.5
3 0.74
5 0-86
7 0.89
This is presumably an estimated average curve, as no numerical data are quoted, and it may be assumed to refer to bare steel. This conclusion is not ~ ~ , main interest was in the supported by the results of V ~ l k e n i n g whose effect of chlorination and who shows that although corrosion increased with velocity of chlorinated sea water, when plain sea water was used velocity had little effect. There can be no doubt that painting will very much reduce the effect of water speed, as also will marine fouling or slime. Excessive corrosion rates are commonly observed on those parts of a ship's hull which are exposed to high and turbulent flow of water, e.g. leading edges of rudders and shaft-brackets. The pitting found in these places is stimulated by selective local damage to paint films and possibly also by the proximity of a bronze propeller. The contribution of high water speed to this accelerated corrosion cannot be separately assessed. An indirect relation between water speed and corrosion arises from the fact that marine fouling organisms (in particular shell-fouling)do not settle if the water speed is more than about 1 5-2 m/s. Fouling may somewhat restrict general corrosion by its shielding effect, but may also cause crevice corrosion. Impingement Attack
This form of attack, especially as affecting copper alloys in sea water, has been widely studied since the pioneer work of Bengough and May25. Impingement attack of sea water pipe and heat exchanger systems is considered in Sections 1.6 and 4.2. In such engineering systems the water flow is invariably turbulent and the thickness of the laminar boundary layer is an important factor in controlling localised corrosion. At very high water speeds cavitation-damage (Section 8.8) is sustained by any metal; high-speed bronze propellers, for instance, may suffer seriously. This form of attack is mainly mechanical, although an element of true corrosion may be present, and is not specifically associated with sea water. With respect to general corrosion, once a surface film is formed the rate of corrosion is essentially determined by the ionic concentration gradient across the film. Consequently the corrosion rate tends to be independent of water flow rate across the corroding surface. However, under impingement conditions where the surface film is unable to form or is removed due to the shear stress created by the flow, the corrosion rate is theoretically velocity (V)dependent and is proportional to the power I "' for laminar flow and
SEA WATER
2:71
V2’3under turbulent flow”. Under cavitation conditions the loss of metal in addition to corrosion may be mechanically induced and the velocity dependence has a higher power r e l a t i ~ n s h i p ~ with ” ~ ~values between V 3and V’, a popular mean value being v6. It has been shown that impingement resistance is not just a simple property of a material with respect to turbuIent flow but is dependent on polarisation characteristics under flow conditionsmand hence is very susceptible to the bimetallic effect of coupling to more noble materials. J. C. ROWLANDS REFERENCES 1. Sverdrup, H. V., Johnson, M. W. and Fleming, R. H.,The Oceans, Prentice-Hall, N.Y., 173 (1942) 2. Compton, K. G., Corrosion, 26, 448 (1970) 3. As Reference 1. but p. 72 4. As Reference 1, but p. 127 5. Central Dockyard Laboratory, Portsmouth (unpublished) 6. LaQue, F. L., Proc. Amer. SOC. Test. Mater., 51, 541 (1951) 7. Cotton, J. B. and Downing, B. P., Trans. Inst. Mar. Engrs., 69, 311 (1959) 8. Southwell C. R. and Alexander A. L., Materials Protection, 9, 14 (1970) 9. Fink, F. W., Corrosion ofMetals in Sea Water, Battelle Memorial Inst., PB 171 344 (1960) 10. LaQue, F. L., The Corrosion Handbook, (Ed. H. H. Uhlig) Wiley, New York; Chapman and Hall, London, (2nd edition) 391 (1948) 11. LaQue, F. L., Corrosion, 6 , 162 (1958) 12. Ambler, H. R. and Bain, A. J., J. Appl. Chem., 5 , 437 (1955) 13. Humble, H. A., Corrosion, 5, 292 (1949) 14. Hudson, J. C., J. Iron Steel Inst., 166, 123 (1950) 15. Boudot, H. and Chaudron, G., Rev. Mktall., 43, 1 (1946) 16. Forgeson B. W., Southwell, C. R. and Alexander, A. L., Corrosion, 16, 10% (1960) 17. Evans, U. R. and Rance, V., Corrosion and its Prevention at Bimetallic Contacts, H.M.S.O., 3rd edn. (1963) 18. Southwell, C. R., Hummer, C. W. and Alexander, A. L., Materials Protection, 4, 30 (1965) 19. Southwell, C. R., Alexander, A. L. and Hummer, C. W., Materials Protection, 7, 41 ( 1968) 20. Bradley, J. N., Inst. Metals MetallurgicalReview (1971) 21. Kirk, W. W., Proc. Workshop Conf. on High Pressure Aquarian Systems, N.A.C.E. (1971) 22. Schreiber, C. F., Osborn, 0. and Coley, F. H.. Mat. Prof., 7 , 24 (1968) 23. Drisco, R. W. and Brouillett. C. V . . Mat. Prof., 32 (1966) 24. Volkening. V. B.. Corrosion, 6, 123 (1950) 25. Bengough, G. D. and May, R., J. Insf. Met., 32, 204 (1924) 26. Dexter, S. C. and Culberson. C., Mat Perf. 19, 16 (1980) 27. Dietrich, G., Atlas of theHydmgraphy of theNorthern North Atlantic, Pub. Conseil Inter-
28. 29. 30. 31.
32. 33. 34.
national Power I’Exploration de la Mer. Service Hydrographique, Charlottenlund Slot, Denmark (1%9) Min. of Agriculture, Fisheries and Food, Atlas of the Seas Around the British Isles, HMSO ISBN 0 907545 00 9 (1981) Schumacher, M.. Seawater Corrosion Handbook. Noyes Data Corporation, Park Ridge, USA (1979) Elliott, S. Metal Construction, 16, 20 (1984) Knuckey, P. J.. Private Communication (1984) Morley, J., I Corr STBulIetin. 19, 2 (1981) Katz, W., Corrosion Data Sheets-Seawater, DECHEMA, Frankfurt/Main (1976) Oldfield, J. W. and Sutton, W. H., Br Corros J , 13, 13 (1978)
2172
SEA WATER
35. Oldfield, J. W. and Sutton, W. H., Br Corr J , 13, 104 (1978) 36. Kain, R. M. and Lee,T. S., 5th Int Conf Marine Corrosion and Fouling, Barcelona (1980) 37. Lush, P. A. et ai., Eurucor, 77, Soc Chem Ind, 137 (1977) 38. Hutton, S. P., 2nd Int Conf on Cavitation. Edinburgh, I Mech E, 41 (1983) 39. Pylaev. N. I. and Sonikov. A. A.. Energomastinostroenie. 12. 4 (1972) 40. Rowlands, J. C. and Angell, B. A., UK Corr, 83, I Corr ST. 133 (1983)
2.5
Soil in the Corrosion Process
Introduction Soil has been defined in many ways, often depending upon the particular interests of the person proposing the definition. In discussion of the soil as an environmental factor in corrosion, no strict definitions or limitations will be applied; rather, the complex interaction of all earthen materials will come within the scope of the discussion. It is obvious only a general approach to the topic can be given, and no attempt will be made to give full and detailed information on any single facet of the topic. Soil is distinguished by the complex nature of its composition and of its interaction with other environmental factors. No two soils are exactly alike, and extremes of structure, composition and corrosive activity are found in different soils. Climatic factors of rainfall, temperature, air movement and sunlight can cause marked alterations in soil properties which relate directly to the rates at which corrosion will take place on metals buried in these soils.
Soil Genesis The condition of any soil represents a stage in the changing process of soil evolution. Soils develop, mature and change with the passage of time. Whereas the time required for a true soil to develop from the parent rock of the earth may be thousands of years, rapid changes can result in a few years when soils are cultivated, irrigated, or otherwise subjected to man’s manipulation. The type of soil that develops from the parent material will depend upon the various physical, chemical and biological factors of the environment. The weathering process which eventually reduces the rock of the parent material to the inorganic constituents of soil comprises both physical and chemical changes. Size reduction from rocks to the colloidal state depends not only upon the mechanical action of natural forces but also on chemical solubilisation of certain minerals, action of plant roots, and the effects of organic substances formed by biological activity. Interrelated with change in particle size and changes in type and kind of soil minerals present, organic matter is formed and accumulates as an integral part of the soil. Organic-matter content varies from practically none in sands to almost 100%, as exemplified by peat formations. The amount of organic matter present thus reflects the interaction of all environmental 2:73
2:74
SOIL IN THE CORROSION PROCESS
factors influencing chemical and biological activity. Whether the percentage of organic matter increases or decreases depends then upon the relation between the rate at which it is being formed by growth, death and accumulation of plant material, and that at which the microbiologicalactivities within the soil are causing the decomposition of the complex organic molecules. Moisture must be considered of primary importance in soil formation, in weathering, and in all of the changes taking place within the soil. The types of soil that form depend to a great extent upon the rainfall situation. Too little rainfall will prevent development of plant and animal life with their soil-building action. Too much moisture has a similar effect in preventing normal soil formation. Closely associated with rainfall and climate is the acid or basic reaction which develops as a soil matures. When rainfall is high, water percolates through the soil, dissolving the soluble components, and leaching out alkaline minerals of the weathering rock. This happens whether a soil is developing from a naturally acidic or a naturally alkaline parent material. The end result is a shift in reaction to an acid condition. The degree to which this acidity develops depends upon many factors such as the parent minerals, biological activity, and temperature, related to the moisture situation. Should the loss of water from a soil be mainly by surface evaporation (as in arid regions), the dissolved salts tend to accumulate near the surface and alkaline conditions usually develop. Although conditions of high rainfall and moderate to warm temperatures usually lead to an overall decrease in organic matter (particularly in cultivated soils), exceptions occur when the amount of water is great enough to prevent the adequate aeration necessary for maximum microbial activity. Swampy areas with peat and muck soils are the result. In a parallel manner, low temperatures of sub-polar regions slow down decomposition of organic materials and again highly organic soils develop.
The Corrosion Process in Soil Although the soil as a corrosive environment is probably of greater complexity than any other environment, it is possible to make some generalisations regarding soil types and corrosion. It is necessary to emphasise that corrosion in soils is extremely variable and can range from the rapid to the negligible. This can be illustrated by the fact that buried pipes have become perforated within one year, while archaeological specimens of ancient iron have probably remained in the soil for hundreds of years without significant attack. Corrosion in soil is aqueous, and the mechanism is electrochemical (see Section 1.4), but the conditions in the soil can range from ‘atmospheric’ to completely immersed (Sections 2.2 and 2.3). Which conditions prevail depends on the compactness of the soil and the water or moisture content. Moisture retained within a soil under field dry conditions is largely held within the capillaries and pores of the soil. Soil moisture is extremely significant in this connection, and a dry sandy soil will, in general, be less corrosive than a wet clay.
SOIL IN THE CORROSION PROCESS
2:75
Although the mechanism will be essentially electrochemical, there are many characteristic features of soif as a corrosive environment which will be considered subsequently; it can, however, be stated here that the actual corrosiveness of a soil will depend upon an interaction between rainfall, climate and soil reaction. A characteristic feature of the soil is its heterogeneity. Thus variation in soil composition or structure can result in different environments acting on different parts of the same metal surface, and this can give rise to differing electrical potentials at the metal/soil interface. This will result in the establishment of predominantly cathodic or predominantly anodic areas, and the consequent passage of charge through the metal and through the soil. Differences in oxygen concentration (differential aeration), or differences in acidity or salt concentrations may thus give rise to corrosion cells. The distance of the separation of the anodic and cathodic areas can range from very small to miles (‘long-line’ corrosion). The conductivity of the soil is important as it is evident from the electrochemical mechanism of corrosion that this can be rate-controlling; a high conductivity will be conducive of a high corrosion rate. In addition, the conductivity of the soil is important for ‘stray-current corrosion’ (see Section ] O S ) , and for cathodic protection (Chapter 10).
Properties of Soils Related to Corrosion Soil Texture and Structure
Soils are commonly named and classified according to the general size range of their particulate matter. Thus sandy, silt and clay types derive their names from the predominant size range of inorganic constituents. Particles between 0-07and about 2 m m are classed as sands. Silt particles range from 0.005mm to 0-07, and clay particle size ranges from 0-005mm mean diameter down to colloidal matter. The proportion of the three size groups will determine many of the properties of the soil. Although a number of systems have been used to classify soils as to texture, the one shown in Fig. 2.17 represents commonly used terminology for various proportions of sand, silt and clay. Since soils contain organic matter, moisture, gases and living organisms as well as mineral particles, it is apparent that the relative size range does not determine the whole nature of the soil structure. In fact most soils consist of aggregates of particles within a matrix of organic and inorganic colloidal matter rather than separate individual particles. This aggregation gives a crumb-like structure t o the soil, and leads to friability, more ready penetration of moisture, greater aeration, less erosion by water and wind, and generally greater biological activity. The loss of the aggregated structure can occur as the result of mechanical action, or by chemical alteration such as excess alkali accumulation. Destruction of the structure or ‘puddling’ greatly alters the physical nature of the soil. Mention should be made of the soil profile (section through soil showing various layers) because it is important to recognise that the soil’s surface
2:76
SOIL IN THE CORROSION PROCESS
100
Fig. 2.17
BO
60 40 PER CENT SAND
20
Proportions of sand, silt and clay making up the various groups of soils classified on the basis of particle size.
gives a very poor indication of the underlying strata. Pipe-lines are buried several feet below surface soils and corrosion surveys based on surface observations give little information as to the actual environs of the pipe when buried. The Clay Fraction
Clays make up the most important inorganic constituents of soil. They consist of various minerals depending on the mineral composition of the parent material, and on the type and degree of weathering. Often clays may be grouped in a family series, depending upon the weathered condition, as, for example, montmorillonite illite -, kaolinite. Weathering of montmorillonite causes loss of potassium and magnesium which alters the crystalline structure, and eventually kaolinite results. In this example (and also for other clay mineral groups) marked changes occur in the physical properties of a soil as clay minerals undergo the weathering process. Montmorillonite clays absorb water readily, swell greatly and confer highly plastic properties to a soil. Thus soil stress (Section 14.8) occurs most frequently in these soils and less commonly in predominantly kaolinitic types. Similarly, a soil high in bentonite will show more aggressive corrosion than a soil with a comparable percentage of kaolinite. A chalky soil usually shows low corrosion rates. Clay mineralogy and the relation of clays to corrosion deserves attention from corrosion engineers. Many important relationships are not fully understood and there is need for extensive research in this area. +
SOIL IN THE CORROSION PROCESS
2:77
Aeration and Oxygen Diffusion
The pore space of a soil may contain either water or a gaseous atmosphere. Thus the aeration of a soil is directly related to the amount of pore space present and to the water content. Soils of fine texture due to a high clay content contain more closely packed particles and have less pore capacity for gaseous diffusion than an open-type soil such as sand. Oxygen content of soil atmosphere is of special interest in corrosion. It is generally assumed that the gases of the upper layers of soil are similar in composition to the atmosphere above the soil, except for a higher carbon dioxide content. Relatively few data are available showing oxygen content of soils at depths of interest to the corrosion engineer. Judging by the fact that plant roots require oxygen to penetrate a soil, however, it may be assumed that soil gases at depths of 6 m or more contain significant amounts of oxygen. Diffusion of gases into soil is enhanced by a number of climatic factors. Temperature changes from day to night conditions cause expansion and contraction of the surface-soil gases. Variation in barometric pressure has a bellows-like effect on gaseous diffusion. To illustrate the magnitude of this diffusion rate on a large scale, it may be recalled that air within the more than 43 km of underground passages of the Carlsbad Caverns in New Mexico undergoes a complete change each day, despite the fact that the single opening of these caverns to the surface is only a metre or so in diameter. Biological activity within the soil tends to decrease the oxygen content and replace the oxygen with gases from metabolic activity, such as carbon dioxide. Most biological activity occurs in the upper 150 mm of soil, and it is in this region that diffusion would be most rapid. Factors which tend to increase microbial respiration, such as the addition of large amounts of readily decomposed organic matter, or factors which decrease diffusion rates (water saturation) will lead to development of anaerobic conditions within the soil. The significant microbiological relationships to corrosion under both aerobic and anaerobic situations are discussed in Section 2.6. Water Relations
No corrosion occurs in a completely dry environment. In soil, water is needed for ionisation of the oxidised state at the metal surface. Water is also needed for ionisation of soil electrolytes, thus completing the circuit for flow of a current maintaining corrosive activity. Apart from its participation in the fundamental corrosion process, water markedly influences most of the other factors relating to corrosion in soils. Its r61e in weathering and soil genesis has already been mentioned. Types of Soil Moisture 1. Free ground water. At some depth below the surface, water is constantly present. This distance to the water table may vary from a few metres to hundreds of metres, depending upon the geological formations present.
2:78
SOIL IN THE CORROSION PROCESS
Only a small amount of the metal used in underground service is present in the ground water zone. Such structures as well casings and under-river pipelines are surrounded by ground water. The corrosion conditions in such a situation are essentially those of an aqueous environment. 2. Gravitational water. Water entering soil at the surface from rainfall or some other source moves downward. This gravitational water will flow at a rate governed largely by the physical structure regulating the pore space at various zones in the soil profile. An impervious layer of clay, a ‘puddled’soil, or other layers of material resistant to water passage may act as an effective barrier to the gravitational water and cause zones of water accumulation and saturation. This is often the situation in highland swamp and bog formation. Usually gravitational water percolates rapidly to the level of the permanent ground water. 3. Capillaty wafer.Most soils contain considerable amounts of water held in the capillary spaces of the silt and clay particles. The actual amount present depends upon the soil type and weather conditions. Capillary moisture represents the important reservoir of water in soil which supplies the needs of plants and animals living in or on the soil. Only a portion of capillary water is available to plants. ‘Moisture-holding capacity’ of a soil is a term applied to the ability of a soil to hold water present in the form of capillary water. It is obvious that the moisture-holding capacity of a clay is much greater than that of a sandy type soil. Likewise, the degree of corrosion occurring in soil will be related to its moisture-holdingcapacity, although the complexities of the relationships do not allow any quantitative or predictive applications of the present state of knowledge.
Si@nTxanceof Fluctuations in Weter Content
Except for zones below the level of permanent ground water where the environment is water-saturated, and for zones of dry surface sand, continual variation may be expected to occur in the water content of soils. This is usually dependent on rainfall, snow, flooding and such climatic influences, though irrigation practices in many agricultural areas influence water content and hence the corrosion rates. Water losses from the soil represent the sum of downward movement of gravitational water and surface losses by evaporation. Man’s activities, other than drainage procedures or long-term water use from pumps in industrial areas, do not usually influence the downward movement of water. On the other hand, agricultural practices have a great effect on surface evaporation losses. As mentioned earlier, there is an inverse relationship between water volumes and oxygen concentration in soil. As soils dry, conditions become more aerobic and oxygen diffusion rates become higher. The wet-dry or anaerobic-aerobic alternation, either temporal or spatial, leads to higher corrosion rates than would be obtained within a constant environment. Oxygenconcentration-cell formation is enhanced. This same fluctuation in water and air relations also leads to greater variation in biological activity within the soil.
SOIL IN THE CORROSION PROCESS
2:79
Chemical Properties of Soils
Soil reaction (pH) The relationship between the environment and development of acid or alkaline conditions in soil has been discussed with respect to formation of soils from the parent rock materials. Soil acidity comes in part by the formation of carbonic acid from carbon dioxide of biological origin and water. Other acidic development may come from acid residues of weathering, shifts in mineral types, loss of alkaline or basic earth elements by leaching, formation of organic or inorganic acids by microbial activity, plant root secretions, and man-made pollution of the soil, especially by industrial wastes. As with other factors, no direct statements can be made relating the reaction of a soil to its corrosive properties. Extremely acid soils (pH 4.0 and lower) can cause rapid corrosion of bare metals of most types. This degree of acidity is not common, being limited to certain-bog soils and soils made acid by large accumulations of acidic plant materials such as needles in a coniferous forest. Most soils range from pH5.0 to pH8-0, and corrosion rates are apt to depend on many other environmental factors rather than soil reaction per se. The 45-year study of underground corrosion conducted by the United States Bureau of Standards' included study of the effect of soils of varying pH on different metals, and extensive data were reported. Soluble salts of the soil Water in the soil should most properly be considered as the solvent for salts of the soil; the result being the soil solution. In temperate climates and moderate rainfall areas, the soil solution is relatively dilute, with total dissolved salts ranging from 80 to 1 500 p.p.m.*. Regions of extensive rainfall show lower concentrations of soluble salts as the result of leaching action. Conversely, soils in arid regions are usually quite high in salts as these salts are carried to the surface layers of the soil by water movement due to surface evaporation. Generally, the most common cations in the soil solution are potassium, sodium, magnesium and calcium. Alkali soils are high in sodium and potassium, while calcareous soils contain predominantly magnesium and calcium. Salts of all four of these elements tend to accelerate metallic corrosion by the mechanisms mentioned. The alkaline earth elements, calcium and magnesium, however, tend to form insoluble oxides and carbonates in nonacid conditions. These insoluble precipitates may result in a protective layer on the metal surface and reduced corrosive activity. The anionic portions of the soil solution play a rBle of equal importance to the cations. The anions function in the manner outlined for cations in conductivity and concentration-cell action, and have an additional action if they react with the metal cation and form insoluble salts. Thus, if the metal is lead and the predominant anion is sulphate, a layer of insoluble lead sulphate may precipitate on the metal surface and form an effective barrier against further loss of metal. Another important relationship between the salts of the soil and corrosion has to do with biological activity. Since the growth of plants and microorganisms depends upon the proper inorganic mineral nutrients, the action of these forms of life varies with the mineral content of the soil. While many of the possible indirect effects, such as the role of various nitrogenous
2:80
SOIL IN THE CORROSION PROCESS
materials in bacterial growth and corrosion, have not as yet been studied, one well-documented situation is known. This relationship of sulphur and sulphates to bacterial activity in corrosion i s fully discussed in Section 2.6. The salts content of soils may be markedly altered by man's activities. The effect of cathodic protection will be discussed later in this section. Fertiliser use, particularly the heavy doses used in lawn care, introduces many chemicals into the soil. Industrial wastes, salt brines from petroleum production, thawing salts on walks and roads, weed-killing salts at the base of metal structures, and many other situations could be cited as examples of alteration of the soil solution. In tidal areas or in soils near extensive salt deposits, depletion of fresh ground-water supplies has resulted in a flow of brackish or salty sea water into these soils, causing increased corrosion.
The Environment of the Pipe-line Ditch A comprehensive study of the soil and microbial situation in the backfilled zone of pipe-line ditches has shown a number of significant facts3". The results of over a thousand bell-hole studies along operating oil and gas pipelines in widely separated geographical areas of the United States has led to the conclusion that the pipe-line ditch represents a marked disturbance of the
Fig. 2.18 Cross-section of soil and backfill areas surrounding underground pipe-lines
SOIL IN THE CORROSION PROCESS
2:81
natural soil situation. Figure 2.18 indicates the general zones of interest. The operation of ditching and back-filling has resulted in a zone above the pipe (B) which never settles but remains less compact than the undisturbed soil. In this zone, water may penetrate and leave more rapidly. Aeration is more efficient, as shown by the presence of strictly aerobic bacteria in abundant population. The bottom of the pipe-line ditch (Cand D, Fig. 2.18) has a higher moisture content than undisturbed soil at comparable depths. Many instances of free water at the ditch bottom were reported. Differences at the surface between backfill (A) and soil of the right-of-way (E)were less and tended to decrease with passage of time. Conclusive evidence was obtained indicating a greatly increased activity of bacteria in the backfill zone. Some of this may have resulted from the mixing of surface and sub-surface soils during ditching and back-filling. High populations of bacteria adjacent to the organic matter of coating or coating and wrapped-in systems on the external surface of the pipe indicated that these organic compounds served as an available food supply. The presence of hydrocarbon-utilising bacteria was a common finding particularly when the soil was in contact with asphalt protective coatings. These concepts of the altered soil situation in the pipe-line ditch have important implications to the corrosion process. The increased aeration, the high moisture of the ditch bottom, presence of organic matter in coatings, and high microbial populations all lead to greatly increased possibilities for the development of heterogeneity and the formation of zones differing in oxidation-reduction potentials. The action of moisture and micro-organisms on asphaltic coatings, unbonding of the coating6, and formation of cathodic and anodic areas on metal surfaces are all directly related to the disturbed environment of the pipe-line ditch.
Cathodic Protection and Soil Properties
The modern procedure to minimise corrosion losses on underground structures is to use protective coatings between the metal and soil and to apply cathodic protection to the metal structure (see Chapter 11). In this situation, soils influence the operation in a somewhat different manner than is the case with unprotected bare metal. A soil with moderately high salts content (low resistivity) is desirable for the location of the anodes. If the impressed potential is from a sacrificial metal, the effective potential and current available will depend upon soil properties such as pH, soluble salts and moisture present. When rectifiers are used as the source of the cathodic potential, soils of low electrical resistance are desirable for the location of the anode beds. A protective coating free from holidays and of uniformly high insulation value causes the electrical conducting properties of the soil to become of less significance in relation to corrosion rates (Section 15.8). Effect of cathodic protection on soils Long-term application of an electrical potential to the metal structure with resulting flow of electrical current through the soil has two noticeable effects, the magnitude of which will be in proportion to the time and amount of current passing through the soil.
2:82
SOIL IN THE CORROSION PROCESS
The most commonly observed effect of current flow is the development of alkaline conditions at the cathode. On bare metal this alkaline zone may exist only at the metal surface and may often reach pH values of 10 to 12. When the soil solution contains appreciable calcium or magnesium these cations usually form a layer of carbonate or hydroxide at the cathodic area. On coated lines the cations usually move to holidays or breaks in the coating. On failing asphalt or asphalt mastic type coatings, masses of precipitated calcium and magnesium often form nodules or tubercles several centimetres in diameter. The existence of an electrical potential causes not only cation and anion movement but also migration of moisture toward the cathode. This movement of water (electroendosmosis) is due to the asymmetrical nature of the polar groups of the water molecule. In arid regions water leaving the anode area may cause the soil surrounding the anodes to become so dry that proper current densities cannot be maintained along the line. To alleviate this, some pipe-line companies have had to transport water into desert areas to re-moisten anode beds. Moisture films are frequently found under unbonded protective coatings of asphalt and plastic tapes. The nature and origin of this water is still unknown but is of great interest because of its relationship to bond failure, microbial utilisation of asphalt and hydrocarbons, and efficiency of cathodic protection6. Long-line currents As the result of the use of protective coatings and cathodic protection, present-day pipe-lines are usually constructed of welded joints and the line forms a continuous conductor rather than a series of insulated sections. This situation led to the finding of the so-called long-line currents. Often low currents of medium to high voltage have been observed. The cause and significance of this phenomenon is not known. Theories as to the origin of these currents are: 1. Pick-up ofstray current ( a x . or d e . ) (Section 10.5). Decreased use of d.c. in many areas has led to less possibilities of pick-up of direct current from utilities, mines, etc. The importance of grounded a.c. systems has been discounted, but Waters’ has shown that alternating currents can accelerate corrosion. Furt‘hermore the rectifying effects of oxide films, clay minerals and other soil factors are not understood. 2. Current induction due to earth’s rotation. The long lines act as conductors, and variations in magnetic flux could cause induced currents. A few studies have shown long-line current activity to be greater at high activity of the aurora borealis, which is known to be related to earth’s magnetism. The existence of extremely long-length waves of electromagnetic radiation’ gives another possibility. 3. Atmospheric lightning. Discharges of static electricity in the various forms of lightning represent high potentials of extremely short duratioo. The dissipation of this potential through the earth’s crust may well be the origin of the long-line currents. 4. Diflerences in soil potential. Since pipe-lines pass through zones of aerated and unaerated soil and possibilities for electrolytic-cellformation are great, the observed currents may have resulted from soil dissimilarities.
SOIL IN THE CORROSION PROCESS
2:83
Those interested in long-line currents are referred to the publications of Gish9, Gill and Rogers lo and Mudd ' I .
Methods Proposed for Evaluation of Soil Corrosivity Because corrosion rates of metals buried in soil show extreme variation, a test procedure to indicate the expected corrosion activity of a given soil would be extremely valuable. The discussion on the heterogeneity of soils, however, indicates the complex nature of the situation and thus also suggests that the likelihood of success of a single survey-type procedure would be slight. Many types of tests have been suggested. Certain ones are in use by corrosion engineers, and others remain to receive further study. The various types of survey vary from the observations of Denison and Ewing'' that corrosivity of Ohio soils could be related to colour and texture, to complex laboratory testing equipment. It is obvious that a useful test procedure should be relatively rapid and capable of field use, show small changes in environmental relations, and give results predicting relative corrosion rates. In the paragraphs which follow, only the general nature of the test will be discussed, and the reader is referred to Sections 10.4 and 10.7. Soil resistivity The r61e of soil in the electrical circuitry of corrosion is now apparent. Thus the conductivity of the soil represents an important parameter. Soil resistivity has probably been more widely used than any other test procedure. Opinions of experts vary somewhat as to the actual values in terms of ohm centimetres which relate to metal-loss rates. The extended study of the US Bureau of Standards' presents a mass of data with soil-resistivity values given. A weakness of the resistivity procedure is that it neither indicates variations in aeration and pH of the soil, nor microbial activity in terms of coating deterioration or corrosion under anaerobic conditions. Furthermore, as shown by Costan~o'~, rainfall fluctuations markedly affect readings. Despite its short comings, however, this procedure represents a valuable survey method. Scott l4 points out the value of multiple data and the statistical nature of the resistivity readings as related to corrosion rates (see also Chapter 10). Oxidation-reduction potential Because of the interest in bacterial corrosion under anaerobic conditions, the oxidation-reduction situation in the soil was suggested as an indication of expected corrosion rates. The work of Starkey and Wight", McVey16, and others led to the development and testing of the so-called redox probe. The probe with platinum electrodes and copper sulphate reference cells has been described as difficult to clean. Hence, results are difficult to reproduce. At the present time this procedure does not seem adapted to use in field tests. Of more importance is the fact that the data obtained by the redox method simply indicate anaerobic situations in the soil. Such data would be effective in predicting anaerobic corrosion by sulphate-reducing bacteria, but would fail to give any information regarding other types of corrosion.
2:84
SOIL IN THE CORROSION PROCESS
Electrolytic method This procedure is also known as the Williams Corfield test ”. It is based on loss of metal from iron electrodes buried in a watersaturated soil through which current from a 6-V battery is passed. It does not reflect field conditions and depends upon soil conductance under saturated conditions. Polarisation-curve procedures The Denison’’ method is to measure the current at various degrees of polarisation of metal in soil in a special cell. While this test is considered quite accurate, it has the disadvantage that the measurements are made in the laboratory and cannot be made in the field.
Combination electrical methods Tomashov and Mikhailovsky describe a method developed in the Soviet Union. This test is essentially a combination of resistivity measurement and polarisation rates on iron electrodes in soil in situ. The usefulness and value of this procedure has not as yet been determined by practical application by corrosion engineers. The development of this combination test does, however, represent an attempt to integrate some of the complex factors controlling corrosion rates in soil. Much more research on these factors and methods of measurement should in the future enable the corrosion engineer to evaluate soil properties with respect to application of corrosion-alleviating operations. Polarisation-resistance method The polarisation-resistance method (see Section 20.1) has been used for determining corrosion rates of metals buried underground. Soil Corrosivit y Assessment
The development of soil corrosivity assessment techniques has largely been due to the pipeline industry’s requirements for better corrosion risk assessment and the reduction of pipeline failures. Corrosion in soil is a complex process and over the years several parameters have been identified as having a significant effect on the corrosion rate in a given soil. Measurement of some of these parameters identifies the risk of a particular type of corrosion, for example pH measurements assess the risk of acid attack and redox potential measurements is used to assess the suitability of the soil for microbiological corrosion, a low redox potential indicates that the soil is anaerobic and favourable for the life cycle of anaerobic bacteria such as to sulphate-reducing bacteria. Other measurements are more general, resistivity measurements being the most widely quoted. However, as yet no single parameter has been identified which can confidently be expected to assess the corrosion risk of a given soil. It is therefore common practice to measure several parameters and make an assessment from the results. Most of the accepted corrosivity assessment techniques have been outline above. Some of the techniques are used widely, others are more controversial. However, it must be accepted that even with a combination of available techniques no corrosivity assessment survey will accurately predict the corrosion rate for metals in every soil.
SOIL IN THE CORROSION PROCESS
2:85
Table 2.20 Soil Corrosivity Assessment Technique from the German Gas and Water Works Engineers’ Association Standard (DVGW GW9)
Item
Measured Value
Soil composition
Calcareous, marly limestone Sandy marl, not stratified sand Loam, sandy loam (loam content 7 5 % or less) marly loam, sandy claysoil (silt content 7 5 % or less) Clay, marly clay, humus, Peat, thick loam, marshy soil None Exist Vary 10,OOO ohm. cm or more 10.000-5,OOO
Ground-water level at buried position Resistivity
5,000-2,300 2,300- 1,OOO 1,OOO or less 20% or less 20% or more 6 or more 6 or less
Moisture content PH
None Trace Exist 5 % or more
Sulphide and hydrogen sulphide Carbonate
5-1 1 or less 100 mg/kg or less 100 mg/kg more
Chloride Sulphate
Cinder and coke
200 mg/kg or less 2m500 500-1,oOO 1,OOO or more None Exist
Soil is regarded as non-corrosive if the total of the above is 0 or higher; Slightly corrosive if 0 to -4; corrosive if and very corrosive if -10 or less.
Marks +2
0 -2 -4 0 -1 -2 0 -1 -2 -3
-4 0 -1 0 -2 0 -2 -4 +2 +I
0 0 -1
0 -1 -2 -3
0 -4
- 5 to - 10
Corrosion when it occurs can take many different forms from general uniform attack to pitting corrosion. In a given situation some forms of corrosion are more deleterious to the metal structures than others. Pitting corrosion, although overall weight loss is small, is more likely to lead to early failure of a pipeline than uniform corrosion, with a considerably higher overall weight loss. Although certain conditions very often lead to a particular type of attack, attempts to categorise soil corrosion in this way cannot be made on a general basis and most corrosivity assessment techniques categorise the soil as reacting to bare steel or iron in one of four ways: Non aggressive Mildly aggressive Aggressive Very aggressive Varying degrees of emphasis can be placed on certain parameters and this
2:86
SOIL IN THE CORROSION PROCESS
results in a variety of techniques. This preference for some tests over others is due to a number of reasons including: 1. The metal under consideration, although usually iron or steel, may vary
in its resistance to attack. 2. Some industries are interested only in certain aspects of soil assessment and do not require a detailed assessment. 2. Some of these tests are not applicable to soils encountered in that industry. 3. The responsible engineer does not have confidence in some of the techniques. Perhaps the most widely known measurement technique is that adopted by the West German Gas Industry” and developed by Steinrath’l for buried pipework. This assigns a value (See Table 2.20) to each parameter measured; the summation of these values determines the corrosivity of the soil. The parameters measured are shown in Table 2.20. Although this technique was developed for the pipeline industry it can be used with some success for general soil corrosivity assessment.
J. 0.HARRIS D. EYRE REFERENCES 1. Romanoff, M., UndergroundCorrosion, Nat. Bur. Stand., Circular No. 579,Washington
(1957) 2. Russell, E.J., Soil Conditions and Plant Growth, Longmans, London (1932) 3. Harris, J. 0..Kunsas Agric. Exp. Stu. Tech.Bull., No. 102, Manhattan (1959) 4. Harris, J. O.,Corrosion, 16, 149 (1%0) 5. Harris, J. O.,Proceedings, Sixth Ann. Appalachian Underground Corrosion Short Course, West Virginia Univ., Morgantown, 198 (1961) 6. Harris. J. O.,Proc. Pacific Cst. Gas Ass., 52, 109 (1961) 7. Waters, F. 0.. Muter. Protect., 1No. 3, 26 (1962) 8. Heirtzler, J. R., Sci. Amer., u)6, 128 (1962) 9. Gish, 0. H., Sci. Mon., N. Y., 32, 5 (1930) 10. Gill, S. and Rogers, W., Physics, 1, 194 (1931) 11. Mudd, 0. C., Oil Gas J., 38, 48 (1939) 12. Denison. I. A. and Ewing, S. P.. Soil Sci., 40,287 (1935) 13. Costanzo, F.E., Corrosion, 14, 363 (1958) 14. Scott, G. N.. Corrosion, 14, 3% (1958) 15. Starkey, R. L. and Wight, K. M., Anaerobic Corrosion of Iron in Soil, Amer. Gas Assoc., New York (1945) 16. McVey, R. E., Proceedings of the Fifth Ann. Appalachian Underground Corrosion Short Course, West Virginia Univ., Morgantown, 23 (1960) 17. Corfield. G., West. Gm.. 7, 123 (1930) 18. Denison, I. A., Not. Bur. Stand. J., 17, 363 (1936) 19. Tomashov, N.D. and Mikhailovsky, Y. N.,Corrosion, 15, 77 (1959) 20. German Gas and Water Works Engineers’ Association Standard, Merkblatt f i r die Beurteilung der Korrosiongefahrdung von Eisen und Stahl im Erdboden, DVGW GW9, Frankfurt, DVGW (1971) 21. Steinrath. H.,Untemchungsmethoden Zur Beurteilung der Aggrmivitat yon Boden, Frankfurt, DVGW (1%6)
2.6 The Microbiology of Corrosion
The role of microbes in the corrosion of metals is due to the chemical activities (metabolism) associated with the microbial growth and reproduction'. Under favourable growth conditions doubling times of 60-120 min are common. By reason of such rapid growth the onset of changes may be sudden, and even when apparently supressed by mechanical or chemical cleaning often return because a residual low number of living organisms rapidly grow again when favourable conditions are restored'. These characteristics are typical of widespread biodeterioration caused by microbes in all industries of which corrosion is a special case. With a few exceptions such as synthetic polymers, all materials including natural products such as cotton, wood, rubber and oils, and man-made materials such as concrete, complex organic chemicals and metals, can be attacked. Rarely a single microbial species is involved, but usually biodeterioration, including corrosion, results from an association of a number of different microbes. For instance, a rapid growth of an aerobic organism may so deplete oxygen that strictly anaerobic sulphate-reducers associated with cathodic depolarisation then appear, and metallic corrosion results. In many years the complicated associations in such microbial ecosystems have become increasingly recognised. Their corrosive effects on metals can be attributed to the removal of electrons from the metal and formation of a corrosion products by: (a) Direct chemical action of metabolic products such as suIphuric acid,
inorganic or organic sulphides and chelating agents such as organic acids. (b) Cathodic depolarisation associated with anaerobic growth. (c) Changes in oxygen potential, salt concentration, pH, etc. which establish local electrochemical cells. (d) Removal of corrosion inhibitors (oxidation of nitrite or amines) or protective coatings (bitumen on buried pipes). (e) The presence of the biomass itself or residues of biomass such as hygroscopic salt deposits from cells burnt-on in annealing. It must be stressed that the identification of the causative organism(s) may be extremely difficult since it often depends on the quantitative determination of numbers of each microbial type in a complex mixture together with an assessment of its chemical and physical activities in that particular environment. 2:87
2:88
THE MICROBIOLOGY OF CORROSION
Although bacteria may predominate, moulds, yeasts and protozoa may be associated with bacteria, or, under some conditions, may either cause corrosion by themselves, or modify it drastically. Although many of the effects of microbes on metal are associated with growth this is not necessarily so because a biomass once established may cease to increase but continue its chemical activities often at an accelerated rate, once the controls on growth are relaxed. Methods of protecting materials against microbial corrosion include: (a)Coatings, particularly of resistant synthetic polymers or paints containing inhibiting salts (e.g. Cu2+,Cr3+and Zn”). (b) Controlled dosing with appropriate biocides. (c) Changes in environmental conditions unfavourable to microbial growth, e.g. removal of water from lubricating or fuel oils, good industrial housekeeping, temperature changes. (d) Designs based on fundamental knowledge of microbial ecology. This implies co-operation between engineers and biologists aimed at reducing infection and maintaining unfavourable conditions for microbial growth.
Acid Corrosion Massive and rapid corrosion of metal, concrete and limestones under aerobic conditions is usually caused by the action of sulphuric acid formed by the oxidation of sulphur or sulphide by members of the genus Thioba~illus~. The majority of this genus grow by assimilating CO, at the expenditure of energy produced by the oxidation of sulphur, sulphite, thiocyanate and trior tetrathionate; some strains are sensitive to low concentrations of H,S. The oxidation of sulphur may produce a concentration of up to 1-2% H,SO, and it is this that produces corrosion; these organisms are also exploited for ore leaching4 and in the biological treatment of coke oven effluents’. Thiobacillus thio-oxidans commonly occurs in soil and water and is to be suspected where corrosion is associated with very low pH in the immediate vicinity of the metal6. It may be isolated on acid media and reliably estimated by plate counting on a solid medium. Acid production and corrosion associated with pyritic deposits is caused by Ferrobacillus ferro-oxidans’. Detection T. thio-oxidans is best detected by the strongly acid conditions it generates in a mineral salt solution on which sulphur is floating. Prepare and sterilise by steaming a medium of (NH4),S04, 0.2 to 0.4 g; KH,PO,, 3 to 4g; CaCI,, 0.25g; MgS04, 0.5g; FeSO,, 10mg; sulphur log; tap water to 1000 ml; pH 5 50.3 (some authorities recommend a trace metal mixture in place of the ferrous salt). Add 1 ml or 1 g of material to be tested to 100ml of medium in a conical flask and incubate in air at 30°C. From four days to two weeks the pH of samples should be measured at intervals. An abrupt drop to 2.5 or lower indicates growth of T. thiooxidans. Little turbidity appears in the medium; under a microscope the sulphur particles are seen to be surrounded by motile stubby Gram-negative rods.
THE MICROBIOLOGY OF CORROSION
2: 89
Estimation’ The above medium is reinforced with IOgA of thiocyanate, sulphur is omitted and it is prepared as pour plates by the addition of 3% agar. Organisms other than Thiobacilli will grow from spread samples, but the Thiobacilli are easily distinguished by sulphur haloes (see Fig. 2.19).
Fig. 2.19 Thiobaciiiwthiooxidans (NCIB 8 342). Usually stubby rods, but a few elongated forms can be seen (these are most common in old cultures). x 260 (Crown copyright courtesy Microbiological Research Establishment)
r“
Examples Parker’ described the role of Thiobacilli in the corrosion of concrete sewer pipes; evolution of hydrogen sulphide from the sewage leads to the deposition of inorganic sulphur compounds on the roof of the pipe, and these are oxidised to sulphuric acid by the bacteria, causing a characteristic corrosion pattern in which the roof of the pipe becomes decayed. Iron pipes carrying polluted effluents, and concrete manhole covers examined at the National Chemical Laboratory have been corroded for similar reasons; corrosion of Mouchel cooling towers” had a like origin, as had corrosion of buried iron gas mains in south London”. Corrosion of statues in France has been partly attributed to Thiobacilli”, corrosion of buildings and stonework probably has a similar origin, the source of sulphur for the bacteria being atmospheric pollution. Occasionally, in the experience of the National Chemical Laboratory, ornamental cements containing sulphur have been used for facing buildings, and these form ideal substrates for Thiobacilli. Materials containing sulphur have been used for jointing water mains and Thiobacilli may cause corrosion by forming acid from them”. Instances in which vulcanised rubber has been corroded by these bacteria are knownL4.Corrosion by Thiobacilli is probably more widespread than the published examples would indicate. Prevention Methods of prevention already summarised may be used singly or in combination. Elimination of sulphur and certain of its compounds are most effective but more recently more resistant materials such as polythene or asbestos are used to replace iron and concrete where acid corrosion in severe.
2:90
THE MICROBIOLOGY OF CORROSION
Corrosion by Ferrobacillus F. ferro-oxidants is capable of accelerating the oxidation of pyritic (FeS2) deposits at acid pH values. It is usually found in association with Thiobacillus and was known as Thiobacillus ferroxidans before the distinction between the two organisms was appreciated. It is responsible for pollution problems arising from acid waters in gold and bituminous coal mines 16* ”; such waters are corrosive to pumping machinery and mining installations (see Fig. 2.20).
Fig. 2.20 Ferrobacillus ferro-oxidans (NCIB 8 451), bacteria and encrustations of ferric oxides; the proportion of bacteria was much increased by filtering and centrifugation, x 260. (Crown copyright courtesy Microbiological Research Establishment)
Detection Corrosive waters formed by these bacteria have a pH in the region of 2 to 3, show a brown deposit of basic ferric sulphate, and contain free sulphuric acid. Prepare and sterilise by heating a medium of (NH,),SO,, 0.15g; KCI, 0.05 g; MgS047H20,0.5 g; KH2P04,0.05 g; Ca(NO,),, 0.01 g in lo00 ml of tap water. Prepare a 10% solution of FeS0,7H,O and sterilise by filtration. Add 1 ml FeSO, solution to 100ml of medium in a 250ml conical flask, innoculate with the material being examined, check that the pH is approximately 3.5, and incubate in air at room temperature. Growth of F. ferro-oxidans is indicated by formation of a brown precipitate as compared with slow, slight browning in an uninoculated culture; the organism grows very slowly and may take up to a month to show unequivocal results; the responsible bacteria are straight rods, often difficult to see microscopically amid the debris. Prevention Neutralisation of acid waters with lime has been recommended but the resulting sulphate-laden water may present a disposal problem. Use of acid-resistant machinery and pipes is more satisfactory.
2:91
THE MICROBIOLOGY OF CORROSION
Mechanism and sulphur oxidation Apart from its intrinsic interest the economic importance of acid corrosion and more lately interest in ore leaching, has stimulated considerable work on the oxidation of sulphur, Fe2+ and Mn2+. It must be stressed that the Thiobacilli are obligate aerobes, i.e. that depend on molecular oxygen as a terminal electron acceptor. Possible reactions for the oxidation of sulphur are”: 4s-
+
2S20;
-,S,O; -,SO; + S,Os
+
40; + eo.-
The central role of SO; is apparent. Reactions Ieading to SO:- formation from pyrites are possibly: 2FeS,
+ 70, + 2H,O
-
2FeS0,
+ 2H,O
+ O2+ H2S04-,2Fe,(SO,), + 2H20 3FeS0, + 2 s FeS2 + Fe, (SO,),
4FeS0,
2s + 30,
-
+
2H,O
+ H,SO,
. . . (2.8) . . .(2.9)
. . .(2.10) . . . (2.11)
some of these reactions, e.g. 2.8 and 2.10, take place slowly in the absence of bacteria but are accelerated in their presence.
Other Acid Corrosion Less well studied than the effects of the Thiobacilli are corrosion reactions due to the formation of acids from the oxidation of organic materials. These may include the products of microbial attack on protective coatings such as hesssian sacking and bitumen coatings used for iron pipes initiated by the cellulose-decomposing bacteria. Paper and synthetic rubber coatings for insulation cables may also be attacked. Under strongly aerobic conditions COz is the end product of the oxidation of the organic material, and lead carbonate has been detected as a corrosion product of lead-coated underground cables. Under semi-anaerobic conditions organic acids accumulate and these may lead to simple acid corrosion or alternatively may accelerate corrosion by chelation of passive layers on metal. Besides bacteria, moulds and yeasts may accumulate organic acids even under aerobic conditions and in some cases may synthesise complex ‘secondary metabolites’ some of which, although only weakly acid, are powerful chelating agents. These may be of special significance when microbial slimes accumulate on metal surfaces, as relatively high concentrations of potentially corrosive products may be trapped in them, and corrosion pits result. It is possible that massive pitting in aluminium fuel tanks in aircraft may originate in this wayz0.
Microbial-accelerated Cathodic Depolarisation of Ferrous Metals Corrosion of iron and steel, especially in anaerobic conditions such as waterlogged soils, is usually caused by sulphate-reducing bacteria of which the genus Desulphovibrio is the most commonly occuring. The presence of organic materials such as acetate often stimulates these organisms’ reducing
2:92
THE MICROBIOLOGY OF CORROSION
power whereby sulphate is reduced to sulphide, but some at least of this genus appear to grow as essentially as chemolithotrophs, and reduce sulphate as follows:
so:-
-k 4H2 +
s2--k 4H2O
. . .(2.12)
Simple corrosion by H, S should yield exclusively FeS:
Fe
+ H,S
+
FeS + H,
. . .(2.13)
whereas both hydroxide and sulphide would be expected if the cathodic reaction was due to the mechanism proposed by von Wolzogen Kuhr and van der Vlugt 2’. 4Fe
+ 4H20+ SO:-
+
3Fe(OH),
+ FeS + 20H- . . .(2.14)
It can be seen from equation 2.14 that the ratio of iron corroded to iron in the form of sulphide should be 4: 1, but values from 0.9 to 48 are commonly obtained experimentally. Subsequently it was shown by Booth,’ and his co-workers that the ratios of the corrosion products were dependent on the particular strain of Desulphovibrio and on their rates of growth. Later the activity of the enzyme hydrogenase which bring about the reaction: H,
+ 2e- + 2H+
. . .(2.15)
was correlated directly to the ratios of corrosion products. Further work, especially with methods of continuous or semi-continuous growth of pure cultures coupled with appropriate enzymic and electrochemical measurement, largely confirms the important role of hydrogenase in cathodic dep~larisation~’, but also suggests rates are affected by precipitated ferrous sulphide which can, in the presence of excess ferrous ions, form films on the metal with the possibility of setting up local concentration cells. The role of ferrous sulphide and ferrous salts is as yet unexplained, but the concentration of ferrous ions appears to have an effect even on the anaerobic corrosion of buried pipes in the field as well as in the laboratory. The proposed mechanisms emphasise the role of environmental conditions on anaerobic microbial corrosion. It is noteworthy that rates of corrosion in the field are often much higher than those in the laboratory with pure cultures. This emphasises the complexity of ‘natural’ eco-systemsM. In considering this, two aspects concern engineers, firstly in any new engineering venture involving buried iron or steel and concrete structures. The soil conditions must be evaluated by boreholes and pits with in situ and laboratory testing. Any evaluation must take into account seasonal and exceptional water regimes in the soil. Off site changes in these must also take account of the changes that are likely to occur by reason of engineering works which are expected to alter soil structure and consequently microbial populations and activities. The term aggressive is often used to imply some approximately quantitative estimate of the likelihood of corrosion and depends on measuring factors such as soil water (resistivity), pH, redox potential, salt concentrations and bacterial populations in order to establish criteria for the prediction of corrosion rates,’. Similar measurements for predicting corrosion
THE MICROBIOLOGY OF CORROSION
2:93
in rivers and bottom deposits have been described. Confirmation of these prognostic tests was made on buried or immersed metal plates with periodic measurements of metallsoil or metal/water potentials with reference to copper/copper sulphate electrodes; weight losses were determined after burial for up to three years. Aggressive soils are characterised by resistivity of less than 2000 cm or a redox potential of less than 400mV (see Table 2.21 for evaluation of Eh);when a soil is ‘borderline’and its resistivity and redox potential approximate to the values above, the water content provides a further criterion and it is regarded as aggressive when the water content is greater than 20%. These criteria have been confirmed to be valid for mild steel with a few exceptions, and have been found to be fairly satisfactory for lead and zinc but invalid for aluminium. The metal/soil potential gave little information of value. Tabk 2.21 Redox potentials* of soils in relation to corrosiveness Range of Eh
Class@cationof mrrosiventm
100mV
Severe
100-200 mV 200-400 mV
Moderate Slight Non-corrosive
400 mV
The redox potential is determined with a probe consisting of a platinum electrode and a Hg/Hg$12-CI or Ag/Ag CI-CI reference electrode. If E is the potential of the platinum probe, E, the potential of the reference electrode and Eh tie redox potential of the soil (in mV on the hydrogen scale) then E,, = Ep
+ E, + @(pH-7)
where pH is the pH of the soil
Detection of Anaerobic Corrosion This is immediately recognised in smooth pitting with a black cormion product and smell of hydrogen sulphide when the metal object isfirst exposed; cast iron shows graphitisation. The iron sulphide corrosionproduct oxidises rapidly on exposure to air and should be examined quickly; in doubtful cases acidificationon exposure is confirmatory. Isolation and counting the bacteria depends on establishing strict anaerobiosis. Methodsfor examining soil and water are given below.
Soil samples from the levels in which structure or pipes are to be laid are filled to the top of screw capped bottles, and bacteriological tests are made within 24 hours. Isdetion and Enumeration
Methods have recently been evaluated by Mara and Williams26 and for most purposes the modified ISA medium is suitable: iron sulphite
2:94
THE MICROBIOLOGY OF CORROSION
agar (oxoid) 23 g/1; FeS0,7H20, 0.5 gll; 7% sodium lactate 5.0gA; MgS0,7H20, 2.OgA. The medium is adjusted to pH 7.5 with sterile NaOH after autoclaving at 121°C for 15 min. For liquid cultures, screw-cap test tubes are filled to the brim, for enumeration sterile test tubes (150 x 16 mm) are filled to within 5 mm of the top with inoculated agar medium, covered with a cap of sterile 1.5% w/v agar and closed with a polypropylene cap. Incubation is at 30°C and colonies are counted until the maximum number has developed. These methods are prone to give false positives and large errors, and should only be attempted where adequate microbiological backing is available. Sulphate-reducing bacteria are present in virtually all soils and the qualitative procedure is valuable because it works only when relatively large numbers are present. Hence a positive result with this test is a rough indication of a particularly aggressive soil, though a negative result does not necessarily mean that the soil is innocuous. Quantitative procedures for enumerating sulphate-reducing bacteria in soils and water have been available only for the last few years and data on populations in aggressive and non-aggressivesoils are therefore scantly. Highly polluted waters and soil are known to contain lo5 and lo6 viable sulphate-reducing bacteria per millilitre or gram; waters with less than lo2 bacteria per millilitre are usually innocuous from the pollution point of view. It likely that the aggressiveness of soils would follow broadly the population of viable sulphate-reducing bacteria within these limits. As already stated, aggression will also depend on soil EHaccording to the scale given in Table 2.21. Temperature is also a controlling factor and both psychrophylic (cold) and thermophilic (hot) forms are known, e.g. in electrical transformers, hot water systems.
Hydrogenase Determination This enzyme is of wide occurrence in bacteria where it is concerned with the reduction of nitrate and CO, as well as sulphur. Methods for its estimation depend on measuring some activity of hydrogenase by (a) dye reduction (benzyl viologen or methylene blue), (b) isotopic exchange and (c) evolution of molecular hydrogen. Interpretation of quantitative results is diflcult due to the complex relationshipbetween the enzyme cell structure and theparticular method selected2’.
Sulphate Reducing Bacteria The range of bacteria species capable of sulphate reduction has been greatly expanded since the studies of Pfennig and Widde12*.By using a wide range of growth media and physical conditions well over twelve new species have been added to this genus, including Desulfosarcina, Desulfobacter and Desulfococcus etc. As their names suggest these differ in their morphology but in addition their growth requirements and conditions also differ. One characteristic of these new species is that their growth rates are much slower than Desulfovibrio. In addition to these true bacteria Stetter (1987) has
THE MICROBIOLOGY OF CORROSION
2:95
isolated an Archaebacterium (Archaeglobus) able to grow and reduce sulphate at 65 to 80°C. This newly characterised group are thought to have merged with the eubacteria and evolved as a distinct family including Methanogens and Halophiles which may occur in consortia associated with metal corrosion. Most ‘biosulphur’ deposits originated in Permian and Jurrasic times, periods in which there was a burst of sulphur reduction as yet unexplainedz9.
Corrosion Due to Microbes Other Than Sulphur Metabolisers In recent years it has become apparent that widespread microbial infections of materials in the manufacturing industries can lead to corrosion for the reason briefly outlined above. Examples include the instant rusting of machined parts, corrosion of machine tools, aircraft fuel tanks, hydraulic systems, strip steel etc. The precise role of any specific organism in these instances is diffiicult to determine and will probably remain the province of specialists, largely because of the ubiquitous occurrence of microbes. However, a number of simplified tests have recently been devised to assist engineers and chemists in diagnosing whether or not a particular corrosion is biological in origin. These are based on (1) direct microscopy, (2) measurements of microbial metabolism (oxygen uptake, dye reduction, extracellular enzyme activity) and (3) direct enumeration of specific species of selective media. These are briefly discussed below: 1. Microscopy3’. A binocular phase-contrast microscope equipped with low-power lens and an achromatic x 40 objective and x 10 compensating eyepiece. An oil-immersion lens ( x 100) is required for stained material. Dark ground illumination is also probably worthwhile. Size measurements can be made with an image splitting eyepiece (Vickers Instruments, York). The significance of microscopal examination depends on familiarity with the material and long experience of its microbial populations. The scanning electron microscope is increasingly used for examining the association of microbes and metals under corrosion attack. When coupled with electron probe analysis it becomes a powerful analytical tool. 2. Two simple metabolic tests may be used for evaluating infections by aerobic and anaerobic organisms. (a) Red spot test3’. This relies on the reduction of a soluble colourless tetrazolium dye to an insoluble coloured formazan by respiring microbes. This can be done on plates (Oxoid Limited), ampoules or slides (BDH Limited). The test can be calibrated by estimating by eye the extent of red formazan deposition and correlating it with elaborate counting methods or merely by taking account of any particular situation. Anaerobic bacteria give rapid responses, slower or little response is given by most moulds and yeasts32.Anaerobes33 can be enumerated in the solidifying media for Desulphovibrio ( p .2.93).
2:96
THE MICROBIOLOGY OF CORROSION
(b) A test based on an enzyme released by microbial growth is illustrated by the Avtur Test (BDH Limited) for jet fuel infections. This is based on an approximate assay of the enzyme acid phosphatase. 3. The medium already described for enumerating Thiobacilli (p. 2.88) is a typical selective media and depends on selecting for a mixed population an organism with a specific growth requirement sufficiently different from most of its fellows in that environment. Further selectivity can be designed into a medium by the addition of suitable growth inhibitors, e.g. bile salts (to select coliforms) penicillin (to select Gram negatives), etc. Temperature and other physical parameters can also be used for selection pressure on a mixed community. The use of media enriched with biocide is particularly important for estimating the likelihood of resistant species occurring during treatment of an industrial plant to stop microbial corrosion. Selection media are available for isolating and enumerating fungi (Corn metal agar; Oxoid Limited) and yeasts (Tryptone soya broth; Oxoid Limited) containing 10% w/v sucrose, which inhibits the growth of most bacteria and fungi34.Sampling can be very important especially if false positive results are to be avoided. The usual practice is to collect directly into sterilised bottles, e.g. plastic (Sterilin Limited), from appropriate sections of a system assuming that, for instance, in multi-phase systems representative samples of each phase are collected, e.g. oil and water lubricating lines. Both liquids should be run to waste for at least 30s before collecting. Samples are assayed as quickly as possible or stored at 2-4°C in the dark. After testing, samples are disposed by autoclaving or immersion in a suitable disinfectant overnight. The advent of Biotechnology now mainly directed at medical diagnostics and more recently to the food industry is likely to yield more rapid and simple tests for measuring microbial mass, enzymes etc. and these, e.g. a clip slide measuring ATP, adapted to corrosion diagnosis3’.
Biofilms Most laboratory studies on microbial corrosion have been made in growth chambers such as chemostats with pure cultures; loss of metal from strips immersed in such ‘homogeneous’ systems has been followed by a variety of methods. However, it is apparent that the natural systems under which corrosion occurs are much more complex than this. In soil, for instance, the microbial population is complex and far from uniform as is the supporting soil structure. Corrosion is a surface phenomenon and it is those microbes at the surface with which the name ‘biofilm’ is now applied36.Even though such films may be 10-20 micron in depth, ingenious studies with computer driven micro-probes shows that condition through the film may vary from oxygen saturation at the outer surface to complete anaerobiosis at the surface of the metal”. This is reflected by a layered composition of microbes each type selected by their responses to differences in environmental conditions. Growth and metabolism and their chemical effects, not least corrosion
THE MICROBIOLOGY OF CORROSION
2:97
rates will be affected. There is no doubt that the ability to form such films in a controlled manner and to investigate their properties will increasingly give many incites to the mechanisms involved in the field. This applies especially to the low rates of corrosion found in model systems compared to those in nature. It is worthwhile drawing attention to health hazards associated with film infected water systems which also cause corrosion. Two of the most common are Legionnaires disease and so called ‘humidifier fever’. Because of strong adhesion of biofilms and diffusion rates through the film treatment based on cleaners and chemical sterilisers such as chlorine often fail; similar considerations apply to other systems in industry, e.g. food, paint, oil and gas are examples where biofilm activities have given massive problems. To conclude it must be stressed that recent work has directed attention to the interplay between different microbial species in most of the corrosion effects described. Microbia1corrosion is therefore one special instance of the rapidly developing field of Microbial
D.E. HUGHES REFERENCES 1. Hughes, D.E. and Hill, E. C., Mining and Metallurgical Congress Paper No. 24 (1969) 2. Hill, E. C. in Microbial Aspects of Corrosion, Ed. Miller, J.D.A., Medical and Technical Publishing, Aylesbury (1971) 3. Miller, J. D. A. and Tiller, A., ibid. 4. Le Roux, N. W., ibid. 5. Stafford, D.A. and Callely, A. G., Symposium on Effluent Treatment, B.C.R.A., Chesterfield, to be published. 6. Butlin, K. R. and Postgate, J. R.. in ‘Autotrophic Micro-organisms’, Syrnp. SOC. Gen. Microbiol., Cambridge University Press (1954) 7. Fletcher, A. W., in Micro6iul Aspecfs of Corrosion, Ed. Miller, J.D.A.. Medical and Technical Publishing, Aylesbury (1971) 8. Williams, A. R., Stafford. D. A., Callely, A. G. and Hughes, D. E., J. Bucf., 33, 656 ( 1970) 9. Parker, C. D., Ausl. J. Exp. Biol. Med. Sci., 23, 81, 91 (1945) 10. Taylor, C. 8. and Hutchinson, G. H., J. SOC.Chem. Ind., London, 66, 54 (1947) 1 1 . Chemistry Research 1956, H.M.S.O., London, 16, 61 (1956) 12. Pochon, J., Coppier, 0. and Tchan, Y. T., Chim. et Industr., 65, 496 (1951) 13. Frederick, L. R. and Starkey, R. L., J. Amer. Waf. Wks. Ass., 40, 729 (1948) 14. Thaysen, A. C., Bunker, H. J. and Adams, M. E., Nature, London, 155, 322 (1945) 15. Leathen, W. W., Kinsel, N. A. and Braley, S. R., J. Bucf., 72, 700 (1956) 16. Braley, S. A., Min. Engng., N. Y . , 8, 314 (1956) 17. Braley, S. A., Min. Engng., N. Y., 9, 76 (1957) 18. Butlin, K . R . and Vernon, W.H.J., J. Inst. Water Engineers, 3, 627-637 (1949) 19. Trudinger, P. A., Advances in Microbial Physiology, 3, 1 I 1 (1%9) 20. Elphick, J., in Microbial Aspectsof Metallgrgy, Ed. Miller, J.D.A., Medical and Technical Publishing, Aylesbury (1970) 21. von Wolzogen Kuhr, C.A.H. and van der Vlugt, L. S., Wafer, Den Haag, 18. 147 (1934) 22. Booth, G. H.. MicrobiologicuICorion. M & B Monographs, Mills and Booth Ltd.. London (1971) 23. Panhania, I. P.. Moosavia. A. N., Hamilton. W. A., J. Gen. Microbiol., 132. 3357 24. Hamilton. W. A., Ann. Rev. Microbial., 39, 195 25. Starkey, R. L. and Wight, K. M., Anaerobic Corrosion of Iron in Soil, Amer. Gas. Ass., New York (1945) 26. Mara, D. D. and Williams, D.J.A., J. Appl. Bucf., 33, 543, (1970) 27. San Pietro, A., Methods in Enzymology, Vol. I I , Academic Press, N.Y. 861
2:98
THE MICROBIOLOGY OF CORROSION
28. Pfennig, N., Widdel Trupor, H. G., TheProkuryotes, Ed. Starr, M . P., Springer-Verlag. Berlin (1981) 29. Postgate. J. R.. The Sulphute Reducing Bucteriu. Cambridge University Press, 2nd Ed. (1984) 30. Barer. R.. The Microscope, Blackwell, Oxford (1959) 31. Hill. E. C. and Pemberthy. I.. Metuls und Muteriuls, 2, 359 (1968) M e t d Finishing. 15. 395 (1969) 32. Hill, E. C. and Gibbon. 0.. 33. Hill, E. C.. Aircruft Engineering, 24 July (1970) 34. Callely. A. G.. Process Biochemistry. 3, 11 (1967) 35. Biologically Induced Corrosion. Proc. Int. Conf.. Gaitlersbury, Nat. Ass. Corrosion Eng., University of Delaware (1983) 36. Hamilton, W.A.. Ecology of Microbial Communities. 41 S.G.M.. Symp.. Cambridge University Press (1984) 37. Wimpenny, J. W. T..Lovitt. R. W.. Coombes. J. P., Microbes In Their Natural Environment, Cambridge University Press (1983) 38. Experimentul Microbiol Ecology, Ed. Burns, R. G., Slater, J. H.,Blackwell, Edinburgh ( 1982)
2.7
Chemicals
For the purposes of this section a chemical may be defined as a substance useful technologically and containing over 95% of the principal chemical. In general most textbooks on corrosion control, including this one, give data on the physical and mechanical properties of a particular metal or alloy and then outline its suitability in various environments, rather than considering a particular chemical in relation to its effect on various materials of construction. This approach may be difficult to appreciate, since the design engineer will be given, at first, the environment he has to work with and then is required to choose the most suitable materials. To date this information is grossly inadequate because of the enormity of collating such a large amount of data. For example, if some 400 chemicals are identified as being handled and processed on a large scale and if there are 10 suitable materials, then 4000 systems would have to be considered. Since temperature, concentration and solution velocity are important in determining corrosion rate, and if only five levels of each of the three variables are considered, then the number of experiments to be carried out would be 4000 x 53 = 5 0 0 0 0 0 . There are in addition several other factors that accelerate corrosion and must be taken into account; these include crevices, galvanic coupling, tensile stress, aeration, presence of impurities, surface finish, etc. If these were also taken into consideration then several million experiments would have to be performed to compile such data. There are many instances where two or more chemicals exert a marked synergistic action such that low dissolution rates obtained in either environment become much greater in the presence of both. Further, the corrosivenessof a chemical will be affected by the presence of certain impurities, which may act as either accelerators or inhibitors. TO take all these factors into account would add to an already impossible task and as Evans has remarked’, ‘There are not enough trained investigators in the world to obtain the empirical information to cover all combinations of conditions likely to arise’. Unfortunately corrosion science has not yet reached the stage where prediction, based on a few well established laws, allows selection of materials to be made without recourse to a vast amount of data.
2:99
2: 100
CHEMICALS
Sources of Information In the chemical industry, many processes are required to contain and handle solutions of complex composition containing many aggressive ingredients, but there exists a readily available supply of materials that can be used in the most aggressive environments. The purpose of this section is to indicate the procedure that an engineer could adopt if he was called upon to design plant items for chemicals for which he had little or no prior knowledge or experience. This is a situation that exists within many engineering industries that do not utilise metallurgists or materials scientists and are generally unaware of the existence of the specialist corrosion engineer or scientist and his importance at the planning stage. The most important sources are: 1. The ‘ideal’ source book for designers, which is the one in which the
individual chemicals are listed together with the corrosion rates for a variety of materials under different conditions of temperature, pressure, velocity, etc. 2. Corrosion resistance data lists on specific materials offered for sale. However the engineer must also consider mechanical and physical properties and, last but not least, the cost of the material, its fabrication and protection. Thus reference must be made to books, journals and data that provide this information. 3. Information based on experience, which includes national standards, specifications and codes of practice, and also the technical and scientific literature reviews and reports. Corrosion engineers, technologists and scientists employed by specialist organisations or as consultants form the most important source of information, and their advice should be sought wherever possible.
Selection Based on Chemical Environment The design engineer who requires full information will be disappointed because such books, tables and monographs do not exist. In the present situation the engineer will turn to official standards, etc. However, they are relatively few in number and, though an important source of information, suffer from the lack of reliable data for many popular materials in a wide range of environments that may lead to pitting, stress-corrosion cracking, crevice corrosion and corrosion fatigue. For the engineer who cannot find a suitable specification for a particular chemical environment various sources of information exist. Perhaps the most comprehensive text presently available is the Corrosion Guide’. In this book a great deal of information has been gleaned from the world’s literature in an attempt to fulfil the engineers’ requirements. Because of the magnitude of the problem, three or four levels of corrosivity are given, but only a small but detailed amount of data are reported for certain chemicals with regard to temperature, concentration and velocity. The data given may be useful for making an initial survey of materials prior to further detailed planning or testing. Many of the 800 or so chemicals mentioned do not corrode all of the materials (metals, ceramics and plastics) and this book could
2: 101 be considered as the most useful source of preliminary information. The guide gives corrosive effects of a large number of individual inorganic and organic chemicals and also some important groupings such as fatty acids, cements, concrete, mortar, plaster, oil and varnishes. However, it must again be emphasised that the Corrosion Guide provides no information other than corrosion behaviour, and that this alone is insufficient in a rational approach to materials selection. The corrosion resistance of mild steel is poor in most environments, but this is frequently outweighed by the fact that it has good mechanical properties, and is readily available, cheap and readily fabricated by welding, etc. When the information is available, the Corrosion Guide provides detailed corrosion data on the preparation of various chemicals. For example, in the section on sulphuric acid, the corrosion rates for several alloys are given when used at various stages of an actual process involving that acid. In the section on phosphoric acid, cognisance of the method of production shows also the influence of the minor constituents as well as the major chemical on the corrosion of various materials. Nevertheless, it must be emphasised that even a book as comprehensive as the Corrosion Guide can only cover a limited number of all the possible chemicals used in practice. CHEMICALS
1
'.
\
I
I
20
I
#
a
I
60 Concentration (%HN03 1
I
80
Fig. 2.21 Safe operating conditions for various steels in nitric acid solutions. Isocorrosion lines at 0.1 g rn-'h-' Areas I + 2 + 3: Fe-ltCr-8Ni Area 1: Fe-13Cr Areas I + 2 + 3 + 4, but excluding area within Areas 1 + 2: Fe-17Cr broken line: Fe- I5Si (after Berg3)
2: 102
CHEMICALS
Where a large collection of data exists then it may be effectivelycondensed in the form of diagrams. A popular method is the use of ‘iso-corrosion rates’ plotted on co-ordinates of temperature and concentration for one material and one chemical. Because of the large amount of data on the common acids there are many examples of this type of diagram, e.g. the work of Berg3 who has chosen metals and alloys that are readily available. He has excluded many metals and alloys on the grounds that they are either ‘Non-resistantor can be substituted by cheaper materials.. . .’ From these diagrams an initial selection can be made for materials to be used for the chemicals considered, and although temperature and concentration are included, the effects of flow and the effect of crevices, galvanic coupling or stresses are not indicated. Thus for sulphuric acid, ‘chemical lead’ and ‘silicon iron’ appear to cover a wide range of temperature and concentration; however, the mechanical weakness of the former and the brittleness and difficulties in fabrication of the latter, might mitigate against their use in say pumps, valves and unsupported pipes. For these purposes the more expensive nickel-base alloys have also been listed which may be preferred on various grounds which might include ease of fabrication and tolerance to thermal cycling. An interesting interpretation of these temperature-concentration diagrams has been given by Nelson4. He has combined information of all the materials on the one diagram so that the best materials for the whole range of conditions are seen together. The diagrams taken from Berg’ have been combined in Fig. 2.21 to show the value of this method of presentation.
Selection Based on the Properties of the Material The more usual method of presenting data on the corrosiveness of various chemicals is by reference to a specific class of metal or material. Thus in the present book, the sections devoted to individual classes of materials contains lists of chemicals and in some cases details of their behaviour under various conditions of concentration, flow and temperature (see in particular the sections devoted to metals and alloys). Manufacturers’ and specialist materials development associations publish extensive corrosion data in the form of monographs, and this form of presentation is also used in national standards6. The most recent comprehensive text in this category is perhaps the publicatiop by the Zinc Development Association’. The work is important in that the section on chemicals also deals with common, though complex, chemical formulations, e.g. fireextinguisher fluids, soaps and syndets, agricultural chemicals such as pesticides and fertilisers. This publication also demonstrates the mammoth task of recording all the available data for just one material. A comparable book for mild steel would probably be much larger, whereas for many other materials the information has not yet been determined. Thus at best, only very incomplete data are available in this form.
CHEMICALS
2: 103
Selection Based on Experience The scientific and technical literature' abounds with much information, adding slowly to the massive factual data required for the design engineer who relies solely on his own literature search. The material development associations and manufacturers have, by their own research and development, accumulated a great deal of information about their own product and this is transmitted directly to potential users of their materials,
Selection Based on Scientific Principles Ideally the design engineer requires an equation which condenses all this information and from which he can calculate the effect of a particular chemical upon a range of materials, and the limiting conditions of say temperature, concentration and velocity. To achieve this objective he needs to know which of the properties of the chemical and the material are the most important in determining the interaction leading to corrosion.
5
-
*
4
%
E E
2 3
e
s b
HCI W
N a O H
2
V
1
0 PH
Fig. 2.22 Effect of pH on metals relying on passive films for protection, e.g. zinc
The majority of metals and alloys available depend for their resistance to corrosion on the properties of an oxide film or corrosion product which is formed initially by the corrosion process. In many cases the protectiveness of the oxide film is determined by its stability in aqueous solutions in a specific pH range, either chemically dissolving to form aquocations at lower pH values or complex anions (aluminate, ferroate, plumbate, zincate, etc.) at higher pH values (Fig. 2.22). An important property of the chemical is therefore the pH value that it develops when dissolved in water. For many materials and many chemicals this is the overriding factor and in many cases
2: 104
CHEMICALS
the oxide film is found to dissolve uniformly by interaction with hydrogen ions, the rate varying with concentration in accordance with equations such as: Rate = constant x (conc. of H + ions)" As the film dissolves more oxide film is formed, i.e. the metal/oxide interface progresses into the metal, and the overall rate may be low enough to be acceptable for a particular process. In other cases, the corrosion products precipitate on the surface of the oxide and either accelerate the overall rate by enhancing diffusion of ions through the porous outer layers or, when less porous layers are formed, access of hydrogen ions to the inner oxide surface is reduced thus decreasing the rate. The ability to form a second barrier film when the conditions are such that the oxide film is dissolving rapidly, is another important property of the chemical, i.e. its ability to form highly insoluble corrosion products which would allow the physical blockage of the surface. There are, however, a certain number of chemicals that have the ability to cause non-uniform dissolution of oxide films and the more serious form of pitting corrosion occurs. Certain anions have the ability to attack the oxide film in small areas and allow corrosion at high rates confined to these areas. This often leads to deep penetration of the metal although overall weight-loss is very low. Many chemicals appear to have this ability but the most insidious are the halides, particularly chloride, which occurs as impurity in many chemicals, raw materials and water supplies. Many chemicals have a reputation of being difficult to handle but this is often due to the presence of these types of impurity. For example, many chlorinated hydrocarbons in the presence of water hydrolyse to form hydrochloric acid which is responsible for the corrosiveness of these non-ionic substances and the possibility of attacks on ships' cargo compartments.
Rationalisation of Data For this purpose a chemical may be said to either (a) dissolve a material uniformly, the rate depending on pH or (b) non-uniformly leading to pitting corrosion. There are, of course, examples of intermediate behaviour but this simple division leads to a method of assessing the probable behaviour of a chemical. Only (a) will be discussed in detail since pitting corrosion is dealt with elsewhere in this book. (See Section 1.6 and Chapter21.) UniformCorrosion The corrosion rate of a metal, which depends for its protection on a passive oxide film, may be predicted from a simple empirical adsorption law (Freundlich): C, = AC" where C, = the corrosion rate, A = the specific rate constant,
2: 105
CHEMICALS
C = the concentration of hydrogen ions and
n = an integer, which may be modified by taking logarithms and substituting pH for C, when the equation becomes log C R = K - n ( p H ) where K = log A . Reliable pH data and activities of ions in strong electrolytes are not readily available. For this reason calculation of corrosion rate has been made using weight-loss data (of which a great deal is available in the literature) and concentration of the chemical in solution, expressed as a percentage on a weight of chemicaVvolume of solution basis. Because the concentration instead of the activity has been used, the equations are empirical; nevertheless useful predictions of corrosion rate may be made using the equations. In strong acid solutions many common structural materials dissolve uniformly and this assumption is reasonable in many real situations. The data given in the monograph by Berg3 are used in order to demonstrate the universal application of the technique. Four main types of behaviour may be identified for metals and alloys in various acids at different temperatures and concentrations. Type 1. Increasing corrosion rate with increasing concentration and teinperature In this case the equation obeyed is A
IogC, = n l o g C - - + K, T where A , K and n are constants specific to a particular material and chemical. This relationship predicts that corrosion rate C, increases continuously with temperature and concentration.
Example I. Hard lead (antimoniacal) can be used in sulphuric acid to quite high concentration but it displays an increasing corrosion rate with increasing temperature and concentration. Relationships are complex, but the general form of the equation may be used: 2 500 10gCR=410gC--T
1.25
The prediction is not very accurate, but in Fig. 2.23 comparison between the actual and calculated value shows the effectiveness of the empirical equations especially at higher temperatures. The iso-corrosion line obtained by substituting 0.1 gm-*h-' in the equation gives a reasonable guide to the temperature and concentration limitations of this material. The activation energy, which may be obtained from the temperature coefficient, indicates an activation-controlled reaction, so that flowing solutions should have a small effect on corrosion rates. Providing that the designer allows for the low mechanical strength of lead, the material could be considered for situations where high flow is involved, but abrasion of the protective film of corrosion products must be avoided.
2: 106
CHEMICALS
i
m
f 6001
+ LO
-
20 -
0
1
I
20
1
I
I
60 80 Concentration (% H,SO,)
GO
I
100
Fig. 2.23 Type 1 behaviour; increasing corrosion rate with increasing temperature and concentration, e.g. lead in sulphuric acid. Iso-corrosion lines at 0.1 g m-*h-'
Type 2. Decreasing corrosion rate with increasing concentration and
temperature Metals and alloys that show this behaviour are important because they can be used for more concentrated acids. Mild steel is a good example in that it resists concentrated H2S04and HNO,. This behaviour is associated with true passivity in the case of HNO,, but in H,SO, there is also a precipitation of corrosion products which appear to block the surface and prevent further attack. For higher temperatures and less concentrated solutions, mild steel is not resistant but other ferrous alloys have been developed that retain this characteristic over a much wider range of temperature and concentration. The empirical equations may be applied, and although an increase in temperature increasesthe rate, the form of the equation must be changed to allow for the decrease in corrosion rate with increasing concentration: IogC, = K-nlogC
-A T
From the temperature factor A , an activation energy may again be calculated which gives a useful indication of the influence of flow-rate on corrosion rate.
2: 107
CHEMICALS
&le 2. The corrosion rate of silicon iron (Fe-15Si) in static H2S04 may be predicted from: IogC, = 6-54 - 1-71lOgC--
1540
T
The calculated and experimental iso-corrosion lines are given in Fig. 2.24 and show reasonable agreement with experimental values when the corrosion rate was 0.1 gm-*h-'. However, at a concentration of 50% there is a sudden decrease in rate and the material is useful at all temperatures beyond this concentration. The activation energy has been calculated as approximately 30kJ/m and suggests that the corrosion rate will increase with increasing flow-rate, which could increase considerably the corrosion rates for static solutions quoted above. A diffusion-controlled reaction would be in accord with a salt passivation or blocking-type mechanism on a surface otherwise protected by SO,. The material is, however, used for pumps, stirrers, etc. in high concentrations of sulphuric acid where a thick anodic film is formed.
7
Calculated from:-
0
1
20 LO 60 80 1 3 Concentration (*/a H2S0,) (a)
0
20 LO 60 80 1 0 Concentratioh ( % H3P0,) (b)
Fig. 2.24 Type 2 behaviour; decreasing corrosion with increasing temperature and concentration, e.g. (a) Fe-15Si in H,SO, and (b) aluminium bronzes in H,PO,. Iso-corrosion lines at 0.lg rn-*h-'
Example 3. Aluminium bronze (Cu-7A1-3Fe) in H,PO, follows the equation: IOgC, = 5.38 - 1*3410gC--
1430
T
This alloy is suitable for high concentrations of de-aerated phosphoric acid providing the concentration is above a certain maximum, which varies with temperature in accordance with the equation. The calculated activation
2: 108
CHEMICALS
energy suggests that diffusion control may be operating and that flowing solutions should influence corrosion rate to a marked extent. The corrosion rate is influenced by aeration which is perhaps responsible for the observed diffusion control. Type 3. Chemicals that show a maximum rate at a certain concentration A great many metals and alloys are unattacked in dilute solution but the
rate increases with increase in concentration up to a maximum and then decreases with further increase in concentration. Thus the rates in the very dilute and highly concentrated acid may be acceptable. This behaviour has been observed for many metals and alloys in acid, alkali and neutral solutions of many salts. Brasher9 et al. have shown that for many chemicals the maximum rate occurs at an acceptably low concentration of chemical followed by a low rate with further increase in concentration. Therefore, chemicals may be classed as ‘aggressive’ or ‘inhibitive’, depending on the position of the maximum corrosion rate in relation to anion concentration. The falling off in corrosion rate arises from the formation of an anodic film that is not affected by an increasing concentration of hydrogen ions. This has often been identified as a crystalline metal salt or oxide associated with the low solubility of the corrosion product. The observed behaviour is therefore a combination of both Types 1 and 2 manifesting itself over the range of temperatures and concentrations.
Example 4. Nickel-chromium steel (Fe-18Cr-lONi-2Mo) in H2SO4. In the range where the rate increases with concentration the relationship is: l o g C R = 8 * 2 +1.6710gC--
O Fig. 2.25
20
3 245 T
20 60 80 Concentration (%H25O4)
loo
Type 3 behaviour; maximum corrosion rate with change in concentration, e . g . nickel and chromium steel in H,SO,. Iso-corrosion lines at 0.1 g d h - ’
CHEMICALS
2: 109
when C is in the range 0-40% and T i n the range 20-60OC. For the range where there is a decreasing rate:
when C is in the range 70-100% and T i s in the range 1O0C-100"C. In the region of ascending rate the reaction is clearly governed by a diffusion process and is susceptible to flow-rate, whereas at high concentrations temperature is of greater importance than solution velocity which would have very little effect on the corrosion rate. At the high rates of corrosion (above about 10 g m-*h-') these relationships do not apply between 40 and 60% H2S04because of the transition between the two types of behaviour. The use of the equations is most effective when some acceptably low rate is chosen, say 0.1 g rn-*h-' and the iso-corrosion line is calculated over a range of temperatures and concentrations (see Fig. 2.25). Type 4. Decrease in corrosion rate with increase in temperature An increase in temperature generally increases reaction rates, and the previous examples show that this applies to corrosion rates. However, at certain high rates of corrosion a decrease in rate can occur when the solubility of certain anodic products is exceeded, owing to surface coverage by such films. Because an increase in temperature leads to an increase in corrosion then it should be possible to reduce corrosion when the dissolution is greater than a certain high rate. This type of behaviour has been found and gives an insight into the seemingly uncharacteristic behaviour of some materials in chemical processing plant.
Cmccn tration (% H3P0,, 1
Fig. 2.26 Type 4 behaviour; decreasing corrosion rate with increase in temperature, e.g. F e - 1 7 - 0 in H,PO,. Iso-corrosion lines at 0.1 g m%'
2: 110
CHEMICALS
Example 5. Chromium steel (Fe-17Cr) in phosphoric acid at low concentrations shows a decreasing rate with increasing temperature (see Fig. 2.26) presumably due to surface coverage by metal phosphates. It is apparent that each metal/chemical system should conform with one or more of the four types of behaviour already mentioned, and the behaviour should therefore be capable of prediction by means of one of the following: A
(i) logC, = K + nlogC - T A (ii) logc, = K - nlogC - T
A T A ( i v )log C R = K - nlog C + T
(iii) log C, = K + nlog C + -
Example 6. Hastalloy B (Ni-26Mo-4Fe) in formic acid shows all the previous types of behaviour and includes the falling off in rate with increase in temperature (see Fig. 2.27). Boiling point curve
100
80 V
?I
60
d .
e
n
E
LO
I-
20
0
40 60 80 Concentration (% formic acid I
20
100
Fig. 2.27 Example of Types 1, 2, 3 and 4 behaviour. Hastalloy B in formic acid 1. Increasing corrosion rate with increasing concentration and temperature 2. Decreasing corrosion rate with increasing concentration and temperature 3. Maximum corrosion rate at a certain concentration 4. Decrease in corrosion rate with increase in temperature
In the main there exists, for each system of a chemical in contact with those metals and alloys that rely on a passive film, the possibility of an increase in corrosion rate with increasing concentration but reaching a maximum and followed by a decrease in rate. If the concentration when this maximum is reached is low, then the chemical is ‘inhibitive’. The effect of temperature on corrosion is dependent on the position of the maximum concentration. FOTmany chemical/metal systems this maximum may be at a temperature
CHEMICALS
2:111
beyond the boiling point of the solution, in which case only part of the behaviour described in Types 1, 2 and 3 is observed. The boiling point line is an artificial barrier, since constant temperature prevails only at constant pressure and concentration. However, many chemical processes are carried out under pressure and the onset of the behaviour described for Type 4 will manifest itself for many other metals and alloys. Thus serious corrosion at higher temperatures resulting from higher pressures may not necessarily occur.
Non-uniform Corrosion The arbitrary division of behaviour has been made because of the extreme behaviour of some chemicals that initiate small areas of attack on a wellpassivated metal surface. The form of attack may manifest itself as stresscorrosion cracking, crevice attack or pitting. At certain temperatures and pressures, minute quantities of certain chemicals can result in this form of attack. Chloride ions, in particular, are responsible for many of the failures observed, and it can be present as an impurity in a large number of raw materials. This has led to the development of metals and alloys that can withstand pitting and crevice corrosion, but on the whole these are comparatively expensive. It has become important, therefore, to be able to predict the conditions where more conventional materials may be used. The effect of an increase in concentration on pitting corrosion follows a similar relationship to the Freundlich equation where CR = K c " ,
but concentrations below which pitting does not occur are generally very low. The effect of temperature may again be represented by the Arrhenius equation: CR= Bexp( - A / T ) )
where A and B are constants. The corrosion rate C, is meaningful when expressed as l/y, where y is the time required for the onset of pitting. Conditions may therefore be chosen when y is large. In other sections of this book the anions and other chemicals which enhance pitting, crevice and stress corrosion are discussed in greater detail.
Equilibrium Potential-pH Diagrams with Anions at Various Temperatures There are now many diagrams available for metals and alloys which have been calculated not only for metal-H,O systems, but also for metal-H,Oanion equilibria (Sections 1.4 and 7.6). Many authors have now produced diagrams in which the effects of temperature have been calculated usually over the range 20°C to 200°C. In many cases, experiments have been over the range carried out to verify the predictions made in the diagrams.
2: 112
CHEMICALS
The following list is a guide to the availability of some of these diagrams: Anion effects Fe-S-H20 10, 11 Fe-Cl-S-H,O 12, 13 FeO -Cl-H20 12 Fe-C0,-H,O 14 Fe-Moo, -H, 0 15 Ni-S-H20 12 Ni-P-H20 16 Cr-S-H,O 17 Cr-Cl-H,O 17 Cu-S-HZO 18 Cu-CO,-HZO 14 Pb-CO3-HzO 14 Zn-CO,-H,O 14 Sn-CI-H,O 19
Temperature Mn-H,O 20 MO-HZO 21 Pt-H,O 21 Ti-H,O 21 AI-HZO 22 Fe-H,O 23 Ag-HZO 23 Cr-H,O 21, 24
25°C to 300°C 25°C to 300°C
25°C to 250°C 25°C to 90°C 25°C to 250°C 25°C to 200°C 25°C 25°C 25°C 25°C to 300°C 25OC to 25°C to 25°C to 25°C to 25°C to
300°C 300°C 300°C 300°C 300°C
Alloys
70YoCu 30YoZn-Cl-H2O 25 MgZn,-C1-HzO 26 2%Mg, 6-4%Zn, CU-CI-H, 26 P. J. BODEN REFERENCES 1. Evans, U. R., The Corrosion and Oxidation of Metals, Arnold, London (1960) 2. Rabald, E.. Corrosion Guide. Elsevier, London, 2nd edn (1968)
3. Berg,
F. F., Corrosion Diagrams, V.D.1 .-Verlag, G.m.b.H ., Diisseldorf (1965)
4. Nelson, G . A., Corrosion Data Survey. Shell Development Co., Emeryville. Calif. (1960)
5 . Examples are: WigginAlloys 100. Henry Wiggin Co. Ltd., Hereford; Copper in Chemical Plunt, Copper Development Association, London (1960); Corrosion Resistance of Stainless Steel, Publication 112/1, Firth Vickers Stainless Steel Ltd.. Sheffield; Corrosion Resistunce of Titanium, Imperial Metals Industries (Kynoch) Ltd., Birmingham 6. British Standard Code of Practice C.D. 3 003: Linings of Vessels und Equipment for Chemical Processes. Part 1: Rubber, Part 2: Glass Enamel, Part 3: Lead, Part 4: P.V.C., Part 5 : Epoxy Resins, Part 6: Phenolic Resin, Part 7: Corrosion and Heat Resistant Materials, Part 8: Precious Metals, Part 9: Titanium and Part 1 0 Brick and Tile 7. Slunder, C. J. and Boyd, W.K., Zinc: Its Corrosion Resistunce, Zinc Development Association, London (1971) 8. Examples are: Conference on Materials Selection, Gen. Ed. Verink, E. D., Jr., Gordon Beach, Met. SOC.,40,London (1966). See particularly, Gachenbach, J . E.,‘Material Selec-
CHEMICALS
9. 10. 11. 12. 13. 14.
15. 16. 17. 18. 19.
20. 21. 22. 23. 24. 25. 26.
2:113
tion in Chemical Industry’ and Koelbl, H. and Schulze, J., Selection of Materials for Chemical Zndustty. Proceedings 4th International Conference on Metallic Corrosion, Amsterdam, N.A.C.E.. New York (1972) Brasher, D. M.. Reichenberg. D. and Mercer, A. D.. Brit. Corr. J.. 3, 144 (1%8) Barry, T. I. in Diagrams of Chemical and Electrochemical Equilibria. CEBELCOR, 142, Brussels (Aug. 1982) Biernat, R. J. and Robins, R. C., ElecfrochimicaActa, 17, 1261 (1972) Macdonald, D. D., Syrett, B. C., Corrosion, 35, 10, 471 (1979) Yang, X. Z., E.P.R.I., Palo Alto, California (1981) Masuko, N., Inque, T., Kodama, T., Proc. Int. Cong. Mef. Corr., 2, 280, Nat. Res. Comm. Canada. Toronto (1984) Kodama, T., Ambrose, J. R., Corrosion, 33, 155 (1977) Gool, A., Boden, P. J.. Harris, S. J., Trans ZMF, 66. 67 (1988) Macdonald. D. D., Syrett. B. C.. Wing. S . S., Corrosion, 35, 1, 1, (1979) Kwok, 0.J., Robins, R. G.. Int. Symp. Hydrometallurgy, AIMMPE, Chicago (1973) House, C. I., Kelsall. G. H.. Electrochimica Acta, 29, 10, 1459 (1984) Macdonald, D.D., Cow. Sci., 16. 461 (1976) Lee, J. B., Corrosion, 37, 467 (1981) Macdonald, D. D., Butler, P., Cow. Sci., 13, 259 (1973) Pound, B. G., Macdondd, D. D., Tomlinson, J. W., ElectrochimicaActa, 24, 929 (1979) Radhakrishnamurty, P. et al., Corr. Sci., 22, 753 (1982) Verink, Jr., E. D. in Electrochemical Techniques for Corrosion Studies, 43 N.A.C.E., Houston, Texas, (1977) Ugiansky, M., Kruger, J., Staehle, R., Proc. 7th Congr. Mef. Corr., ABRACO, Brazil. 605 (1979)
2.8 Corrosion by Foodstuffs
Foodstuffs, like other chemical substances, are responsible in all phases of processing (including packaging) for corrosive effects of different kinds on constructional materials. These effects are, in addition, influenced by environmental conditions of processing, i.e. by the temperatures involved, by the rates of flow (in particular, erosion is associated with very high rates of flow), and by alternating stresses which may be present in component parts of process machinery (this effect is specifically called stress corrosion). The presence or absence of oxygen is generally important in relation to the extent of corrosion produced. The corrosive effects to be considered (mainly simple corrosion of metals) are, as would be expected from the edible nature of foodstuffs which are not excessively either acidic or basic but which may contain sulphur, less severe than those often encountered with inedible materials containing reactive substances. The importance of corrosive effects where foodstuffs are concerned lies not so much in the action of the foodstuffs on the metal involved as in the resultant metal contamination of the foodstuff itself, which may give rise to off-flavours, in the acceleration of other undesirable changes (by the Maillard reaction' for example), and in the possible formation of toxic metallic salts. Metal ions generally have threshold values of content for incipient taste effect in different liquid foodstuffs. Except in the case of the manufacture of fruit juices and pickles, process plant failure through corrosion must be rare. Nevertheless all foodstuffs, particularly liquid ones, should be regarded as potentially corrosive and capable of metal pick-up which may be undesirable.
Construction Materials The most suitable construction material should usually be selected on the basis of prior experience or direct tests. An exact analysis of the constituents of foodstuffs is not always possible, since they are often of complex composition, but it is usually the organic acids present which are the important corrosive agents. Corrosion rates not in excess of 10-100gm-2 d-' are generally to be sought in food process plant. It is also necessary to bear in mind that in many cases the plant will have to be regularly disinfected by cleaning and sterilising solutions, These should either be non-corrosive or 2: 114
CORROSION BY FOODSTUFFS
2: 115
contain inhibiting additives. As always, the choice of construction material is determined in practice in relation to ease of fabrication and cost, which depend upon complexity of construction (cf. jacketed pans with a h.t.s.t. pasteuriser). Electrolytic couples should be avoided. Since bacteria may be pathogenic or, like certain metals, cause off-flavours, the constructional material for foodstuffs plant should be of such a nature that its surfaces can easily be kept clear and free of bacterial lodgment. In selection of construction materids, the tendency is therefore to play safe; stainless steel is on this account generally specified’ and is always used in cases of doubt. The suitability of other materials should not, however, be overlooked3. Prior to the general advent of stainless steel in 1924, tinned copper was much in vogue. Today titanium, although expensive, represents a possible alternative4. The precise effect of different quantities of various trace metals should be considered in relation to the specific type and variety of foodstuff in question. Differing quantities of inhibiting substances may be present in different varieties of the same foodstuff. Thus for example different quantity levels of trace metals are reported as causing off-flavours in various fruit juices, and again the quantity level of trace metals causing rancidity in edible oils depends upon whether the oils are crude or refined. The possible toxicity of these trace metals in different foodstuffs has been investigated, recommendat i o n ~by~ the Food Standards Committee of the Ministry of Agriculture, Fisheries and Food are made from time to time, and there are statutory regulations in force in different countries’ for some foodstuffs. To state the position briefly, aluminium and tin in the quantities normally encountered are considered as non-toxic. Tin, however, in excess is considered undesirable, and may be responsible for gastro-intestinal upsets, so that the accepted limit for canned foods is taken at 250p.p.m. Particular groups of foodstuffs which create corrosive environments, and the processing of these foodstuffs, will now be briefly discussed. Indications will also be given of construction materials recommended to meet these situations.
Foodstuffs Liquid ioodstuffs
a considerable degree of processing, they are probably the most important group. Within this group, there are certain foodstuffs which are slightly acid in reaction, such as instant coffeeextracts and the fruit juices (for example, lemon juice at a pH of 4, and blackcurrant at 2.5). The processing6.’ of these fruits involves milling, pressing, concentration by evaporation as necessary (temperatures are of the order of 40 to SOOC), pasteurisation (h.t.s.t.), subsequent storage and canning. The common metals, together with aluminium, will be attacked; the degree to which they are susceptible to corrosion is determined by oxygen content and the presence of hydrogen acceptors. Some published information on rates of attack is e.g. boiling lemon juice and tomato juice give figures of 1 400 and 180 g m-*d-’ respectively on aluminium. The use of sulphur As these demand
2:116
CORROSION BY FOODSTUFFS
dioxide for the protection of fruit juices against moulds and bacteria presents an additional corrosion hazard. The important commercial feature of these juices, especially significant with blackcurrant and tomato juices, is their ascorbic acid (or vitamin C) content, of which loss by oxidation is known to be accelerated both by heat and by metal (particularly copper) contamination. The effect of copper has been carefully investigated for pure ascorbic acid", and more recently ascorbic acid in blackcurrant juice and model systems'*. There are, however, oxidation inhibitors of different kinds (which may themselves be heatsensitive) present in various fruits, which give differing results. The presence of metals will also affect flavoursL3,may cause discoloration, and may give rise to clouding effects, as in apple juiceI4. The first choice of construction material for processing fruit juices is stainless steel (BS En 58A or En 58B-the B type being similar to the A type but providing in addition for resistance to weld decay-or En 58J). Where sulphur dioxide preservation is conducted, high-molybdenum stainless steel (type En 585) is used. Both mild steel and copper should be avoided. There are, however, other materials available, e.g. tinned copper (provided that the coating is satisfactory, and that it is shown to be otherwise suitable), enamelled metal (less popular than formerly on account of the danger of chipping), and Pyrex glass for piping and linings. Gun-metal and aluminium bronze are useful metals for various parts of pressing equipment. Plastic materials are suitable for storage-tank linings ", provided that the surfaces can be sterilised. Nickel and Monel have certain applications; Monel for example offers good resistance to erosion. Aluminium, though known to give fair resistance to organic acids, is not generally used for normal-temperature work; certain alloys may be found more suitable than the pure metal. For canning, conventional tin cans are available"*", but it is generally recommended that the coating be protected by acid-resisting lacquers, and, when sulphur dioxide has been used'*, by an acid- and sulphur-resisting lacquer. The effect of nitrates in water causing detinning is of further importance. Milk is subjected to the process operations' of pasteurisation (h.t.s.t., 71 1-72.2"C for 15 s), evaporation (temperatures of the order of 4O-5O0C), homogenisation, sterilisation and drying. In addition, milk is processed into other dairy products such as butter, cream and cheese by what are essentially normal-temperature operations. Milk is approximately neutral in reaction, although lactic acid is present; the lactic acid content is increased by natural souring or by the artificial souring necessary for cheese and butter manufacture. This is perhaps the only constituent of milk which is responsible for any metal attack. Protective films of precipitated colloids may be formed by heating. Milk is sensitive in flavour to the presence of such metals as copper and iron, the chief result being a 'castor oil' or 'tallowy' taste which probably arises from fat or lecithin turning rancid. Tin is regarded as not affecting the flavour of milk. Since the flavour of milk is also influenced to a great extent by bacterial action, cleansing and sterilising solutions are regularly used in milk process plant. Stainless steel is now generally recommended for all phases of processing (type En 58J is not normally necessary except in cases where the process liquid is specially acid)". Aluminium is widely used but suffers from the
CORROSION BY FOODSTUFFS
2:117
defect of pittingMat higher temperatures. Tinned copper is suitable, where the construction permits its use. Nickel is satisfactory for cold milk and for milk that is being heated up, but is stated to be unsuitable for milk which is cooling down. Glass or enamelled steel is suitable for storage materials and for pipe-lines. Where rapid temperature fluctuations in process operations are involved, however, glass is at a disadvantage. Plastics are used for lining milk-storage vessels and for piping up to temperatures of say 50°C. There are certain reservations; with P.v.c., certain plasticisers used may impart an objectionable taste, and certain brands become opaque. Alkaline cleansing agents (up to strengths of 5% caustic soda solution equivalent) do not affect P.v.c., whereas phenolformaldehyde linings are affected. There are, however, other agents available for such linings. The edible oils and the margarine emulsions derived from them are large tonnage foodstuffs. High- (Le. 190°C in deodorisation) and mediumtemperature operation, and low-temperature storage are involved. The refining of crude oils is carried out in the presence of caustic soda at 60°C. Refined oils would not be expected to attack the common metals, but in fact edible oils, particularly in the crude state, contain varying amounts of fatty acids, and margarine emulsions contain salt, both of which are capable of attacking metals. Here again, the effect of the metal contaminant on the foodstuff is more important than the corrosive effect of the foodstuff on the metal. It is now common knowledge that copper and iron impurities act as catalysts in the oxidation of oils, to cause slight or marked rancidity. Various oils, however, contain different inhibiting substances. Stainless steel is the prescribed material (En %A, En 58B; or En 585 for emulsions containing salt). Deodorising plant handling refined oils may be fabricated in mild steel. Glass-lined mild-steel tanks are available for storage. The manufacture by fermentation and/or distillation of alcoholic beverages containing a wide range of organic materials including acids, whose effect it would be difficult to assess, is traditionally carried out in copper plant, and this has found full consumer acceptance.
Sugar Products The foodstuff sugar is relatively inert as a chemical, though when it is processsed as an aqueous solution slightly acid conditions may be present and boiling temperatures may be involved. There are no deleterious effects due to trace metals. Mild steel is generally recommended for sugar processing and for the handling of aqueous solutions in allied industries. The processing of fruits into jams and purtes, which consists essentially in boiling in open pans, is closely related. The environmental factors here are the natural acidity of the fruits and the possible presence of sulphur dioxide from stored fruit pulps. Traditionally, copper plant, clad on cast iron, uncoated, tinned or even silvered, is used, but stainless steel is now widely adopted. In the manufacture of sugar confectionery, including chocolate, the main ingredients are sugar and glucose, milk (including condensed milk), and cocoa fat, and the essential operations those of boiling and compounding. Cocoa fat, like other edible fats, is liable to oxidative rancidity. The modern choice of constructional material is again stainless steel.
2:118
CORROSION BY FOODSTUFFS
v8@86/6 hOCeSSi??g
In pickle and sauce manufacture, vinegar (an approximately 5% solution of acetic acid), and salt at temperatures up to boiling point are the important corrosive agents. Sodium chloride, particularly in the presence of oxygen and acids, is known to cause rapid attack. Metal contamination (particularly iron) is likely to cause darkening by reaction with tannin present, which is either leached from wooden vessels or derived from certain spices. In pickle and sauce manufacture, stainless steel (En 585) is recommended; it is interesting to note, however, that it is claimed that the flavour of piccalilli liquor is more readily brought out in the presence of iron”. This belief finds expression in the practice of using cast iron process vessels. Meat and Fish Products
Despite attention to hygiene and cleanliness, corrosion of base metals by meat juices and deterioration of meat and fish owing to metal contamination is liable to occur. Stainless steel is recommended in soup and paste manufacture, and aluminium has a certain application. Mild steel is however used in the corned beef industry, for meat pre-cooking. Other Food Products
Many foodstuffs are in the form of solids or processed powders, and do not offer serious corrosion problems, though mild steel equipment in infrequent use, or after washing down, can develop slightly rusted surfaces. This material is usually undesirable if it finds its way into food products. Scouring batches of dry foodstuffs is one solution to the problem if stainless steel is not used or affordable. Hygiene and cleanliness are, however, dominant factors. Meat and fish are very liable to bacterial putrefaction; in this connection an interesting innovation is the increasing use of easily cleaned aluminium fish boxes. It is also possible that copper should be avoided in contact with herrings, which have a high fat content. Mention should also be made of glutamic acid and invert sugar which are used in foodstuffs and demand the use of hydrochloric acid-resistant material in manufacture, and of the essential flavouring oils which should preferably be stored and prepared in stainless steel and aluminium equipment. R. J . CLARKE REFERENCES 1 . Reynolds, T. M.. ‘The Chemistry of Non-enzymic Browning’, Advances in Food Science, 14, 168 (1%5)
2. Gilroy, P. E.. Food, 21. 255 (1952) 3. Various authors. Ann. Falsg., 213 (1950) 4. O’Keefe. J., Be(l’s Food und Drugs, Butterworths, London, 14th edn. (1968)
CORROSION BY FOODSTUFFS
2:119
5. Reith. J. F.. Ann. Bromatologia, 8, 145 (1956) 6. Clarke, R. J.. Process Engineering in the Food Industries. Butterworths. London (1957) 7. Tressler, D. K. and Joslyn, M. A., Fruit and VegetableJuice Processing Technology, Avi. Westport (1968) 8. Blount, A. L. and Bailey, H. S., Trans. Amer. Inst. Chem. Eng., 18, 139 (1926) 9. Gilroy, P. E. and Champion, F. A., J. Soc. Chem. Ind.. 67, 407 (1948) 10. McKay, R. J. and Worthington, R., CorrosionResistance ofMetals and Alloys. Reinhold, New York (1936) 11. Joslyn, M. A. and Miller, J., Food Res.,14. 325 (1949) 12. Timberlake, C. F., J.S.F.A., 8, 159 (1957) and 2. 268 (1960) 13. Schrader, J. H. and Johnson, A. H., Industr. Eng. Chem. (Industr.), 26, 179 (1934) 14. Kieser, M. E. and Timberlake, C.F., J.S.F.A.. 8, 151 (1957) 15. Docherty. A. C. and Hughes, H.,Chem. Ind., 1 171 (1959) 16. Hartwell. R. R.. Advances in Fowl Reseorclr, 3, 327 (1950) 17. Alderson. M. G.. Food Manuf., 67 (1970) 18. Board, P. W.,Holland, R. V. and Elborne. R. G.. Journal Sci. Food. Agric., 18, 232 (1967) 19. Bottom, G. H., J. Soc. h i r y Tech., 6, 179 (1953) 20. Aziz, P. H.and Goddard. H. P., Industr. Eng. Chem. (Industr.). 44, 1 791 (1952) 21. Smith, E., Corrosion Resisring Steels-Application in the Pickle and Souces Industry, B.F.M.I.R.A., Scientific and Technical Survey, December (1950)
2.9
Mechanisms of Liquid-metal Corrosion*
The corrosion of metals and alloys by liquid metals generally follows the pattern of the formation of metallic alloys, i.e. solution and intermetallic compound formation and the corrosion process is often one of simple dissolution in the liquid metal. In some special cases electron-transfer processesinvolving reducible impurities in the liquid metal -may modify or even override the simple dissolution process. This is especially true when, as with liquid alkali metals, the solubility of structural metals is very low. Very often under isothermal conditions equilibrium between an alloy and liquid can be approached. Continued corrosive attack is then possible only if the equilibrium is disturbed by removing in some way the dissolved corrosion product from the system. Thus the nature of liquid-metal corrosion varies depending on whether the fluid is static or is moving relative to its container, on whether the temperature is constant or varying throughout the system, and on whether the container is a single metal or a composite of two or more metals. Most processes, however, involve solution as a first step. No adequate theory is available to explain the variation in the solubilities of metals in molten metals. Both Kerridge' and Strachan and Harris' noted that plotting the solubilities of metals in a number of solvent metals showed a periodic variation with the solute and not the solvent (see Fig. 2.28), i.e. a given metal such as manganese showed a consistently high solubility in molten magnesium, tin, bismuth and copper, compared with iron or chromium. Kerridge correlated this variation with the solute lattice energy and hence with the latent heat of fusion. This correlation is useful in making qualitative predictions. In the more practical sense solution may be uniform or localised. Preferential solution can take two forms: 1. Leaching- one component of an alloy is preferentially dissolved, an example being nickel which is leached from stainless steels by molten lithium or bismuth, sometimes to such an extent that voids are left in the steel. 2. Intergranular attack -the liquid penetrates along the grain boundaries, owing either to the accumulation of soluble impurities in the boundarjes or to the development at the junction of a grain boundary with the metal surface of a low dihedral angle to satisfy surface-energy relationships. * Testing procedures for liquid-metal corrosion are given in Chapter 19. 2: 120
MECHANISMS OF LIQUID-METAL CORROSION
Ti
V
22
23
Cr
Mn Fc Co Ni Cu Zn Ca 24 25 26 27 28 29 3 0 31 ATOMIC NUMBER OF SOLUTE METAL
2: 121
Ge
32
Fig. 2.28 Solubilities of the first row of transition elements in five liquid metals. (To avoid overlap of points, the graphs have been set at different positions on the solubility axis) Solvent metal Temperature (“C) For true solubility, deduct,from axis reading:
0
e
a
v
H
20
-
Hg Mg Sn Bi
700 450 400
8% 12%
cu
1 200
16%
4%
I I
WITH LIQUID METAL)
DIRECTION OF CRACK GROWTH
-
Fig. 2.29 Energies involved in the growth of a crack
When this process is accompanied by stress, catastrophic failure can occur, a classical example being the action of mercury on brass. The situation may be described in terms of the surface-energy changes when a crack propagates through a solid metal as shown in Fig. 2.29 where ys is the solid-gas interfacial energy, ysr is the solid-liquid interfacial energy and ye is the grain-boundary energy. Tabulated below are the energy changes involved for different cracking modes, with numerical values for the case of copper in contact with liquid
2: 122
MECHANISMS OF LIQUID-METAL CORROSION
lead with a dihedral angle at a copper grain boundary of 90" and ys = 1.8 J/mZ, ye = 0-6 J/m2 and ysL = 0 - 4 J/m2: Transgranular: Grain-boundary cracking: Grain-boundary cracking in the presence of liquid metal. Transgranular cracking with the crack filled with liquid metal:
2ys = 3.6 J/m2
2ys - ye = 3.0 J/m2
2ysL - yB = 0.2 J/m2
2ysL = 0.8 J/mz
It is seen that the presence of the liquid metal greatly lowers the surfaceenergy change for grain-boundary cracking3.
Temperature Gradient The solubility of metals, S,in molten metals generally varies with temperature according to the relationship logs = A
- (B/T)
where A and B are constants for a given system. It is therefore possible for more material to dissolve from a container at its highest-temperature end than at the low-temperature end, and if the melt flows round the container by natural or forced convection the liquid arriving at the cold region will be supersaturated and will precipitate solute until equilibrium is attained. If it is then recycled to the hot end it dissolves more metal until saturated and then returns to the cold end to precipitate this excess. This process is termed 'thermal gradient mass transfer'. It can best be illustrated by circulating a corrosive metal such as bismuth round a thermal convection loop of the type shown in Fig. 2.30. After a prolonged period of operation, thinning of the inner wall of the hot section and a precipitate on the wall of the cold section can be clearly observed. (The latter is often termed aplug since it eventually blocks the pipe to liquid flow.) This process has been analysed in some detail and the various stages are detailed in Fig. 2.31. The overall rate-controlling step has been shown by Ward and Taylor4 to be diffusion through the boundary film of solute atoms into the flowing stream. To be precise, they found that the solution of solid copper in liquid lead and bismuth obeyed the following equations. Under static conditions, at temperature T
n, = no[ 1 - exp( - K S / V ) ] where n, = concentration of solute after time t, no = saturation concentration of solute, S = surface area of solid exposed to liquid of volume V, K = KOexp [ - AE*/RT] (M*= activation energy for solution). Under flowing conditions dnt = K ( S / V )(no - n,) dt If, therefore, the solute atoms can be prevented from entering the boundary film from the solid the process will be halted. A method for doing this
MECHANISMS OF LIQUID-METAL CORROSION
2: 123
STAINLESSSTEEL
x -THERMOCOUPLE LOCATIONS
Fig. 2.30 Thermal convection loop
Fig. 2.31 Stages in thermal gradient mass transfer 1 . Solution 2. Diffusion 3. Transport of dissolved metal
4. Nucleation
5. Transport of crystallites 6. Crystal growth and sintering (plug formation)
(after Brookhaven National Laboratory)
was discovered by workers at the U.S.General Electric Company some time ago when developing a mercury boiler for electric power generation’. They found that small quantities of dissolved titanium, zirconium, chromium, nickel and aluminium were effective as inhibitors of the corrosion of steels by hot mercury, the first two being particularly so. Later interest in the use of liquid bismuth as a carrier of uranium in a liquid-metal-fuelled reactor led to the extension of the use of zirconium inhibitor to bismuth in steel circuits
2: 124
MECHANISMS OF LIQUIDMETAL CORROSION
and to an elucidation of the inhibiting mechanism6*'.The zirconium reacts with the nitrogen, which is always present in steel to the extent of about lWp.p.m., to form a surface layer of ZrN which is thermodynamically a very stable compound and is an effective diffusion barrier. Furthermore, as long as there is residual zirconium in solution in the bismuth (or mercury) and dissolved nitrogen in the steel, the film is self-healing. Mercury boilers have operated successfully for thousands of hours relying on this principle. In recent years there has been a continued interest in the use of alkali metals, notably sodium and lithium, as heat exchange media in nuclear reactors and fusion systems respectively and as chemical reactants in fuel cells. This interest is reflected in the proceedings of several major conferences which are referenced in the bibliography (see p. 2.109). Generally speaking corrosion processes in liquid alkali metals are either concerned with dissolution of the component (general or selective), chemical reaction between the component and non-metallic impurities O , , c , N,, H,, which are soluble in liquid sodium at the ppm level, or a combination of both processes where dissolution is followed by chemical reaction in the liquid phase. Solubilitiesof constructional materials -refractory metals and the components of iron and nickel base alloys-in liquid alkali metals are much less than in more noble metals, mercury and bismuth, and solubilities in liquid sodium at 650°C can range from a fraction of a ppm (refractory metals) to 1-10 ppm for metals like Fe, Cr and Ni. Elements such as Ni, Cu and the precious metals have appreciable solubilities in liquid lithium and it is generally considered that alloys of high nickel content have limited use in lithium systems operating with a temperature gradient. In heat exchange systems corrosion processes are more concerned with dynamic, not static, liquid alkali-metal environments, consequently the previously quoted corrosion rate equation, is more applicable in this type of system. Resistances to mass transport may be present in either the hot or cooler parts of the circuit and the kinetics of mass transfer may be rate controlled either by dissolution or deposition at the solid surface, by transport of material through the adjacent laminar sub-layer or by phase boundary reactions in the solid metal. In diffusion-controlled mass-transfer situations involving turbulent fluids Epstein' has suggested that mass transfer equations can be derived from heat transfer analogies and expressions relating corrosion rate to the dimensionless groups. Reynolds No. (Re) and Schmidt No. (Sc) have been found to have some application where corrosion rates sensitive to changes in flow velocity or diffusivity in the liquid phase. The equation suggested by Epstein to meet this situation is of the form: Rate = 0.023 (D/d)Cw(Re)'.' S C ' . ~ ~ in which C, is the concentration of the dissolved species at the wall, D is the liquid phase diffusivity for the soluble species and d is the pipework diameter. The velocity term is incorporated in the Re No. expression. Non-metallic impurities in liquid alkali metals play a major role in the corrosion of materials either by affecting metal solubilities, forming spallible corrosion products on the metal surface, promoting liquid metal embrittlement or bulk embrittlement of the surface or by sensitising the structure for further attack by other impurities e.g. 02.As in other corrosive environments the direction and magnitude of these impurity reactions
MECHANISMS OF LIQUID-METAL CORROSION
2: 125
are dictated by free energy and solubility relationships for both solid and liquid phases. In some metal components it is possible to form oxides and carbides, and in others, especially those with a relatively wide solid solubility range, to partition the impurity between the solid and the liquid metal to provide an equilibrium distribution of impurities around the circuit. Typical examples of how thermodynamic affinities affect corrosion processes are seen in the way oxygen affects the corrosion behaviour of stainless steels in sodium and lithium environments. In sodium systems oxygen has a pronounced effect on corrosion behaviour' whereas in liquid lithium it appears to have less of an effect compared with other impurities such as C and N,. According to Casteels lo Li can also penetrate the surface of steels, react with interstitials to form low density compounds which then deform the surface by bulging. For further details see non-metal transfer. One important and perhaps unique feature of corrosion in the alkali metals is the formation of corrosion products based on complex ternary oxides. Horsley" has shown that oxides of the type (Na,O), FeO can form when iron is corroded in sodium under conditions where the standard state binary oxides FeO, Fe,O, and Fe, 0, are thermodynamically unstable. Weekes" in his analysis of those factors which affect corrosion behaviour, has also suggested this type of oxide may play a role in the corrosion of stainless steels in sodium. Addison l3 has shown that many transition metals form complex oxides with Na,O and corrosion products based on the ternary oxide Na,NbO, have been identified on the surfaces of niobium after exposure to the sodium containing oxygen14. This type of oxide is mechanically unstable in flowing sodium environments and therefore it is relatively easy to promote fresh surfaces for further attack by oxygen impurities. Complex oxides of the type NaCrO, also feature in the corrosion of stainless steels in alkali metals. NaCrO, for example may exist either as an oxide film on the surface of the steel during the initial stages of corrosion or it may, under more adverse conditions, penetrate the grain boundaries and become a precursor for grain detachment. Impurity reactions can be controlled or eliminated by adequate purification of the liquid metals and in pumped loop systems this can be achieved by using techniques known as cold trapping and hot trapping. Cold trapping involves taking a small percentage of the main loop flow and by-passing it through a container which is cooled to the required temperature to precipitate out the impurities. Hot trapping on the other hand involves removal of impurities by chemical reaction between the soluble species and a material which has a higher thermodynamic affinity for the impurity than the liquid metal or its containment. In sodium systems cold trapping can remove oxygen impurities down to the 1-3 ppm level whereas hot trapping using Zr heated to 600°C can take the levels down to 75% Au and are based on the ternary Au-Ag-Cu alloys with the mechanical properties improved through additions of Pt, Pd and Ir. These high gold alloys possess near ideal casting characteristics as well as excellent corrosion resistance. In recent years, however, the increasing price of gold has resulted in greater use of alloys containing less gold and greater amounts of silver and palladium as well as various base metal alloys (see later) for dental castings. Lower gold content alloys have good mechanical properties and can be accurately cast but they do not possess the oral corrosion resistance of high gold alloys. Further, these alloys are often metallurgically heterogeneous or at least less homogeneous than the high noble-metal alloys, which will affect their corrosion susceptibility due to galvanic coupling effectsz3. Cast restorations often exhibit little corrosion but rather an unaesthetic tarnishing or discolouration. There appears in fact to be an inverse
2: 158
CORROSION IN THE ORAL CAVITY
relationship between oral corrosion and tarnish for many low gold and copper-rich alloys in potential ranges where the corrosion rate is limited by the rate of the cathodic reaction’””. This effect is possibly due to the uniform distribution of anodic and cathodic sites in single phase alloys with no preferential deposition of corrosion products and consequently little tarnish. In contrast, the anodic and cathodic reactions in multiphase alloys are separated and the different nobilities of these phases may give rise to numerous bimetallic galvanic cells. Other systems, typically Ag-Pd alloys, do appear to show a correlation between tarnish and corrosionz5. Most cast dental restorations are subjected to some form of subsequent heat treatment such as annealing, hardening or soldering. This often induces changes in the structural state or in the phases present and may establish local galvanic cells. Potentio-dynamic polarisation studiesx have shown that high gold alloys are unaffected by their thermal history but the corrosion susceptibility of low golds (containing 850
(part B in Fig. 3.15)
one using a solution of 0.6 M sodium chloride and 0.1 M sodium bicarbonate at 25°C. The pitting potential was that to give a current of 10-'A/crn2. The beneficial effects of chromium, molybdenum and nitrogen are clearly apparent. Not so obvious from these data are the effects of nickel which is slightly beneficial, and manganese which is somewhat detrimental. The markedly adverse effect of substantial sulphur additions (normal range as impurity is 0.005-0.02Vo)is clearly shown. Other elements also have beneficial or detrimental effects, but only chromium, molybdenum and nitrogen are used extensively to promote resistance to localised corrosion. It is becoming increasingly common to describe the pitting corrosion resistance of stainless steels in terms of these elements using formulae, of which the following is, perhaps, the most popular:
3:52
STAINLESS STEELS
Yo Cr
+ 3.3 x Yo Mo + 16 x Yo N
There clearly must be effects from other elements present in the steel incidentally or functionally, but a reasonable relationship between this factor and the measured pitting potential is shown for a large number of steels in Fig. 3.19. Of interest, but not yet explained, is the shape of the curve as discussed elsewhere".
m .-c
C Q c
0
Q
0
C .-c
0
._ 4-8
a -500 0
10
20 Cr
30
40
+ 3.3Mo + 16N
~ at 2S°C) versus a composition factor. Cr, 0.021-28.5; Fig. 3.19 Pitting potential ( 0 . 6 NaCl Mo, 0.01-4.23; N, 0.01-0.455; Ni, 0.22-20.39; Mn, 0.22-4.53; Si, 0.13-0.71
Metallurgical Considerations and Forms of Corrosion
The corrosion resistance of any of the stainless steels is at its best when the material is single phase and in a homogeneous state. In acid solutions corrosion is then usually essentially uniform in nature except for grain orientation effects and possibly etch pitting. In very strongly oxidising solutions (i.e. conditions of transpassivity) however, apparently homogeneous austenitic steels may exhibit intercrystalline corrosion, which is believed to be associated with segregation of certain interstitial elements at grain boundaries, but is of little practical importance since stainless steels are not suitable for use in such solutions. Pitting or localised corrosion is rare in correctly treated steels in acid solutions, but can be obtained in some circumstances. In near neutral or alkaline halide-bearing solutions, however, any corrosion is by pitting and the consequences may be serious. The mechanism of pitting has been discussed in Sections 1. 5 and 1.6 and will not be considered further here. In practice, of course, few alloys are homogeneous in composition, the segregation produced during solidification being reduced but not
STAINLESS STEELS
3:53
eliminated during subsequent working and heat treatment. This is the case with stainless steels, and while microsegregation of the major alloying elements does not usually lead to any significant effect on overall corrosion resistance, some differential corrosion rates and etching effects of various surfaces can be noted under environmental conditions causing significant corrosion. The presence of second phases also implies partitioning of the elements with consequent alloying element segregation. In some martensitic steels there can be retained austenite following cooling from the hardering temperature while conversely, with some austenitic steels, there can be some martensite present in the softened state or, more commonly, induced by cold deformation. Since the austenite-martensite transformation is rapid and occurs at low temperatures, there is little chance of alloying element segregation while the crystallographic state itself has little effect on corrosion resistance. Thus the presence of austenite in martensite and vice versa has little practical effect, although substantial acid attack can reveal a difference. When austenite is formed in a martensitic steel from prolonged heating below the A,, temperature it does have a different composition from the now tempered martensite matrix. This situation can apply with the precipitation hardening grades when overaged, but the practical effect on corrosion resistance is very small. Austenite may form in some, but not all, ferrite steels (depending upon composition) during high-temperature heat treatment. In the case of the less highly alloyed types (e.g. 430817) this transforms to martensite on cooling, while with the more highly alloyed types it remains as austenite. In both cases there is reversion to ferrite and homogenisation with the correct heat treatment but even if such treatment is omitted, the effect on corrosion resistance is small. The most likely second phase in martensitic and in austenitic steels is delta ferrite although, usually because of other considerations, care is taken to balance the composition so that it is avoided. In the cases of castings and weld metal, however, a small amount of ferrite is usually present in the interests of soundness. The ferrite phase is higher in chromium and molybdenum but lower in nickel, nitrogen and copper (for instance) than the austenitic phase but the differentials between the ferrite and the martensite or austenite in the near equilibrium (i.e. correctly heat-treated) state are not great, so that both the phases in a two-phase structure have quite similar corrosion resistances. There are differences, however, and the phases can be revealed on etching, while in certain media there can be selective attack in practice. There is little practical hazard except in the case of weldments in the as-deposited state. The ferrite content at very high temperature is greater than after normal heat treatment and the rapid cooling following welding can lead to retention of an increased amount of ferrite. This leads to more marked element partitioning with a more noticeable effect an corrosion resistance, especially if the ferrite content is such as to form continuous paths in the matrix. This danger is avoided by control of composition or by post-weld heat treatment. However, as-deposited weld metal proves perfectly satisfactory for many applications. In duplex steels, ferrite is a major intentional constituent. The features noted above are relevant; the two phases do have differing analyses but not to such an extent as to cause serious corrosion problems. A further
3:54
STAINLESS STEELS
structural feature, theoretically possible in other grades but usually only of any consequence in the duplex steels, is the formation of sigma or chi phases. These are rich in chromium and molybdenum and form on heating in the 550-950°C temperature range. The upper temperature limit for their formation and the time-temperature relationship necessary to produce a significant amount depends on steel composition and for most steels there is no practical danger (unless prolonged periods at elevated temperature are expected in service). Chi and sigma formation can be much more rapid with the duplex steels and can occur during cooling following heat treatment. The steels more highly alloyed with chromium and molybdenum are most at risk and quenching after heat treatment is advisable. Even with quenching the problem can arise with larger sections. Both toughness and corrosion resistance can be affected. Sigma and chi phases themselves are very resistant to corrosion but their formation leads to depletion from the adjacent material of chromium and molybdenum (the phases form preferentially in ferrite). Thus the full corrosion resistance of the alloy is diminished. While due care and attention must be paid to this aspect, it does not normally represent a serious limitation with this group of steels. Precipitation of Carbides
The most marked structural effect on corrosion resistance is that associated with chromium-bearing carbides, which can occur in all three groups of steels. The martensitic types of hypo-eutectic composition have essentially all carbon in solid solution in the as-transformed (i.e. the hardened) condition. Tempering reduces strength as shown in Fig. 3.11 and causes the precipitation of carbides; at lower temperatures the carbides are essentially of iron, but at higher temperatures they are chromium-rich. The diffusion rate of carbon which is an interstitial element is much greater than that of chromium, and as a result chromium gradients are set up adjacent to the growing carbide particles. If the carbide particle distribution is such that the chromium-reduced regions can overlap, a continuous low-chromium path can be formed which may lead to selective attack. Such attack can occur along prior austenite boundaries and also along martensite lath boundaries. The treatments giving susceptibility to such attack are indicated in Fig. 3.20; at low temperatures there is insufficient chromium in the carbides to have an effect, while at higher temperatures diffusion of chromium is rapid enough to prevent severe gradients, although the overall effective chromium content is reduced somewhat, as is the corrosion resistance as a result. Obviously the effects are more pronounced with higher carbon steels. As the effects are associated with a precipitation reaction, the time of tempering has an effect as well as temperature*. For discussion of the tempering processes see References 2 and 3. *Martensitic stainless steels are usually used in the softened (tempered at or above 65OoC) or in the fully hardened condition (tempered at or below 25OOC) so that there is no substantial reduction in corrosion resistance resulting from carbide precipitation. However, the hard soldering of knife blades can result in carbide precipitation and pitting of the blade at the area adjacent to the handle, and care must be taken in the soldering process to avoid this danger.
3:s
STAINLESS STEELS
550 OC
Tempering time ( 5 )
Fig. 3.20 Effect of tempering treatment on corrosion resistance of 420S45 (air cooled, 980°C). Corrosion tests in 10% nitric acid solution at 20°C
Austenitic steels of the 304S15 type are normally heat treated at 1 050°C and cooled at a fairly rapid rate to remove the effects of cold or hot working, and in this state much of the carbon is in supersaturated solid solution. Reheating to temperatures below the solution treatment temperature leads to the formation of chromium-rich M,,C, precipitates predominantly at the grain boundaries with the production of chromium gradients and reduced corrosion resistance as is the case with the martensitic steels. Any attack is
-u
0, 8000 I
2 750-
e
Time at temperature (h)
Fig. 3.21 Temperature-time-sensitisation diagrams for three austenitic Cr-Ni steels solution treated at 1 050°C. The curves enclose the treatments causing susceptibility to intercrystalline corrosion in a boiling CuSO, + H2S04test reagent
3:56
STAINLESS STEELS
intergranular in this instance. The effect of temperature and the time of reheating on establishing whether steels are susceptible to this form of attack or not, is shown in Fig. 3.21. The corrosive medium used was that specified in ASTM A262E and it should be noted that the precise position and area of the curves are relevant to this test medium only, although the general form is similar for other test media. Much has been written regarding the relative merits of various accelerated tests, but it is generally accepted in this country that the copper sulphate-sulphuric test gives an adequate guide to steel condition for most purposes. For other test methods ASTM 262 should be consulted. It should also be noted that not all corrosive media produce intergranular attack on ‘susceptible’ steels. The metallurgical condition leading to susceptibility to intercrystalline corrosion can be caused by cooling from the solution treatment temperature at a sufficiently slow rate as well as by isothermal treatment. Large sections in certain grades may thus require quenching in oil or water to ensure optimum corrosion resistance. The heating introduced by welding can also produce the undesirable metallurgical condition in bands a short distance on each side of the weld. This may lead to the localised attack known as weld decay (see Section 9.5). Since the susceptibility to intercrystalline attack is due to carbide formation it follows that low carbon steels are more resistant to the phenomenon, as illustrated in Fig. 3.21. Thus the alloy with a carbon content suitable for the thermal treatment to be encountered in fabrication should be selected. An alternative method for ensuring resistance to intercrystalline corrosion is to ‘stabilise’ the carbon, that is to restrict its solubility at heat treatment temperature by incorporating a strong carbide-forming element in the steel, and titanium or niobium are utilised for this purpose (321831 and 347331). Whether low-carbon grades or stabilised grades are selected for a given application where intercrystalline corrosion may be a hazard depends on a variety of factors. The stabilised grades are especially useful where service may involve periodic heating into the critical temperature range followed by contact with some aqueous corrodent. As shown in Fig. 3.21, even the low carbon grades can be sensitised in time. Low carbon grades may be preferred where welding is followed by heating into the critical range for a relatively short time for the purpose of stress relief. The zone immediately adjacent to the weld is subjected to a very high temperature in welding which causes solution of the titanium or niobium carbides so that chromium-rich carbides may form on reheating. Thus a narrow band immediately adjacent to the weld may corrode in service. This is known as knife-line attack to distinguish it from weld decay where the corrosion zones are some slight distance (= 2 mm) from the weld. The ferritic steels may also undergo intercrystalline corrosion as a result of grain boundary carbide formation. In the normal softened state (treated rr 800°C) the carbon is largely precipitated and the ferrite composition homogenised so that further heating at lower temperatures has no adverse effect. During solution treatment above 950”C, however, carbon is redissolved. Sensitisation can then occur at lower temperatures but the rate is SO rapid that it can only be suppressed by very rapid cooling which is not practically feasible. Thus weld decay is very possible in service unless a remedial
STAINLESS STEELS
3:57
heat treatment (%800”) is applied. Unlike the austenitics, the weld decay band is the higher temperature zone immediately adjacent to the fusion zone and, of course, the fusion zone itself if matching filler is used. This rapid sensitisation behaviour is due to the lower solubility of carbon and higher difussion rates in ferrite. T.T.S curves as those in Fig. 3.21 can be developed for material cooled very rapidly following solution treatment but the ‘Cycurve range is 4 0 0 550°C and the ‘nose’ is at very short times. Freedom from sensitisation in welding can be obtained by ensuring extremely low carbon (and nitrogen) but such levels are not commercially feasible. Stabilisation by niobium and titarium is feasible, but higher ratios are needed than with austenitic steels. With most of the ‘super ferritic’ group a combination of a practical low carbon level and titanium addition is used. Precipitation of Nitrides
Nitrogen can dissolve at elevated temperatures and precipitate as chromium nitrids at lower temperatures with similar effects to those described in the previous section, although there are some differences in detail. Nitrogen is not an alloying addition with most of the martensitic group but is present, usually at about 0.02%, as an incidental. It can be a deliberate addition (up to 0.05Vo) with some specials and higher values are being proposed for some creep-resisting steels. Similar effects as with carbon may be anticipated but adverse effects in corrosion behaviour should not arise from good heat treatment practice avoiding tempering between 350 and 650°C. Nitrogen is also present as an incidental in austenitic steels, usually at about 0.03’70, but the only practical consequence is that allowance for its presence should be made when calculating the amount of titanium added for stabilisation. Nitrogen is being used to an increasing extent as a deliberate addition because of the improvement in strength’* and resistance to some forms of corrosion. Modest additions to the 18/10 types actually retard carbide sensitisation and at 0.2% nitrogen the T.T.S. curves for low-carbon steels are very similar to those for low-nitrogen equivalents. Further alloying with nitrogen is possible by melting and freezing under increased pressure or, for modified steel analysesI4, at atmospheric pressure. At these higher nitrogen levels sensitation due to C,N formation is possible under adverse heat treatment conditions. This aspect is discussed in Reference 13. In the ferritic steels the effects of nitrogen and carbon are indistinguishable one from the other and the normal incidental level is sufficient to cause weld decay susceptibility. Thus in the ‘super ferritic’ group both carbon and nitrogen are controlled to a low practicable level and sufficient titanium is added to stabilise both elements. Stress-acceleratingEffects
It is worthy of note that the simultaneous presence of a sustained tensile stress and a corrodent with a sensitised steel may result in rapid cracking
3:58
STAINLESS STEELS
rather than relatively slow, general intergranular penetration. Perhaps stress-accelerated intercrystalline corrosion in a better term for this than stress-corrosion cracking. In its worst manifestations, the presence of stress may cause failure of steel at a degree of sensitisation that would have been insufficient to give problems in the environment without stress.
Corrosion in Natural Environments Atmospheric Corrosion
One of the major assets of stainless steels has proved to be their resistance to discoloration in the atmosphere and even the least corrosion-resistant alloys have been used indoors with success. Knives, for instance, must be made from the 13% chromium martensitic types to give the required hardness yet they retain their bright attractive appearance without special cleaning other than ordinary washing. Many other domestic and kitchen articles have been made from the simple ferritic and austenitic steels with similar success. Outdoor service is more arduous, however, and the martensitic steels are not used where appearance is important. The ferritic steels 430817 and 434817 have been and are used for motor car trim (the extent varying according to fashion) and retain their appearence well. In one way this is surprising as simple exposure tests out of doors show loss of reflectivity, especially with 430817. It is thus presumed that thin films of wax and grease applied in the normal coarse of operation and cleaning have a beneficial effect. The mechanism of atmospheric corrosion of stainless steel has not been widely explored. Good service indoors, even of the lesser alloyed grades, may be attributed at least in part to the predominantly low relative humidity in heated buildings, but short time tests indoors in chambers giving high humidity have shown at least delayed initiation relative to outdoor exposure. Since the gaseous composition of air indoors must be similar to that outdoors it may be presumed that the species causing the difference are transported to the surface as airborne solids, as airborne free liquid droplets or in rain. The last of these is improbable since samples sheltered from the rain outdoors invariably corrode faster than samples exposed alongside but not sheltered. Exposure tests have shown that steel corrodes more rapidly at coastal and industrial sites than at rural, inland sites. Thus, it is likely that airborne chlorides and sulphurous gases are major causes; some laboratory tests have confirmed the adverse effects of both these acting separately and together. From such testing there is also evidence that deposited carbon is detrimental. Any corrosion is in the form of very fine pitting, the degree being reflected in pit density rather than size. The pits cannot usually be resolved with the naked eye, and any degradation is perceived as a loss in reflectivity or a ‘misting’of the surface. Unless the steel is cleaned regularly, a more immediate visual effect on a corroding surface is rust staining. The effects of some alloying elements on relative behaviour in an industrial atmosphere (Sheffield, U.K.) are shown in Table 3.21A. For comparison, data for simultaneous tests on carbon steel and some non-ferrous material are given. Results are as weight loss over a five-year period and data from
3:59
STAINLESS STEELS
two series are given, these being at the same site but with 24 years separating them. The wide differencesbetween results in the two series are attributed to changes in degree and possibly type of atmospheric pollution. As with pitting under immersed conditions, a markedly beneficial influence of molybdenum is obvious, and this is further demonstrated by the more recent data of Table 3.21B in which testing was for 18 years at a ‘heavy industrial’ site and assessment was by pit density and pit depth measurements. Table 3.21C shows the effect of geographical location, the sites being classified as ‘severe marine, heavy industrial, semi-industrial and rural’”. Other prolonged tests on various steels at various sites have been described elsewhere. In these tests assessment was by appearanceI6. In all the test programmes it was found that smooth surfaces give better results than rough ones and that the quality of abrasive used can be of significance for abraded surfaces. Table 3.21A
Atmospheric exposure test, Sheffield, 5 years. Results: loss weight (g/m2); means of multiple samples
Material
193811943 series
1962/1967 series
Carbon steel 410,521 43OS17 4348 17 180-8Mn 12Cr- 12Ni 304315 25Cr-20Ni 321331 30% Cr 316331 AI
3 700 270 135 165 160
45 15.5
85
4.5
cu
185
Phosphor Zn 54 Cu-44Ni
200 385
Table 3.21B
80 70 70 1 165
440 Effect of molybdenum content on atmospheric corrosion. Heavy industrial site. 18 years
Material 304s 15 315S16 316333 317316
Mo content
Pit density (number/cm ’)
Pit depth (ctm)
3 870 lo00 625 290
0.31 1.44 2-70 3.45
81 52 35.5 17.5
Table 3.21C Effect of environment type on atmospheric corrosion. 18 years
Site Rural Semi-industrial Heavy industrial Marine
Pit depth b m )
Pit density (number/cm’) 304s 15
3 16S33
304315
316333
2030 3030 3870 3 160
225 420 625 355
20 21 81
17.5 18 35.5 24
85
3:60
STAlNLESS STEELS
Where retention of appearance is of prime importance, the molybdenumbearing steels are used almost exclusively in the U.K. Different opinions have been expressed as to whether regular cleaning is necessary to obtain the best results. Data have now become available which show that the absence of such clearing is of no detriment far fully exposed steel (although clearing obviously removes any staining) but can be of value where steel is sheltered from rain. In Table 3.22A are given results from a series of tests in Sheffield, U.K. (1962-1968), the values being total area of pitting as a percentage as approved to number of fits. Table 3.22B contains results for 316833 steel exposed to a marine atmosphere either sheltered or unsheltered.I5 Table 3.22A
Frequency of washing
Effect of periodic working on the behaviour of several stainless steels exposed to an industrial atmosphere (Sheffield, 1962-1%8) Area corroded ('70) 430s 17
434317
304s 16
316Sll
23.8 27.9 47.3 26.5
3.66 4.20 4.74 3.60
3.60 4.19 4.59 2.0
102-
c .In C W
-0 c
10’L
3
u
100 -
lo-‘
~
10-2L -0.5
0
0.5
1.0
1.5
2.0
Potential (S.C.E.) Fig. 3.71 Change in polarisation curve of amorphous Fe-IOCr-13P-7C alloy in 1 N HCI with the time of heat treatment at 723 K. The time of heat treatment is expressed in the figure in minutes3’
3: I54
AMORPHOUS (FERROUS AND NON-FERROUS) ALLOYS
effect of heat treatment in the case of Fe-IOCr-l3P-7C3’. A microcrystalline metastable phase is formed in the amorphous matrix by heat treatment at 703 K for 100 min. The alloy becomes no longer spontaneously passive in 1 N HCl as soon as the microcrystalline phase appears in the amorphous matrix, and the anodic dissolution current continues to increase with increasing time of heat treatment. This occurs because of the introduction of chemical heterogeneity into the homogeneous single phase of the amorphous alloy. Rapidly solidified microcrystalline stainless steels also have high pitting corrosion and detrimental defects on which a stable passive film does not form are mostly precipitates and segregates of impurities 34. The chemically homogeneous single-phase nature of amorphous alloys which are free of defects resulting in the formation of a uniform passive film is responsible for the high corrosion resistance of these alloys.
Fast Passivation When the chromium-enriched passive film is formed on amorphous and crystalline iron-chromium alloys, containing no noble metals such as nickel, the composition of the alloy surface just under the chromium-enriched passive film is almost the same as that of the bulk alloy24.Hence, the formation of a chromium-enriched passive film results from selective dissolution of alloy constituents unnecessary for passive film formation. When an alloy is able to passivate, fast active dissolution of the alloy results in rapid enrichment with beneficial ions. The passivating ability is, therefore, closely related to the activity of the alloyI4. The thermodynamically metastable nature of amorphous alloys is responsible for their high reactivity when they are not covered by a passive film, and hence is responsible for the fast passivation by the formation of the film in which the beneficial ions are highly concentrated. As shown in Fig. 3.67 for iron-, cobalt- and nickelbased alloys, when the alloy chromium content is not high enough to cause spontaneous passivation, the more active iron-based alloys dissolve rapidly and the more noble nickel-based alloys dissolve slowly. The fast dissolution in iron-based alloys is effective in concentrating the chromic ions, so that iron-based alloys passivate spontaneously with the addition of a small amount of chromium. In contrast, the slowly dissolving noble nickel-based alloys require the addition of larger amounts of chromium for spontaneous passivation.
Metastable Nature Amorphous alloys are in a thermodynamically metastable state, and hence essentially they are chemically more reactive than corresponding thermodynamically stable crystalline a l l ~ y ’ * ”If~ an ~ ~amorphous . alloy crystallises to a single phase having the same composition as the amorphous phase, crystallisation results in a decrease in the activity of the alloy related to the active dissolution rate of the alloy35. Since amorphous alloys can be regarded as metallic solids with a frozenin melt structure, the liquid structure freezes at different temperatures
3: 155 depending upon quenching conditions with the consequent formation of different amorphous states. Accordingly, even for amorphous alloys of the same composition, anodic dissolution currents are not always identical owing to different structural relaxation inten~ities~~-~'. AMORPHOUS (FERROUS AND NON-FERROUS) ALLOYS
Effect of Metalloids As can be seen in Fig. 3.67, the corrosion resistance of amorphous alloys changes with the addition of metalloids, and the beneficial effect of a metalloid in enhancing corrosion resistance based on passivation decreases in the order phosphorus, carbon, silicon, boron39 (Fig. 3.72). This is attributed partly to the difference in the speed of accumulation of passivating elements due to active dissolution prior to passivationm. 20 L
Fe-I OCr-136-7X
m W >.
. E E
a,
5
10
c 0 .u)
2
L
0 0
0
Si
B
0.2-
C
P
I
I
L
m
Fe-10 Cr-13P-7X in 0 . 1 HzSO, ~
W
,z E E a; c
I
.-0 In
9 0
0
0
a
a
The effect of metalloids on the corrosion resistance of alloys also varies with the stability of polyoxyanions contained in their films. Phosphorus and carbon contained in iron-chromium-metalloid alloys do not produce passive films of phosphate and carbonate in strong acids, and so do not interfere with the formation of the passive hydrated chromium oxyhydroxide
3: 156
AMORPHOUS (FERROUS AND NON-FERROUS) ALLOYS
film20*40. In contrast, boron-containing alloys require the addition of large amounts of chromium to increase the passivating ability by concentrating the chromium oxyhydroxide in the surface films because the films contain chromium The thickness of the passive films discussed above is up to 3-4 nm. In contrast, the surface film on an amorphous Cu-40Zr alloy continues to grow to over 100nm in 1 N H 2 S 0 4 ,but the addition of only 2 at% phosphorus is effective in depressing the film growth to a few tens of n a n ~ m e t r e s ~The '~~~. addition of a few atomic percent of phosphorus to amorphous Ni-30Ta alloys results in a decrease in the corrosion rate in boiling 6 N HCI by about four orders of magnitude ". The corrosion rate of amorphous Ni-P alloys in 1 N HCI is lower than those of crystalline nickel metal and amorphous Fe-P-C alloy by factors of about 5 and 250, respectively, and is further decreased by the addition of various elements43(Fig. 3.73). I
-
1
Amorphous Ni-M- (18 20) P
-
1M HCI 30°C 10-7
Crystalline 600
Ni
-
7
Iu) N.
Fig. 3.73 Average corrosion rates of amorphous Ni-P alloys measured in 1 N HCI at 30°C. Included are average corrosion rates of crystalline nickel and nickel-base alloys43
Silicon-containing amorphous metal-metalloid alloys form surface films. Sputter-deposited Fe-Si alloys containing 25 at.% or more silicon are passivated by anodic polarisation in dilute sulphuric acid owing to the formation of a S O , film". Melt-spun amorphous Fe-39Ni-lOB-12Si alloy is more resistant against pitting corrosion than the amorphous Fe-40Ni-20B alloy
AMORPHOUS (FERROUS AND NON-FERROUS) ALLOYS
3: 157
owing to the formation of a silicon-enriched surface film45.An increase in the silicon content of amorphous Fe-B-Si alloys extends the passive potential rangeM. Increasing the silicon content of amorphous Fe-1OCr-5MoB-Si alloys leads to a decrease in current densities in both the active and passive regions in 6 N HCl at 25°C without changing the open circuit corrosion potential owing to the formation of a SiOJike substance along with the hydrated oxyhydroxide film4’.
Passivity Breakdown The chemically homogeneous amorphous alloys with high passivating ability form uniform passive films in which beneficial ions are highly concentrated. Passivity breakdown occurs in the form of general corrosion only when the whole film is dissolved under very aggressive condition^^^. The high passivating ability also provides high resistance against crevice c o r r o s i ~ n ~ ~ * ~ ~ . The crevice corrosion potentials and protection potentials of these alloys are very high.
Active Path Corrosion Stress-corrosion cracking based on active-path corrosion of amorphous alloys has so far only been found when alloys of very low corrosion resistance are corroded under very high applied stressess’~52. However, when the corrosion resistance is sufficiently high, plastic deformation does not affect the passive current density or the pitting potential”, and hence amorphous alloys are immune from stress-corrosion cracking.
Hydrogen Embrittlement Amorphous alloys are capable of absorbing far higher amounts of hydrogen than conventional crystalline steelss4.Thus, some amorphous alloys fail by hydrogen embrittlement when they are corroded under tensile-stressed conditions. However, increasing corrosion resistance by alloy modifications, such as increasing the chromium and/or molybdenum contents of amorphous iron-based alloys, reduces hydrogen absorption and hence hydrogen embrittlement 5 5 .
Oxidation The oxidation behaviour of amorphous alloys studied below their crystallisation temperature is not greatly different from that of crystalline metals, although the presence of large amounts of metalloids complicates the situation 56-58. The amorphous structure favours internal oxidation unless a protective oxide film is formed as, for example, under low oxygen partial pressuress9.
3: I58
AMORPHOUS (FERROUS AND NON-FERROUS) ALLOYS
Production Methods The thickness of amorphous alloys is dependent upon production methods. Rapid quenching from the liquid state, which is the most widely used method, produces generally thin amorphous alloy sheets of 10-30 pm thickness. This has been called melt spinning or the rotating wheel method. Amorphous alloy powder and wire are also produced by modifications of the melt spinning method. The corrosion behaviour of amorphous alloys has been studied mostly using melt-spun specimens. Laser and electron beam processing are effective methods for preparing amorphous surface alloys covering conventional crystalline bulk metals60-61. Sputter deposition is capable of producing thick alloys. The corrosion behaviour of amorphous sputter deposits is similar to that of their melt-spun amorphous counterpart^^^*^^. However, sputter deposits prepared using conventional sputtering apparatus have never been defect-free, and hence the substrate metals are corroded in aggressive environments@. Technological improvements to the sputter deposition process have enabled the preparation of defect-free sputter deposits. Sputtering is particularly suitable for the production of special amorphous alloys such as Cu-Ta"', A1-NbM and A1-Ta', which cannot be prepared even in crystalline phase mixtures by conventional methods, e.g. melting, because the boiling points of copper and aluminium are far lower than the melting points of the valve metals. These alloys containing tantalum or niobium have very high corrosion resistance. Various amorphous alloys can be prepared by plating6'. Plating is particularly suitable for the preparation of thinner amorphous alloys than is possible by melt spinning, e.g. < 1 pm, although production of defect-free alloys is difficult. Ion implantation and ion mixing produce amorphous alloys as thin as only several tens of nanometres. Implantation of metalloids such as phosphorus in austenitic stainless steel has been known to produce amorphous surface alloys having high corrosion r e s i ~ t a n c e ~ * * ~ ~ . K. HASHIMOTO
REFERENCES 1. Naka, M., Hashimoto, K. and Masumoto, T., J. Japan Inst, Metals, 38, 835 (1974) 2. Hashimoto, K., Naka, M. and Masumoto, T., Sci. Rep. Res. Inst. Tohoku University, A-24,48 (1976) 3 . Naka, M., Hashimoto, K. and Masumoto, T., Sci. Rep. Res. Inst. Tohoku University, A-26,283 (1977) 4. Naka, M., Hashimoto, K. and Masumoto, T., J. Non-Crysl. Solids, 29, 61 (1978) 5 . Hashimoto, K., Asami, K., Naka, M. and Masumoto, T., Sci. Rep. Res. Insf. Tohoku University, A-21,237 (1979) 6. Masumoto, T., Hashimoto, K. and Naka, M., Proc. 3rd I n f . Conf. Rapidly Quenched Metals, The Metals Society, London, 435 (1978) 7. Naka, M., Hashimoto, K. Inoue, A. and Masumoto, T., J. Non-Cryst. Solids, 31, 347 ( 1979)
AMORPHOUS (FERROUS AND NON-FERROUS) ALLOYS
3: 159
8. Cadet, P., Keddam, M. andTakenouti, H., Proc. 41hInl. Conf.Rapidly QuenchedMetals, The Japan Institute of Metals, Sendai, 2, 1477 (1982) 9. Kovacs, K., Farkas, J., Kiss, L., Lovas, A. and Tompa, K., ‘Rapidly Quenched Metals’,
Proc. 4th Int. Conf. Rapidly Quenched Metals, The Japan Institute of Metals, Sendai, 2, 1471 (1982) 10. Kobayashi, K., Hashimoto, K. and Masumoto, T., Sci. Rep. Res. Inst. Tohoku University, A-29,284 (1978) 11. Hashimoto, K., Naka, M., Noguchi, J., Asami, K. and Masumoto, T., in Passivity of
12. 13. 14.
15. 16. 17. 18.
Metals, (Frankenthal, R. P. and Kruger, J., eds.), The Electrochemical Society, Princeton, N.J., 156 (1978) Naka, M., Hashimoto, K. and Masumoto, T., Proc. 3rd I n f . Conf. Rapidly Quenched Metals, The Metals Society, London, 449 (1978) Hashimoto, K., Kasaya, M., Asami, K. and Masumoto, T., Corros. Engng. (Boshoku Gijutsu), 26, 442 (1977) Naka, M., Hashimoto, K. and Masumoto, T., J. Non-Cryst. Solids, 34, 257 (1979) Hashimoto, K., Kobayashi, K., Asami, K. and Masumoto, T., Proc. 8th Int. Cong. Metallic Corrosion, DECHEMA, Frankfurt/Main, I, 70 (1981) Kawashima, A., Shimamura, K., Chiba, S., Matsunaga, T., Asami, K. and Hashimoto, K., Proc. 4th Asian-Pac.$c Corrosion Control Conference, Tokyo, 2, 1042 (1985) Shimamura, K., Kawashima, A., Asami K. and Hashimoto, K., Sci. Rep. Inst. Tohoku Univ., A-33,196 (1986) Mitsuhashi, A., Asami, K., Kawashima, A. and Hashimoto, K., Corros. Sci., 27, 957
(1987) 19. Hashimoto, K., Masumoto, T. and Shimodaira, S., in Passivityand Its Breakdown on Iron and Iron-Base Alloys, (Staehle, R. W. and Okada, H., eds.), NACE, Houston, 34 (1975) 20. Asami, K., Hashimoto, K., Masumoto, T. and Shimodaira, S., Corros. Sci., 16,909(1976) 21. Hashimoto, K., Asami, K., Naka, M. and Masumoto, T., Corros. Engng. (Boshoku Gijutsu), 28, 271 (1979) 22. Kawashima, A., Asami, K. and Hashimoto, K., Corros. Sci., 24, 807 (1984) 23. Asami, K., Hashimoto, K. and Shimodaira, S., Corros. Sci., 18, 151 (1978) 24. Hashimoto, K., Asami, K. and Teramoto, K., Corros. Sci., 19,3 (1979) 25. Hashimoto, K. and Asami, K., Corros. Sci., 19, 251 (1979) 26. Asami, K. and Hashimoto, K., Corros. Sci., 19, 1007 (1979) 27. Hashimoto, K., Osada, K., Masumoto, T. and Shimodaira, S., Corros. Sci., 16,71 (1976) 28. Diegle, R. B. and Slater, J. E., Corrosion, 32, 155 (1976) 29. Kulik, T., Baszkiewicz, J., Kaminski, M., Latuszkiewicz, J. and Matyja, H., Corros. Sci., 19, 1001 (1979) 30. Naka, M., Hashimoto, K. and Masumoto, T., Corrosion, 36, 679 (1980) 31. Kapusta, S. and Heusler, K. E., Z. Metallkd., 72, 785 (1981) 32. Diegle, R. B., Proc. 4th Int. Conf. on Rapidly Quenched Metals, The Japan Institute of Metals, Sendai, 2, 1457 (1982) 33. Tsuru, T. and Latanision, R. M. J. Electrochem. SOC., 129, 1402 (1982) 34. Kawashima, A. and Hashimoto, K., Corros. Sci., 26, (1982) 35. Huerta, D. and Heusler, K. E., Proc. 9th Int. Cong. Metallic Corrosion, National Research Council Canada, Ottawa, 222 (1984) 36. Masumoto, Y.,Inoue, A., Kawashima, A., Hashimoto, K., Tsuai, A. and Masumoto, T., J. Non-Cryst. Solids, 86, 121 (1986) 37. Nagarkar, P. V., Searson, P . C. and Latanision, R. M., Proc. Symp. on Corrosion,
Electrochemistry and Catalysis of Metallic Glasses, (Diegle, R. B. and Hashimoto, K., eds.), the Electrochemical Society, Pennington, 118 (1988) 38. Habazaki, H., Ding, S.-Q., Kawashima, A., Asami, K., Hashimoto, K., Inoue, A. and Masumoto, T., Corros, Sci., 29, (1989) 39. Naka, M., Hashimoto, K. and Masumoto, T., J. Non-Crysl. Solids, 28,403 (1978) 40. Hashimoto, K., Naka, M., Asami, K. and Masumoto, T., Corros, Engng. (Boshoku Gijutsu), 21, 279 (1978) 41. Burleigh, T. D. and Latanision, R. M., in Passivity of Metals and Semiconductors, (Froment, M., ed.), Elsevier, Amsterdam, 317 (1983) 42. Burleigh, T. D. and Latanision, R. M., Proc. 9th Inl. Cong. Metallic Corrosion, National Research Council of Canada, Ottawa, 2, 645 (1984) 43. Kawashima, A., Asami, K. and Hashimoto, K., J. Non-Crysl. Solids, 70, 69 (1985)
3: 160
AMORPHOUS (FERROUS AND NON-FERROUS) ALLOYS
44. Brusic, V., Maclnnes, R. D., and Aboaf, J., in Passivity of Metals, (Frankenthal, R. P. and Kruger, J . , eds.), The Electrochemical Society, Princeton, N.J., 170 (1978) 45. Janik-Czachor, M., Werk. u. Korr., 34, 47 (1983) 46. Janik-Czachor, M., Werk. u. Korr., 34, 451 (1983) 47. Hashimoto, K., Asami, K. and Kawashima, A., Proc. 9th Int. Cong. Metallic Corrosion, National Research Council of Canada, Ottawa, 1, 208 (1984) 48. Hashimoto, K., in Passivity of Metals and Semiconductors, (Frornent, M., ed.), Elsevier, Amsterdam, 1983, 235 (1983) 49. Diegle, R. B., Corrosion, 35, 250 (1979) 50. Diegle, R. B., Corrosion, 36, 362 (1980) 5 1 . Pampillo, C. A., J. Mater. Sci., 10, 1194 (1975) 52. Archer, M. D. and McKim, R. J., Corrosion, 39, 91 (1983) 53. Devine, T. M., J. Electrochem. SOC., 124, 38 (1977) 54. Kawashima, A., Hashimoto, K. and Masumoto, T., Corros. Sci., 16, 935 (1976) 55. Kawashima, A., Hashimoto, K. and Masumoto, T., Corrosion, 36, 577 (1980) 56. Hunderi, 0. and Bergerson, R., Corros. Sci., 22, 135 (1982) 5 1 . Thomas, M. T. and Bear, D.R., Proc. 4th Int. Conf. Rapidly QuenchedMetals, The Japan Institute of Metals, Sendai, 2, 1453 (1982) 58. Ley, L. and Riley, J. D., Proc. 7th Int. Vacuum Cong., 2031 (1977) 59. Bigot, J. Calvayrac, Y., Harmeline, H., Chevalier, J-P. and Quivy, A., Proc. 4th Int. Conf. Rapidly Quenched Metals, The Japan Institute of Metals, Sendai, 2, 1463 (1982) 60. Yoshioka, H., Asami, K., Kawashima, A. and Hashimoto, K., Corros. Sci., 27,981 (1987) 61. Kumagai, N., Samata, Y., Jikihara, S., Kawashima, A., Asami, K. and Hashimoto, K., Mater. Sci. Engng, 99, 489 (1988) 62. Wang, R., J. Non-Cryst. Solids, 61 and 62, 613 (1984) 63. Diegle, R. B. and Merz, M. M., J. Electrochem. Soc., 127, 2030 (1983) 64. Anderson, R. A., Dobisz, E. A., Perepezko, J. H., Thomas, R. E. and Wiley, J. D., in Chemistry and Physics of Rapidly Solidified Materials, (Berkowitz, B. J. and Scattergood, R. 0.. eds.), the Metallurgical Society of AIME. Warrendale, 1 I 1 (1983) 65. Shimamura, K., Miura, K. Kawashima, A., Asami, K. and Hashimoto, K., Proc. Symp. on Corrosion, Electrochemistry and Catalysis of Metallic Glasses, (Diegle, R. B. and Hashimoto, K., eds.), the Electrochemical Society, Pennington, 232 (1980) 66. Yoshioka, H., K. Kawashima, A., Asami, K. and Hashimoto, K., Proc. Symp. on Corrosion, Electrochemistry and Catalysis of Metallic Glasses, (Diegle, R. B. and Hashimoto, K, eds.), the Electrochemical Society, Pennington, 242 (1988) 67. Watanabe, T. and Tanabe, Y., J. Metal Finishing SOC.Japan, 32, 600 (1981) 68. Grant, W. A., Nuclear Instruments and Methods, 182/183, 809 (1981) 69. Clayton, C. R.. Wang, Y-F. and Hubler. G. K., in Passivity of Metals and Semiconductors, (Froment, M., ed), Elsevier, Amsterdam, 235 (1983)
4
NON-FERROUS METALS A N D ALLOYS
4.1 Aluminium and Aluminium Alloys 4.2 Copper and Copper Alloys 4.3 Lead and Lead Alloys 4.4 Magnesium and Magnesium Alloys 4.5 NiLkel and Nickel Alloys 4.6 Tin and Tin Alloys 4.7 Zinc and Zinc Alloys
4: 1
4:3 4:38 4:77 4:99 4:l17 4:158 4:169
4. I
Aluminium and Aluminium Alloys
Aluminium and the aluminium alloys lend themselves to many engineering applications because of their combination of lightness with strength, their high corrosion resistance, their thermal and electrical conductivity and heat and light reflectivity, and their hygienic and non-toxic qualities. The variety of forms in which they are available also enhances their utility.
Composition and Mechanical Properties Pure aluminium has good working and forming properties, high resistance to corrosion, low mechanical strength, and high ductility. The diverse and exacting technical demands made on aluminium alloys in different applications are met by the considerable range of alloys available for general and specific engineering purposes (BS 1470-75, 1490), each of which has been designed and tested to provide various combinations of useful properties. These include strength/weight ratio, corrosion resistance, workability, castability, or high-temperature properties, to mention but a few. The compositions and properties of these standard alloys are given in Tables 4.1 to 4.4. More specialised alloys are covered by the DTD and L series for aircraft applications and include the high strength Al-Zn-Mg alloys. The medium strength weldable Al-Zn-Mg compositions are finding increasing utility in engineering and a national specification may be anticipated in the near future. Where free machining characteristics are required, this may be achieved by additions of cadmium, antimony, tin or lead (e.g. BS 4300/5). Materials for electrical use are of special composition (BS 2627,3988), while bearings are manufactured from AI-Sn alloys. Composites of aluminium alloy with a thin cladding on one or both surfaces of a more anodic aluminium alloy or pure aluminium, enable sheet, plate and tube to be produced with special combinations of strength and corrosion resistance appropriate to service conditions. Although originally applied to high strength aircraft alloys, this principle of cladding is now utilised in several important industrial applications. The I.S.O. designations may be correlated directly with the British Standard General Engineering series and partially with the American Aluminium Association designations. The nearest equivalents for the three systems are given in Table 4.5, although differences in alloying practice in America 4: 3
P .. P
Table 4.1
Some wrought British Standard aluminium alloys for general engineering purposes (non-heat treatable alloys)
::
Suitability for designation
BS 1470-75 1
1A IB IC N3 N4 N5
N6(') N8
I.S.O.
AI 99-99
Al99-8
Al 99.5 A1 99-0 Al Mnl Mg2 AI Mg3.5 Al Mg5 AI Mg4.5 Mn
Major alloying constituents (nominal Vo)
Tensile strength range* (N/mm2)
Resistance to atmospheric attack
99.99 AI 99.8 AI 99.5 Al 99 A1 1.25 Mg 2.25 Mg 3.5 Mg
100 max. 125 max.
V
5 . 0 Mg
4.5 Mg
V
55-135
V
60-140 90-175
V
160-225
V V
215-275 250 min. 275-345
V G
V
Cold forming@) Machining
V V V
Fusion welding ('1 oxy-gas
1""' gas shielded arc
V
P P
V
V V
F F
V
V
V V
V
F
V
V V
G G G G
G G G G
G F F F
G G G G
Resistance spot
Nom: 1. Rivet and screw stock only. 2. Rating are for material in the optimum condition for forming. 3. Ratings are given for correct technique and filler rod and take into account the properties of material after welding. Y = very good. G = good. F = fair, P = poor and U = unsuitable. Where strength varies with temper the specified minima are quoted. Similarly where properties arc influenced by the fabrication process. the lowest minima are given.
C
Protective anodising
5
2 F F G G
V
V
V V
V
G
V V
V
V V
V G G
P i! U
E5 z
C
3
Fr
s
v1
Table 4.2
Heat treatable alloys in the solution treated condition, but naturally aged where applicable Suitability f o r
Material designation BS 1470-75
H9 HI2 HI5 HCl5 H20
H30
I.S.O.
P
Resistance
Major alloying constituents Tensile strength * Fusion welding(” Resisrance (nominal To) Protective (N’mmz) atmospheric ‘Old Machining attack forming(*’ Oxy-gas Inert-gasanodising shielded arc
AI Mg Si 0 . 7 Mg, 0-5 Si AlCu2Nil 2 . 3 C u , l.ONi, 0.9Mg. 0.9 Fe, 0.9 Si Mg Fe Si A1 Cu4 Si Mg 4.5 Cu,0.8 Si, 0.5 mg, 0 - 8 Mn AI Cu4 Si Mg As HIS, clad with 1B Al Mgl Si Cu 1 .O mg, 0.6 Si, 0-25 Cu (0.5 Mn or 0.25 Cr) A1 Si Mg Mn 1 .O Si, 0.8 Mg,0-7 Mn
125 min
V
G
G
G
G
G
V
310 min
F
F
G
U
G
U
F
370 min 375 min
F G
F F
G G
U
G
U
F
V V
V
215 min 200 min
G G
G G
G G
F
G G
G G
G G
F
Nom: 2. Ratings are for material in the optimum condition for forming. 3. Ratings are given for correct technique and filler rod and take into account the properties of material after welding. Where strength varies with temper the specified minima are quoted. Similarly where properties are influenced by the fabrication process. the lowest minima are given.
F
Table 4.3
Material designation
BS 1470-75 H9 HI2
HCl5 H16
H30 Noless:
Suitability for Major alloying constituents Tensile strength ’ to (nominal ‘70) (N’mmz) atmospheric attack
0.7 Mg. 0.5 Si 2.3 Cu, 1.0 Ni, 0.9 Mg, 0.9 Fe, 0.9 Si AI Cu4 Si Mg 4 - 5 Cu. 0.8 Si, 0.5 Mg, 0 - 8 Mn AI Cu4 Si Mg As H15, clad with 1B AlCu2Mgl.5 2 . 3 C u , 1.5Mg, 1.2Fe. Fel Nil 1.1 Ni AI Mg Si Cu 1 S OMg, 0.6 Si, 0.25 C u (0.5 Mn or 0-25 Cr) A1 Si Mg Mn 1 .O Si, 0.8 Mg, 0.7 Mn
AI Mg Si AI Cu2 Nil
HI5
H20
I.S.O.
Heat treatable alloys in the fully heat treated condition
Fusion welding(’)
forminge) Machining
Resirtanre
oxy-gas
Inert-gasshielded arc
spot
Protective anodising
150 rnin
G
G
V
G
G
G
V
385 rnin
F
P
V
U
G
U
F
400 min 400 min
P
P
V G
G F
F
P
U U
V
G
V
V
430 rnin
P
P
V
U
G
U
F
280 min 280 rnin
G G
F F
V V
F F
G G
G G
G G
2. Ratings arc for material in the optimum condition for forming. 3. Ratings are given for c o r m technique and filler rod and take into account the properties of material after welding. ’ Where strength varies with temper the specified minima arc quoted. Similarly where properties are influenced by the fabrication process. the lowest minima are given
Table 4.4
BS 1490
Major alloying constituents (nominal lo)
Cast British Standard aluminium alloys for general and special purposes
Min. resile strength Condition
(N’mm ’) Fluidity Sand Chill cast cast
Resistance pressure to hot tightness tearing
Machinability
Resistance Protective lo anodising corrosion
Comments
GENERAL PURPOSE ALLOYS
LM2
IOSi, 1.5Cu
M
-
150
G
5Si,3CU 12Si
M TF M
140
LM4 LM6
230 160
160 280 190
v
M
-
190
-
180
130 230
160 280
LM20 12Si
LM24 8.5Si. 3 . 5 C u M M LM25 7 Si, 0 . 3 Mg TF LM27 7 Si, 2 CU, 0.4 Mn M Nore:
140
160
V
V
G
G
V
G
F
V
V
F
V
P
v
V
V
G
G
P
G
V
V
G
F
P
V
V V
G
V
G
G
G
V
G
P
P
General purpose die casting, particularly for thin sections Widely used, general purpose alloy Excellent castability for thin sections and intricate shapes, good corrosion resistance Excellent castability, but less corrosion resistant than LM6 General purpose die casting High strength with good corrosion resistance General purpose alloy
* Where strength varies with t e m p r the spcihed minima are quoted. Similarly where properties are influenced by the fabrication process. the lowest minima are given.
Table 4.4 (continued) SPECIAL PURPOSE ALLOYS
LMO
9 9 . 5 AI min.
M
-
-
F
F
F
F
V
V
LM5
4.5Mg, 0.4Mn
M
140
170
F
F
P
G
V
V
LM9
12 Si, 0.4 Mg, 0.4Mn
M
240
295
G
V
G
F
G
P
LMlO 10Mg
TB
280
310
F
G
P
G
V
F
LM12 10Cu.0.3Mg
M
-
I70
F
G
G
V
P
P
TF
170 I40 170 230 120
280
G
V
F
F
F
P
G
G
G
F
140
G
V
G V
F
V
P P
150
170
G
G
G
G
P
P
-
245
G
G
G
G
P
P
210 170 190 190
G
V
F
F
P
P
Elevated temperature applications Used for food handling equipment High strength with good engineering characteristics High shock resistance, used for structural applications Used for pistons
F
G
F
P
G
P
Special piston applications
190 130 160
F
G
F
P
G
P
Special piston applications
G
F
F
F
P
P
Automobile cylinder blocks
LM13 11 Si, 1 Cu, 1 Mg LM16 5Si, 1.3Cu, 0.5 Mg LM18 5Si
TF7 TB TF M
M LM21 6 Si, 4 Cu, 0 . 4 Mn,0.2 Mg LM22 5 Si, 3 Cu, 0.4 Mn TB LM26 9.5 Si, 3 Cu, 1 Mg TE TE LM28 18Si, 1.5Cu. TE IMg, 1Ni TE LM29 23 Si, I Cu, 1 Mg, TF 1 Ni M LM30 17Si. 4 * 5 C u , TS 0.5 Mg
120 120 120 -
Now Y = very good, G = good. F = fair and P = poor
190
200 230 280
Suitable for corrosive environments and electrical applications Suitable for corrosive environments and for decorative applications Good combination of castability, corrosion resistance and strength Combines toughness with shock resistance Specially suitable for hydraulic equipment Used for pistons
4:9
ALUMINIUM AND ALUMINIUM ALLOYS Table 4.5
British Standard, I .S.O. and comparable aluminium association designations
Nearest A A designation
Major alloy constitutents (nominal To)
99.8
1099 I080
99.5
1050
99.0
I I IO
99.99 AI 99.8 AI 99.5 AI 99 AI
Mnl Mg2 Mg3.5 Mg5 Mg4.5 Mn
3003 5052 5154 5456 5083
I .25 Mn 2.25 Mg 3.5 Mg 5.0 Mg 4.5 Mg
BS 1470-75
I
I.S.O.
99.99
IA 1B IC
AI AI AI AI
N3 N4 N5 N6 N8
AI AI AI AI AI
H9 HI2
AI Mg Si A1 Cu2 Nil Mg Fe Si
HI5 HC15 HI6 H20
AI AI AI AI
H30
AI Si Mg Mn
Cu4 Si Mg Cu4 Si Mg Cu2 M g l . 5 Fel Nil Mgl Si Cu
6063
-
2014 AI clad 2024
6061 635 1
0.7 Mg, 0.5 Si 2 . 3 Cu, 1 .O Ni, 0 . 9 Mg, 0 . 9 Mg, 0 . 9 Fe, 0 . 9 Si 4 . 5 Cu, 0.8 Si, 0.5 Mg, 0.8 Mn As HIS, clad with I B 2 . 3 C u . 1.5Mg. 1.2Fe. I . I N i 1 .O Mg, 0.6 Si, 0.25 Cu (0.5 Cu (0.5 Mn or 0.25 Cr) I .O Si, 0.8 Mg, 0 . 7 Mn
make a direct correlation impossible in several cases. Notably, where European practice utilises a minor addition of manganese, a similar effect is frequently achieved by a chromium addition in American practice; also, there is a greater tendency to use small copper additions in American alloys than in European alloys. The British Standard alloys use a systematic letter notation to indicate the form and heat treatment of the material. Details are given in Table 4.6. For example, the strongest condition is H8 for a non-heat-treatable alloy, TB or TD for a single heat-treatment alloy, and TE, T F and T H for a double heattreatment alloy. The non-heat-treatable alloys (prefixed N) are hardened by cold work and attain the desired properties by a combination of annealing and cold work. The hard material has markedly increased strength with only slightly reduced corrosion resistance. The heat-treatable alloys (prefixed H)- notably the AI-Cu-Mg and the Al-Mg-Si types-can be heated at 480-535°C for a period between 20min and some hours to obtain solution of the alloying elements, and then rapidly quenched. This solution treatment gives increased strength, and may also give slightly increased corrosion resistance. Further strengthening of certain alloys is achieved by an additional lower temperature heat-treatment for longer periods (1-20 h or more, according to the alloy) which promotes precipitation of the alloying elements within the metal crystal structure. With some alloys this ageing treatment takes place at room temperature. The ageing or precipitation treatment slightly reduces the corrosion resistance of most alloys. Comprehensive details of alloy properties and characteristics are provided in the publications of the major aluminium companies and independent organisations I .
4: 10
ALUMINIUM AND ALUMINIUM ALLOYS
Table 4.6
Temper and heat treatment symbols for aluminium alloys-suffixes
British Standard designation
Meaning WROUGHT MATERIALS
M
0 H1, H3, H5, H7,
H2 H2 H6 H8
TB
TD TE
TF TH
As manufactured. Material which acquires some temper from shaping processes in which there is no special control over thermal treatment o r amount of strain hardening Annealed. Material which is fully annealed to obtain the lowest-strength condition Strain hardened. Material subjected to the application of cold work after annealing (or hot forming) or to a combination of cold work and partial annealing/stabilising in order t o secure the specified mechanical properties. The designations 1-8 are in ascending order of tensile strength Solution heat treated and naturally aged. Material which receives no cold work after solution heat treatment except as may be required to flatten o r straighten it. Properties of some alloys in this temper are unstable Solution heat treated, cold worked and naturally aged Cooled from an elevated temperature shaping process and precipitation treated Solution heat treated and precipitation treated Solution heat treated, cold worked and then precipitation treated CAST MATERIALS
M TS TE TB TB7 TF TF7
As cast Stress relieved only Precipitation treated Solution treated Solution treated and stabilised Solution treated and precipitation treated Full heat treatment plus stabilisation
Physical Properties Some of the more useful physical and mechanical properties of aluminium are given in Tables 4.7 and 4.8. The common wrought forms are rolled plate (prefixed P), clad plate (PC), sheet and strip (S), clad sheet and strip (C), bars, rods, and sections (E), extruded round tube and hollow sections (V), drawn tubes (T), wire (G), rivet stock (R),bolt and screw stock (B), and forgings and forging stock (F). Castings are made in sand moulds or in metal moulds known as dies; the most widely used methods involve casting either under gravity or under pressure. Aluminium and aluminium alloys are fabricated into products such as roll plate, sheet, extruded sections, drawn tube, etc. by all the familiar processes, with modifications appropriate to the temper or condition of the material. Joining may be carried out by mechanical methods (such as riveting and bolting), brazing, soldering, adhesive bonding, or welding. The argon-shielded arc welding methods (mig and tig) are particularly appropriate where corrosion resistance of welded joints is of importance*.
> r
Electrical
Thermal
C
(K)
Sp. heat at 293 K (JlkgK)
Mean sp. heat (293-931 K) (J/kgK)
Latent heat of fusion (kJ/kg)
93 I
8%
1 047
387
m.p.
Coeff. of linear exp. (293-393 K) (m/K) 0.61 x
Thermal conductivity at 273 K (W/m K) 214
Elec. Vol. resistivity at 293 K (pn cm) 2.7-3.0
Elec. vol. conductivity at 293 K
(wI.A.c.s.) 63-57
Temp’ ‘Oeff‘ of elec. resistance per for 293 K 0-0041
Thermoelectric power vs platinum (mV/100 K)
E
+0*41
5;
z
?!
2
4: 12
ALUMINIUM A N D ALUM1
Table 4.8
Mechanical propertiei of aluminium
Young's modulus (MN/rn*)
Torsion modulus (MN/rnZ)
59 x lo3
24
X
UM ALLOYS
Poisson's
IO3
Compressibiliry ( dv/vo Q ) at 293 K
0.34
1.45 x IO-('
at 4 0 0 K 1.70
X
Selection of Purity or Alloy Type It will be noted that the materials covered by the BS specifications fall into several distinct groups, sometimes with apparently small differences within the group. Characteristics which could influence the selection of the most appropriate material for a specific application are tabulated in Table 4.1 for wrought products, but some elaboration is desirable since the successful utilisation of aluminium begins with the selection of alloy. Additionally, mention should be made of materials not covered by the BS General Engineering series. Pure Aluminium
Within the BS series the corrosion resistance of unalloyed aluminium increases with increasing metal purity. The use of the 99.8% and 99.99% grades is usually confined to those applications where very high corrosion resistance or ductility is required. The chemical industry can advantageously use these purities for handling some products, but because of their low mechanical strength they are sometimes used as a cladding material for a stronger substrate. Decreasing the purity results in increased strength for the 99% and 99.5% grades, which still retain a high resistance to corrosion. The 99% pure metal may be considered the more useful general purpose metal for lightly stressed applications such as panelling and cooking utensils. Aluminium-Manganese Alloy
The alloy N3 is the sole BS alloy of this binary system. In sheet form, the combination of good corrosion resistance with adequate mechanical strength results in large tonnages being used in building, cooking utensils and sundry general applications. Aluminium-MagnesiumAlloys
For general use, the AI-Mg system is represented by N4, N5 and NS with increasing magnesium content respectively. The corrosion resistance of all these alloys is extremely good, while the level of mechanical properties obtainable makes them ideally suited for structural use in aggressive conditions.
ALUMINIUM A N D ALUMINIUM ALLOYS
4 : 13
The characteristics of these alloys make them ideal for boat and shipbuilding, for which a long history of satisfactory performance is on record for the higher magnesium alloys. Where strength is less critical the lower magnesium alloys may be used with similar success and are recommended for aqueous conditions. Elevated temperatures should be avoided with N5 and N8, since the precipitation of Mg,Al, over a period of time can lead to serious structural corrosion. In case of doubt regarding this aspect, the manufacturer should be consulted. Aluminium-Magnesium-SiliconAlloys
The heat-treatable AI-Mg-Si alloys H9, H20 and H30 are predominantly structural materials, all of which have a high resistance to corrosion. The low Mg Si content of H 9 facilitates the production of complex extrusions with a good surface finish making H9 a natural choice for glazing sections and other architectural features. Higher mechanical properties are obtainable with the H20 and H30 compositions, which are therefore more suitable for load bearing structures. The corrosion resistance of the AI-Mg-Si alloys is slightly inferior to that of the AI-Mg alloys, but where maximum obtainable strength is required then a fully heat-treated AI-Mg-Si alloy would generally be preferable to an AI-Mg alloy with comparable properties obtained by cold working.
+
Aluminium-Copper AIIo ys
The composition of these alloys extends beyond the binary system and they may be categorised as the Duralumin type H 15 and the complex types H 12 and H16. The mechanical properties and characteristics of H15 cause it to be used for those applications where high strength is the prime criterion, outweighing its poor resistance to corrosion. Protection by anodising or painting is desirable, when satisfactory performance may be expected except in the most severe conditions. Alternatively, the HC15 clad version has a corrosion resistance similar to its pure cladding, provided that repeated heat treatments have not caused excessive copper diffusion into the pure cladding. The use of HI5 for machined components is fairly common, but cannot be recommended where the service conditions will be aggressive. Where the retention of strength at elevated temperatures is required, then the alloys H12 and H16 should be considered. Because of their copper content the corrosion resistance is mediocre and for service in aggressive environments the Al-1Zn clad version to DTD 5070 would generally be preferred to the unclad metal. Aluminium-Zinc-MagnesiumAlloys
The AI-Zn-Mg alloy system provides a range of commercial compositions,
4: 14
ALUMINIUM A N D ALUMINIUM ALLOYS
primarily for those areas where strength is a major consideration. None of these alloys are yet included in the BS General Engineering series although their use in Europe and America is quite well established. Essentially the range of compositions may be conveniently divided into two categories. The high strength alloys contain a Zn Mg content well in excess of 6% and are used in specialist structures such as aircraft. The risk of stress corrosion cracking in these alloys may be accentuated by incorrect heat treatment or composition and they cannot be recommended for general use (Section 8.5). The other group of alloys are those with a Zn + Mg content not exceeding 6%. These have been used for general engineering, when natural ageing after welding can be utilised to permit the fabrication of strong welded structures. In particular, these medium strength Al-Zn-Mg alloys have been successfully used for transport applications and it seems probable that this will increase in the near future. With correct manufacturing procedures the risk of stress corrosion with these alloys is negligible and the resistance to unstressed corrosion is only slightly inferior to the AI-Mg-Si structural alloys.
+
Corrosion Behaviour in Aqueous Environments Theoretical Considerations of Corrosion Behaviour
Aluminium is a very reactive metal with a high affinity for oxygen. The metal is nevertheless highly resistant to most atmospheres and to a great variety of chemical agents. This resistance is due to the inert and protective character of the aluminium oxide film which forms on the metal surface (Section 1.5). In most environments, therefore, the rate of corrosion of aluminium decreases rapidly with time. In only a few cases, e.g. in caustic soda, does the corrosion rate approximate to the linear. A corrosion rate increasing with time is rarely encountered with aluminium, except in aqueous solutions at high temperatures and pressures. The corrosion resistance of aluminium and its alloys is largely due to t t e protective oxide film which within seconds attains a thickness of about 10 A on freshly exposed metal; continuation of growth is markedly influenced by the environment, being accelerated by increasing temperature and humidity. Immersion in water results in rapid oxide thickening. The behaviour of the oxide may be modified by impurities or alloying additions. In aluminiummagnesium alloys the presence of magnesia in the oxide imparts a characteristic bloom to metal stored under humid conditions. The possible effects of minor impurities or additions is well illustrated in the case of tin, whose modifying effect upon the oxide’ is utilised in obtaining a highly electronegative potential in aluminium sacrificial anodes (see Section 10.2). The oxidation of aluminium at room temperature is reported to conform to an inverse logarithmic equation for growth periods up to 5 years duration4. At elevated temperatures, oxidation studies over shorter periods illustrate conformity to parabolic, linear and logarithmic relationships according to time and temperature. These kinetic variations are attributed to different mechanisms of film The various equilibria of the AI-H, 0 system have been collated by
4: 15
ALUMINIUM A N D ALUMINIUM ALLOYS 5
6
7
8
--f.l--L--L
Q I
.I
0.80.6-
AI^+
COLOURLES!
0.20--
-.0.2-
-a. --
CORROSION
> -
I
I
J-0.4-
I
AI+?
a -1
16
I
0.8
CO LOU A l OR ; LESS
0.4
0.2
0
.-\ -.
I
W
6-0.8-
I5
I PASS IVAT ION I
g-0.6-
;I 14
0.6
HYDRARGILLIT I WHITE
-L I
5
-. .
. 7
I A120,.3H,0
0.4-
13
I .2
I
-
I2
I .A
I
1
1
I
-0.4
.--
-0.6
- 0.8
I
-1 . 2 -
-I .2
I I I I
4
I
I
-I .8 -2
I
AI I M M U N IT Y
-2.2
-%
-2.4 -2.6 15
Fig. 4.1
16
Potential versus pH diagram for AI/H20 system at 25°C (after Corrosion, 14, 4961 (1958))
Pourbaix et ai. in a potential versus pH diagram (Fig. 4.1). This diagram indicates the theoretical circumstances in which aluminium should show corrosion (forming A13+ at low pH values and A10; at high pH values), passivity due to hydrargillite, i.e. Al,O, . 3 H 2 0 (at near-neutral pH values) and immunity (at high negative potentials). The nature of the oxide actually varies according to temperature, and above about 75"C, boehmite (A1203*H,O) is the stable form. It should be noted that the potential-pH diagram does not indicate one of the most important properties of aluminium, i.e. its ability to become passive in strongly acid solutions of high redox potential such as concentrated nitric acid (see also Section 1.4).
Characteristic Features of Corrosion Behaviour General Dissolution
This occurs in strongly acid or strongly alkaline solutions, but there are specific exceptions. Thus in concentrated nitric acid the metal is passive and the kinetics of the process are controlled by ionic transport through the
4: 16
ALUMINIUM A N D ALUMINIUM ALLOYS
oxide film, while inhibitors such as silicates permit the use of some alkaline solutions up to pH 11 a 5 to be used with aluminium. Even where corrosion may occur to a limited extent aluminium is often preferred to other metals because its corrosion products are colourless. Pitting
This is the most commonly encountered form of aluminium corrosion. In certain near-neutral aqueous solutions a pit once initiated will continue to propagate owing to the fact that the solution within the pit becomes acid, and the alumina is no longer able to form a protective film close to the metal. When the aluminium ions migrate away from the areas of low pH, alumina precipitates as a membrane, further isolating and intensifying local acidity, and pitting of the metal results (see Section 1.6). Solutions containing chlorides are very harmful, particularly when they are associated with local galvanic cells, which can be formed for example by the deposition of copper from solution or by particles such as iron unintentionally embedded in the metal surface. In alkaline media pitting may occur at mechanical defects in the oxide. Pits usually have no crystallographic shape although structurally indicative etch pits can be produced on aluminium. Where perforation is the criterion of failure, statistical analysis may be judiciously applied to the distribution and depth of pits. Aziz' shows that the maximum pit depth on comparatively small test pieces can be related linearly with the maximum depths to be expected in service on large areas over the same period of time. This involves the use of special probability paper (graph paper ruled in such a way that data involving random probabilities may be plotted to give straight-line relationships). Other work from the same laboratory indicates that the use of a small size of panel or of an insufficient number of panels may invalidate pitting test results. Media which are capable of causing pitting may produce no attack when the panels are too small or may attack only a percentage of the panels. Intercrystalline Corrosion
This is also electrochemical in nature, the galvanic cell being formed because of some heterogeneity in the alloy structure, which may arise from major or trace alloying additions or from minor elements present. In the aluminiumcopper type alloys, precipitation of CuAI, particles at the grain boundaries leaves the adjacent solid solution anodic and prone to corrosion'. With aluminium-magnesium alloys the opposite situation occurs, since the precipitated phase Mg,Al, is less noble than the solid solution. However, serious intercrystalline attack in these two alloys is not usual, provided that correct manufacturing and heat treatment conditions are observed. In the case of the aluminium-magnesium system, most commercial alloys are usually supersaturated, so that elevated service temperatures and inexpert heat treatment are inadvisable, since any resultant grain boundary precipitation may induce susceptibility to intercrystalline attack. The extent of this susceptibility may be approximately deduced from the continuity of
ALUMINIUM AND ALUMINIUM ALLOYS
4: 17
Mg,Al, at the boundaries, continuous or nearly continuous films being extremely detrimental and discrete widely spaced particles being relatively harmless. Trace elements which adversely affect intercrystalline attack are normally controlled at a safe level. Copper is particularly pertinent in this respect since relatively small additions can cause a marked increase in intercrystalline attack in some alloy systems (Sections 1.3 and 1.7). Stress Corrosion
This form of corrosion is of limited occurrence with only a few aluminium alloysg, in particular the higher strength materials such as the Al-Zn-MgCu type and some of the AI-Mg alloys, wrought and cast, with the higher magnesium contents, notably after specific low-temperature heat treatments such as occur during stove enamelling. Stress corrosion is intergranular on aluminium alloys (see Section 8.5). Filiform Corrosion
This appears as a random non-branching white tunnel of corrosion product either on the surface of non-protected metal or beneath thin surface coatings. It is a structurally insensitive form of corrosion which is more often detrimental to appearance than strength, although thin foil may be perforated and attack of thin clad sheet (as used in aircraft construction) may expose the less corrosion resistant aluminium alloy core. Filiform corrosion is not commonly experienced with aluminium, as reflected by the insignificance afforded it in reviews on the phenomena" (Section 1.6). Layer Corrosion
This may occur on material which has a marked fibrous structure caused by rolling or extrusion. The attack is rapid and very selective, forming partly detached layers of relatively uncorroded material. It is regarded by some authorities as a form of stress corrosion, the stress being either inherent in the metal or produced through the pressure of the larger volume of the corrosion product. It is rare, occurring mainly in copper-bearing alloys, but can occur in a number of environments, including some regarded as only mildly corrosive. Suitable adjustments of ageing treatments and copper content may largely overcome the effect in the higher-strength AI-Cu type alloys I ' (Section 1.3). Effect of Composition
Few general statements can be made regarding the effect on corrosion resistance of alloying elements or impurities. A useful summary of the information has been prepared by Whitaker'*. Copper is usually harmful causing increased susceptibility to intercrystalline or general attack, so that alloys
4 : 18
ALUMIMUM A N D ALUMINIUM ALLOYS
containing copper should be regarded as less corrosion resistant than copperfree materials. There are however exceptions to this generalisation, such as an improved stress corrosion resistance in AI-Zn-Mg alloys obtained by a small copper addition I 3 * l 4 . Alternatively, the presence of copper may be utilised to delay perforation at the expense of increased general corrosion. With increasing purity of aluminium, greater resistance to corrosion is developed. On high-purity materials, however, any pits which develop are likely to be deeper though fewer in number than those formed in more impure metal. In some special applications, notably in contact with ammonia solutions or pure water at elevated temperatures and pressures, the iron and silicon present in commercial-purity metal are beneficial and retard corrosion. Up to about 5% magnesium improves the corrosion resistance to sea-water. Bimetallic Corrosion (Section 1.7)
Aluminium is anodic to many other metals and when it is joined to them in a suitable electrolyte-which may even be a damp porous solid-the potential difference causes a current to flow and considerable corrosion can result. Corrosion is most severe when the resistance of the electrolyte is low, e.g. sea-water. In some cases surface moisture on structures exposed to an aggressive atmosphere can give rise to galvanic corrosion. In practice, copper, brasses, and bronzes in marine conditions cause the most trouble. The danger from copper and its alloys is enhanced by the slight solubility of copper in many solutions and its subsequent redeposition on the aluminium to set up active local cells. This can occur even when the copper and aluminium are not originally in contact, e.g. when water running over cuprous surfaces subsequently comes into contact with aluminium. Similarly, water washings from lead can cause pitting of aluminium. The controlling factor with lead and cuprous washings is the solvency of the water, so that soft waters are the most damaging in this respect. The successful utilisation of these metals in close proximity to aluminium, e.g. in plumbing and roofing, therefore requires careful design to avoid the transfer of a harmful solute to the aluminium. Contact with steel, though less harmful, may accelerate attack on aluminium, but in some natural waters and other special cases aluminium can be protected at the expense of ferrous materials. Stainless steels may increase attack on aluminium, notably in sea-water or marine atmospheres, but the high electrical resistance of the two surface oxide films minimises bimetallic effects in less aggressive environments. Titanium appears to behave in a similar manner to steel. Aluminium-zinc alloys are used as sacrificial anodes for steel structures, usually with trace additions of tin, indium or mercury to enhance dissolution characteristics and render the operating potential more electronegative. Aluminium may accelerate attack on zinc alloys; this is particularly noticeable when there is an unfavourable area ratio, as with galvanised fittings in aluminium shects. In alkaline solutions, however, the aluminium may be preferentially attacked. The copper-bearing aluminium alloys are more noble than most other aluminium alloys and can accelerate attack on these, notably in sea-water. Mercury and all the precious metals are harmful to aluminium.
ALUMINIUM A N D ALUMINIUM ALLOYS
4: 19
Mechanical and Design Factors Stress below the proof stress does not normally affect corrosion rates. Cyclic stresses in combination with a corrosive environment (corrosion fatigue) can produce failure at below the ordinary fatigue limit. Alloys susceptible to intergranular attack may corrode faster when stressed (see Section 8.5). Soldered or brazed joints will usually have lower corrosion resistance than the parent metal, but sound welded joints with resistance to attack equal to that of the parent metal can be obtained in most alloys2. Many assemblies contain angles, pockets or crevices which attract moisture originating either from extenal sources or from condensation. The corrosion so caused could often be avoided by slight redesign of the assembly, the provision of drain holes of at least 5 mm dia., and the avoidance of horizontal surfaces being among the more important features. Crevices may be filled with jointing compounds. In static assemblies these compounds may be of the setting variety, but in assemblies subject to vibration or movement, as on ships, it is essential that the mastic used should not become too rigid as it might crack in service. It is advisable to incorporate chromates in jointing compounds to inhibit attack by any moisture that may penetrate.
Corrosion in Natural Environments Atmospheric
The aluminium alloys as a group weather outdoors to a pleasant grey colour, which deepens to black in industrial atmospheres. Superficial pitting occurs initially but gradually ceases, being least marked on high-purity aluminium. With some alloys, including the copper-bearing alloys and the mediumstrength AI-Zn-Mg alloys, additional protection, e.g. painting, is desirable in the more aggressive atmospheres to avoid any risk of intercrystalline corrosion. Gases such as hydrogen sulphide and carbon dioxide do not increase the corrosivity of the atmosphere towards aluminium 15. Service experience extends over 70 years and includes such well-known examples as Eros, Piccadilly Circus, London, which is in excellent condition, although cast in a low purity (98%) aluminium, and a cupola of San Gioacchino, a church in Rome which was covered in 1897 with sheet 1.25 mm thick and now shows attack to a depth of less than 0.13 mm. Twenty-year tests at selected marine, industrial and rural sites in the U.S.A.l6 have shown that the greater part of the attack takes place in the first year or two and that thereafter the rate of attack maintains a low value. Results from typical environments are shown in Fig. 4.2, and it is apparent that clad alloys give the best results. The relatively high percentage strength losses are due to the extremely thin test specimens. After 20 years the average measured depth of attack for an aluminium-copper alloy at a sea coast test site did not exceed 0.15 mm. The falling-off in rate of pitting with time is in sharp contrast to the behaviour of the older-established structural metals which have a fairly uniform corrosion rate throughout their life, and indicates that the relative merit of aluminium increases with scheduled life.
4:20
:
ALUMINIUM AND ALUMINIUM ALLOYS
:
,
0
(a)
LO v)
6 0
9 40
20
0 ,
0
2
4
6
8
IO
12
14
16
,
18
I
,
20
/
22
EXPOSURE P E R l O O ( y e a r s )
(C)
Fig. 4.2 A.S.T.M. 20-year corrosion tests: .2107-T3 0 3003-Hl4 A 6051-T4 0 1100-HI4 Alclad 2017-T3 ( a ) State College, Pa. (rural). Premachined tension specimens 0.89 mm thick. Curves for
1100-Hl4, 3003-HI4 and Alclad 2017-T3 fall below curve shown; ( b ) New York, N.J. (industrial); ( c ) La Jolla, Calif. (seacoast) (after A.S.T.M. Symposium on Atmospheric Corrosion of Non-ferrous Metals, 27 (1955))
Aggressive environments include marine conditions and particularly industrial atmospheres containing high concentrations of acid gases such as sulphur dioxide; rain washing is beneficial in both environments, while dampness and condensation alone can accentuate the rate of attack in the presence of chlorides and acidic sulphates. The relative aggressivity of industrial, marine and rural conditions has been clearly demonstrated by the results of seven year tests in the U.S.A. and British Isles”, and in this work the benefit from rain washing was especially manifest for the industrial sites in the British Isles (Fig. 4.3). The combination of acidic sulphates and condensation experienced in some industrial conditions, can cause a particularly voluminous loose corrosion product on some alloys, such as NS3. Where this is likely to be
ALUMINIUM A N D ALUMINIUM ALLOYS
4:21
troublesome, cladding with high purity metal is recommended and has been successfully employed, for example on the underside of aluminium-roofed industrial buildings.
12 -
c
a-
c ln d
VI
d
.
2 6.01
Sites
E
7
- 5.0a,
v
-
1.00 .-
3.00,
-
c
2.0-
3
1
m
0
0
-
d
3
1.0-
0
0-
A-Kure beach 0 - Newark C -Paint Reyes D -State college S Sheffietd L - London H- Hoyling Is. BAN- Banbury ANG- Anglesey
-
VI
a, 0
a
-
1 , , , 1 L
H
BAN ANG
Fig. 4.3 Comparison of weathering effectsat United States and English sites- 1199-HI8 (and SI-H) alloy-1 year data (after Metal Corrosion in the Atmosphere, A . S . T . M . Pub. STP435, 151 (1967))
While the continual removal of atmospheric pollution by rain washing is beneficial, the removal of the protective corrosion product is obviously undesirable. The retention of the weathered surface is therefore usually preferred unless aesthetic considerations are of major importance, in which case abrasive or specialist chemical cleaning are effective. In urban areas, atmospheric fall-out of carbon from partially burned fuel can cause severe localised pitting by galvanic action, although this is not commonly encountered.
4:22
ALUMINIUM A N D ALUMINIUM ALLOYS
Indoors, aluminium retains its appearance well, and even after prolonged periods may show no more than slight dulling or on aluminium-magnesium alloys a slight bloom. This superficial deterioration can be accelerated by the presence of moist conditions and condensation which in extreme cases may lead to staining. The presence of condensation in confined spaces, such as the turns of a coil or stacked sheets, can cause a more severe staining accompanied by a thick bloom. Oiling or the use of interleaving is sometimes successful in preventing this damage in marginal cases, but improved storage conditions or the elimination of crevice conditions are preferable. Natural Waters
Immersed aluminium and its alloys have excellent resistance to attack by distilled or pure condensate water, and are used in industry in condensing equipment and in containers for both distilled and deionised water, as well as in steam-heating systems '*. Of the more important British Standard alloys, only those which contain copper as a major alloying constituent are likely to corrode in unpolluted sea-water, but pollution of the sea-water may cause localised pitting attack to occur on other aluminium alloys. The AI-Mg alloys containing up to about 4.5% magnesium offer particularly good combinations of corrosion resistance and strength. Fouling collects readily on aluminium alloys, as on other materials, and where it may be necessary to use paints containing cuprous oxide for anti-fouling purposes the risk of bimetallic corrosion can be substantially inhibited by a chemical pretreatment of the aluminium followed by a chromate priming paint. Mercury-containing anti-fouling compositions must never be used, as serious bimetallic corrosion will result. The behaviour of aluminium in natural fresh water and tap waters may vary as these waters differ widely in their dissolved solid content. No corrosion occurs immediately on immersion of aluminium and its alloys in these near-neutral waters, and aluminium gives satisfactory service with all types of tap water provided regular cleaning and drying can take place, as occurs with aluminium hollow-ware. In some waters, black or brown stains which are largely due to optical effects associated with the oxide film on the metal surface, occur. Although somewhat unsightly, the film is quite harmless and can be removed by simple methods such as boiling of fruit (e.g. rhubarb). Alternatively. preliminary boiling with pure water provides some protection against the staining, but can hardly be considered justifiable in most cases. The combination of carbonate, chloride and copper is more damaging than if they are present singly or if one of them is absent ' 9 * 2 0 , so that some supply waters are naturally more aggressive than others. The role of copper is of particular relevance, since as little as 0.02 parts per million can initiate pitting in hard waters2', although more is required in soft waters which are otherwise less aggressive. In this context however it must be remembered that soft waters are inherently more cupro-solvent than hard waters; consequently the conjoint use of aluminium and copper fittings is rarely advisable irrespective of the necessity for avoiding galvanic interaction when the two metals are in direct contact.
ALUMINIUM A N D ALUMINIUM ALLOYS
4:23
Once pitting has started it may continue in solutions which would themselves be incapable of initiating corrosion. In waters of all types, the rate of increase in the depth of pitting falls off rapidly with time. Water movement (of the order of 0.3 m/s or more) will reduce pitting or prevent its initiation. A rise in temperature tends to lead to higher corrosion rates at existing pits, but even with the most aggressive hard waters, above about 50°C the oxideforming mechanisms act to prevent the initiation of pitting, as shown by the long and satisfactory service given by aluminium hollow-ware which is assisted in some waters by scale formation. Where aluminium is to be used in direct contact with cold natural waters with no possibility of regular cleaning, clad aluminium alloys are the preferred materials. An Al-1.2 Mn alloy clad with AI-1.2 Zn is suitable. The cladding is anodic to the core and corrosion is therefore restricted to the surface cladding, thus obviating the risk of perforation. Cladding with super-purity aluminium is preferable where it is important to have the minimal degree of total corrosion, but in this case the potential relationship with the core is more critical and in some circumstances the cladding can actually become cathodic. Sacrificial protection may also be obtained from sprayed coatings of appropriate composition which can be applied to extrusions and castings as well as to sheet, rod, plate and tubes. In practice, unclad aluminium-manganese alloys have been used for piping soft waters in this country and, more widely, in the USA.
Underground Corrosion by Soils
This is largely related to the presence of moisture which can leach out soluble constituents from the soil. As is the case with natural waters, the nature of the corrosive environment is a more important factor than the alloy used, provided that copper-bearing alloys are avoided. At the present time it is impossible to produce a satisfactory classification of soils in respect of their aggressive action on aluminium alloys. Made-up ground, particularly when it includes cinders, is usually extremely corrosive, while neutral clays are often least corrosive. It is desirable that protection should be given to all aluminium materials buried in except where there is previous experience of satisfactory service from aluminium in a given soil. Pipe wrappings based upon bitumen or chromates are effective, while for cable sheathing a continuous plastic coating provides both electrical and corrosion protection. Cathodic protection has been utilised for pipeline^^^'^^ but is not widely practised; close control is necessary since over-protection can result in alkali attack. Potentials in the region of - 1 . O V vs saturated Cu/CuSO, are favoured, although some divergence of opinion exists in this respect.
Corrosion in Chemical Environments Detailed information about behaviour of specific chemicals is given in several works of r e f e r e n ~ e ~ ~ . ~ ' .
4124
ALUMINIUM AND ALUMINIUM ALLOYS
Acids
Fig. 4.4 Action of nitric acid of various concentrations on commercial-purity aluminium at 20°C (after Reference 26)
Fig. 4.5
Action of sulphuric acid of various concentrations on commercial-purity aluminium (after Reference 26)
4:25
EXPOSURE TIME ( d )
Fig. 4.6 Action of dilute (0.I N) solutions of inorganic acids on commercial.-purity aluminium at 25°C (after Reference 26)
4000
-- 3000 U N
-'k VI
2
2000
I
w
W
3
1000
Fig. 4.7
Action of 40% hydrochloric acid on aluminium of various purities at 20°C (after Reference 26)
4:26
ALUMINIUM AND ALUMINIUM ALLOYS
Most acids are corrosive to aluminium-base materials. The oxidising action of nitric acid at concentrations above about 80%, however, causes passivation of aluminium. Very dilute and very concentrated sulphuric acid dissolves aluminium only slowly. Figures 4.4 and 4.5 give corrosion data at various concentrations for these two acids. The corrosion rates of aluminium in other inorganic acids in dilute solution are shown in Fig. 4.6. Boric acid also exerts little attack on aluminium, while a mixture of chromic and phosphoric acids can be used for the quantitative removal of corrosion products from aluminium without attacking the metal.
30
20
0
0
40
20
60
80
100
EXPOSURE TIME (days)
Fig. 4.8 Action of diIute(0.1
solutions of organic acids on commercial-purity aluminium at 25'C (after Reference 26)
N)
The effect of commercial metal purity (impurities mainly iron and silicon) on corrosion by 40% hydrochloric acid is shown in Fig. 4.7. This curve is typical of that obtained with many acids.
ALUMINIUM AND ALUMINIUM ALLOYS
4:27
Organic acids usually have low rates of attack on aluminium, notable exceptions to this generalisation being formic acid, oxalic acid and some chloride-containing acids such as trichloroacetic acid. Corrosion rates for dilute organic acid solutions are given in Fig. 4.8. Glacial acetic acid (pH 3) has no significant corrosive effect on aluminium but the rate of attack increases rapidly with decreasing concentration or in the absence of the traces of water normally present. The rate of corrosion in an acid solution rises rapidly with temperature, often doubling or more with a 10°C rise.
Alkalis
Alkalis are generally corrosive to aluminium; caustic soda is in fact used for chemical milling of aluminium. 99.0% aluminium is, however, resistant to ammonium hydroxide, even at pH 13, while the action of more dilute caustic alkalis can be inhibited by the use of silicates. Mild alkalis such as sodium carbonate are moderately corrosive and are not recommended for washing aluminium hollow-ware. Synthetic detergents, in general, give satisfactory service in cleaning aluminium, but those containing uninhibited sodium carbonate may give some surface roughening. Inhibitors such as silicates can prevent attack by the more dilute solutions.
METAL PURITY I%)
Fig. 4.9
Action of 5.6% potassium hydroxide solution o n aluminium of various purities at 20°C (after Reference 26)
Alloys of aluminium with magnesium or magnesium and silicon are generally more resistant than other alloys to alkaline media. The corrosion rate in potassium and sodium hydroxide solutions decreases with increasing purity of the metal (Fig. 4.9), but with ammonium hydroxide the reverse occurs.
4:28
ALUMINIUM A N D ALUMINIUM ALLOYS
Inorganic Salts
Most simple inorganic salt solutions cause virtually no attack on aluminiumbase alloys, unless they possess the qualities required for pitting corrosion, which have been considered previously, or hydrolyse in solution to give acid or alkaline reactions, as do, for example, aluminium, ferric and zinc chlorides. With salts of heavy metals - notably copper, silver, and gold - the heavy metal deposits on to the aluminium, where it subsequently causes serious bimetallic corrosion. Some salts, notably chromates, dichromates, silicates, borates and cinnamates, have marked inhibitive power and are very effective in closed-circuit water systems. Care must be taken to ensure that a sufficient quantity of such anodic inhibitors as chromates is added, as otherwise attack, though occurring at fewer points, may be more severe at these points. Chromates and dichromates have little inhibitive power in strongly acid solutions. Aluminium is used in hydrogen peroxide (H.T.P.) processing and storage equipment partly because of its high corrosion resistance but also because it does not cause degradation of the peroxide.
Organic Compounds
With many organic compounds, aluminium shows high corrosion resistance either in the presence or absence of water. The lower alcohols and phenols are corrosive when they are completely anhydrous - rarely encountered in practice - since repair of breaks in the natural protective oxide film on aluminium cannot take place in the absence of water. Amines generally cause little attack unless very alkaline. Processing and storage equipment for many chemicals, including acetaldehyde, formaldehyde, nylon salt, methyl methacrylate, carbon tetrachloride, glycerol, triacetin, proprionic acid, acetic acid and acetic anhydride, is manufactured from aluminium alloys, primarily because of their excellent corrosion resistance. Antifreeze solutions based on ethylene glycol additions to water have been standardised (BS 3150-3152), the standard differing in the type of corrosion inhibitor present. Inhibition of antifreeze with sodium mercaptobenzothiazole and triethylamine phosphate (NaMBT TEP) has been used for many years with complete success in contact with aluminium, e.g. in aeroengines, but difficulties with graphitisation of case iron engine components in the solution have led to the introduction of two other types of inhibitors: ( a ) benzoate plus nitrite, and ( b )borax, usually with soluble oils. Service experience has indicated that corrosion of aluminium components in these inhibited solutions occasionally takes place, though most trials give satisfactory results. In refrigerating systems, halogen derivatives of methane and ethane marketed under the trade names of Arctons and Freons are without action on pure aluminium and its copper-free alloys in dry conditions, but in wet conditions monochlorodi-, dichloromono-and trichloromonofluoromethanes can hydrolyse to produce slight attack on the aluminium.
+
ALUMINIUM A N D ALUMINIUM ALLOYS
4129
Aluminium has good resistance to petroleum products, and an AI-2Mg alloy is used for tank heating coils in crude-oil carriers. Caked-on deposits must be removed from the coils by hot sea-water cleaning in order to maintain effective heat transfer and prevent corrosion. Aluminium is also used in the petroleum industry for sheathing for towers, heat exchangers, transport and storage tanks and scrubbers. Many industries use aluminium alloys for heat exchangers, clad alloys being used where pitting corrosion is liable to be initiated by one of the contacting materials. Heat exchangers in the gas industry have utilised duplex tubes, with aluminium on the water side and steel on the gas side in cases where aluminium is unsuitable owing to the presence of catechol which can attack it. Aluminium does not become brittle at low temperatures and for this reason (and because of its corrosion resistance) it has been adopted for the carriage and storage of liquefied methane.
High-temperature Corrosion Dry Atmospheres
When exposed at high temperatures in dry atmospheres aluminium is highly resistant to corrosion by most of the common gases, other than the halogens or their compounds. High-temperature Aqueous Systems
When aluminium corrodes at temperatures below 90°Cin aqueous systems, attack is usually by pitting. At temperatures between 90 and 250°C (for the attainment of which considerable pressures are needed) uniform attack is the commonest form of aqueous corrosion. Above about 250"C, uniform attack is merely the prelude to highly destructive intergranular attack. The corrosion products from the uniform attack form a film which includes a barrier layer and a bulk film analogous to those formed during anodising (Section 15.1); it is the bulk film which controls the corrosion rate, which is not significantly affected by most common dissolved ions3'. The onset of intergranular attack occurs at about the same time as the crystallisation of the amorphous barrier layer oxide. Kinetic studies indicate that over the temperature range from 100 to 363°C the oxidation rate law is successively inverse logarithmic, parabolic and linear '*. The requirements of nuclear energy application fostered an interest in special alloys for service in high temperature aqueous environments, but their utilisation has not been widespread. Encouraging results have been reported for alloys of 2Ni-0.5Fe33 and 1 -2Ni-1-8Fe34. It has been suggested that the role of nickel (as NiAI,) is to provide sites of low hydrogen overvoltage, where cathodically liberated hydrogen may be liberated without disrupting the protective oxide35.The distribution of such sites is apparently critical however, since high corrosion resistance is associated with a fine dispersion of the second phase, while the electronic conductivity of the film is probably also i m p ~ r t a n t ' ~ .
4:30
ALUMINIUM A N D ALUMINIUM ALLOYS
Steam forms a protective white film at temperatures up to about 250"C, but above this temperature steam can, under some conditions, react with aluminium progressively to form aluminium oxide and hydrogen. Sintered aluminium powder (S.A.P.) has relatively good resistance to steam at 500"C, but at about 300°C an addition of 1% nickel to the S.A.P. is needed to prevent rapid disintegration. Molten Salts and Metals
Aluminium-base alloys resist the action of many molten salts which are nearly neutral in reaction. Molten sodium nitrate or mixtures of sodium nitrate and potassium nitrate are used for salt bath heat treatment of some aluminium alloys. Molten metals generally attack aluminium alloys. Both zinc and tin form alloys by dissolution of aluminium, although the aluminium does not melt. Molten lead is inert to aluminium, and molten lead baths can be used for heating aluminium alloys. Mercury, molten at room temperature, amalgamates readily with aluminium alloys if their naturally formed oxide films are temporarily removed by scratching, and rapid corrosion occurs on subsequent exposure to moist air or water. Under stressed conditions cracking will frequently result, since mercury penetrates into the aluminium alloy selectively at grain boundaries. Contact of aluminium with mercury is extremely dangerous and severe corrosion can occur with a very small amount of mercury.
Aluminium in Contact with other Materials The aluminium alloys recommended for building purposes (not including the high strength alloys containing copper) have good resistance to concretes, mortars, plasters and asbestos cement products. When freshly mixed, some of these materials release traces of alkaline products which may be sufficient to stain aluminium or to etch it slightly. As soon as the mixture is set, however, the attack ceases and even after many years' service, attack on embedded aluminium is found to be negligible 38. With cement and concrete under continually wet conditions, there may be some surface attack. This decreases rapidly with time, and the strength of components is not significantly affected. Under embedment conditions, bituminous protection is advisable, to avoid risk of cracking of the concrete due to stresses set up owing to the bulk of the corrosion product. Plasters are generally even less aggressive than Portland cement. In damp environments, some corrosion of aluminium may arise in contact with the more open-grained building stones and brickwork, but the hard stones, such as granite, are inert. With building stone and brickwork, as with soils (Section 9.3), it is the nature of the products which can be leached out which will determine whether aluminium corrodes. Unprotected aluminium, in the form of nails for example, can be satisfactorily used in contact with precast concrete blocks, which are usually non-corrosive to aluminium. Magnesium oxy-chloride compositions (used for flooring), on the other hand, stimulate corrosion of aluminium under moist conditions, as will many insulation materials based on magnesium and calcium silicates. 379
ALUMINIUM A N D ALUMINIUM ALLOYS
4:31
Plastics are generally without action on aluminium and are widely used to provide insulation between other metals and aluminium, while the use of aluminium/plastic laminates is increasing. Rubber has no effect upon aluminium. A few acid woods, such as oak, chestnut and Western red cedar, accelerate surface weathering of aluminium, but do not usually give rise to serious attack 39. Timber preservatives containing soluble copper compounds should be avoided; creosote and zinc napthenate are satisfactory preservatives for wood in contact with aluminium. Common packaging materials are a potential source of aggressive substancea, and careful selection is recommended to avoid surface deterioration. Where paper is in contact with aluminium, the chloride content should be below 0.05%, sulphate content below 0.25%, copper content below 0.01% and the pH of aqueous extracts in the range pH 5.5-7.5, in order to avoid corrosion in damp conditions. Papers and felts used in building applications should also conform to this specification as a minimum requirement and be of the highest quality, since metallic copper found in materials of inferior origin can result in severe local galvanic attack of aluminium. Tarpaulins are sometimes treated with copper-containing preservatives and water leached from these has been found to cause corrosion of underlying aluminium sheets. Fibreglass insulation produced from soda glass can cause pitting in conditions where leaching of alkali occurs, for example, by condensation: the use of Fibreglass produced from Pyrexglass is therefore preferred. Common putties of whiting/linseed oil composition d o not attack aluminium; adhesion is obtained by allowing the metal surface to weather, or by applying an etch primer treatment to the metal. Both thermosetting adhesives (e.g. the phenolic types) and thermoplastic adhesives (such as paraffin and microcrystalline waxes, or bitumen) are non-corrosive to aluminium. In general, adhesives applied to aluminium should not contain chlorides in excess of 0.05% (as NaCI) of the solid content, and should be free from copper- or mercury- containing anti-fungicides. The presence in the adhesive of borax or sodium silicate is beneficial when one of the adhesive components is of an acid character.
Recent Literature Survey The corrosion resistance of aluminium in a variety of media has been reported. It has been observed that mono-chloroacetic acid has no corrosive effect and di-chloroacetic acid has negligible effects up to 5 m but tri-chloroacetic acid produces a vigorous reaction4'. The effects of some transition metal and heavy metal cation on dissolution of aluminium in neutral and acid chloride containing solutions has been reported by Khedr and Lashien4* while the corrosion rate of aluminum conductors in integrated circuits has received the attention of Lerner and Eldredge43. An experimental pHpotential diagram for aluminium in seawater is available". The environmental chemistry factors affecting surface film destruction have received some and the energy transfer in aluminium dissolution is represented by a potential energy surface diagram 46. The cathodic protection of aluminium in seawater in considered by Gundersen and Nisancioglu 47,48.
4:32
ALUMINIUM A N D ALUMINIUM ALLOYS
The sources of characteristic emission generated in aluminium alloys exposed to various environments is presented by A r ~ r while a ~ ~Graver and Wiedmer have undertaken an electrochemical investigation into AI-rich intermetallics so, Larsen-Basse has reported on the corrosion of aluminium alloys in ocean thermal energy conversion seawaters5', Ahmed on corrosion and its prevention in de-salination plants 52, Lashermes on marine environment effectss3,similarly Huppartz and Krajewskis4, Kunze on the corrosivity of various foodstuffs to aluminium packaging materials ", Rogozhina etal. in the corrosion of a range of aluminium alloys used in agricultural enclosuress6, while reactions with nitric acid have been covered by Singh et and Horn'". Singh has shown that surface roughness has an important effect on the well-water staining of aluminium and its alloyss9. The addition of magnesium to A-Li-Cu-Zr-Ge alloy results in a more consistent corrosion behaviour 60. Overageing increases the exfoliation corrosion resistance as well as the resistance to electrochemical corrosion. On the other hand, the addition of germanium to AI-Li results in the underaged alloy being more stable in terms of breakdown and repassivation compared to pure aluminium or the binary alloy6' although ageing can produce a grain boundary structure and associated precipitation effects that reduce the corrosion resistance62. Mauret and Lacaze studied the water corrosion of AI-Mg and AI-Cu-Mg using gas chromatography of hydrogen63,Huppartz and Wieser report on the electrochemical behaviour of AI-Mg-Mn and AlMg-Si in seawater and hard water with the relevant E-Z diagramsu and Shirkhanzadeh and Thompson provide information concerning the corrosion of AI-Ga in alkaline solutions6s. As seen above Moran et al., has commented upon the exfoliation corrosion of Al-Li-Cu-Zr-Ge@'. The mechanism has been investigated by Reboul and BouvaistM and a mathematical model suggested by Robinson67. The influence of alloy elements in Al-Zn-Mg has been reported by Rebou168and it has been shown that exfoliation corrosion of AI-Mg-Si in irrigation water is also governed by alloy and impurity c o n ~ e n t r a t i o n ~ ~ . A number of studies have been undertaken concerning the suitability of various inhibitors including p-quinone and acetic acid7', the former being of little value in the case of AA1060 alloy in aqueous potassium nitrate, carboxyliacids 71, Diamines72, Complexons such as zinc phoshate^^, oxalates 74, and morpholine and thiosemi-carbozide derivatives 75. An XPS investigation of dichsomate and molybdate in chloride ioncontaining solutions showed that, under the conditions used, chromium exists on the surface primarily as Cr(1II) whereas molybdenum exists as Mo(1V) and M o ( V I ) ~ ~Zanzuichi . and Thomas report on the use of inhibitor for aluminium films in integrated Pitting corrosion always remains a worthy subject for study, particularly with reference to mechanism, and the problem conveniently divides into aspects of initiation and growth. For 6061 alloy in synthetic seawater, given sufficient time, pit initiation and growth will occur at potentials at or slightly above the repassivition potential7". In an electrochemical study, it was found that chloride ions attack the passive layer as a chemical reaction partner so that the initiation process becomes one of cooperative chemical and electrochemical A focused laser beam was used by Alkine and FeldmanEO to create local
ALUMINIUM A N D ALUMINIUM ALLOYS
4:33
depassivation thus providing a novel and specific approach to initiation of found that pitting resisvalue in fundamental studies; yet Bonora et tance may be improved if the entire surface area be irradiated with such a beam; it was argued that a chemically inert surface was produced by the report on the involvement of preirradiation treatment. Thompson et existing flaws or weak spots in the surface film and Fokin and Koteneus3 describe an ellipsometric study or pit formation and repassivation. Increasing the hydrostatic pressure can increase pitting susceptibility and decrease the passivation range as a result of the decreased thickness and increased number of defects in the oxide layers4. Hunkeler and Bohnis5 quantitatively examined pit growth as a function of time, potential and electrolyte conductivity. They considered that growth occurred as a result of a primary change in the properties of the surface area of the pit caused by adsorption of chloride ions while growth is ohmically controlled which, under ideal conditions, results in a square root growth laws6. Alwitt e t ~ l . ' ~ found what they consider is a unique form of pitting corrosion during the anodic dissolution of aluminium in hot chloride ion containing solution. A high density of fine etch tunnels were produced extending along the (100) directions and evolving from cubic etch pits when all but one wall of a pit becomes passivated; dissolution rate is high from this active pit surface. Computer simulation of etch pit morphology provided good agreement with experiment for Idemoto and K o ~ r a morphology ~~, in chloride ion and nitrate ion containing solution also having been investigated by Klinger and Fellers9 while both Mansfield et a1.90and Sharland and Tasker9'v9' have been involved with mathematical modelling of pitting corrosion. Furthermore reports are available concerning the pitting of aluminium foils93, the effect of molybdenum 94, rapid solidification p r o c e ~ s i n g ~both ~ , of the latter being beneficial, alternating current%, brazing9' and in Al-Zn-Mg-Cu alloy98and 5083/6061 alloys w. Hitzig et al. have produced a simplified model of the aluminium oxide layer(s) to explain impedance data of specimens prepared under different layer formation and sealing conditions loo. The model also gives consideration to the formation of active and passive pits in the oxide layer. Shaw et al. have shown that it is possible to electrochemically incorporate molybdenum into the passive film which, as previously noted"', improves the pitting resistance. Interest has been aroused in connection with the formation of electrochemical films on aluminium covered with a thermally grown film Both the thermally and the anodically grown film are amorphous normally but growing ananodic film on top of a thermal film results in the anodic film being crystalline. Less charge seems to be required compared with the anodising of a 'clean' aluminium surface and so presumably the crystallised film can withstand a higher electric field than the amorphous film. Fundamental studies elucidating the growth and properties of barrier-type films have been reported by Skeldon et al.'", Csanady et a1.Io7,Ebihara et al.lo8, Fukuda and FukushimaIw, Menezes et al."o,Wittberg et al."', and Thompson et al."'. Strazzi has reviewed methods of sealing oxide films and Omata et al. find that adhesion of paint films to anodised layers depends on penetration
4134
ALUMINIUM A N D ALUMINIUM ALLOYS
of the paint into the micro-pores of the anodic layer'I4. Faller has made a comparison of a number of anodising processes and process parameters after weathering the anodised specimens for five years in industrial and marine atmospheres 'IJ. Elevated temperature oxidation behaviour of Al-Mg-Li and AI-Zn-Mg 'I7 have also been reported. Film dissolution in acetic acid-acetate buffers has been investigated by Valand using the potential step method"* in neutral and acid solutions using high potential cathodic polarization 'I9 and dissolution in KF solution has been shown to follow an empirical relation incorporating a film thickness parameter I*'. A change in dissolution rate occurred indicating a duplex oxide with the inner layer dissolving more easily than the outer layer. The volume of reported work concerning the environmentally assisted cracking of aluminium alloys, particularly the Al-Zn-Mg type, is quite phenomenal and cannot adequately be reviewed in this general update. Surface reactions and their relation to environmentally assisted cracking of AI-Mg has been reported by FordI2' and Pathania and Trumaris'22while Lee and Pyum 123 have undertaken an electrochemical study of Al-Cu-Mg showing that its SCC rate is affected by prior metallurgical history. Dietzer e t ~ l . "determined ~ Kiscc of 2024 in 3 . 5 % sodium chloride solution finding that the value was little affected by the three different loading methods used in their study. With the AI-Li-Cu alloy fatigue bonded in 3 . 5 % sodium chloride solution, pitting was found to be important to crack initiation but this was dependent on the strain rate range involved'". For the Al-Li-CuMg'26-8Ki,, decreased with increasing ageing time, no doubt a result of the precipitation of S-(AI-Li) at high angle boundaries and associated PFZ (precipitate-free-zone) formation. The AI-Zn-Mg alloy'22v 129-142 studies have been mainly concerned with Kiscc determination for different bonding modes, environmental conditions and sample metallurgical history. Failure is very much a function of grain size, grain boundary precipitation and formation of PFZs; thus, work that attempts to improve the situation by alloy/microstructure modification is prominent. This involves compositional changes and heat treatment designed to affect segregation phenomena and precipitate type, morphology and location as a result of ageing. There are two generally accepted mechanisms for SCC; film rupture with anodic dissolution and hydrogen-assisted cracking. Which of these occurs appears to be dependent on environmental chemistry and the potential of the alloy in the environment 1 3 1 * 1 4 0 although bonding mode can also be important 13'. A number of studies have addressed the role of hydrogen in SCC of aluminium alloys '*'* 1 3 ' , 139. 143, 145. Although there is still no general consensus of opinion, it does seem that hydrogen affects the plastic deformation properties of the aluminium matrix in the crack tip zones. The related alloy system, Al-Zn-Mg-Cn is also well documented 146-152. Overageing is reported to be beneficial since modification of the grain boundary precipitate aspect ratio occurs 148. Bucci1I3has produced a useful and extensive report of value for the selection of suitable aluminium alloys to resist both SCC and corrosion fatigue while Khobaib discusses a range of beneficial inhibitors suitable under conditions of corrosion fatigue.
ALUMINIUM AND ALUMINIUM ALLOYS
4:35
Finally, reports are available on the durability of adhesively bonded aluminium joints 155- 156. J . C. BAILEY F. C. PORTER A. W. PEARSON R.A. JARMAN REFERENCES 1 . The Properties of Aluminium and its Alloys, 6th edn, The Aluminium Federation, Bir-
2. 3. 4. 5. 6. 7. 8.
mingham (1968) Blewett, R. V. and Skerry, E. W., Melallurgia, 71, 73 (1965) Keir, D. S., Pryor, M. J. and Sperrey, P. R., J. Electrochem. Soc., 114, 777 (1967) Godard, H. P., J. Electrochem. Soc., 114, 354 (1967) Aylmore, D. W., Gregg, S. J. and Jepson, W. B., J. Inst. Met., 88, 205 (1959-60) Bartlett, R. W., J. Electrochem. Soc., 111, 903 (1964) Aziz, P. M. and Godard, H. P., Corrosion, 12, 495t (1956) Hunter, M. S., Frank, G. R. and Robinson, D. L., 2nd International Congress on Metallic Corrosion, 66 (1963)
9. Champion, F. A., J. Inst. Met., 83, 385 (1954-55) 10. Barton, J. F., Paint Manufacture, Nov., 53 (1%) and Dec., 47 (1964) 1 I . Bell, W. A. and Campbell, H. S., J. Insl. Met., 89, 464 (1960-61) 12. Whitaker, M. E., Metal Ind., 80, 183, 207, 227, 247, 263, 288, 303, 331, 346, 387 (1952) 13. Chadwick, R., Muir, N. B. and Grainger, H. B., J. Inst. Met., 85, 161 (1956-57) 14. Bushy, J . , Cleave, J. F. and Cudd, R. L., J. Inst. Met., 99, 41 (1971) 15. Aziz, P . M. and Godard, H. P., Corrosion, 15, 529t (1959) 16. Symposium on Atmospheric Corrosion of Non-Ferrous Metals, Amer. SOC.Test. Mat ., Special Technical Publication No. 175 (1956) 17. Metal Corrosion in the Atmosphere, Amer. SOC.Test. Mat., Special Technical Publication No. 435 (1968). Papers by McGeary, et a/., p. 141 and Ailor, J . R., p. 285 18. Symposium on Corrosion by High Purity Water, Corrosion, 13, l 5 l t (1957) 19. Davies, D. E., J. Appl. Chem., 9, 651 (1959) 20. Rowe, L. C. and Walker, M. S., Corrosion, 17, 353t (1961) 21. Porter, F. C. and Hadden, S. E., J. Appl. Chem., 3, 385 (1953) 22. Gilbert, P. T. and Porter, F. C., Iron andSteelInst., Special Report No. 45, 55-74 (1951) 23. Raine, P. A,, Chem. and Ind. (Rev.), 1102, 1196 (1956) 24. Sprowls, D. 0. and Carlisle, M. E., Corrosion, 17, 125t (1961) 25. Day Chemische Verhalten uon Aluminium, Aluminium-Verlag GmbH, Dusseldorf (1955) 26. Aluminium in the Chemical and Food Industries, British Aluminium Co. Ltd., London (1959) 27. Aluminium with Food and Chemicals, Alcan Booth Industries Ltd. (1966) 28. Ritter, F., Korrosionstabelien Metallischer Werkstofle, Springer-Verlag. Vienna (1944) 29. Aluminium with Food and Chemicals, The Aluminium Association, New York (1967) 30. Process Industries Applications of Alcoa Aluminium, Alcoa, Pittsburgh, U.S.A. 31. Troutner, V. H., Corrosion, 15, 9t (1959) 32. Dillon, R. L., Corrosion, 15, 13t (1959) 33. Perryman, E. C. W., J. Inst. Met., 88, 62 (1959) 34. Dillon, R. L. and Bowen, H.C., Corrosion, 18, 406t (1962) 35. Draley, J. E. and Ruther, W. E., Corrosion, 12, 480t (1956) 36. Greenblatt, J . H. and Macmillan, A. F., Corrosion, 19, 146t (1963) 37. Porter, F. C., Metallurgia, 65, 65 (1962) 38. Jones, F. E. and Tarleton, R. D., Effect of Embedding Aluminium and Aluminium Alloys in Building Materials, National Building Studies Research Paper No. 36, H.M.S.O., London 39. Farmer, R. H. and Porter, F. C., Metallurgia, 68, 161 (1963) 40. Scott, D. J. and Skerrey, E. W., Br. Corros. J., 5, 239 (1970) 41. Mansour H. etal., Bull. of Electrochemistry, 2 , 449-451 (1986) 42. Khedr M. G . A. and Lashien, A. M. S., J. Electrochem. Soc., 136, 968-72 (1989)
4:36
ALUMINIUM AND ALUMINIUM ALLOYS
43. Lerner, I. and Eldridge, J. M., ibid., 129, 2270-73 (1982) 44. Gimenez P . etal., Rev. Aluminium, 518, 261-72 (1982) 45. Godard, H. P., Materials Performance, 20, 9-15 (1981) 46. Foley, R. T. and Nyuyen, T. H., J . Electrolem. Soc., 129, 464-7 (1982) 47. Gundersen, R. and Nisancioglu K., Corrosion, 46, 279-85 (1990) 48. Nisancioglu, K., Lunder, 0. and Holtan, H., ibid., 41, 247-57 (1985) 49. Arora, A., ibid., 40, 459-65 (1984) 50. Graver, R. and Wiedmer, E., Wekstofle Korros., 31, 550-5 (1980) 5 1. Larsen-Basse, J., Materials Performance, 23, 16-21 (1984) 52. Ahmed, Z., Anti-Coros. Wleth, Mar.. 28,4-10 (1981) 53. Lashermes, M., Rev. Aluminium, 523, 505-1 1 (1982) 54. Huppatz, W. and Krajewski, H., Weistrofle Korros., 30, 673-84 (1979) 55. Kunze, E., Aluminium, 52, 296-301 (1976) 56. Rogzhina, E. P., Koltunova, G. A., Pashkova, 0. A. and Goluber, A. I., Zashch, Met., 25, 120-24 (1989) 57. Singh, D. D. N., etal.. J. Electrochem. SOC., 129, 1869-74 (1982) 58. Born, E-M., Werkstofle Korros.. 41, 32-3 (1990) 59. Singe, T., Aluminium, 57, 187-9 (1981) 60. Moran, J. P., eta/., Corrosion, 43, 374-82 (1987) 61. Colvin, E. L., Cahen Jr., G. L., Stoner, G. E. and Starke, E. A., Corrosion, 42,416-21 ( 1986) 62. Kumai, C., Kusinski, J., Thomas, G. and Devine, T. M., ibid., 45, 294-302 (1989) 63. Mauret, P. and Laraze, P., Corros Sci., 22, 321-9 (1982) 64. Huppartz, W. and Wieser, D., Werkstofle Korros, 40. 57-62 (1989) 65. Shirkhanzadeh, M. and Thompson, G. E., Electrochim. Acta, 33, 939-40 (1989) 66. Reboul, M. G. and Bouvaist, J., Werksrofle Korros., 30, 700-12 (1979) 67. Robinson, M. J., Corros. Sci., 22, 775-90 (1982) 68. Reboul, M. G. and Bouvaist, J., Rev., Aluminium, 491, 41-55 (1980) 69. Zahavi, J. and Yahalom, J., J. Electrochem. SOC., 129, 1181-5 (1982) 70. Onuchukwa, I. and Oppong. F. W., Corros. Sci., 26, 919-26 (1986) 71. Moussa, M. N. and El-Togoury, M. M., AntiCorros, Meth. Mat., 37, 4-8 (1990) 72. AI-Suhybani, A. A., Corros. Prev. Control, 37, 11-16 (1990) 73. Kuznetsov, Yu. I. and Bardasheva, T. I., Zashch. Met., 24, 234-40 (1988) 74. Wilhelmsen, W. and Grande, A. P.. Electrochim. Acta, 33, 927-32 (1988) 75. Anon., Anti-corros. Merh. Mat., 35,4-8 (1988) 76. Bairamow, A. K., Corros. Sci.. 25,69-73 (1985) 77. Zanzuichi, P. J. and Thomas 111, J. H., J. Electrochem. SOC., 135, 1370-1376 (1988) 78. Aylor, D. M. and Moron, P. J., ibid,, 133. 868-72 (1986) 79. Tomisanyi, L., Varga, K. and Bartik, I., Electrochim. Acta, 34, 855-9 (1989) 80. Alkine, R. and Feldman, M., J. Electrochem. SOC., 135, 1850-51 (1988) 81. Bonora, P. L., etal., Thin Solid Films, Lausanne, 81, 339-45 (1981) 82. Thompson, G. E., et a/., J. Electrochem. Soc.. 129, 1515-17 (1982) 83. Fokin, M. N. and Kotenev, V. A., Zashch. Met., 24, 1 11-4 (1988) 84. Beccaria, A. M. and Poggi, G., Corrosion, 42, 470-75 (1986) 85. Hunkeler, F. and Bohrii, H., Werkstofle Korros., 34, 593-603 (1983) 86. Idem., Corrosion, 40, 534-40 (1984) 87. Alwitt, R. S., etal.. J. Electrochem. SOC., 131, 13-7 (1984) 88. Idemoto, Y. and Koura, N., J. Metal Finishing SOC,Japan, 37, 30-5 (1986) 89. Klinger, R. and Feller, H. G., Aluminium, 57, 224-7 (1981) 90. Mansfield, F., Lin, S., Khim, S. and Shih, H., J. Electrochem. Soc., 137,78-82 (1990) 91. Sharland, S. M. and Tasker, D. W., Corros. Sci., 28, 603-20 (1988) 92. Sharland, S. M., ibid.. 28, 621-30 (1988) 93. Aylor, D. M. and Moran, P. J., J. Electrochem. Soc.. 133, 949-51 (1986) 94. Mosher, W. C., eta/., ibid., 133, 1063-4 (1986) 95. Yoshioku, H., eta/., Corros. Sci., 26, 795-812 (1986) 96. Vu Quang, K., etal., J. Electrochem. SOC., 130, 1248-52 (1983) 97. Hattori, T. and Sakamota, A., Welding J., 61, 3395-425 (1982) 98. Maitra, S. and English, G. C., Met. Trans., 13A, 161-6 (1982) 99. El-Boujclaini, M., Ghali, E. and Galibois, A., J. Appl. Electrochem., 18,257-64 (1988) 100. Hitzig, J., etal., J. Electrochem. SOC., 133, 887-92 (1986) 101. Shaw, B. A., Davis, G. D., Fritz, T. L. and Olver, K. A., ibid., 137, 359-60 (1990)
ALUMINIUM AND ALUMINIUM ALLOYS
4:31
102. Kobayashi, K., etal., ibid, 133, 140-1 (1986) 103. Crevecour, C. and de Wit, H. J., ibid.. 134, 808-16 (1987) 104. Partridge, P. G. and Chadbourne, N. C., J. Mater. Sci., 24, 2765-74 (1988) 105. Skeldon, P., et a/., Thin and OIid Films, 123, 127-133 (1985) 106. Xu, Y., etal., Corros Sci., 27, 83-102 (1987) 107. Csanady, A., elal., Corros Sci., 24, 237-248 (1984) 108. Ebinhara, K., et al., J. Metal Finishing Soc., Japan, 33, 156-64 (1982) 109. Fukuda, Y. and Fukushima, T., Electrochim. Acta., 28,47-56 (1983) 110. Menezes, S., Haak, R., Hagen, G., Kendig, M., J. Electrochem SOC.,136, 1884-6 (1989) 11 I . Wittberg, T. N., Wolf, J. D. and Wang, P. S., J. Mater, Sci., 23, 1745-7 (1988) 112. Thompson, G. E., et a/., Trans. Inst. Metal Finishing, 58, 21-5 (1980) 113. Strazzi, E., Alluminio, 50, 4%-9, 520-5 (1981) 114. Omata, K., etal., Aluminium, 57, 811-3 (1981) 115. Faller, F. E., ibid., 58, E23-5 (1982) 116. Csanady, C. and Kurthy, J., Mat. Sci., 16,2919-22 (1981) 117. anon., Corros. Sci., 22, 689-703 (1982) 118. Valard, T., Electrochim. Acta, 25, 287-92 (1980) 119. Cabot, P . L., et al., Corros. Sci., 26, 357-9 (1986) 120. Abou-Romia, M. M. and El-Basiouny, M. S., Corrosion, 42, 324-8 (1986) 121. Ford, P., et al., J. Electrochem. SOC., 127, 1325-31 (1980) 122. Pathania, R. S. and Trumans, D., Met. Trns. 12A, 607-12 (1981) 123. Lee, K. W. and Pyum, S. I., Metell., 36, 280-3 (1982) 124. Dietzel, W. D., Schwalbe, K. H. and Wu, D., Fatigue Fract. Eng. Mater. Struct., 12.. 495-510 (1989) 125. Rebiere, M. and Magnin, T., Mater. Sci. Eng., A128, 99-106 (1990) 126. Dorward, R. C. and Hasse, K. R., Corrosion, 43, 408-13 (1987) 127. Dorward, R. C., ibid., 46, 348-52 (1990) 128. Ahmad, M., Mater. Sci. Eng. A125, 1-14 (1990) 129. Lunarska, E. and Szklarska-Smialowski, Z., Corrosion, 43, 414-24 (1987) 130. Trumans, D., ibid., 42, 601-8 (1986) 131. Kim, Y. S. and Pyum, S. I., Brit. Corros. J . , 18, 71-5 (1983) 132. Mankowski, G. and Dabosi, F., Corrosion, 40,552-8 (1984) 133. Mudlee, M. P., Thompson, A. W. and Bernstein, 1. M., ibid., 41, 127-36 (1985) 134. Rajan, K., et ai., J . Mat. Sci., 17, 2817-24 (1982) 135. Richter, J. and Kaesche, H., Werkstofle Korros.. 32, 289-95 (1981) 136. Scamans, G. M., Aluminium, 57, 268-74 (1981) 137. Holroyde, N. J. H. and Hardie, D., Met. Technol., 9, 229-34 (1982) 138. Rahman, M. S., et al., Z . Metallkunde, 73, 589-93 (1982) 139. Christocloulou, L. and Flowers, H. M., Acta Metall., 28, 481-7 (1980) 140. Lotto, C. A. and Cottis, R. A., Corrosion, 45, 136-41 (1989) 141. Ratke, L., Z. Mefallkd., 81, 144-8 (1990) 142. Onoro, J., Moreno, A. and Ranninger, C., N . Mater. Sci., 24, 3888-91 (1989) 143. Bond, G. M., Robertson, I. M. and Birnbaum, H. K., Acta Metell., 36, 2193-7 (1988) 144. Zeides, F., Mater. Sci. Eng., A125, 2-30 (1990) 145. Watson, J . W., Shen, Y. 2. and Meshi, M., Met Trans. 19A, 2299-304 (1988) 146. Hasse, K. R. and Dorward, R. C., Corrosion, 41, 663-9 (1986) 147. Hermann, J., J. Mal. Sci., 16, 2381-6 (1981) 148. Narasimha Rao, B. V., Met Trans., 12A, 1356-9 (1981) 149. Cordier, H., etal., Meroll., 36, 33-40 (1982) 150. Dorward, R. S. and Hasse, K. R., Corros. Sci., 22, 251-7 (1982) 151. Swanson, R. E., etal., Scripta Metell., 16, 321-3 (1982) 152. Sarker, B. e t a / . , Met Trans., 12A, 1939-43 (1981) 153. Bucci, R. J., Eng. Fruct. Mechanics, 12,407-44 (1979) 154. Khobaib, M., Lynch, C. T. and Vahlcliek, F. W., Corrosion, 37, 285-92 (1981) 155. Minford, J. D., Int. J. Adhesion Adhesives, 2, 25-8 (1982) 156. Cotter, J. C. and Kohler, 61.. ibid., 1, 23-8 (1980)
4.2
Copper and Copper Alloys
Copper and copper alloys are amongst the earliest metals known to man, having been used from prehistoric times, and their present-day importance is greater than ever before. Their widespread use depends on a combination of good corrosion resistance in a variety of environments, excellent workability, high thermal and electrical conductivities, and attractive mechanical properties at low, normal and moderately elevated temperatures. A wide range of cast and wrought alloys is available. For detailed expositions of properties and uses the reader is referred to publications on many specialised aspects obtainable from the Copper Development Association offices in various countries. Relevant publications of the British Standards Institution include BS 1400, Copper Alloy Ingots and Castings‘ and BS 2870-5, Copper and Copper Alloy Wrought Products2. All standards of the American Society for Testing and Materials relating to copper and copper alloys are included in a volume published annually3.
Composition and Properties The mechanical properties of wrought alloys4 depend on composition and metallurgical condition. At the extremes, annealed pure copper has a tensile strength of 180 MN m-’ and a hardness of 40 Hv, and heat-treated beryllium copper can have a tensile strength of 1 300 MN m-2 and a hardness of 390Hv. Summaries of typical properties of some of the more important wrought and cast copper alloys are given in Tables 4.9 and 4.10. Coppers The purest grade of copper commercially available, and that with the highest electrical conductivity, is oxygen-free high-conductivity copper. The minimum copper content required by some specifications is 99.99%, and the method of manufacture is such that no residual deoxidant is present. Oxygen itself has very little effect on conductivity, and the ‘tough pitch’ coppers (either electrolytic or fire-refined), containing about 0.04% oxygen, are high-conductivity materials. One disadvantage of tough pitch coppers is the embrittlement that is liable to occur when they are heated in atmospheres containing hydrogen. For many purposes, therefore, and particularly where fabrication is involved, deoxidised coppers are preferred. The usual deoxidising agent is phosphorus, and specifications require residual phosphorus contents of between 4:38
Table 4.9
( d c m '1
Melting p t . (liquidus, a C)
Coefficient of expansion x 106
Electrical conductivity 070 I.A.C.S.
Thermal conductivity (W/m"K)
8.94
1083
18
103
390
8.93 8.93 8-93 8.2 8.74 8.53 8.38 8.33 8.41 8.35 8-69 8.87 8.89 8-70 8.52 7.95 8-91 8.94
I 682 1 080 1075 955 1025 955 905 980 890 890 1060 1190 1 050 995 1 030 I 050
18
80 45
340 175 360 85
Demity Alloy
H.C. copper Deoxidised nonarsenical copper Arsenical copper Tellurium copper Beryllium copper 85/ 15 brass 70/30 brass 60/40brass Aluminium-brass Naval brass H.T. brass 15% nickel silver 30% nickel silver 59'0 tin-bronze 12% tin-bronze Silicon-bronze 7% aluminium-bronze 90/10 cupro-nickel 70/30 cupro-nickel
Typical properties of wrought alloys
I150 1 240
17 18 18 19 20 21 19 21 21 16 17 18 19 18 18 16 16
96 23 35 27 29 23 25 23 7 5 17 8 8 15 10 5
155
125 I25 100 110 105
35 20 80 50
40 80 50 30
Tensile strength (MN/m *)
To
hardness
180-340
10-60
40-1 10
180-340 220-360 230-320 500-1 300 280-540 280-600 370-600 320-700 370-620 520-770 350-700 390-700 340-740 460-830 260-630 430-770 3 10-620 370-700
10-55 10-55
40-120
10-so 2-40 8-70 5-75 5-45 6-75 5-45 8-35 4-55 4-50
5-70 5-65 5-75 4-65 8-55 5-55
40- 125 40-110 110-390 65-170 55-180 75-180 70-200 75- 180 90-220 70-220 90-220 70-220 110-250 60-200 80-2 10 90-200 95-210
4:40
COPPER AND COPPER ALLOYS Table 4.10
Properties of cast alloys
Alloy
1OSn-O.2P 10Sn-1OPb 10Sn-2Zn 5Sn-5Zn-5Pb 9.5A1-2Fe Silicon-bronze 30Zn-2Pb Naval brass H.T. brass (up to 2.5AI) H.T. brass (up to 5.OAl) H.T. p brass
Min. tensile strength (MN/rn2)
Min. elongation
230
7
(TO)
190
5
260 200 490 310 I90
15
310 460
590 740
I5 20 15
12 20 20 15 12
0.004 and 0 05%. Phosphorus-deoxidised coppers commonly have electrical conductivities about 80% of those of pure copper. Arsenical coppers containing about 0.4% arsenic (tough pitch or deoxidised) are used where increased strength at elevated temperatures is required. Additions of cadmium (1 .O%), chromium (05%), and silver (0-1To) also give improved high-temperature properties, but without any gives improved serious loss of electrical conductivity. Tellurium (1 machinability. An addition of about 2% beryllium gives a heat-treatable alloy that can develop extremely high strength. e o % )
Brasses Brasses are basically alloys of copper and zinc, containing between about 10 and 45% Zn, but many other additions are made and the resulting alloys are the most complicated of all the copper-base series. The singlebrasses, containing up to about 37% Zn in the binary alloys, may phase (CY) have additions of 1070 Sn (Admiralty brass), 2% AI (aluminium-brass), or 1-2070 P b for ease of machining. Duplex (a-p) brasses containing more than 37% Zn, may have additions of 1% Sn (Naval brass), or 1-3% P b to assist machining. Both CY and a-/3brasses, with and without lead, are used in the cast as well as the wrought form. (or, occasionally, /3) alloys containing up to High-tensile brasses are CY-@ 5% AI and 1-2'70 of one or more of the following: Sn, Pb, Fe, Mn. These alloys also are used in both wrought and cast form. Copper/nickel alloys AIIoys containing 5-30% Ni, used mostly in the wrought condition', have a very good combination of properties. For optimum corrosion resistance, additions of 0.5-2.0Vo each of Fe and Mn are made. Tin-bronzes and gunmetals Alloys containing 3.0- 12.5 070 Sn and 0-020.04% P, known as phosphor-bronzes, are widely employed. Cast as well as wrought alloys are used, and cast leaded bronzes are also available. Gunmetals are alloys of copper, tin and zinc, with or without lead, used in the cast condition. Commonly used alloys are (a) 10Sn-2Zn, and (b) 5Sn-5Zn-SPb. Aluminium-bronzes Aluminium-bronzes usually contain 5- 10% AI, the structure being duplex when more than about 8% AI is present. Plain Cu-AI
COPPER A N D COPPER ALLOYS.
4:41
alloys are sometimes used, but wrought (single-phase) alloys may have additions of about 0.25-2% of one or more of the following: Ni, Fe, Mn, Ag, Sn, As. Cast alloys of high strength and complex structure usually contain about 10% AI and additions of Fe, Mn and Ni.
Silicon-bronzes Silicon-bronzes usually contain 1 5-3'7'0 Si and 0.5-1 To Mn6. They are used in wrought or cast form, though the cast alloys may also contain some Zn and Fe. Nickel silvers These wrought alloys consist essentially of Cu, Zn and Ni, with Ni in the range 10-30%. Leaded nickel brasses are also used, usually where some machining is involved.
General Considerations of Corrosion Behaviour Copper is the first member of Group IB of the periodic table, having atomic number 29 and electronic configuration 2.8.18.1. Loss of the outermost electron gives the cuprous ion C u + , and a second electron may be lost in the formation of the cupric ion C u 2 + . Copper occurs in the uncombined state in nature and is relatively easily obtained by the reduction of its compounds. It is not very active chemically and oxidises only very slowly in air at ordinary temperatures. In the electrochemical series of elements, copper is near the noble end and will not normally displace hydrogen, even from acid solutions. Indeed, if hydrogen is bubbled through a solution of copper salts, copper is slowly deposited (more rapidly if the process is carried out under pressure). (See Section 1.2 for thermodynamic considerations.) As copper is not an inherently reactive element, it is not surprising that the rate of corrosion, even if unhindered by films of insoluble corrosion products, is usually low. Nevertheless, although the breakdown of a protective oxide film on copper is not likely to lead to such rapid attack as with a more reactive metal such as, say, aluminium, in practice the good behaviour of copper (and more particularly of some of its alloys) often depends to a considerable extent on the maintenance of a protective film of oxide or other insoluble corrosion product. Many of the alloys of copper are more resistant to corrosion than is copper itself, owing to the incorporation either of relatively corrosion-resistant metals such as nickel or tin, or of metals such as aluminium or beryllium that would be expected to assist in the formation of protective oxide films. Several of the copper alloys are liable to undergo a selective type of corrosion in certain circumstances, the most notable example being the detincification of brasses. Some alloys again are liable to suffer stress corrosion by the combined effects of internal or applied stresses and the corrosive effects of certain specific environments. The most widely known example of this is the season cracking of brasses. In general brasses are the least corrosion-resistant of the commonly used copper-base alloys. The various grades of copper available do not differ to any marked extent in their corrosion resistance, and a choice is usually based on other grounds. Subsequent references to the corrosion behaviour of copper may therefore be taken to apply broadly to all types of copper.
4:42
COPPER A N D COPPER ALLOYS
The choice of alloy for any particular application is determined by the desired physical, mechanical and metallurgical properties. Within these limits, however, a range of materials is usually available. It is essential that at the very earliest stage the choice of materials and the details of design of the installation should be considered from the point of view of corrosion, if the best performance is to be obtained in service. This is particularly true of copper alloys, where protective measures are not normally applied. Several books contain general summaries of the corrosion behaviour of copper and its and the formation of copper corrosion products and methods for their identification have been described in a number of papers 16. Electrode Potential Rehtionships
The standard potentials for the equilibria Cu2++ 2 e e Cu Cu++e=Cu C u 2 ++ e 2 C u t
+
+
. . ,(4.1)
. . .(4.2) . . .(4.3)
+
0-34V, 0.52 V and 0.17 V respectively, based on values in the are book by W. M. Latimer". For the equilibrium
+
2cu+ e c u 2 + c u K = acu2+/(a$u+)2
. . .(4.4)
K has the value of about 1 x lo6 at 298 K, and in solutions of copper ions in equilibrium with metallic copper, cupric ions therefore greatly predominate (except in very dilute solutions) over cuprous ions. Cupric ions are therefore normally stable and become unstable only when the cuprous ion concentration is very low. A very low concentration of cuprous ions may be produced, in the presence of a suitable anion, by the formation of either an insoluble cuprous salt or a very stable complex cuprous ion. Cuprous salts can therefore exist in contact with water only if they are very sparingly soluble (e.g. cuprous chloride) or are combined in a complex, e.g. [Cu(CN),]-, [Cu(NH,),]+ . Cuprous sulphate can be prepared in non-aqueous conditions, but because it is not sparingly soluble in water it is immediately decomposed by water to copper and cupric sulphate. The equilibrium between copper and cuprous and cupric ions is disturbed by the presence of oxygen in solution, since the reaction shown in equation 4.3 is facilitated, the oxygen acting as an electron acceptor. Behaviour of Copper Electrodes
The electrode potential behaviour of copper in various solutions has been investigated and discussed in considerable detail by Gatty and Spooner I". According to these workers a large part of the surface of copper electrodes in aerated aqueous solutions is normally covered with a film of cuprous oxide and the electrode potential is usually close to the potential of these filmcovered areas. The filmed metal simulates a reversible oxygen electrode at
COPPER AND COPPER ALLOYS
4:43
the existing oxygen concentration and pH, less an overvoltage determined by the existing current density. The principal factors that affect the electrode potential are thus the nature of the solution and the way in which this influences the area of oxide film, and the supply of oxygen to the metal surfaces. In solutions containing chloride there is a tendency for the establishment of the Cu/CuCI/Cl- electrode potential, so that the activity of chloride ions is an important factor in determining the electrode behaviour. From a knowledge of the solubility products of cuprous chloride and cuprous oxide it is possible to predict under what conditions chloride or hydroxyl ions are the potential-determining ions. According to Gatty and Spooner, chloride determines the potential if uoH- < lo-*'' x acl- and hydroxyl if aoH-> lo-*''x acl-. This will not hold in concentrated solutions, however, since complex [CuCI,]- ions as well as simple ions will be present. A further factor to be considered is the ready formation of insoluble basic compounds. In solutions not containing chloride (e.g. sulphate or nitrate solutions), corrosion rates are usually lower and the electrode potential is more steady over
Fig. 4.10 Potential-pH equilibrium for the system copper-water at 25°C (courtesy M . J . N. Pourbaix o f Centre Beige d'Etude de la Corrosion. after Delhez, R . , Depommier, C. and van Muylder, J . , Report RT 100, July (1962))
4:44
COPPER AND COPPER ALLOYS
a wide range of conditions. In this case Gatty and Spooner consider that the rate of corrosion is probably determined by the rate at which metal ions can escape through pores in the protective oxide film, and this is supported by the results of experiments on the anodic and cathodic polarisation of copper. The potential-pH equilibrium diagrams devised by Pourbaix are considered in detail in Section 1.4, and one of the diagrams for the Cu-H,O system is shown in Fig. 4.10. Such diagrams are of considerable assistance in discussing many problems related to the chemistry, electrochemistry, electrodeposition and corrosion of copper. It is well recognised, of course, that the thermodynamic approach has limitations, the most important of which is that though predictions can be made about the possibility of a given reaction proceeding in certain circumstances no information can be gained about the rate at which it will proceed. A method of representing the behaviour of copper in dilute aqueous solutions by means of corrosion-current/pH diagrams has been given by RubiniC and MarkoviC”. A study of the behaviour of copper when anodically polarised has been made by Hickling and Taylor2’, using an oscillographic method that records the variation of potential with quantity of electricity passed. In alkaline solutions the main stages of polarisation were (a) the charging of a double layer, and (b) the formation of a film of cuprous oxide which was almost at once oxidised to cupric oxide. In 0.1 N NaOH the film was about four molecules thick when oxygen evolution first commenced. In buffer solutions of decreasing pH, the formation of sparingly soluble salts preceded or accompanied the formation of the oxide film and in acid solutions giving soluble copper salts no passivity developed, the anodic process being merely dissolution of copper. Other workers have also studied the anodic behaviour of copper or copper alloys in alkalinez3and in acid2, solutions. I9s2O
Atmospheric Corrosion Copper has a high degree of resistance to atmospheric corrosion, and is widely used for roofing sheets, flashings, gutters and conductor wires, as well as for statues and plaques. The resistance of copper and its alloys is due to the development of protective layers of corrosion products, which reduce the rate of attack. The formation, in the course of time, of the typical green ‘patina’ gives copper roofs a pleasing appearance; indeed efforts have been made to produce it artificially or to accelerate its f ~ r m a t i o n ~The ~ . ~nature ~. of the corrosion products formed on copper exposed to the atmosphere was exhaustively studied by Vernon and Whitby*’.’’. In the early periods of exposure the deposit contains sulphide, oxide and soot. By the action of sulphuric acid and by the oxidation of sulphide, copper sulphate is formed, and this hydrolyses and forms a coherent and adherent basic copper sulphate. This approximates initially to CuSO, .Cu(OH), , but gradually increases in basicity until after 70 years or so it becomes CuSO, .3Cu(OH), and is identical with the mineral brochantite. In some cases small quantities of basic carbonate, CuCO, .Cu(OH), (malachite), are also present, and near the sea coast basic chloride CuCI, .3Cu(OH), , (atacamite) is produced.
COPPER A N D COPPER ALLOYS
4:45
Even very near the sea coast, however, sulphate usually predominates over chloride. The presence of atmospheric pollution is thus an essential factor in the development of green patina. In laboratory tests Vernon2*showed that the relative humidity and the presence of sulphur dioxide have a profound effect on the rate of corrosion of copper, as of many other metals. When the relative humidity was less than 63%, there was little attack even in the presence of much sulphur dioxide, but when the relative humidity was raised to 75%, corrosion became severe and increased with the concentration of sulphur dioxide present. By exposing specimens to the atmosphere at different times of the year, Vernon found that the rate of attack on copper was determined by the conditions prevailing at the time of first exposure. For specimens first exposed in winter there was a linear relationship between increase in weight and time of exposure, indicating that the layer of corrosion product formed under these conditions was non-protective. For specimens first exposed in summer the square of the increase in weight was proportional to the time of exposure, indicating that the coating formed in summer, when the atmospheric pollution was relatively low, was protective. The parabolic law holds when the corrosion product layer obstructs the access of the corrosive agent to the metal, the rate of attack then being inversely proportional to the thickness of the layer. The protective character of the layer persisted through subsequent periods when the pollution was relatively high. Copper tarnishes rapidly when exposed to atmospheres containing hydrogen sulphidez9and the reaction is not dependent on the presence of moisture. Atmospheric corrosion tests on copper and several copper alloys were carried out by Hudson3' at a number of sites in Great Britain. Corrosion damage was assessed by one or more of the following methods: gain in weight, loss of weight after cleaning, loss of electrical conductivity, and loss of tensile strength. Hudson found that the resistance to atmospheric corrosion was high and that the rate of attack tended to decrease with time of exposure. Little difference was found between the behaviour of arsenical copper and high-conductivity copper, and most of the alloys tested behaved very similarly except for the brasses, which deteriorated more rapidly owing to dezincification. Several series of atmospheric exposure tests have been carried out since Hudson's work, and the loss in weight data obtained in five of the most important investigations are summarised in Table 4.1 1. In all cases, losses in tensile strength were also determined, and the results from the two methods were, in general, in good agreement. However, for alloys suffering selective attack (such as dezincification of brasses), change in mechanical properties usually provided a more reliable indication of deterioration than weight loss. Some other findings common to all the tests were that (a) corrosion rates decreased with time, (b) least attack occurred at rural sites and most in urban and industrial atmospheres, (c) corrosion was uniform and with few exceptions there was no significant pitting. Tracy, Thompson and Freeman3' exposed specimens of 1 I different grades of copper in the form of sheet and wire to rural, marine and industrial atmospheres in the USA for periods up to 20 years. The differences in the behaviour of the materials were small and of little, if any, practical significance. Very similar results for various types of copper were found by
4:46
COPPER AND COPPER ALLOYS
Mattsson and Holm34in Sweden and Scholes and in the UK (see Table 4.1 1). The results of tests on copper alloys have been given by Tracy3*, Thompson”, Mattsson and Holm34and Scholes and Jacob35,the first two of these investigations being made under the aegis of the American Society for Testing and Materials. The tests of Tracy, and Scholes and Jacob were both for periods up to 20 years; in those of Thompson, and Mattsson and Holm specimens have been removed after 2 years and 7 years and further specimens remain exposed for removal after 20 years. The numbers of materials tested are given in Table 4.1 1; they included brasses, nickel silvers, cupro-nickels, beryllium coppers and various bronzes. Mattsson and Holm tested 14 alloys in the form of rod in addition to the sheet materials, the results for which are given in Table 4.11. In the tests described by Tracy, a high-tensile brass suffered severe dezincification (Table 4.11). The loss in tensile strength for this material was 100% and for a non-arsenical 70/30 brass 54%; no other material lost more than 23% during 20 years’ exposure. In Mattsson and Holm’s tests the highest corrosion rates were shown by some of the brasses. Dezincification caused losses of tensile strength of up to 32% for a /3 brass and up to 12% for some of the a-B brasses; no other materials lost more than 5% in 7 years. Dezincification, but to a lesser degree, occurred also in the a! brasses tested, even in a material with as high a copper content as 92%. Incorporation of arsenic in the a! brasses consistently prevented dezincification only in marine atmospheres. In the tests described by Thompson, the alloy showing the lowest rate of attack at all sites was a bronze containing 7Al-2Si. Relatively high corrosion rates were shown by Cu-5Sn42P at a marine site, and Cu-2.5Co-O.5 Be in the industrial environment. The beryllium-copper alloys were the only materials to show measurable pitting, the deepest attack being 0.06mm after 7 years. In Scholes and Jacob’s tests some pitting, intergranular or transgranular penetration, or selective attack occurred on some of the alloys. The maximum depth of attack exceeded 0 . 2 m m in 20 years on 6 of the 21 materials (three brasses, two nickel silvers and Cu-2ONi-ZOMn), but exceeded 0.5 mm in 20 years only on Cu-2ONi-20Mn and 60140 brass. These two latter alloys lost up to 73% and 13% respectively of their tensile strength; no other alloys lost more than 10% in 20 years. From the work described and other investigations’99,it is evident that copper and most copper alloys are highly resistant to atmospheric corrosion. In general, copper itself is as good as, or better than, any of the alloys. Some of the brasses are liable to suffer rather severe dezincification and it is unwise to expose these to the more corrosive atmospheres without applying some protection. When unusually rapid corrosion of copper and its alloys occurs during atmospheric exposure, it is likely to be for one of the following reasons: 1. Extreme local pollution by products of combustion.
2. Bad design or construction, e.g. the presence of crevices where moisture
may lodge for long periods. 3. Constant dripping of rain water contaminated by atmospheric pollution (e.g. from near-by chimney stacks) or by organic acids from lichens, etc.
Table 4.11 Atmospheric corrosion tests on copper and copper alloys Of
No. of
types
dflerent
copper
alloys
No. of
11 2 1
-
Mattsson and Holm34
I-"
-
Scholes and Jacob"
[-"
Tracy, Thompson and Freeman3' Tracv 32 Thompson33
Rates of attack for a high-tensile brass were 45 x IO-' to I I5
9 17 18 17 X
IO-' rnm/y.
No. sitesOf
4 7 4 3 3 2 2
Period of exposure
Average rates of attack from weight losses
(mm/y x
io4)
8
(years)
Rural sites
Marine sites
Urbadindustrial sires
20 20 7 7 7 20 20
3.6-4.3 0.5-7.6 3.3-10 5-6 2-5 -
6.9-9.4 1.3-23; 4.3-25 7-8 6-1 I 6-10 8-26
8.6-12 13-30; 13-27 10-12 9-22 11-20
-
0
14-38
W
> 2 W 0
8-a m yr
Fr
s
v)
4:48
COPPER AND COPPER ALLOYS
4. Corrosion fatigue due to inadequate allowance for expansion and contraction with consequent buckling as the temperature fluctuates. Most of these can be avoided by attention to design. Corrosion in tropical environments has been the subject of several paper^'^, some of which deal with corrosion at bimetallic contacts”.
Soil Corrosion Several extensive series of soil-corrosion tests have been carried out by the National Bureau of Standards in the Uhited States, and the results have been summarised by Romanoff3*.In one series two types of copper and ten copper alloys were exposed in fourteen different soils for periods up to 14 years. The results for the copper specimens are summarised in Table4.12. The behaviour of the phosphorus-deoxidised and tough pitch coppers was in general very similar. At the less corrosive sites, copper was, with few exceptions, the best material, but most of the alloys lost not more than about twice as much weight, with maximum depths of attack usually not more than two or three times as great as with copper. At the other sites copper was also usually rather better than the alloys, but some of the alloys were occasionally superior. The three most corrosive sites were rifle peat (pH 2-6), cinders (pH 7-6) and tidal marsh (pH 6.9). Corrosion of some of the alloys was particularly severe in the cinders. The behaviour of the brasses tested, particularly those high in zinc, was rather different from that of the other materials. In most cases dezincification occurred and the brasses were the worst materials; in
Table 4.12
Soil-corrosion tests on copper by National Bureau of Standards and British Non-ferrous Metals Research Association
BNFMRA 1st series: 5 least corrosive soils BNFMRA 2nd series: 4 least corrosive soils Nat. Bur. Standards: 9 least corrosive soils Nat. Bur. Standards: 2 next most corrosive soils BNFMRA 1st series: acid clay and acid peat BNFMRA 2nd series: cinders Nat. Bur. Standards: 3 most corrosive soils: rifle peat, cinders, tidal marsh
Period Of exposure
A verage rate of attack from loss in weight
(years)
(mm/y x 1 0 4 )
Maximum rate of pitting (mm/y)
10
0.5-2.5
Nil
5
5.0-25
0.040
14
4.0-25
0.043
14
25- I 3 0
0.033
10
53-66
0.046
5
66
0.32
14
160-355
0.115
COPPER AND COPPER ALLOYS
4:49
the cinders, for instance, several brass specimens were completely destroyed by dezincification. In some of the soils rich in sulphides, however, the brasses were the best materials. The British Non-Ferrous Metals Research Association carried out two series of tests, the results of which have been given by Gilbert 39 and Gilbert and Porter4; these are summarised in Table 4.12. In the first series3’ tough pitch copper tubes were exposed at seven sites for periods of up to 10 years. The two most corrosive soils were a wet acid peat (pH 4.2) and a moist acid clay (pH 4.6). In these two soils there was no evidence that the rate of corrosion was decreasing with duration of exposure. In the second seriesm phosphorus-deoxidised copper tube and sheet was exposed at five sites for five years. Severe corrosion occurred only in cinders (pH 7.1). In these tests sulphides were found in the corrosion products on some specimens and the presence of sulphate-reducing bacteria at some sites was proved. It is not clear, however, to what extent the activity of these bacteria is a factor accelerating corrosion of copper. Cinders and acid peaty soils are obviously among the soils most corrosive toward copper. There is, however, no direct relationship between the rate of corrosion and any single feature of the soil composition or constitution4’. For instance, in the American tests corrosion in several soils with either low pH or high conductivity was not particularly severe, while the British tests show that high chloride or sulphate contents are not necessarily harmful. The above-mentioned tests show that bare copper can safely be buried in a wide range of soils without fear of excessive corrosion. Experience of the behaviour of copper water service pipes, which are used widely, confirms this. Trouble is confined to ‘made-up’ ground containing cinders, etc. and a few other aggressive soils, and in these circumstances it is necessary to apply protection such as bitumen-impregnated wrappings or plastic coatings. Tin coatings cannot be recommended since experience shows that accelerated attack is liable to occur at pores and scratches in the coating, leading to premature failure. Copper water pipes have been known to fail by the action of stray electric currents but this is not a common cause of trouble. There is also agreement between the soil-corrosion tests carried out by the National Bureau of Standards and practical experience of the behaviour of hot-pressed brass fittings used for joining copper water service pipes. These duplex-structure brass fittings are liable to suffer attack by dezincification in many soils in which copper behaves satisfactorily, and for burial underground fittings of copper or gun metal are to be preferred. In general, it may be said that unless there is some special reason for using a copper alloy, it is preferable to choose copper for applications involving service underground.
Copper and Copper Alloys in Natural Waters Copper and copper alloy pipes and tubes are used in large quantities both for conveying fresh and salt waters and in condensers and heat exchangers where fresh or salt waters are used for cooling. Pumps, screens, valves and other ancillary equipment may also be largely constructed of copper alloys. Large tonnages of these materials are therefore used in power stations, on
4:50
COPPER AND COPPER ALLOYS
board ship, in sugar factories and in oil refineries, as well as in hot-and coldwater circuits and heating and cooling systems in hospitals, hotels, factories and homes. Corrosion problems that arise are frequently discussed under the headings (a) sea-water, and (b) fresh waters, but there is, in fact, no sharp dividing line, since some harbour, estuarine and brackish well waters are mixtures of sea-water and fresh water and are often variable in composition. In the past, corrosion problems were serious, particularly in sea-water service, but resistant alloys have been developed and although trouble still occasionally arises this is more frequently due to poor design or operation rather than to lack of materials suitable for the application. There are several distinctive types of corrosion that copper and copper alloys may suffer, particularly in sea-water, but also on occasion in fresh waters. The more important of these are discussed briefly below. Impingement attack When moving water flows over copper or copper alloys the turbulence may be sufficient to cause breakdown of the surface film. This is particularly likely to happen if air bubbles entrained in the water break as they hit the metal surface. The resulting corrosion is characteristic, producing clean-swept pits, often of a ‘horseshoe’ shape as shown in Fig. 4.11. This type of attack was first described by Bengough and May42s43. The action can be very rapid, the local anodes being depolarised by the continuous removal of metal ions and corrosion product, and the local cathodes by the dissolved oxygen in the rapidly moving well-aerated water stream. Factors that tend to increase the severity of impingement attack are increase of water speed and particularly of local turbulence, pollution of the water, and, within certain limits, increase in the size and the amount of entangled air bubbles (see also Section 1.6). A laboratory test designed to simulate the conditions occurring in condenser tubes in practice was devised by May43and newer versions of this ‘jet impingement apparatus’ have been described”, as has some of the testing equipment in use in the USA45.Use of the jet impingement apparatus has been an important factor in the development of alloys resistant to impingement attack, but it has to be borne in mind that the results obtained when the water is recirculated may be different from those obtained when it is passed once through the apparatus, as shown by Gilbert and LaQue4.
Fig. 4.1 I
Typical impingementattack on Admiraltybrass condenser tube. Magnification x 2
COPPER AND COPPER ALLOYS
4:51
Details of jet impingement tests will be found in Section 19.1. Alloys resistant to impingement attack will be considered subsequently.
Dezincification of brasses When dezincification occurs, regions of the brass become replaced by a porous mass of copper which, though retaining the shape of the original article, has virtually no strength. There has long been discussion as to whether there is selective corrosion of the zinc in the brass, which leaves the copper behind, or whether complete dissolution of the brass occurs, followed by re-deposition of copper. Possibly both processes occur in different circumstances. The mechanism has been investigated and discussed by Evans7, Fink47,Lwey4*,Feller4’ and Heidersbach”, and is referred to in many other papers”. With a single-phase brass the whole of the metal in the corroded areas is affected. Dezincification may proceed fairly uniformly over the surface, and this ‘layer type’ takes much longer to cause perforation than the localised ‘plug type’ that more often occurss2. With a two-phase brass the zinc-rich fl phase is preferentially attacked as shown in Fig. 4.12. Eventually the CY phase may be attacked as well. The zinc corrosion products that accompany dezincification may be swept away, or in some conditions may form voluminous deposits on the surface which may lead to blockages, e.g. in fittings. In general, the rate of dezincification increases as the zinc content rises, and great care needs to be exercised in making brazed joints with copperhinc brazing alloys, particularly if they are to be exposed to sea-water. Under these conditions, a properly designed capillary joint may last for some time, but it is preferable to use corrosion-resistant jointing alloys such as silver solders (e.g. BS 1845, Type AGZ or AG5)53.
Fig. 4.12 Dezincification of two-phase brass showing preferentialattack of the 8 phase (upper half of photomicrograph). Magnification x 133
4:52
COPPER AND COPPER ALLOYS
Factors that cause increased rates of dezincification are high temperature, high chloride content of water and low water speed. Dezincification is also likely to occur preferentially beneath deposits of, for instance, sand or silt on the metal surface, or in crevices where there is a low degree of aeration. Addition of about 0.04% arsenic will inhibit dezincification of a brasses4‘ in most circumstances and arsenical a brasses can be considered immune to dezincification for most practical purposes 54. There are conditions of exposure in which dezincification of these materials has been observed, e.g. when exposed outdoors well away from the sea34,or when immersed in pure water at high temperature and pressure, but trouble of this type rarely arises in practice. In other conditions, e.g. in polluted sea-water, corrosion can occur with copper redeposition away from the site of initial attack, but this is not truly dezincification, which, by definition, requires the metallic copper to be produced in situ. The work of Lucey4*goes far in explaining the mechanism by which arsenic prevents dezincification in a brasses, but not in a-0 brasses (see also Section 1.6). An interesting observation is that the presence of a small impurity content of magnesium will prevent arsenic in a brass from having its usual inhibiting effect”. Additions of antimony or phosphorus, in amounts similar to arsenic, are claimed to be also capable of preventing dezincification of a brasses. Most manufacturers use arsenic, however, and it certainly appears desirable to avoid phosphorus, since has shown that this element can, in some circumstances, lead to an undesirable susceptibility to intercrystalline corrosion. The same appears to be true of excessive amounts of arsenic (over about 0.05%). N o reliable method of inhibiting dezincification of two-phase brasses has been discovered. Various additions, including arsenic, have been advocated from time to time, but nothing is known that will render a-0 brasses immune to dezincification under all conditions of exposure. The addition of 1Yo tin can markedly reduce the rate of dezincification, and naval brass (61Cu38Zn-1Sn) is attacked considerably more slowly than 60140 brass in seawater, though there may be virtually no difference in most fresh waters. Some of the cast complex high-strength two-phase brasses containing tin, aluminium, iron and manganese appear to have relatively good resistance to dezincification, but they are by no means immune to it. Selective attack in other alloys Selective attack analogous to dezincification can occur in other copper alloys, particularly aluminium bronzes and less frequently tin bronzes5’, cupro-nickels 5 8 , etc. In recent years de-aluminification of aluminium bronzes has been studied extensivelys9and the results indicate that whilst a-phase alloys suffer such attack comparatively rarely, alloys of higher aluminium content can be more susceptible. The electrochemical relationships are such that preferential corrosion of the second phase is liable to occur in a-y, alloys, but a-fl alloys are relatively resistant to attack. Retention of fl phase is favoured by rapid cooling after casting or after high-temperature heat treatment, and also by incorporating manganese in the alloy. Deposit attack and pitting When water speeds are low and deposits settle on the surface (particularly at water speeds below about 1 m/s), pitting of copper and copper alloys is liable to occur by differential aeration effects.
COPPER A N D COPPER ALLOYS
4:53
In sea-water systems such attack may occur under dead barnacles or shellfish, the decomposing organic matter assisting corrosion. Pitting is most likely to occur in polluted in-shore waters, particularly when hydrogen sulphide is present. In such contaminated waters non-protective sulphide scales are formed and these tend to stimulate attack.
Corrosion of condenser tubes and related equipment There have been many surveysm of the problems of corrosion of condenser and heatexchanger tubes and related components in marine service and others6’ dealing with oil refinery service. Corrosion of condenser tubes was a problem of great magnitude during the first quarter of this century. Its solution was based on research carried out for the Institute of Metals by Bengough e t ~ l . ~ ’ * ~one ‘ * ~of~ whom, , May43*54, remained associated with the research when it was transferred to the auspices of the British Non-Ferrous Metals Research Association in 1930. A history of condenser tubes up to 1950 has been published’”. In early times 70130 brass condenser tubes failed by dezincification and Admiralty brass (70Cu-29Zn-1Sn) was brought into use. This proved little better, but some time later the addition of arsenic was found to inhibit dezincification. Failures of Admiralty brass by impingement attack became a serious problem, particularly as cooling water speeds increased with the development of the steam turbine. The introduction of alloys resistant to this type of attack was a great step forward and immediately reduced the incidences of failure. The alloys in most common use today are aluminium brass (76Cu-22Zn2A1-0.04As) and cupro-nickels containing appropriate iron and manganese additions I5v6’. Three cupro-nickel alloys are in widespread use containing (approximately) (i) 30Ni-0*7Fe-0.7Mn, (ii) 30Ni-2Fe-2Mn and (iii) 10Ni1-5Fe-1Mn. These materials are extensively and successfully used in ships, power stations, oil refineries, etc., in condensers and heat exchangers, with nominal water speeds through the tubes of up to about 3 m/s, sometimes with much entangled air present. At the highest water speeds there is a rather greater factor of safety with 70/30 cupro-nickel, and this alloy is also usually to be preferred under most polluted water conditions, although occasionally other alloys are as good or even better. There is evidence to indicate that when the operating conditions involve relatively high temperatures, aluminium-brass or 90/10 cupro-nickel is to be preferred to 70130 cupronickel”. Admiralty brass is no longer considered a suitable alloy for seawater service, except possibly where water speeds are very low, i.e. not more than about I m/s. In some oil refineries, for instance, Admiralty brass is preferred because it has good resistance to corrosion by oil products, and in these installations the heat exchangers are designed to have low coolingwater speeds, so that corrosion from the sea-water side is kept within reasonable bounds. Tin-bronze containing about 12% Sn has been shown to have good resistance to impingement attackw, to attack by acid cooling waters, and to abrasion in cooling waters containing suspended solid particles, but the alloy has so far only been used on a limited scale. The alloy most commonly used when suspended abrasive matter is a problem is the 30% Ni alloy containing 2% each of Fe and Mn. Aluminium-bronze tubes have sometimes given
4:54
COPPER AND COPPER ALLOYS
good results, but their use has been limited because of their susceptibility to pitting attack. The occasional failures that still occur in condenser tubes are usually due to one (or sometimes several) of the following factors: 1. Localised attack or pitting in badly contaminated waters. 2. Pitting under decaying barnacles, shell fish or other deposits. 3. Impingement attack due to high local water velocities, e.g. at partial obstructions in a tube such as pieces of coke, shell fish, etc. 4. Erosion due to sand or other abrasive particles suspended in the water 65. 5 . Use of tubes of the wrong alloy, or of incorrect composition, or containing manufacturing defects. A difficult condenser-tube corrosion problem arises from the use of polluted cooling waters from harbours and estuaries that may be severely contaminated. All condenser-tube materials are liable to suffer corrosion in these circumstances, and the choice of materials is made difficult by the fact that different orders of merit apply at different locations and even at the same location at different times. The state of the water when the tubes first enter service may well determine whether or not a satisfactory life will be obtained&. The most corrosive waters are those containing free hydrogen sulphide produced by the action of sulphate-reducing bacteria. Waters may also be rendered abnormally corrosive by the presence of small amounts of organic sulphur compounds produced by bacterial action, as shown by Rogers67. Corrosion of power-station condenser tubes by polluted waters has been particularly troublesome in Japan and efforts have been made to study the problem by electrochemical methods69 and by exposing model condensers at a variety of power station sites7'. Improved results have been reported using tin brasses7' or special tin bronzes7*. Pretreatment with sodium dimethyldithiocarbamate is reported to give protective films that will withstand the action of polluted waters7', though the method would be economic only in special circumstances. Electrochemical studies, including the determination of polarisation curves, have been carried out in recent years by many authors74 in endeavours to understand more fully the mechanism of protective film formation on copper alloys in sea-water. Other authors have described experiences with condenser tubes in fresh or brackish waters75.Methods of maintaining tube cleanliness include ~ h l o r i n a t i o n ~ use ~ , of high molecular weight water-soluble and use of the Taprogge system of circulating sponge-rubber ball^^**^^. Condenser tube-plates of Naval brass usually undergo some dezincification in sea-water service, but this is normally not serious in view of the thickness of metal involved. Attack can, however, be more serious with 60/40 brass (Muntz metal) and such plates may have to be renewed during the life of the condenser. Increasing use is now being made of tube-plates of more resistant materials such as aluminium-brass, silicon-bronze, aluminiumbronze or cupro-nickel. Plates that are too large to be rolled in one piece can be fabricated by welding together two or more pieces. In some special applications the tubes are fusion welded or explosively welded to the tube-plates.
COPPER AND COPPER ALLOYS
4:55
Fusion welding operations are rather more difficult with brasses than with other copper alloys (because of evolution of zinc fume from brasses). Condenser water-boxes were hitherto usually made of unprotected (or poorly protected) cast iron and these afforded a measure of cathodic protection to the tube-plates and tube ends. This beneficial effect has been lost with the general adoption of water-boxes completely coated with rubber or some other impervious layer, or of water-boxes made from resistant materials such as gunmetal, aluminium-bronze or cupro-nickel, or steel clad with cupronickel or Monel. To prevent attack on tube-plates and tube-ends in these circumstances, it is highly desirable to install either a suitable applied-current cathodic-protection systemsso, or sacrificial soft-iron or mild-steel anodes. Ferrous wastage plates have the additional advantage that the iron corrosion products introduced into the cooling water assist in the development of good protective films throughout the length of the tubes. This is particularly important in the case of aluminium-brass tubes; indeed, with such tubes it may be desirable, as an additional preventive measure, to add a suitable soluble iron salt (such as ferrous sulphate) regularly to the cooling water. Cases of the success of such treatment in power station condensers have been described by Bostwick" and Lockhart ", and other worker^'^'^^ have since studied the effects of ferrous sulphate treatment on tube behaviour. As it has become increasingly necessary to supplement natural sources of fresh water in various parts of the world, processes for producing fresh water from sea-water have been intensively studied and the literature dealing with the subject is very extensive. Distillation is currently the process most widely used and during recent years increasing numbers of multi-stage flashdistillation plants have been installed in various countries, many of the larger units being capable of producing several millions of gallons of fresh water per day. In these plants, sea-water passes through horizontallydisposed tubes and steam 'flashed' from the brine condenses on the outside. In some parts of the plant the conditions are similar to those in steam condensers, but in other parts the sea-water has been treated to remove dissolved gases and is at much higher temperatures. In another distillation process receiving considerable study, films of sea-water fall down the inner walls of comparatively large-diameter vertically-disposed tubes, usually of fluted configuration, and evaporation takes place due to the heating effect of steam condensing externally. Copper-alloy tubes are used in large numbers for the heat-exchange units in distillation plants, mainly aluminium-brass and the various cupro-nickel alloys, and the factors affecting choice of materials have been considered in several papersE4. For ships' cooling-water trunking and salt-water services in the engine room and elsewhere, including fire mains, plumbing and air-conditioning systems, more resistant alloys are taking the place of copper or galvanised steel, which were formerly extensively used, but which do not have adequate resistance to attack by sea-water. Both aluminium-brass and the Cu-1ONi1-5Fe alloy are widely used and, being highly resistant to impingement attack, normally give excellent service. In some special naval applications pipelines of 70/30 cupro-nickel are used. It is important that correct fabrication and installation techniques are used. Carbonaceous residues from fillers used in bending operations must be avoided or pitting corrosion may occur in service. Jointing materials of low corrosion resistance should not be used,
4:56
COPPER AND COPPER ALLOYS
silver brazing or appropriate welding methods being the correct techniques. Residual stresses, if present, can cause stress-corrosion failures of aluminium-brass pipelines in service. Copper alloys in wrought or cast form are used for other purposes in ships and other marine installations, such as for propellerss5, bearings, valves and pumps. One widespread application of aluminium-brass is its use for heating coils in tankers carrying crude oil or petroleum products. Some corrosion problems encountered in this and other applications on board ship have been described by Gilbert and Jenner".
Fresh Waters
Fresh waters are, in general, less corrosive towards copper than is sea-water, and copper is widely and satisfactorily used for distributing cold and hot waters in domestic and industrial installations '5*s7. Copper and copper alloys are used for pipes, hot-water cylinders, fire-back boilers, ball floats, ball valves, taps, fittings, heater sheaths, etc. In condensers and heat exchangers using fresh water for cooling, tubes of 70/30 brass or Admiralty brass are usually used, and corrosion is rarely a problem. Joints in copper components may be a source of trouble. Copper/zinc brazing alloys may dezincify and consequently give rise to leaksw. In some waters, soft solders are preferentially attacked unless in a proper capillary joint. Copper/phosphorus, copper/silver/phosphorus, and silver brazing alloys are normally satisfactory jointing materials. Excessive corrosion of copper is sometimes produced by condensates containing dissolved oxygen and carbon dioxide. Rather severe corrosion sometimes occurs on the fire side of fire-back boilers and on electric heater element sheaths under scales deposited from hard waters9'.
Dezincifcation of brasses This may occur, particularly in stagnant or slowly-moving warm or hot waters relatively high in chloride and containing little carbonate hardness". Dezincification of a brasses is inhibited by the usual arsenic addition (see Fig. 4.12), but two-phase brasses are liable to severe attack in some waterss9. In such waters the use of duplex-structure brass fittings should be avoided. Impingement attack Copper may occasionally suffer this form of attack in systems where the speed of water flow is unusually high and the water is one that does not form a protective scale, e.g. a soft water containing appreciable quantities of free carbon dioxide9*. Ball valve seatings may also suffer an erosive type of attack. The corrosion of ball valves, including the effect of chlorination of the water, has been studied by several workers9'. Dissolution Some waters continuously dissolve appreciable amounts of copperw. Factors that favour this action are high free carbon dioxide, chloride and sulphate contents, low hardness, and increase of temperature. The trouble is therefore most prevalent in hot, soft, acid waters. The corrosion is general and the resulting thinning is so slight that the useful life of the pipe or component is virtually unaffected (unless impingement attack
COPPER AND COPPER ALLOYS
4:57
occurs). Trouble is usually confined to (a) stimulation of the corrosion of components of zinc-coated light alloys%, and sometimes bare steel with which the water subsequently comes into contact; and (6)the formation of green stains in baths, sinks, etc. owing to the combination of copper with soaps. In de-aerated conditions, for instance in most central heating systems, little if any attack on copper O C C U ~ S ~ ~ As * ~ ' far . as drinking waters are concerned, copper is not classified as a toxic substance or hazardous to health. To avoid any difficulties due to unpalatability, the maximum continuous copper content should not exceed 1.0 p.p.m., with a limit of 3 p.p.m. in water after standing overnight in copper pipes. A review of the subject by G r ~ n a makes u ~ ~ reference to 394 published papers. Piffing Occasionally copper water pipes fail prematurely by pitting. This most often occurs in cold waters originating from deep wells and boreholes and has been shown by Campbellw to be associated with residues of carbon produced in the bore of the tubes during bright annealing, as a result of decomposition of residual drawing lubricant. It is therefore necessary for manufacturers to take steps to avoid these harmful residues. This trouble has occurred in many different countries loo. Failures of this type are confined to certain districts, and Campbell has shown that in many supply waters in Great Britain pitting cannot proceed, even in tubes containing dangerous cathodic films, owing to the presence in the water of small amounts of a naturally-occurring inhibitor, probably an organic colloidal material, that stifles pitting of copper. Trouble therefore only occurs in waters that contain little or no inhibitor. Pitting failures also occasionally occur in copper water cylinders lo' and as a result of a study of this problem LuceyIo2has made suggestions about the mechanism of pitting of copper in supply waters. In hot-water pipes, failures sometimes occur in certain areas supplied with soft waters from moorland gathering grounds. The waters concerned contain a few hundredths of a part per million of manganese, and in the course of several years' exposure, a deposit rich in manganese dioxide is laid down in the hottest parts of the system. This may cause pitting and eventually lead to failure. Hot-water pitting of another type is sometimes experienced in soft waters having a high sulphate content in relation to the carbonate hardness and a relatively low pH value103.
Behaviour in Chemical Environments Detailed information on the action of a wide range of chemicals on copper and copper alloys is given in a number of publications, particularly those listed under References 7-12, 104 and 105. When contemplating the use of copper-base materials for industrial purposes it is necessary to bear in mind that even though a satisfactory life of the component may be obtained, trouble can arise from other causes: 1. Copper compounds can be tolerated only in small amounts in potable
waters or substances that are to be consumed.
4:58
COPPER A N D COPPER ALLOYS
2. Copper compounds are highly coloured, and a very small amount of corrosion may lead to staining and discoloration of products. 3. stimulation of the corrosion of vital parts made of more anodic metals may occur if they are connected to copper. 4. Very small amounts of copper taken into solution may cause considerable corrosion of more anodic metals elsewhere in the system, particularly zinc9', aluminium%, and sometimes steel IW. Small particles of copper deposited from solution set up local cells that cause rapid pitting. Despite these qualifications copper and its alloys are used extensively and successfully in much chemical equipment. Uses include condensers and evaporators, pipelines, pumps, fans, vacuum pans, fractionating columns, etc. Tin-bronzes, aluminium-bronzes and silicon-bronzes are used in some circumstances because they present better corrosion resistance than copper or brasses.
Acid solutions Copper does not normally displace hydrogen even from acid solutions, and it is therefore virtually unattacked in non-oxidising conditions. Most solutions that have to be handled contain dissolved air, however, and this will cause cathodic depolarisation and enable some corrosion to take place"'. It is difficult, therefore, to lay down any general recommendations for the use of copper in acid solutions, since the rate of attack depends so greatly on the particular circumstances. Under fairly mild conditions copper or copper alloys are successfully used for handling '04*IO9, sulphuric'w'O* 'lo, phosolutions of hydrofluoric IO4, 'OB, ~ p h o r i c ~ * ~ ~and ~"'* acetic ' ' ' and other fatty acids8-10*1M*112. Rates of corrosion, in general, increase with concentration of acid, temperature, amount of aeration ' I 3 and speed of flow 'I4. Tin-bronzes', aluminium-bronzes 7 . 1 ' s , silicon-bron~es~~' and cupro-nickels' are among the copper alloys most resistant to acids. Brasses should not normally be used. All copper-base materials are attacked rapidly by oxidising acids such as nitric, strong sulphuric, etc. The dissolution of copper and of brasses in acid solutions has been studied by several authorslL6.Various substances have inhibiting effects on the rate of attack of copper or brasses in nitric acid'I7 and in hydrochloric acid"'. Neutral and alkaline solutions Copper-base materials are resistant to alkaline solution^^*^*"^ over a wide range of conditions but may be appreciably attacked by strong solutions, particularly if hot. Copper/nickel alloys usually give the best results in alkaline solutions. Copper and copper alloys should be avoided if a ~ n r n o n i a * is ~ ~present, * ' ~ ~ owing to the danger of both general corrosion and, if components are under stress, stress corrosion. Copper is satisfactory for handling solutions of most neutral salts IO4*12' unless aeration and turbulence are excessive. An exception is provided by those salts that form complexes with copper, such as cyanides, and solutions containing oxidising agents, such as ferric or stannic compounds '. Other chemicals Copper and copper alloys are unsuitable for handling hydrogen peroxide IO4. or molten sulphur IO4, 123. Hydrogen sulphide accelerates corrosion of most copper-base materials. In its presence brasses high in zinc are usually found to behave better than other copper alloys'. 7-99
4:59
COPPER A N D COPPER ALLOYS
Halogens have little action on copper at room temperature when dry, but are corrosive when wet. Hypochlorite solutions are corrosive124.Most organic compounds are without appreciable action8*IW. Copper and copper alloys are extensively and successfully used in refrigeration systems employing organic refrigerants such as CCI,F,. Attack can, however, occur if halogenated compounds hydrolyse in the presence of moisture to give traces of hydrochloric acid. Copper alloys are widely used for handling hydrocarbon oils, though if sulphur compounds are present attack can be serious'. The effects of synthetic detergents on copper have been investigated'25,and several author^^*'@'*^^^ have discussed various aspects of the behaviour of copper and its alloys in the food-processing industry.
Oxidation and Scaling Several authors'***127-129 have reviewed the literature on the oxidation and scaling of metals, including copper. Copper
-
The volume ratio (see Section I .9) for cuprous oxide on copper is 1 7, so that an initially protective film is to be expected. Such a film must grow by a diffusion process and should obey a parabolic law. This has been found to apply for copper in many conditions, but other relationships have been noted. Thus in the very early stages of oxidation a linear growth law has been observed (e.g. at 1 000°C)130. At 180-290°C it was foundI3' that the parabolic law first applied but subsequently changed to a logarithmic relationship of the type y = KlogB(t
+ l/B)
B being a constant. Other workers have reported a cubic relationship under some conditions. Evans132has shown how the effect of internal stresses in growing films may have various effects that can lead to any one of the first three growth laws referred to above. At medium and high temperatures 133v134 copper ultimately follows the parabolic law 128*13s. It has been shown'36using radioactive tracers that the diffusion of copper ions in cuprous oxide is the rate-determining step at 800-1 0oO"C, and there is considerable evidence in favour of the view that metal moves outwards through the film by means of vacant sites in the oxide lattice133. When oxidation is a diffusion process the oxidation rate should be related to the temperature by the Arrhenius equation K = A exp [ - Q / R T ] where K is the rate constant, A a constant, R the gas constant, Tthe absolute temperature, and Q the activation energy. Values that have been obtained for A and Q are summarised by Tylecote"', Pilling and BedworthI3', Feitknecht '" and others give values of Q of about 0.17 MJ for temperatures
4:m
COPPER AND COPPER ALLOYS
of 700-1 OOO"C, while at lower temperatures (up to 500°C) and others obtained values of about 85 kJ. These values are in agreement with calculations by Valensi140based on the assumption that at the high temperatures the oxidation proceeds by the reaction of oxygen with metallic copper to produce cuprous oxide while at lower temperatures the rate is determined by the reaction between oxygen and cuprous oxide to form cupric oxide. At low temperatures (e.g. 300°C) the film consists almost entirely of CuO. As the temperature increases the film consists of a layer of Cu,O beneath a layer of CuO, the proportion of Cu,O increasing until at high temperature the film is almost entirely C u 2 0 . The precise composition of the film depends, however, on a number of factors, including temperature, time, oxygen concentration in the atmosphere, etc. Tylecote has investigated the composition, properties and adherence of scales formed on various types of copper at temperatures between 400 and 900°C. At the higher temperatures the scales formed on coppers containing phosphorus were more brittle and less adherent than those formed on coppers containing no phosphorus. Studies have been carried out of the effects at high temperatures of sulphur and of atmospheres containing hydrogen sulphide 143, steam 144.145, sulphur dioxide '45 and hydrogen chloride 145. Copper AIIOys
With copper alloys containing more noble metals the oxide will be substantially pure copper oxide since the oxides of the noble metals have higher dissociation pressures than the copper oxides. With alloys containing baser metals, however, the alloying element will appear as an oxide in the scale, often in greater concentration than in the alloy itself, and sometimes to the exclusion of copper oxides. The dissociation pressures of many oxides have been calculated by Lustman Whether the rate of oxidation of an alloy of copper with a baser metal is less or more than that of copper will depend on the concentration of the alloying element and the relative diffusion velocities of metal atoms or ions in the oxide layers. There is extensive literature on the oxidation behaviour of copper alloys 12** 129*14' . According to Wagner's theory148 the rate of oxidation will be largely influenced by the electrical conductivity of the film, and the theory is therefore supported by the fact that the alloying elements giving maximum oxidation resistance, i.e. beryllium, aluminium and magnesium, form oxides having very low conductivities, as shown by Price and Thomas 14'. Wagner calculated that when sufficient aluminium was present in copper to cause the formation of an alumina film the oxidation rate should be decreased by a factor of more than 80 OOO. Experiment showed a factor of only 36, but when Price and Thomas carried out initial oxidation under very slightly oxidising conditions, producing only a film of alumina, the oxidation rate on subsequent exposure to full oxidising conditions was decreased by a factor of about 240000. Hallowes and Voce145found that selective oxidation of a 95Cu-5AI alloy by this method gave protection from atmospheric oxidation up to 800°C unless the film was scratched or otherwise damaged, or the atmosphere contained sulphur dioxide or hydrogen chloride.
COPPER AND COPPER ALLOYS
4:61
The effects on oxidation resistance of copper as a result of adding varying amounts of one or more of aluminium, beryllium, chromium, manganese, silicon, zirconium are described in a number of papersk4’. Other authors have investigated the oxidation of copper-zinc I 5 O * Is’ and copper-nickel alloys 1s1,152, the oxidation of copper and copper-gold alloys in carbon dioxide at 1 OOOoCks3 and the internal oxidation of various alloys’54. Copper alloys have been used extensively in high-pressure feed-water heaters in power generating plant. Experience has shown that when such heaters are operated intermittently, 70Cu-30Ni or 80Cu-20Ni alloy tubes suffer fairly rapid and severe steam-side (external) oxidation with the formation of exfoliating scales. This corrosion, which may be associated with ingress of air during shutdowns, has been the subject of several published papersIs5. The behaviour of other alloys for feed-water heater service has also been discussed Is6.
Stress Corrosion (Chapter 8) Failure of copper alloys may occur by cracking due to the combined influence of tensile stress and exposure to a corrosive environment. When the stresses are produced in components during manufacture the trouble is usually known as season cracking and failures of brass components due to this form of stress corrosion have been known for many years1s7-1m. Only certain specific environments appear to produce stress corrosion of copper alloys, notably ammonia or ammonium compounds or related compounds such as amines. Mercury or solutions of mercury salts (which cause deposition of mercury) or other molten metals will also cause cracking, but the mechanism is undoubtedly different ‘ “ I . Cracks produced by mercury are always intercrystalline, but ammonia may produce cracks that are transcrystalline or intercrystalline, or a mixture of both, according to circumstances. As an illustration of this, Edrnundsl6* found that mercury would not produce cracking in a stressed single crystal of brass, but ammonia did. Stress Corrosion of Brasses
Alloys containing only a few per cent of zinc may fail if the stresses are high and the environment sufficiently corrosive. Most types of brass, besides the plain copper/zinc alloys, appear to be susceptible to stress corrosion. An extensive investigation of the effect of additions to 70/30 brass was carried out by Wilson, Edmunds, Anderson and P e i r ~ e ’who ~ ~ , found that about 1070 Si was markedly beneficial. Other additions were beneficial under some circumstances and none of the 36 additions tested accelerated stress-corrosion cracking. Further results are given in later papersIw In general, the susceptibility to stress corrosion appears to increase with increase in zinc content, but in some circumstances alloys containing 64-65% Cu were found to be rather more affected than those containing 60% have investigated the residual stresses introduced Many workers 15’* by different working processes in brasses of various compositions and the
4:62
COPPER A N D COPPER ALLOYS
annealing treatments necessary to remove these stresses or reduce them to a safe level. A 'stress-relief anneal' at about 300°C will usually lower internal stresses to comparatively small values without much effect on the hardness of the material. Specifications for brass products customarily include provision for carrying out a mercurous nitrate test '57 to ensure that unduly high residual tensile stresses are not present, but a satisfactory result in this test does not guarantee freedom from cracking in environments containing ammonia. More searching tests involving exposure to ammonia have therefore been devised. The standardisation of stress-corrosion cracking tests and their correlation with service experience have been described in several papers 167. Other authors have described practical cases of stress-corrosion cracking, usually involving tensile stresses applied in service. Two possible preventive measures are the use of coatings'61 or inhibitors I7O. The behaviour of a wide range of a,a-/3and /3 brasses in various corrosive environments was studied by Voce and Bailey and the results summarised by Whitakeri7'. Penetration by mercury and by molten solder was intercrystalline in all three types of brass. In moist ammoniacal atmospheres the penetration of unstressed brasses of all types was intercrystalline. Internal or applied stresses accelerated the intercrystalline penetration of a brasses and initiated some transcrystalline cracking, and also caused severe transcrystalline cracking of /3 alloys and transcrystalline cracking across the /3 regions in the two-phase brasses. Immersion in ammonia solution, however, caused intercrystalline cracking of stressed 0 brasses. /3 brasses containing 3% or more aluminium failed with an intercrystalline fracture when stressed at about the 0.1% proof stress in air. The behaviour of alloys of this type was subsequently studied by P e r r ~ m a n ' ~ and ~ , by Bailey173,who has shown that cracking in air occurs only when moisture is present. It has been confirmed that /3 brasses are prone to crack in service174. High-strength a-/3 brasses containing up to about 5% AI (with small amounts of Fe, Mn, Sn, etc.) used for propellers, parts of pumps, nuts and bolts, etc. usually give good service but occasionally suffer intercrystalline failure, for instance in contact with sea-water. Examination of such failures usually reveals thin dezincified layers along the cracks, but it is difficult to decide whether the crack or the dezincification occurred first. The theoretical aspects of stress-corrosion cracking have attracted much attention in recent years. Amongst the copper alloys, the behaviour of brasses in ammoniacal environments has been most studied. Whilst cracking has been shown to be possible in contact with some other corrosive agents, ammonia has the most powerful effect. Evansi32suggests that this is because ammonia is a feeble corrodent that produces little attack except at regions such as grain boundaries or other lattice imperfections and because it prevents accumulation of copper ions in the crevices formed owing to the formation of stable complexes, Cu[(NH3),I2+.The mode of cracking (intercrystalline or transcrystalline) can be affected by changes in composition of the brass or by changes in the nature of the e n ~ i r o n m e n t ' ~ ~ . Mattsson i76 found that on immersion in ammoniacal solutions of different pH values, stressed brasses cracked most rapidly at pH 7 - 1 - 7 . 3 and that in this range black surface films formed on the metal. The r81e of tarnish films has been further studied s~bsequently'~'. Many authors have studied
COPPER AND COPPER ALLOYS
4:63
electrochemical 17' or metallurgical 179 aspects of the stress corrosion of copper alloys and discussed theories of the mechanism. Papers on the subject have been included in several symposia or conferences on stress corrosion of metals I59.180.181 . The stress cracking of brasses was reviewed by Bailey'82 contain references to the and subsequent reviews of stress corrosion subject. Stress Corrosion of Other Copper Alloys
Evidence indicates that pure copper is not liable to undergo stress corrosion'84-'86but instances are known of the failure by stress corrosion of copper containing about 0.4% As'87or 0.02% PIa4.Failure can also occur with copper-beryllium IE8, copper-manganese IE9, aluminium-bronzes l9O, tinbronzes, silicon-bronzes, nickel silvers 19' and cupro-nickels 19'- 19*. Most of these alloys are much less susceptible to cracking than brasses 185.193. Under some conditions, however, aluminium-bronzes can be very prone to cracking '%. In ammoniacal environments the cracks tend to be transcrystalline, and in steam atmospheres intercry~talline'~~. Additions of 0.35% Sn or Ag An are claimed to be effective in preventing intercrystalline cracking '%*I9'. aluminium-bronze containing 2% Ni and 0.5-0.75Vo Si is claimed to have good resistance to stress corrosion I%* 19' Thompson and Tracy'@ carried out tests in a moist ammoniacal atmosphere on stressed binary copper alloys containing zinc, phosphorus, arsenic, antimony, silicon, nickel or aluminium. All these elements gave alloys susceptible to stress corrosion. In the case of zinc the breaking time decreased steadily with increase of zinc content, but with most of the other elements there was a minimum in the curve of content of alloying elements against breaking time. In tests carried out at almost 70MN/m2 these minima occurred with about 0.2% P, 0.2% As, 1% Si, 5% Ni and 1% Al. In most cases cracks were intercrystalline.
Protective Measures The good behaviour of copper and copper alloys is dependent upon correct choice of material, good design of equipment, and proper methods of operation. If proper attention is given to these factors there will usually be no need for protective measures. In special cases, however, e.g. to prevent the dissolution of small amounts of copper or to maintain a high-grade finish, metallic coatings of one or more of the following metals may be applied: tin, lead, nickel, silver, chromium, rhodium, gold. In other cases painting, varnishing or lacquering may be desirable, or if the conditions are very severe, as in some corrosive soils, heavier protection such as bituminous or plastic coatings may be necessary. Brasses that are liable to suffer dezincification or stress corrosion may need protection where other copper alloys would be satisfactory unprotected. Sometimes use is made of the principles of cathodic protection, e.g. steel 'protector blocks' in condenser water-boxes. In some circumstances, use of inhibitors may be a desirable remedial measure. For instance. benzotriazole has been found of considerable value
4:64
COPPER AND COPPER ALLOYS
for preventing staining and tarnishing of copper products2"0. Sodium diethyldithiocarbamate also has useful inhibiting properties 201. Other types of inhibitors can be of value in condensate systems202and in acid solutions 1 1 7 * 1 1 8 . Reviews have been given of corrosion inhibitors for copperzo3 and brasses". The danger of accelerated attack on copper-base materials due to coupling with other metals is small since copper is usually the cathodic member of the couple, but precautions are often necessary to prevent excessive corrosion of the anodic member. Surveys of the behaviour of couples involving copper and copper alloys have been One material that has been found capable of accelerating attack on copper in practice is graphite; hence graphitic paints are undesirable. Occasionally the action between different copper-base materials may be appreciable, e.g. gunmetal may stimulate the corrosion of copper or brass in sea-water.
Copper Alloys in Marine Environments Much attention continues to be devoted to the corrosion behaviour of copper alloys in an increasing range of marine application^^^', 244* 245- 246. Many publications have dealt with the long-established uses of copper alloys for condensers and heat exchangers and there have been several reviews of the selection of materials for these applications208and others discussing the factors that may lead to corrosion problems209. Specific aspects covered include the effects of velocity2", ferrous sulphate treatment211. 212. 213 and sponge ball cleaning212* 2 1 3 v 214; the latter may lead to greater corrosion, though ferrous sulphate treatment can offset the effect. At low seawater temperatures2" alloys appear to be more susceptible to attack and the beneficial effects of ferrous sulphate additions to the seawater are reduced. Chlorination212* 244 can cause increased attack in some circumstances, aluminium brass being more susceptible than cupronickels. Presence of sulphide p ~ l l u t i o n4'.1~ ~causes '~ serious corrosion if polluted and aerated conditions alternate, or if oxygen and sulphides are simultaneously present. Attention has been focussed on the cupronickels2I".245. 246 , which have been shown to have extremely low long-term corrosion rates in quiet or slowly-moving seawater2I9.The 90/10 alloy220.221* 246 is of particular interest: 245 and has become widely used it is well established for pipelines on ships222* for piping systems on offshore platforms221* 223. 224. 245. In addition to its good corrosion resistance, 90/ 10 cupronickel is resistant to marine macrofouling219. 225. 244. 245 (providing it is not cathodically protected). This has led 226. 2279 228* 244- 245, to proposals for uses such as construction of ships' hulls223* fish cages for aquaculture233* 227* 229 and cladding of steel offshore structures in the tidal/splash zones244.245. Large-scale use of copper alloys in desalination plant (particularly multi-stage flash units) has continued and much information has appeared on the selection of materials and their performance in service230* '-. Investigations into the effects of arsenic and phosphorus in single-phase brasses on their susceptibility to intergranular attack and stress-corrosion cracking in seawater231have shown that the normal addition of arsenic to 216s
COPPER A N D COPPER ALLOYS
4:65
inhibit dezincification (about 0.04%) has no significant adverse effect 232. Other problems investigated have included de-alloying of aluminium bronzes233* 244, effects at bimetallic contacts 234 and influence of siphonic effects Major efforts have been made to understand the nature of films formed on copper alloys in seawater. These films have quite different characteristics236. 244. 245 to those formed in sodium chloride solution237,the differences being associated with the presence of organic material in the natural environment. Protective films often have a duplex structure, with an outer layer rich in iron providing impingement resistance and an inner layer giving chemicaVelectrochemica1 protection. With aluminium brass a colloidal mixed hydroxide inner layer provides a buffering action; with cupronickel, however, there is a chloride-rich layer which strongly inhibits the cathodic reaction. However, the structure of the films is affected by variables such as water velocity, temperature and oxygen content. The polarisation resistance technique has been used to evaluate the films formed on condenser tubes in service 238.
’”.
Other Topics Methods of avoiding pitting failures in copper cold-water tubes have been further studied’”. Many hot-forged brass water fittings are now made from modified alloys that have an CYOstructure during forging and are then heattreated to a dezincification-resistantCY structure’“. The corrosion resistance of fl aluminium brasses (shape memory effect alloys) has been studiedz4’. Stress corrosion of brasses continues to form the subject of much research242,as does the effect of inhibitors in various circumstance^^^^. Microbiologically-induced corrosion of copper water pipes in institutional buildings has been reported from several countries. The results of research, leading to remedial measures, have been summarised by G e e ~ e y ~ ~ ’ . P. T. GILBERT Acknowledgement This section is based on the article ‘Chemical Properties and Corrosion Resistance of Copper and Copper Alloys’ which formed Chapter XVIII of the American Chemical Society Monograph No. 122,
Copper: The Science and Technology of the Metal its Alloys and Compounds, edited by Professor Allison Butts, and published by the Reinhold Publishing Corporation, New York, in 1954. Acknowledgement is hereby made to the Reinhold Publishing Corporation for permission to use the above-mentioned section as a basis for the present chapter.
REFERENCES 1 . Copper Alloy Ingots and Copper and Copper Alloy Castings, BS 1400 (1973) 2. Specifications for Copper and Copper Alloys: BS 2870, Sheet. Strip und Foil ( 1 980); BS 287 I , Tubes ( 1 97 1); BS 2872, Forging Stock and Forgings (1989); BS 2873, Wire (1969); BS 2874, Rods and Sections (1986); BS 2875, Plate (1969)
4:66
COPPER AND COPPER ALLOYS
3. Copper and Copper Alloys, American Society for Testing and Materials, Philadelphia (1992) 4. Standard Handbook- Copper, Brass, Bronze, Wrought Mill Products, 8th edn, Copper Development Association Inc., New York (1985); ‘Properties and Applications of Wrought Coppers and Copper Alloys’, Metal frog., 98, 85 (1970) 5 . ‘Wrought Cupro-Nickels’, Mater. in Des. Engng., 49, 127 (1959); Katz, W., Werkst. u. Korrosion, 15, 977 (1964) 6. ‘Silicon Bronzes’, Mater. in Des.Engng., 50, 112 (1959) 7. Evans, U. R., Metallic Corrosion Passivity and Protection, Edward Arnold, London (1946); Evans, U. R., The Corrosion and Oxidation of Metals, Edward Arnold, London (1960) 8. Uhlig, H. H. (Ed.), Corrosion Handbook, Wiley, New York and Chapman and Hall, London ( 1948) 9. LaQue, F. L. and Copson, H. R., Corrosion Resistance of Metals and Alloys, 2nd edn, Reinhold, New York (1963) IO. Speller, F. N., Corrosion, Causes and Prevention, McGraw-Hill, New York (1951) 1 I . Burghoff, H. L., Corrosion of Metals, Amer. SOC.Metals Monograph, Cleveland, Ohio, 100-130 (1946) 12. Mefals Handbook, Amer. SOC.Metals, Metals Park, Ohio, 1, 983-1005 (1961) 13. Rogers, T. Howard, Marine Corrosion Handbook, McGraw-Hill Co. o f Canada Ltd., Toronto (1960) 14. Rogers, T. Howard, Marine Corrosion, Butterworths, London (1968) 1 5 . Cairns, J. H. and Gilbert, P. T., The Technology of Heavy Non-Ferrous Metals and Alloys, Butterworths, London (1967) 16. Anon., Corrosion, 15, 199t(1959); 16, 131t(1960); Lasko, W. R. and Tice, W. K., Corrosion, 18, 116t (1962); Erdos, E., Werkst. u. Korrosion, 19, 385 (1968) 17. Latimer, W. M., Oxidation Potentials, Prentice-Hall, New Jersey (1952) 18. Catty, 0. and Spooner, E. C. R., Electrode Pofential Behaviour of Corroding Metals in Aqueous Solufions, Clarendon Press, Oxford (1938) 19. Pourbaix, M. J. N., Thermodynamics of Dilute Aqueous Solutions (Trans. by J . N. Agar), Edward Arnold, London (1949) 20. Pourbaix, M. J. N., Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon, Oxford (1966) 21. RubiniC, L. and MarkoviC, T., Werkst. u. Korrosion, 10, 666 (1959) 22. Hickling, A. and Taylor, D., Trans. Faraday Soc., 44, 262 (1948) 23. Feitknecht, W. and Lehnel, H. W., Helv. Chim. Acta.. 27,775 (1944); Wilde, B. E. and Teterin, G. A., Brit. Corrosion J., 2, 125 (1%7) 24. Bonhoeffer, K. F. and Gerischer, J., Z. Elektrochem., 52, 149 (1948); Stolica, N. D. and Uhlig, H. H., J. Electrochem. SOC.,110, 1215 (1963); Mansfield, F. and Uhlig, H. H., J. Electrochem. SOC., 117, 427 (1970); Varenko, E. S. el al., Zashchita Mefallov., 6 , 103 ( 1970) 25. Vernon, W. H. J., J. Insf. Met., 49,153 (1932) 26. Freeman, J. R. and Kirby, P. H., Metals and Alloys, 5, 67 (1934) 27. Vernon, W. H. J. and Whitby, L., J. Inst. Met., 42, 181 (1929); 44, 389 (1930); 49, 153 (1932); Vernon, W. H. J., J. Chem. Soc., 1853 (1934) 28. Vernon, W. H. J., Trans. Faraday SOC.,27, 255, 582 (1931) 29. Evans, U. R., Trans. Electrochem. Soc., 46,247 (1924) 30. Hudson, J. C., Trans. Faraday SOC., 25, 177 (1929); J. Inst. Met., 44, 409 (1930); J . Birmingham Met. Soc., 14, 331 (1934); Metal Ind., 44, 415 (1934); J. Insf. Met., 56, 91 (1935) 31. Tracy, A. W., Thompson, D. H. and Freeman, J. R.. Special Technical Publication No. 175, A.S.T.M., 77-87 (1955) 32. Tracy, A. W., STP No. 175, A.S.T.M., 67-76 (1955) 33. Thompson, D. H.. Metal Corrosion in the Afmosphere, STP 435, A.S.T.M., 129 (1968) 34. Mattsson, E. and Holm, R., Metal Corrosion in the Atmosphere, STP 435, A.S.T.M., 187 (1 968) 35. Scholes, I. R. and Jacob, W. R., J. Inst. Met., 98, 272 (1970) 36. Compton, K. G., Trans. Electrochem. Soc., 91, 705 (1947); Ambler, H . R. and Bain, A. A. J., J. Appl. Chem., 5,437 (1955); Forgeson, B. W. et a/., Corrosion, 74,73t (1958); Hummer, C. W., Jr., Scuthwell, C . R. and Alexander, A. L., Mater. Protection, 7 No. I , 41 (1968)
COPPER AND COPPER ALLOYS
4:67
37. Cole, H. G., R.A.E. Report MET82 (1954); Compton, K. G., Mendizza, A. and Bradley, W. W., Corrosion, 11, 383t (1955) 38. Romanoff, M., UndergroundCorrosion, Nat. Bur. Stand. Circ. 579, Supt. of Documents, Washington, D.C. (1957) 39. Gilbert, P. T., J . Inst. Met., 73, 139 (1947) 40. Gilbert, P. T. and Porter, F. C., Iron and Steel Inst. Special Report No. 45, 55-74, 127-134 (1952) 41. MarkoviC, T., SevdiC, M. and Rubinif, L., Werkst. u. Korrosion, 11, 87 (1960) 42. Bengough, G. D. and May, R., J. Insr. Met., 32, 81 (1924) 43. May, R., J. Inst. Met., 40, 141 (1928) 44. May, R. and de V. Stacpoole, R. W., J. Inst. Met., 77, 331 (1950) 45. LaQue, F. L . and Stewart, W. C., Milaux et Corros., 23, 147 (1948); LaQue, F. L., Proc. Amer. SOC. Test. Muter., 52, 1 (1952) 46. Gilbert, P. T. and LaQue, F. L., J. Electrochem. SOC., 101, 448 (1954) 47. Fink, F. W., Trons. Electrochem. SOC.,75, 441 (1939); Evans, U. R., ibid., 446 48. Lusey, V. F., Brit. Corrosion J., 1, 9 and 53 (1965) 49. Feller, H-G., 2. Metollkunde, 58, 875 (1967) 50. Heidersbach, R., Corrosion, 24, 38 (1968) 51. Polushkin, E. P. and Shuldener, H. L., Trons. Amer. Inst. Min. (Metoll.) Engrs., 161, 214 (1945); Kleinberger, R., Okuzumi, H. and Perio, P., MProux (Corrosion-lnd.), No. 413, 40-43 (1960); Kenworthy, L. and O'Driscoll, W. G., Corrosion Tech., 2, 247 (1955); Piatti, L. andGrauer, R., Werksf.u. Korrosion. 14, 551 (1963); Hashimoto, K., Ogawa, S. and Shimodaira, S., Trons. Jopon Inst. Met., 4.42 (1963); Robinson, F.P.A. and Shalit, M.,Corrosion Tech., 11, 1 1 (1964); Rowlands, J. C., Proc. 2nd Internat. Cong. Met. Corrosion, New York, 1%3, N.A.C.E., Houston, 795 (1966); Frade, G. and Lacombe, P., Mdm. Sci. Rev. Mktoll., 63,649 (1966); Sugawara, H. and Ebiko, H., Corrosion Sci., 7, 513 (1967); Joseph, G. and Arce, M.T.. Corrosion Sci., 7 , 597 (1967); Langenegger, E. E. and Robinson, F.P.A., Corrosion, 24, 41 1 (1968) and 25, 59 (1969); Horton, Ralph M., Corrosion, 26, 160 (1970); Rothenbasher, P., Corrosion Sci., IO, 391 (1970); Potzl, R. and Lieser, K. H., Z. Metollkunde, 61, 527 (1970) 52. Bengough, G. D., Jones, R. M. and Pirret, R., J . Insf. Met., 23, 65 (1920) 53. Upton, B., Brit. Corros. J., 1, 134 (1966) 54. May, R., Trans. Inst. Mor. Engrs., 49, 171 (1937); Sherborne, H. F. (p. 76), Bailey, G. L. (p. 78). Discussion of Bradbury, E. J. and Johnson, L. W., Trans. Inst. Mor. Engrs., 63, 59 (1951) 5 5 . Breckon, C. and Gilbert, P. T.. Chem. and Ind., Jan. 4, 35 (1964) 56. Bem, R. S., The Engineer, 206, 756 (1958) 57. Clark, W. D., J. Inst. Met., 73, 263 (1947) 58. Breckon, C. and Gilbert, P. T., Proc. 1st Internot. Cong. on Met. Corrosion, London, 1961. Butterworths, London, 624 (1962) 59. Gleekman, L. W. and Swanby, R. K., Corrosion, 17, 144t (1961); Rowlands. J. C., Corrosion Sci., 2, 89 (1%2); Smith, A. A., Corrosion Prev. ond Control, IO, 29 (1963); Upton, B., Corrosion, 19, 204t (1963); Piatti, L. and Grauer, R., Werksf. u. Korrosion, 14, 551 (1963); Neiderberger, R. B., Modern Castings, 45 No. 3, 115 (1964); Gaillard, F. and Weill, A. R., Mem. Sci. Rev. Met., 61, 437 (1964); Maersch, R. E. and Ciesleiwicz, J. M., Muter. Protection, 3 No. 7, 54 (1964); Shinoda, G. and Amano, Y., Trans. Japan Inst. Mer., 4, 231 (1%3); Tanabe, Z., Corrosion Sci., 4, 413 (1964); Arnaud, D. etol., Fonderie, No. 226, 403 (1%4); Arnaud, D., Fonderie, No. 275, 88 and No. 281, 355 (1969); Shibard, P. R. and Balachandra, J., Anti Corrosion Methods Muter., 14 No. 2, 10 (1967) 60. Gilbert, P. T. and May, R., Trans. Insf. Mor. Engrs., 62, 291 (1950); Gilbert, P. T., Trons. Inst. Mor. Engrs., 66, 1 (1954); Breckon, C. and Baines, J. R. T., Trons. Inst. Mor. Engrs.. 67, 1 (1955); Bradbury, E. J. and Johnson, L. W., Trons. Inst. Mor. Engrs., 63, 59(1951); Slater, 1. G., Kenworthy, L. and May, R., J. Inst. Met., 77,309 (1950); Bethon, H. E., Corrosion, 4, 457 (1948); Eichhorn, K., Werkst. u. Korrosion, 8 , 657 (1957) and 21,535 (1970); Nothing, F. W., Metoill, 10,520(1956)and 16, 1089, 1196(1962); Maurin, A. J., Corros. ond Anti-Corros., 5 , 275, 383 (1957) and 6, 15 (1958); Todhunter, H. A., Corrosion, 11, 221t (1955) and Power, 100, 85 (1956); Gilbert, P. T., Chem. ond Ind., July 1 I , 888 (1959); Nowlan, N. V., Corrosion Tech., 7 , 397 (1960); Otsu, T., Sumitomo Light Metal Tech. Rep., 1 No. 1, 62 (1960); Gilbert, P. T., Inst. Mor. Engrs. Materials Section Symposium, March, 14(1968); Gilbert, P. T., Trans. Inst. Mar. Engrs., 82 No. 7 .
4:68
COPPER AND COPPER ALLOYS
6 (1970); Kingerley, D. G., Brit. Chem. Engng., 6 No. 1.20 (1961); Sisson, A. B., Corrosion, 17,18 (1961); Hall, B. N., Corrosion Prev. and Control, 10, 49 (1963); Malcolm, R. R.. Australasian Corros. Engng., 7 No. 3, 17 and 7 No. 10, 25 (1963); Serre, J. and Laureys, J . , Corrosion et Anti-Corrosion, 11, 305, 360 (1963) and Corrosion Sci., 5, 135 (1965); Kenworthy, L., Truns. Inst. Mar. Engrs., 77, 149 (1965); Page, G. C., AntiCorrosion Methods Mater., 14 No. 5, 13 (1967); Yandushkin, K. N., ZashchitaMetallov., 6, 46 ( 1 970) 61. Tracy, A. W., Corrosion, 1, 103 (1945); Mitchell, N. W., Corrosion, 3, 243 (1947); Van der Baan Sj., Corrosion, 6, 14 (1950); Mason, J. F., Corrosion, 12, 199t (1956); Rust, A. D., Corrosion Tech., 3,185 (1956); Gilbert, P. T., SOC. Chem. Ind. Monograph, No. 10, 1 1 1-120 (1960); Bird, D. B. and Moore, K. L., Mater. Protection, 1 No. 10, 70 (1962) 62. Bengough. G. D., J. Inst. Met., 5.28 (1911); Bengough, G. D. and Jones, R. M.,J. Inst. Met., 10, 13 (1913); Gibbs, W. E., Smith, R. H. and Bengough, G. D., J. Inst. Met., 15, 37 (1916); Bengough, G. D. and Hudson, 0. F., J. Inst. Met., 21, 167 (1919) 63. Bailey, G. L., J. Inst. Met., 79,243 (1951); Tracy, A. W. and Hungerford, R. L., Proc. Amer. SOC. Test. Mater., 45, 591 (1945); LaQue, F. L. and Mason, J . F., Proc. Arner. Petrol Inst., 30M 111, 103 (1950); Todhunter, H . A., Corrosion, 16, 226t (1960) and Mater. Protection, 6 No. 7, 45 (1967); LaQue, F. L. and Stewart, W. C., Corrosion, 8, 259 (1952); Krafack, K. and Franke, E., Werkst. u. Korrosion, 4, 310 (1953); Se Ui Y u and Turkovsaya, A. V., TsvetnayaMetall., 4, 145 (1961); May, T. P. and Weldon, B. A., Rev. Nickel, No. 3, 183 and No. 4, 219 (1966) 64. Chapman, J. and Cuthbertson, J. W., J . SOC.Chem. Ind., 58, 100, 330 (1939); Cuthbertson, J . W., J. Inst. Met., 72, 317 (1946) 65. Tanabe, Z., Sumitomo Light Metal Tech. Rep., 9 No. 3, 167 (1968) 66. Baker, L., Trans. Insf. Mar. Engrs., 65, (1953) 67. Rogers, T . Howard, J. Inst. Met., 75, 19 (1948-49) 68. Sato, S., Sumitomo Light Metal Tech. Rep., 6 No. 1, 42 (1965); Tanaka, R., Sumitomo Light Metal Tech. Rep., 3 No. 3, 55 and 3 No. 4, 1 (1962) and 6 No. 3, 152 (1965); Changarnier, J . , Corros. Anti-Corros., 1, 8 (1953) 69. Shimodaira, S., Sugawara, H. and Sato, S., Sumitomo Light Metal Tech. Rep., 4 No. 1, 31 (1963); Tanabe, Z., Sumitomo Light Metal Tech. Rep., 5 No. I , 16 (1964) 70. Otsu, T. and Sato, S., Sumitorno Light Metal Tech. Rep., 2 No. 2, 23 and 2 No. 4, 27 (1961); Otsu, T. and Okawa, M., Sumitomo Light Metal Tech. Rep., 3 No. 3 , 35 (1962); Otsu, T., Sato, S. and Watanabe, T., Sumitomo Light Metal Tech. Rep., 4 No. 2. 21 (1963); Tanaka, R. and Tanabe, Z., Sumitomo Light Metal Tech. Rep., 5 No. I , 9 (1964); Tanaka, R., Sumitomo Light Metal Tech. Rep., 5 No. 3, 188 (1964) and 6 No. 1, 71 (1965); Sato, S. and Sagiska, K., Sumitamo Light Metal Tech. Rep., 11 No. 2, 1 (1970) 71. Tanabe, Z., Sumitomo Light Metal Tech. Rep., 6 No. 2, 119 (1965) 72. Sato, S., Proc. 4th Int. Cong. Met. Corrosion Amsterdam (1969). Nat. Assoc. Corrosion Eng., Houston, 795 (1972) 73. Rowlands, J. C., J. Appl Chem., 15, 57 (1965) 74. Grubitsch, H., Hilbert, F. and Sammer, R., Werkst. u. Korrosion, 17, 760 (1966); Tanabe, 2.. Sumitomo Light Meld Tech. Rep., 8 No. 1, 10 (1967); Meany, J . J . , Jr., Mater. Protection, 8 No. 10, 27 (1969); Mor, E., Scotto, V. and Trevis, A., Corrosion {Paris), 18 No. 2, 67 (1970); Baudo, G. and Giuliani, L., Werkst. u. Korrosion, 21, 332 (1970); North, R.F. and Pryor, M. J., Corrosion Sci., 9, 509 (1969) and 10, 297 (1970); Giuliani, L. and Bombara, G., Brit. Corrosion J . , 5, 179 (1970) 75. McAllister, R. A., etal., Corrosion, 17, 579t (1961); Erdos, E., Schweizer Archiv., 30, 251 (1964); Mifflin, R. C. and Bird, D. B., Muter. Protection, 8 No. 9, 72 (1969) 76. Sato, S., Sumitomo Light Metal Tech. Rep., 3 No. 3, 106 (1962) 77. Sherry, A. and Gill, E. R., Chem. and Ind., Jan. 18, 102 (1964); Edwards, B. C., Corrosion Sci., 9, 395 (1969) 78. Gilbert, P. T., Chem. and Ind., July 11, 888 (1959) 79. Sato, S., Nagata, K. and Ogiso, A., Sumitorno Light Metal Tech. Rep., 11 No. 3 , 1 (1970) 80. Peplow, D. B., Brit. Power Engineering, 1 No. 5, 51 (1960); Attwood, P. G. and Richards, N. G., Corrosion, 17, 8t (1960); Crennel, J. T . and Sawyer, L. J . E., J. Appl. Chem., 12, 170 (1962); Page, G. G., Proc. 2nd Intermat. Cong. Met. Corrosion, New York (1963). Nut. Assoc. Corrosion Eng., Houston, 275 (1966) 81. Bostwick, T. W., Corrosion, 17, 12 (1961) 82. Lockhart, A. M., Proc. Inst. Mech. Engrs., 179, 495 (1964-65)
COPPER A N D COPPER ALLOYS
4:69
83. North, R. F., Corrosion Sci., 8, 149 (1968); Gasparini, R., Della Rocca, C. and Ioannilli, E., Corrosion Sci., 10, 157 (1970) 84. Proceedings of Conference on the Role of Copper and its Alloys in Desalination Equipment, London, December (1966), Copper Development Association. See Stewart, J. M., p. 21, Gilbert, P. T., p. 31, Weldon, B. A. and Tuthill, A. H., p. 39; Tuthill, A. H. and Sudrabin, D. A., Metals Engng. Quart., 7 No. 3, 10 (1967); Fink, F. W., Tech. Rep. 704/6, Copper Development Association, New York (1966) and Mater. Protection, 6 No. 5,40 (1967); Schoraten, A., Metall., 22, 1153 (1968); Cohen, A. and Rice, L., Mater. Protection, 8 No. 12, 67 (1969) and 9 No. 11.29 (1970); Bom, P. R., Brit. Corrosion J., 5, 258 (1970) 85. Campbell, H. S. and Carter, V. E., J. Inst. Metals, 90, 362 (1962); Murphy, T. J. and Jack, J. B., Shipping World and Shipbuilder, Jan. 21, 282 (1965) 86. Gilbert, P. T. and Jenner, B. J., Inst. Marine Engrs. International Marine and Shipping Conference, London, June ( I 969) 87. Campbell, H. S., Chem. and Ind., 692 (1955); Hatch, G. B., J. Amer. Waterworks Assoc., 53, 1417 (1961); N.A.C.E. Tech. Rep. 60-11 and Corrosion, 16, 453t (1960); Schafer, G. J., New Zealand J. Sci., 5, 475 (1962) 88. Tumer, M. E. D., Proc. SOC. Water Treatm. Exam., IO, 162 (1961) and 14, 81 (1965) 89. Baldwin, A. B. and Campbell, H. S., Brit. Waterworks Assoc. J., 43, 13 (1961); Schafer, G. J. and Dall, R. A., Australasian Corrosion Engng., IO No. 3.9 (1966); Ladeburg, H., Metall., 20, 33 (1966); Simmonds, M. A. and Huxley, W. G. S., Australasion Corrosion Engng., 11 No. 11, 9 (1967) 90. Schafer, G. J., Foster, P. K. and Marshall, T., New Zealand J. Sci., 4, 194 ( 1 9 6 1 ) 91. Schafer, G. J . and Dall, R. A., Brit. Corrosion J., 3, 12 (1968); Harrison, P. S., Electrical Times, 153, 219 (1968) 92. Obrecht, M. F., Corrosion, 18, 189t (1962) 93. Ingleson, H., Sage, A. M. and Wilkinson, R., J. Inst. War. Engrs., 3, 81 (1949); Wormwell, F. and Nurse, T. J., 1.Appl. Chem., 2,685 (1952); Solelev, A., J . Inst. Waf.Engrs., 9, 208 (1955) 94. Tronstad, L. and Veimo, R., J . Inst. Met.. 66, 17 (1940); Kenworthy, L., J. Instn. Heat. Vent. Engrs., 8, 15 (1940); Gilbert, P. T., Proc. SOC. Water Treatm. Exam., 15, 165 (1966) 95. Kenworthy, L., J. Inst. Met., 69, 67 (1943) 96. Porter, F. C. and Hadden, S. E., J. Appl. Chem., 3, 385 (1953) 97. Davenport, W. H., Nole, V. F. and Robertson, W. D., J. Electrochem. Soc., 106, 1005 (1959); Ives, D. J. G. and Rawson, A. E., J. Electrochem. SOC., 109,447 (1962); Obrecht, M. F. and Pourbaix, M., J. Amer. Waterworks Assoc., 59, 977 (1967) 98. Grunau, E. B., Stadtehygiene, No. 7, 153 (1967) 99. Campbell, H. S., J. Inst. Met., 77, 345 (1950); J. Appl Chem., 4,633 (1954); Proc. Soc. Wat. Treatm. Exam., 3, lOO(1954) and Proc. 2nd. Internat. Cong. Met. Corrosion, New York (1963), N.A.C.E., Houston, 237 (1966) 100. Schafer, G. J., Australasian Corrosion Engng., 6 No. 8, 15 (1962) and New Zealand Plumbing Rev., 1 No. 9, 10 (1964); Rambow, C. A. and Holrngren, R. S., Jr., J. Amer. Waterworks Assoc., 58, 347 (1966); Pourbaix, M.. Van Muylder, J. and Van Laer, P., Corrosion Sci., 7 , 795 (1967); von Franque, 0.. Werkst. u. Korrosion, 19, 377 (1968); Lihl, F. and Klamet, H., Werkst. u. Korrosion, 20, 108 (1969); Walker, I. K. and Page, G. G., Australasian Corrosion Engng., 13 No. 4, 13 (1969): Kennett, A., Australasian Corrosion Engng., 13 No. 4, 5 (1969); Gilbert, P. T., Australasian Corrosion Engng., 13 No. 5 , 13 (1969) 101. Schafer, G. J. and Dall, R. A., Australasian Corrosion Engng., 7 No. 10, 33 (1963) 102. Lucey, V. F., Brit. Corrosion J., 2 , 175 (1967) 103. Mattsson, E. and Fredriksson, A-M., Brit. Corrosion J., 3, 246 (1968) 104. Lee, J. A., Materials of Construction for Chemical Process Industries, McGraw-Hill, New York (1950) 105. Rabald, E., Corrosion Guide, Elsevier Publishing Co., New York (1951); Carmenisch, K. P., Pro-Metal, 13 No. 73, 288 (1960); Ritter, F., Korrosionstabellen Metallischer Werkstofe, Springer-Verlag. Vienna (1958); LaQue, F. L., Corrosion, 10, 391 (1954); Heim, A. T., Industr. Engng. Chem., 49, 63A, 64A, 66A (1957); Baker, S., Corrosion Tech., 8, 8 (1961); Tracy, A. W., Chem. Engng., 69, 130, 152 (1962); Anon., Industr. Engng. Chem., 40,1827 (1948), 43, 2218 (1951) and 49, 63A (1957) 106. Gould, A. J. and Evans, U. R.,J. Iron St. Inst., 155, 195 (1947)
4:70
COPPER AND COPPER ALLOYS
107. Lacan, M., Markovic, T. and Rubinic, L., Werkst. u. Korrosion, 10,767 (1959) 108. Holmberg, M. E. and Prange, F. A., Industr. Engng. Chem., 37, 1030 (1945);Anon., Indusrr. Chem., 30, 609 (1954);Lingnau, E., Werksr. u. Korrosion, 8,216 (1957) 109. Fontana, M. G., Indusrr. Engng. Chem., 42, 69A (1950) 110. Groth, V. J. and Hafsten, R. J., Corrosion, 10, 368 (1954) 111. Bulow, C. L., Chern. Engng., 53, 210 (1946) 112. Friend, W. Z. and Mason. J. F., Corrosion 5, 355 (1949).N.A.C.E. Report and Corrosion. 13. 757t (1957) 113. Russell, R. P. and White, A., Industr. Engng. Chem., 19,116 (1927);Darnon, G. H. and Cross, R. C., Indusrr. Engng. Chem., 28, 231 (1936) 114. Cornet, I., Barrington, E.A. and Behrsing, G. U., J. Electrochem. Soc., 108,947(1961) 115. Caney, R. J. T., Ausf.Engr., 64,54(1954)and U.K. Pat 718,987;Zitter, H. and Kraxner, G., Werksr. u. Korrosion, 14.80 (1963);Piatti, L. and Fot, E., Werkst. u. Korrosion,15, 27 (1964) 116. Gregory, D. P. and Riddiford, A. C., J. Electrochem. Soc., 107,950(1960);Talati, J. D., Desai, M.N. and Trivedi, A. M., Werksr. u. Korrosion,12,422(1961); Bumbulis, J. and Graydon, W. F., J. Electrochem. SOC., 109, 1130 (1962);Kagetsu, T.J. and Graydon, W. F., J. Electrochem. SOC., 110,856 (1963);Feller, H-G, Corrosion Sci., 8,259 (1968); Otsuka, R. and Uda, M., Corrosion Sci., 9,703 (1969) 117. Rana, S. S. and Desai, M. N., Indian J. Techno/., 5,393 (1967);Desai, M. N. and Shah, Y. C.. Anti-Corrosion Methods Mater., 15 No. 12, 9 (1968);Desai, M.N., Shah, Y. C. and Gandhi, M. H., Corrosion Sci., 9,65 (1969);Padma, D. K. and Rama Char, T. L., Anti-CorrosionMethodsMafer., 16No.4(1969);Desai, M. N., Shah, Y. C.and Punjani, B. K., Brit. Corrosion J., 4, 309 (1969) 1 18. Ammar, I. A. and Riad, S., Corrosion Sci., 9,423 (1969) 119. Anon., Proc. Amer. SOC. Test. Mater.. 35, 161 (1935);Desai, M. N. and Rana, S. S., Werksr. u. Korrosion, 17, 870 (1966) 120. Radley, J. A., Stanley, J . S. and Moss, G. E., Corrosion Tech., 6,229 (1959);Schaefer, B. A., Corrosion Sci., 8,623 (1968);Bartonifek, R., Holinka. M. and LukaSovska, M., Werkst. u. Korrosion, 19, 1032 (1968);Green, J. A. S., Mengelberg, H. D. and Yolken, H. T.,J. Electrochem. SOC.,117,433 (1970);Jenkins, L. H. and Durham, R. B., J. Elecrrochem. SOC., 117, 768 (1970) 121. Dubrisay, R. and Chesse, G., Compt. Rend. Acad. Sci. Paris, 220, 707 (1945) 122. Reichert, J. S. and Pete, R. H., Chem-Engng., 54, 218 (1947) 123. West, J. R., Chem. Engng., 58, 281 (1951) 124. Botharn, G. H. and Dummett, G. A., J. Dairy Res., 16, 23 (1949) 125. Holness, H. and Ross, T. K., J. Appl. Chem., 1, 158 (1951);Bukowiecki, A., Schweizer Archiv. Angew. Wiss., 24, 355 (1958) 126. Mason, J. F., Corrosion, 4, 305 (1948);Inglesent, H. and Storrow, J. A., J. SOC.Chem. Ind., 64,233(1945);Clendenning, K. A., Canad. J. Res. F(TechnologyJ, 26,277 (1948) 127. Review of Oxidation and Scaling of Heated Metal Solids, D.S.I.R., H.M.S.O., London (1935); Vernon, W. H. J . , Chem. and Ind. (Rev.), 59, 87 (1940) 128. Tylecote, R. F., J. Inst. Met., 78, 259 (1950-51);Cabrera, N. and Mott, N. F., Rep. Progr. Phys., 12, 163 (1948-49);Ronnquist, A. and Fischmeister, H., J. Inst. Mer., 89, 65 (1960-61) 129. Kubaschewski, 0.and Hopkins, B. E., Oxidation of Metals and Alloys, Butterworths, London (1953); Hauffe, K., Oxydation von Metallen und Legierungen, Springer-Verlag, Berlin (1956) 130. Wagner, C. and Grunewald, K., Z. Phys. Chem., 40,455 (1938B) 131. Dighton, A. L. and Miley, H. A., Trans. Electrochem. SOC., 81, 321 (1942) 132. Evans, U. R., Symposium on Internal Stresses in Metals and Alloys, Inst. Metals, London, 291-310 (1947). Trans. Electrochem. SOC., 91, 547 (1947)and Research, London, 6, 130 (1953) 133. Mott, N. F., Trans,Farada~Soc.,35.1175(1939);36,472(1940);43,429(1947); Nature, London, 145, 996 (1940);J. Chim. Phys., 44, 172 (1947);Research, London, 2, 162 (1949);Price, L. E., Chem. and Ind. (Rev.), 56, 769 (1937) 134. deCarli, F. and Collari, N., Chim. el Industr., 33,77 (1951);McKewan, W.and Fassell, W. M., J. Metals. N. Y . , 51, 1127 (1953);Paidassi, J., AcraMetallurgica, 6, 216 (1958); Lohberg, K. and Wolstein, F., 2. Metallk., 46,734 (1955);Baur, J. P., Bridges, D. W. and Fassell, W. M., J. Electrochem. Soc., 103,273 (1956);Gulbrausen, E. A., Copan, T. P . and Andrew, K. F., J. Electrochem. SOC., 108, I19 (1961); Ronnquist, A., J. Insf.
COPPER AND COPPER ALLOYS
135. 136. 137. 138. 139. 140. 141. 142. 143. 144.
145. 146. 147.
148.
149. 150. 151. 152. 153. 154. 155.
156. 157. 158. 159. 160.
161. 162. 163. 164.
165. 166.
4:71
Met., 91, 89 (1962); Yoda, E. and Siegel, B. M., J . Appl. Phys., 34, 1512 (1963); Wallwork, G. R. and Smeltzer, W. W., Corrosion Sci., 9. 561 (1%9) Tylecote, R. F., J. Inst. Met., 78, 327 (1950-51) and 81, 681 (1952-53) Bardeen, J., Brattain, W. H. and Shockley, W., J . Chem. Phys., 14, 714 (1946); Castellan, G. W. and Moore, W. J., J. Chem. Phys., 17, 41 (1949) Pilling, N. B. and Bedworth. R. E., J. Inst. Met., 29, 529 (1923) Feitknecht, W., Z. Elektrochem., 35, 142, 500 (1929) Vernon, W. H. J., J. Chem. SOC.,2273 (1926) Valensi, G., Pittsburgh International Conference on Surface Reactions, Corrosion Publishing Co., Pittsburgh, 156-165 (1948) Tylecote, R. F., J. Inst. Met., 78, 301 (1950-51) Oudar, J., M&aux, 35, 397, 445 (1960) Dyess, J. B. and Miley, H. A., Trans. Amer. Inst. Min. (Metall.) Engrs., 133,239 (1939); Vernon, W. H. J., Trans. Faraday Soc., 19, 839 (1924) Preston, G. D. and Bircumshaw, L. L., Phil. Mag., 20, 706 (1935) Hallowes, A. P. C. and Voce, E., Melallurgia, Manchr., 34, 95 (1946) Lustman, B., Metal. Prog., 50, 850 (1946) Dennison, J. P. and Preece, A., J. Inst. Met., 81, 229 (1952-53); Spinedi, P., Metallurg. Ital., 45, 457 (1953); Collari, N. and Spinedi, P., Metallurg. Ital., 46, 403 (1954); Blade, J. C. and Preece, A., J. Inst. Met., 88,427 (1959-60); Maak, F. and Wagner, C., Werkst. u. Korrosion, 12, 273 (1961); Wallbaum, H. J., Werkst. u. Korrosion, 12, 417 (1961); Maak, F., Z. Metallkunde, 52, 538 (1961); Zwicker, U., Metall., 16, 1110 (1962); Kapteijn, J., Couperus S. A. and Meijering, J. L., AcraMetull., 17, 1311 (1%9); Sanderson, M. D. and Scully, J. C., Corrosion Sci., 10, 165 (1970) Dunwald, H. and Wagner, C., 2. Phys. Chem., 22, 212 (1933B); Wagner, C., 2. Phys. Chem., 21, 25 (19338). Pittsburgh International Conference on Surface Reactions, Corrosion Publishing Co., Pittsburgh, 77-82 (1948); Hoar, T. P. and Price, L. E., Trans. Faraday Soc., 34, 867 (1938) Price, L. E. and Thomas, G. J., J. Inst. Med., 63, 21 (1938) Schiickher, F. and Lampe, V., Pro-Metal., No. 105, 192 (1965) Wood, G. C. and Chattopadhyay, B., J. Inst. Met., 98, 117 (1970) Whittle, D. P. and Wood, G. C., J. Inst. Met., 96, 115 (1968) and Corrosion Sci., 8, 295 (1968) Swaroop, B. and Wagner, J. B., Jr., J . Electrochem. Soc., 114, 685 (1967) Rhines, F. N., Corros. Mat. Prot., 4, 15 (1947); Ashby, M. F. and Smith, G. C., J. Inst. Met., 91, 182 (1963); Bolsaitis, P. and Kahlweit, M., Acta Metall., 15, 765 (1967); Potschke, J., Mathew, P. M. and Frohberg, M. G., Z. Metallkunde, 61,152 (1970) Moore, C. and Bindley, D., Proc. 2nd Internal. Cong. Met. Corros., New York (1%3), Nat. Assoc. Corros. Eng., Houston, 391 (1966); Castle, J. E., Harrison, J. T. and Masterson, H. G., ibid, 822; Hopkinson, B. E., A.S.M.E., Paper No. 62-WA274 (1%2); Wiedersum, G. C. and Tice, E. A., A.S.M.E., Paper No. 64-WA/CT-3 (1964); Castle, J. E., Harrison, J. T. and Masterson, H. G., Brit. Corros. J . . 1, 143 (1%) Otsu, T. and Sato, S . , Trans. Jap. Inst. Mer., 2, 153 (1961); Sato. S., Sumitorno Light Metal Tech. Rep., 5 No. 1 , 2 (1964); 5 No. 2.27 (1964); 5 No. 3,231 (1964); 5 No. 4. 290 (1964); Brush, E. G. and Pearl, W. L., Corrosion, 25, 99 (1969) Moore, H., Beckinsale, S. and Mallinson, C. E., J. Inst. Met., 25. 35 (1921); Moore, H. and Beckinsale, S., J. Insf. Mer., 23, 225 (1920) Bibliography on Season Cracking, Proc. Amer. SOC.Test. Mater., 41, 918 (1941) Symposium on Stress-corrosion Cracking in Metals, Amer. SOC.Test. Mater.-Amer. Inst. Min. (Metall.) Engrs., Philadelphia (1944) Nelson, G. A., Bull. Amer. SOC. Test. Mat., No. 240, 39 (1959) Robertson, W. D., Trans. Amer. Inst. Min. (Metall.) Engrs., 191, 1190 (1951) Edmunds, G., Ref. 159, pp. 67-89 Wilson, T. C., Edmunds, G., Anderson, E. A. and Peirce, W. M.. Ref. 159, pp. 173-193 de Jager, W. G. R., Metall., 4, 138, 185 (1950); Steinle, H., Metall., 9, 492 (1955); Kamath, K. V., FreibergerForschungshefre, B56(1%1); Lihl, F. and Hutter, H., Metall., 21, 884 (1967); Sato, S. and Nosetani, T., Sumitorno Light Metal Tech. Rep., 10 No. 2, 83 (1969); Syrett, B. C. and Parkins, R. N., Corrosion Sci., IO, 197 (1970) Ref. 8, pp. 78-79 Sato, S., Sumitorno Light Metal Tech. Rep., 1 No. 3. 45 (1960); Kamath. K. V. and
4:72
COPPER AND COPPER ALLOYS
Erdmann-Jesnitzer, F., Metall., 14, 1061 (1960); Thompson, D. H., Chem. Engng., 68 No. 3, 130 (1961); Laub, H., Metall., 20, 1174 (1966); Adamson, K., Corrosion Sci., 7, 537 (1967); Erdmann-Jesnitzer, F. and Kaeslingk, N., Werkst. u. Korrosion, 20, 493 (1969); Sabbadini, L., Metallurgia Italiana, 62, 228 (1970) 167. Edmunds, G., Anderson, E. A. and Waring, R. K., Ref. 159, pp. 7-18; Bulow, C. L., Ref. 159, pp. 19-35; Jamieson, A. L. and Rosenthal, H., Ref. 159, pp. 36-46; Hellsing, S., Lissner, 0.. Rask, S. and Strom, B., Werkst. u. Korrosion, 8, 569 (1957); Aebi, F.. Z. Metaelk., 49, 63 (1958); Thompson, D. H., Mater. Res. Standards, I, 108 (1961); Szabo, E., Werkst. u. Korrosion, 14, 162(1963); Mattsson, E., Lindgren, S., Rask, S. and Wennstrom, G., Current Corrosion Research in Scandinavia, Kemian Keskusliitto, Helsinki, 171 (1965) 168. Breckon, C. and Gilbert, P. T., Metallnd., 93.89, I14 (1958); Sinclair, N. A. and Albert, H. J., Mater. Protecrion, 1 No. 3, 35 (1962); Fox, D. K., Modern Castings, 42 No. 6, 51 (1962); Baumann, G., Werkst. u. Korrosion, 13, 737 (1962); Sato, S., Sumitorno Light Metal Tech. Rep., 4 No. 1, 48 (1963); Uhlig, H. H. and Sansone, J., Mater. Protecrion, 3 No. 2, 21 (1964); Peters, B. F., Carson, J. A. H.and Barer, R. D, Mater. Protection, 4 No. 5, 24 (1965); Logan, H. L. and Ugiansky, G. M.,Mater. Protection, 4 No. 5 , 79 (1965); Serre, J., Corrosion et Anti-corrosion, 14 No. 1, 9 (1966) 169. Laub, H., Metall.,20,597 (1966)and21,173 (1967); Laub, H., Metalloberpirche,20,413, 453, 493 (1966) 170. Sato, S. and Nosetani, T., Sumitomo Light Metal Tech. Rep., 10 No. 3, 175 (1969) 171. Whitaker, M. E., Metallurgia, Manchr., 39, 21, 66 (1948) 172. Perryman, E. C. W., J. Inst. Met., 83,369 (1954-55); Perryman, E. C. W. and Goodwin, R. J., J. Inst. Met., 83, 378 (1954-55) 173. Bailey, A. R., Metallnd., 80, 519 (1952) and J. Inst. Met., 87, 380 (1959) 174. Sheehan, T . L. and Dickerman, H. E., J. Amer. SOC. Nav. Engrs., 58, 586 (1946) 175. Pugh, E. N., Craig, J. V. and Montague, W. G.. A.S.M. Trans. Quart., 61, 468 (1968) 176. Mattsson, E., Electrochimica Acta, 3, 279 (1961) 177. Forty, A. J . and Humble, P., Phil. Mag., 8 No. 86,247 (1963); Proc. 2nd Internat. Cong. Met. Corrosion, New York (1963). N.A.C.E., Houston, 80 (1966); McEvily, A. J., Jr. and Bond, A. P., J. Electrochem. Soc., 112, 131 (1965) 178. Dix, E. H., Jr., Proc. Amer. Soc. Test. Mafer., 41,928 (1941); Read, T. A,, Reed, J. B. and Rosenthal, H., Ref. 159, 90-1 IO; Chaston, J . C., Sheet Metal Ind., 24, 1395 (1947); Ziirrer, T., Pro-Mefal., 13, 307 (1960); Forty, A. J., Mefol Progr., 75, 154 (1959); Graf, L. and Budki. J., Z. Metallk., 46, 378 (1955); Graf, L., Corros. Anti-Corros.. 6, 151 (1958); Graf, L. and Lacour, H . R., 2.Metalik., 51, 152 (1960) and 53,764 (1962); Graf, L. and Richter, W., Z. Metallk., 52,834(1961); Aebi, F., 2. Metallk., 46, 547 (1955)and 47,421 (1956); Logan, H. L., J. Res. Nat. Bur. Stand., 48.99 (1952) and 56. I59 (1956); Bakish, R. and Robertson, W. D., J. Electrochem. SOC., 103, 320 (1956); Edeleanu, C. and Forty, A. J., Phil. Mag., 5 No. 58, 1029 (1960); Graf, L., Proc. 2nd Internat. Cong. Met. Corrosion, New York (1963). N.A.C.E., Houston,.89 (1966) and Metall., 18, 1163, 1287 (1964); Lynes, W., Corrosion, 21, 125 (1965); Ohtani, N. and Dodd, R. A., Corrosion, 21, 161 (1965); Pugh, E. N. and Westwood, A. R. C., Phil. Mag., 13. 167 (1966); Pugh, E. N., Montague, W. G. and Westwood, A. R. C., A.S.M. Trans. Quart., 58,665 (1965); Hoar, T. P. and Booker, C. J. L., Corrosion Sci., 5,821 (1965); Fairman, L., Corrosion Sci., 6, 37 (1966); Tanabe, Z., Sumitorno Light Metal Tech. Rep., 7 No. 3, 137 (1966); Takano, M. and Shimodaira, S., Trans. Japan Inst. Met., 8,239(1%7)and CorrosionSci..8, 55(1%8);Murakami,Y.andIkai,Y., Trans. Japanlnsl. Met.,8,246(1967); Lahiri, A. K., Brit. Corrosion J., 3,289 (1968); Lahiri, A. K. and Banerjee, T., Corrosion Sci., 8, 895 (1968); Hoar, T. P. and Rothwell, G. P., Electrochimica Acta., 15, 1037 ( 1970) 179. Swann, P. R. and Nutting, J., J. Inst. Met., 88,478 (1960); Swann. P. R., Corrosion, 19, 102t (1963); Swann, P. R. and Pickering, H. W., Corrosion, 19, 369t. 373t (1963); Tromans, D. and Nutting, J., Corrosion, 21, 143 (1965); Ronnquist, A,, Jernkontorets Ann., 149,604(1965); Brown, B. F., Mer. Mater.. 2 No. 12, 171 (1968); Graf, L., Werkst. u. Korrosion, 20, 408 (1969) 180. Symposium on InternalStresses in Metals and Alloys (1947), Institute of Metals, London ( 1948) 181. Stress Corrosion Cracking and Embrittlement (Electrochem. SOC. Symposium), Ed. Robertson, W. D., Wiley, New York (1956); Physical Metallurgy of Stress-Corrosion Fracture (A.I.M.E. Symposium), Ed. Rhodin, T. N.. Interscience, New York (1959);
COPPER AND COPPER ALLOYS
182. 183.
184. 185. 186. 187. 188. 189.
190. 191. 192. 193. 194. 195. 196. 197. 198.
199. 200.
201. 202. 203. 204.
205. 206. 207.
208.
209. 210.
211.
4:73
Conference on Fundomentol Aspects of Stress Corrosion Crocking, Ohio State Univ. (1967). N.A.C.E., Houston (1969) Bailey, A. R., Meloll. Rev., 6 No. 21, 101 (1961) Logan, H. L., Met. Engng. Quart., 5, 32 (1965); Parkins, R. N., Metoll. Rev., 9 No. 35, 201 (1964); Engell, H-J. and Speidal, M. 0..Werkst. u. Korrosion, 20, 281 (1969) Thompson, D. H. and Tracy, A. W., J . Metals, N.Y., 1, 100 (1949) Cook, M., Ref. 180, p. 73 Pugh, E. N., Montague, W. G. and Westwood, A. R. C., Corrosion Sci., 6, 345 (1966); Uhlig, H. H. and Duqette, D. J., Corrosion Sci., 9, 557 (1969) White, L. F. and Blazey, C., Metol Ind., 75, 92 (1949) Sylwestrowicz, W. D., Corrosion, 25, 168, 405 (1969) and 26, 160 (1970) Lahiri, A. K. and Banerjee, T., Corrosion Sci., 5 , 731 (1965); Chatterjee, U. K., Sircar, S. C. and Banerjee, T., Corrosion, 26, 141 (1970); Chatterjee, U. K. and Sircar, S. C., Brit. Corrosion J . , 5 , 128 (1970) Blackwood, A. W. and Stoloff, N. S., A.S.M. Trans. Quart., 62, 677 (1969) Helliwell, B. J. and Williams, K. J.. Metollurgia, 81, 131 (1970) Graf, L., et ol., Z. Metollk., 54, 406 (1963) Thompson, D. H., Corrosion, 15, 433t (1959) Marshall, T. and Hugill, A. J., Corrosion, 13, 329t (1957) Klement, J. F., Maersch, R. E. and Tully, P. A , , Corrosion, 15, 29% (1959) Norden, R. B., Chem. Engng., 65, 194, 196 (1958) Klement, J. F., Maersch, R. E. and Tully, P. A., Met01 Prog., 75, 82 (1959), Corrosion, 16, 519t (1960) and U.S. Pat. 2 829 972 Robertson, W. D., Grenier, E. G., Davenport, W. H. and Nole, V. F., Metol Prog., 75, 152 (1959) and U.K. Pat. 802 044 Wiederholt, W., Werkst. u. Korrosion, 15,633 (1964); Laub, H., Metoll., 22, 1 116(1968) Dugdale, I. and Cotton, J. B., Corrosion Sci., 3,69 (1963); Cotton, J. B., Proc. 2ndInternot. Cong. Met. Corrosion, New York (1963). N.A.C.E., Houston, 590 (1966); Walker, R., Anti-Corrosion Methods and Muter., 17 No. 9, 9 (1970); Cotton, J . B. and Scholes, I. R., Conf. on the Protection of Metal in Storoge and in Transit, Brintex Exhibitions Ltd., London (1970); Poling, G. W., Corrosion Sci., 10, 359 (1970) Bhatt, I. M.,Soni, K. P. and Trivedi, A. M., Werkst. u. Korrosion, 18, 968 (1967) Tinley, W. H., Chem. and Ind., Dec. 12,2036 (1964); Obrecht, M. F., Proc. 2nd Internot. Cong. Met. Corrosion, New York (1963). N.A.C.E., Houston, 624 (1966) Desai, M. N., Rana, S. S. and Gandhi, M. H., Anti-Corrosion MethodsMoter., 17 No. 6, 17 (1970) Desai, M. N., Shah, Y. C. and Gandhi, M. H., Austrolosion Corros. Engng., 12 No. 3, 3 (1%8); Desai, M. N. and Shah, Y.C., Werkst. u. Korrosion, 21, 712 (1970) Rance, V. E. and Evans, U. R., Corrosion and its Prevention at Bimetallic Contocts, H.M.S.O., London (1956). Gilbert, P. T., Historical Metallurgy SOC. Conference, Birmingham, Paper Copper 4 ( 1984) LaQue, F. L., Murine Corrosion, Causes ond Prevention, John Wiley & Sons, New York (1975); Schumacher, M., Sea Wafer Corrosion Hondbook, Noyes Data Corp., Park Ridge, N.J. (1979); Gilbert, P . T., Muter. Performance. 21 (2). 47 (1982) Gilbert, P. T., Proc. 6th Internot. Cong. Met. Corrosion, Sydney, Australia (1975); Joncheray, D. Guegan, F., and Groix, F., Metaux Corrosion Ind., No. 644,140 (1979); Toscer, G., Metaux Corrosion Ind., No. 606,68 (1976); Sato, S. and Nagata, K., Sumitorno Light Metal Tech. Rep., 19 (3,4), 83 (1978) Popplewell, J. M., N.A.C.E. Nat. Conference, Houston, Paper No. 21 (1978); Syrett, B. C., Corrosion, 32, 242 (1976); Syrett, B.C., and Coit, R.L., Muter. Performonce, 22 (2). 44 (1983) Henrikson, S., and Knuttson, L., Brit. CorrosionJ., 10, 128 (1975); Efird, K. D., Corrosion, 33, 3 (1977); Lush, P . A., Hutton, S. P., Rowlands, J. C. and Angell, B., Proc. 6fh Europeon Cong. Met. Corrosion, London, p. 137 (1977); Lush, P. A,, Hutton, S. P., Rowlands, J. C. and Angell, B. Proc. 5th Internal. Cong. Murine Corrosion ondFouling, Barcelona, p. 200 (1980) Effertz, P. H. and Fichte, W., Der Moschinenschoden, 49, 163 (1976); VGB Kraftwerkstechnik, 57, 116 (1977); Sato, s., and Okawa, M., Sumitorno Light Met01 Tech. Rep., 17, (1, 2). 17 (1976); Sato, S., Nosetani, T., Yamaguchi, Y. and Onda, K., Sumitorno Light Metol Tech. Rep., 16 (1, 2). 23 (1975); Gilbert, P. T., Chem. and Ind.
4:14
212. 213. 214. 215. 216. 217.
218. 219. 220. 221. 222. 223. 224.
225. 226. 227. 228.
229. 230.
231. 232. 233.
COPPER AND COPPER ALLOYS
Supplement 2 No. 13,37 (1977); Hack, H. P., and Gudas, J. P., Mater. Performance, I8 (3). 25 (1979); 19 (4), 49 (1980); Heaton, W. B., Brit. Corrosion J., 12, 15 (1977); 13,57 (1978); Henrikson, S., Asberg, M. and Holm, R., Proc. 8th Scandinavian Corrosion Conference, Helsinki (1978) Kawake, A., Ikushima, Y., Iijuma, S., Sato, S. and Nagata, K., Sumitomo Light Metal Tech. Rep., 18 (3,4), I (1977) Elmer, K., Edison Electrical Institute Power Station Sub-committee Conference (April 1975) Lo, B. K., Electrotechnik, 53, 831 (1979) Francis, R., Brit. Corrosion J., 18, 35 (1983) Francis, R., Mater. Performance, 21 (S), 44 (1982) Niederberger, R. B., Gudas, J. P. and Danek, G. J., N.A.C.E. Nat. Conference, Houston, Paper No. 76 (1976); MacDonald, D. D., Syrett, B. C. and Wing, S. S., Corrosion, 35, 367, 409 (1979); Syrett, B. C. and Wing, S. S., N.A.C.E. Nat. Conference Chicago, Paper No. 33 (1980); Efird, K. D. and Lee, T. S., Corrosion, 35. 79 (1979); Gudas, J. P., and Hack, H. P., Corrosion, 35, 67 (1979); Gudas, J. P., and Taylor, D. W., Corrosion, 35, 259 (1979); Efird. K. D. and Lee, T. S., N.A.C.E. Nat. Conference, Houston, Paper No. 24 (1978); De Sanchez, S.R. and Schiffrin, D. J., Corrosion Sci., 22, 585 (1982); Schiffrin, D. J. and De Sanchez, S. R., Corrosion, 41, 31 (1985) Richter, H., Werkstoffe Korrosion, 28,671 (1977); Drolenga, L.J.P., Ilsseling, F. P. and Kolster, B. H., Werkstoffe Korrosion, 34, 167 (1983) Efird, K. D. and Anderson, D. B., Mater. Performance, 14 (1 I), 37 (1975) Gilbert, P. T.. Brit. Corrosion J., 14,20 (1979) Ijsseling, F. P., Krougman, J. M. and Drolenga, L. J. P., Proc. 5th Internal. Cong. Marine Corrosion and Fouling, Barcelona, p. 146 (1980) Vreeland, D. C., Mater. Performance, 15 (IO), 38 (1976); Gilbert, P. T. and North, W., Trans. Inst. Mar. Engrs., 84, 9 (Mater. Section Symposium) (1972) Nicholson, R. B. and Todd, B., Metallurgist Mater. Tech., 12. 302 (1980) Gilbert, P. T., Metallurgist Mater. Tech., 10,316 (1978); Gilbert, P. T., Proc. Internal. Symposium Corrosion and Protection Offshore, Cefracor, Paris (1979); Copper Development Assoc. (UK), Copper Alloys f o r Offshore Tech., Publication CM-L39 (1976); Lim, L. H., Conference Offshore Europe '77, Aberdeen, Paper No. 36680 (1977) Efird, K. D., Mater. Performance, 15 (4), 16 (1976); Chandler, H. E., Metals Progress, 115,47 (1979) Manzolillo, J. L., Thiele, E. W. and Tuthill, A. H., SOC.Naval Arch. and Mar. Engrs. Conference, New York (1976); Obrzut, J. J., Iron Age, 220, 36 (1977); 222, 35 (1979); Anon., Metal Construction, 11, 181 (1979) Moreton, B. B. and Glover, T. J., Proc. 5th Internat. Cong. Marine Corrosion and Fouling Barcelona, p. 267 (Biology) (1980) Prager, M. and Thiele, E. W., Welding J., p. 17 (July 1979); Middleton, L. G., R. Inst. Naval Arch/Copper Development Assoc. Symposium (Jan. 1980), London; Schorsch, E., Bicicchi, R. T. and Fu, J. W., Trans. SOC.Naval Arch. & Marine Engrs., 86 (1978); Moreton, B. B., Metallurgist Mater. Tech., 13,247 (1981) Internat. Copper Research Assoc. Newsletter No. 9, p. 1 (1979) Desalination Materials Manual, Dow Chem. Co. for US Office of Water Research & Tech. (1975); Materials Failure Identification Manual for Sea Water Desalination Plants, Aqua-Chem Inc. for US Office of Water Research &Tech., O.R.N.L. (1976); Todd, B., Middle East Water & Sewage J. (Oct./Nov. 1977); Temperley, T., Desalination, 33, 99 (1980); Hill, K., Desalination, 25, 1 I I (1978); Sato, S. Trans. Japan Inst. Metals, 19, 575 (1978); Sato, S. and Nagata, K., Sumitomo Light Metal Tech. Rep., 18 (1,2), 1 1 (1977); Schrieber, C. F., Boyce, T. D., Oakes, B. D. and Coley, F. H., Mater. Performance, 14(2), 9(1975); Oakes, B. D., Mater. Performance, 15 (I), 44 (1976); Ross, R. W. and Anderson, D. B., Mater. Performance, 14 (9), 27 (1975); Ross, R. W., Mater. Performance, 18 (7). 15 (1979) Sato, S. and Nagata, K., Sumifomo Light Metal Tech. Rep., 15, 174 (1974); Bianchi, G., Maua, F., Sivieri, E. and Torchio, S., Proc. 6th European Cong. Met. Corrosion, London, p. 271 (1977); Torchio, S., Corrosion Sci., 21, 59, 425 (1981) Campbell, H. S., Brit. Corrosion J . , 18, 206 (1983) Rowlands, J. C. and Brown, T. R. H. M., Proc. 4th Internal. Cong. Marinecorrosion & Fouling, Juan-les-Pins, p. 475 (1976); Culpan, E. A. and Rose, G., Brit. Corrosion J . , 14, 160 (1979); Ferrara, R. J. and Caton, T. E., Mater. Performance, 21 (2), 30 (1982)
COPPER A N D COPPER ALLOYS
4:15
234. Gilbert, P. T., Proc. 5th Internat. Cong. Marine Corrosion & Fouling, Barcelona, p. 210 (1980); Scholes. I. R., Astley, D. J . and Rowlands, J. C., Proc. 6fhEuropean Cong. Met Corrosion, London, p. 161 (1977) 235. Page, G. G., New Zealond J. Sci., 26, 415 (1983) 236. Ijsseling, F. P., Krougman, J. M. and Drolenga, L.J., Proc. 5th Internal. Cong. Morine Corrosion & Fouling, Barcelona, p. 146 (1980); Efird, K. D., Corrosion, 31, 77 (1975); Efird, K. D., Corrosion, 33, 347 (1977); MacDonald, D. D., Syrett, B. C. and Wing, S. S., Corrosion, 34, 289 (1978); Ijsseling, F. P. and Krougman, J. M., Proc. 6th European Cong. Met. Corrosion, London, p. 181 (1977); Ilsseling, F. P. and Krougman, J. M., Proc. 4th Internot. Cong. Marine Corrosion & Fouling, Juan-les-Pins (1976); Castle, J. E., Corrosion Sci., 16, 3 (1976); Epler. D. C. and Castle, J. E., Corrosion, 35,451 (1979); Shone, E. B., Brit. Corrosion J., 10, 33 (1974) 237. Kato, C., Ateya, B. G., Castle, J. E. and Pickering, H. W., J . Electrochem. SOC., 127, 1890, 1897 (1980); Kato, C. and Pickering H. W., J . Electrochem. Soc., 131, 1225 (1984) 238. Parker, J. G. and Roscow, J. A., Brit. Corrosion J., 16, 107 (1981) 239. Proc. Internal. Symposium Corrosion of Copper & Copper Alloys in Building, Tokyo, Japanese Copper Development Assoc., 12 papers (1982); Cornwall, F. J., Witdsmith, G. and Gilbert, P. T., Brit. Corrosion J., 8, 202 (1973); A.S.T.M. Spec. Tech. Pub. 576 p. 155 (1976) 240. Bowers, J. E., Oseland, P. W. and Davies, G. C., Brit. Corrosion J . , 13, 177 (1978) 241. Terwinghe, F., Celis, J. P. and Roos, J. R., Brit. Corrosion J., 19, 115 (1984) 242. Sparks, J. M. and Scully, J. C., Corrosion Sci., 16, 619 (1974); Kermani, M. and Scully, J. C., Corrosion Sci., 18, 833 (1978); 19, 89, 489 (1979); Scully, J . C., MerolSci., 12, 290 (1978); Corrosion Sci., 20, 297 (1980); Takano, M. and Staehle, R. W., Trans. Japan Inst. Metals, 19, 1 (1978); Takano, M., Trons. Japan Inst. Metals, 18, 787 (1977); Corrosion, 30,441 (1974); Kawashima, A., Agrawal, A. K.and Staehle, R. W., A.S.T.M. Spec. Tech. Pub., 665, p. 266 (1979); Linder, M. and Mattsson, E., Proc. 6th European Cong. Met. Corrosion, London (1977); Uhlig, H., Gupta, K. and Liang, W., J. Electrochem. SOC., 122,343 (1975); Holroyd, N. J. H., Hardie, D. and Pollock, W. J., Brit. Corrosion J . , 17, 103 (1982) 243. Gupta, P., Chaudhary, R. S. and Prakash, B., Brit. Corrosion J., 18.98 (1983); Walker, R., Corrosion, 20, 290 (1973); Lewis, G., Brit. Corrosion J., 16, 169 (1981); Subramanyan, N. C., Sheshadri, B. S. and Mayanna, S. M., Brit. Corrosion J., 19 (4). 177 (1984) 244. Conference ‘Copper Alloys in Marine Environments’ Birmingham UK. April 1985. Copper Development Association. 20 papers 245. Conference ‘Marine Engineering with Copper-Nickel’ London April 1988. Institute of Metals. 12 papers 246. Parvizi, M. S., Aladjem, A. and Castle, J. E., Internat. Moterials Reviews, 33 (4), 169 (1988) 247. Geesey, G. C., Lewandowski, Z. and Fleming, H-C., Biofou/ing/Biocorrosion in Industrial Water Systems. Lewis Publishers Inc. Chelsea, Michigan (1993) or 4)
4.3 Lead and Lead Alloys
Introduction Lead forms a series of relatively insoluble compounds, many of which are strongly adherent to the metal surface. In conditions where a stable continuous film can form, further reaction is often prevented or greatly reduced. Thus the general good corrosion resistance of lead results from the formation of relatively thick protective films of corrosion product. The major uses of lead in the UK are in batteries, and in sheet and pipe of which the vast majority is sheet for building purposes. These applications account for about one third each of lead used. This situation is unique, since in all other countries batteries account for most of the lead market. A small but very important application is sheet and pipe for the chemical industry. Lead is no longer installed for water services. Lead cable sheathing which accounts for 5% is in general decline, but is valued in niche applications such as on oil rigs where resistance to hydrocarbons is important. The use of lead for anodes accounts for a very small tonnage, but is still of great importance to the industries which use them. Lead sheet is used in the building industry throughout continental Europe and to a lesser extent Australia, but hardly at all in the USA. Other aspects of lead consumption follow the same general trends worldwide.
Composition and Mechanical Properties Lead
While lead of purity in excess of 99.99% is commercially available, it is very rarely used owing to its susceptibility to grain growth and fatigue failure by intercrystalline cracking, and indifferent mechanical properties. Because of its generally superior corrosion resistance, pure lead to BS 334: 1982 type A, shown in Table 4.13, is occasionally used in chemical plant, but only if there is no suitable alternative.
4:76
LEAD AND LEAD ALLOYS
4:77
Lead Alloys Besides Type A lead, nine lead alloys are specified in British Standards for various purposes. Their compositions and impurity limits are given in Table 4.13. In addition, alloys for batteries and for anodes are of importance. In due course it is likely that European standards will supersede the current national ones. Of the elements commonly found in lead alloys, zinc and bismuth aggravate corrosion in most circumstances, while additions of copper, tellurium, antimony, nickel, silver, tin, arsenic and calcium may reduce corrosion resistance only slightly, or even improve it depending on the service conditions. Alloying elements that are of increasing importance are calcium, especially in maintenance-free battery alloys and selenium, or sulphur combined with copper as nucleants in low antimony battery alloys. Other elements of interest are indium in anodes’V2, aluminium in batteries3, and selenium in chemical lead as a grain refiner lo*.
BS 334:1982 Compositional limits of chemical lead defines the composition of five grade of lead (Types A, B1, B2, B3 and C) and also gives guidance on selection, a method for the determination of creep strength and an empirical test for corrosion resistance, the flash test*, which, however, does not guarantee compliance with the standard. Chemical analysis is always to be preferred. It has been shown that flash points rise slightly with increasing copper and tellurium, remain constant with small additions of silver and fall with bismuth, zinc, tin and antimony4. Type A lead should only be used in a vibration-free environment and where the superior corrosion resistance is of paramount importance. For general chemical plant use, type BI copper lead is to be preferred on account of its much greater structural stability, especially at elevated temperatures. Its mechanical properties are also significantly better. Type B2 copper tellurium lead has extremely good fatigue resistance which is retained to a greater extent at elevated temperatures than type B1. The main effect of tellurium is to form a fine-grained uniform grain structure, to enhance work hardening, and to delay recrystallisation. The silver content in type B3 also delays recrystallisation and promotes a large-grained stable structure which is creep-resistant 5 . 6 . Type C antimonial lead is used for valves, pump bodies and fatigue-resistant applications, but is not suitable for use at temperatures above 60°C owing to a rapid increase in creep rate, or in sulphuric acid concentrations above 60%. BS 801:1984 Composition of lead and lead alloy sheaths for electric cables gives compositional requirements for lead and three alloys, B, E and 112 C. The impurity limits for lead are more relaxed than for type A lead, but lead to this grade can also be prone to intercrystalline cracking, which has been observed in the transport of cables as well as in service. Alloy E contains tin and antimony, alloy B 0.85% antimony, and 112 C tin and cadmium. Alloy B is suitable for use in environments where severe vibration is * A test for evaluating the chemical quality of grades A, BI, 82 and B3 by resistance to H 2 S 0 4 . A specimen is placed in 95-96% H 2 S 0 4 and the temperature raised to 300°C in 7 rnin. The
‘flashing’is due to a sudden increase in the rate of formation of PbSO, and should not happen below 28S°C, or 300°C for lead for use at elevated temperatures.
Table 4.13
Specifcation
Composition of types of lead and lead alloys commonly used in the United Kingdom
Alloying element’ or impurities (%)
Use Sn
BS 334:I982 Type A BS 334 Type B1 Copper-Lead BS 334 Type B2 Copper-Tellurium-L .ead BS 334 Type B3 Copper-Silver-Lead BB 334 Type CL
Specialised chemical applications General chemical applications Specialised chemical applications Specialised chemical applications Specialised chemical applications
0.001
Sb
Ag
Cu
Ni
Fe
Bi
0-002 0.002 0-003 0.001 0.003 0.005
Cd
Zn
Te
As
S
Trace’
0.002
-
Trace
Trace 99.99min
Pb
r
F0
> z 0.001
0.001
0.002
Trace
0.002 0.050 0.001 0.003 0.005
Trace
0.002
0.020 Trace 0.050
Trace
0.002 0.003 0.003 0.001 0.003 0.005 Trace
0.002
-
Trace
Trace Balance
0.002
-
0.01
Trace Balance
0.002
0.002
0.070 0.001
0.001
-
0.002 0.050 0.001 0.003 0.005 0.070
2.5 11
0.005
0.005
0.01
0.01
0.005
0.003 0.015 Trace
Trace
Trace
Balance Balance
0 P
F0 F
6
s
BS 801:1984 Lead BS 801:1984 Alloy B BS 801:1984 Alloy E BS 801:1984 Alloy 1/2C BS 1178:1982
Cable sheathing Cable sheathing
0.35 0.01
Cable sheathing
0.35 0.45
Cable sheathing
0 . I8
0.15
0.80 0.95 0.15 0.25 0.005
0.005 0.005
0.06 0.06
-
-
0.05 0.05
0.02
0.002 0.005 0.005
0.02
0.002
0.005
0.005
-
Balance' Balance'
0.005
0.06
-
-
0.05
0.02
0.002 0.005
0.005
-
Balance'
0.005 0.06
-
-
0.05
0.06 0.09
0.002
0.005 0.005
-
Balance'
0.03
-
-
0.05
-
0.05
-
-
Balance'
0.22
Sheet for building
0.005
0.01 0.01
0.06 I . One figure indicates a maximum impurity level. Two figures show maximum and minimum content of alloying additions. 2. Trace is defined as less than 0~005% 3. Total other elements 0.01% max.
-
r m > 0 3-
z
0 r m 30
4:80
LEAD AND LEAD ALLOYS
expected, alloy E is somewhat resistant to vibration and, as previously mentioned, unalloyed lead is not at all resistant. These materials are less corrosion resistant than chemical lead, but their performance is adequate in underground or marine environments.
BS 1178:1982 Milled lead sheet for building purposes lays down requirements for composition, structure, thickness, freedom from defects, width and length, and marking. The specified copper content stabilises the structure of the material, conferring resistance to thermal fatigue cracking caused by grain growth and thermal cycling. Lead acid batteries currently use antimonial alloys of a range of compositions or lead-calcium-(tin) alloys depending on the application. They are proof against the comparatively weak acid used and offer good resistance to oxygen evolved during charging, but have a variety of advantages and disadvantages which are covered later.
Corrosion Behaviour The standard electrode potential, EOpb2+,pb= -0.126 V738,shows that lead is thermodynamically unstable in acid solutions but stable in neutral solutions. The exchange current for the hydrogen evolution reaction on lead is very small (-IO-” - lo-” A crn-,), but control of corrosion is usually due to mechanical passivation of the local anodes of the corrosion cells as the majority of lead salts are insoluble and frequently form protective films or coatings.
Anodic Behaviour Lead is characterised by a series of anodic corrosion products which give a film or coating that effectively insulates the metal mechanically from the electrolyte (e.g. PbSO,, PbCI,, Pb,O,, PbCrO,, PbO, PbO,, 2PbC0, .Pb(OH),), of which PbSO, and PbO, are the most important, since they play a part in batteries and anodes. Lead sulphate is important also in atmospheric passivation and chemical industry applications. In an aqueous electrolyte, the anodic behaviour of lead varies greatly depending on the conditions prevailing. Extensive reviews of the anodic behaviour of lead have been produced . Under certain conditions, the passive film may be converted to lead dioxide which has an electronic resistivity of 1 - 4 x lo-, ohm cm. Two polymorphs of PbO, exist, a and 0. Both are non-stoichiometric and on a lead substrate there is always an oxygen An upper limit of n in PbO, has been given as 1.99 and various lower limits of 1 e938, 1.875 and below have been given (25,30, 137, 138). In practice there is a variation in oxygen content owing to resistance to diffusion of 0 and 0- species through the film20-25. The structure of lead oxide films can be very complex and detailed studies have been undertaken6. 9-1 I . 15-17.25.27,28.39 , often using recently developed techniques 26. Figure 4.13 shows the regions of thermodynamic stability for the compounds which can form in the Pb-H,SO,-H,O system. 19s20321
4:81
LEAD A N D LEAD ALLOYS
PbO, has a low overpotential for the liberation of oxygen from H,S0,3’ and KOHZ8solutions. and for chlorine”.
3
Fig. 4.13
Potential-pH diagram of lead in the presence of sulphate ions at unit activity and 25°C (after Ref. 139), reproduced by permission of Pergamon Press.
Cathodic Behaviour Cathodic disintegration can occur with lead, observable as a grey cloud of fine metal particles. Hydrogen evolved on the surface of the lead can be absorbed if the current density is sufficiently high34.35.Above this level, ‘avalanche penetration’ can occur, leading to the formation of lead hydride, which leads to disintegration in the manner described 37. Electrochemical . implantation of alkali metals can also lead to di~integration’~ Thermodynamics of the Pb-H,0, Pb-H,0 - X systems
Pourbaix et ai.38have studied the Pb-H,O, Pb-H,O-X systems where X is a non-metal, and have established the domains of thermodynamic stability
4:82
LEAD A N D LEAD ALLOYS
of lead, lead cations and anions, and insoluble compounds of lead. Figure 4.15 shows the Pb-H,O system; it can be seen that in the region of high and
low pH, corrosion occurs owing to the amphoteric nature of lead (cf. Zn, Al, Sn). This is a significant factor in the behaviour of lead in actual environments. I
I 1.6
-
1.2
1.4
\
\
I -
0.8 0.6
$O..t 0.2
-
'.--. '.
i
\
\ \ \
PbSO,
PbO -4
-2
-0.8
-1.2
- 1 0
1
2
3
4
5
6
7
8
9 1 0 1 1
1 2 1 3 1 4
PH
Fig. 4.14 PotentiaVpH diagram for the Pb-H,O system. The area between @ and @ corresponds to the thermodynamic stability of water. Light lines represent equilibrium conditions between a solid phase and an ion at activities 1, IO-', Heavy lines represent and equilibrium conditions between two solid phases. Broken lines represent equilibrium conditions between two ions for a ratio of these ions equal t o unity (after Delahay, Pourbaix and van Rysselberghe 38)
In contrast with the P b - H 2 0 system, it can be seen in Figure 4.13 that in the presence of SO:- the corrosion zone in the region of low pH no longer exists, owing to the thermodynamic stability of PbSO,. The Pb-H,O-CO, system has been expressed in a similar pH/potential diagram3* in which account has been taken of insoluble carbonates and basic carbonates of lead. The predictions of the pH/potential diagram are generally fulfilled, but in very concentrated acid solutions, attack may diminish, owing to the relative insolubility of the relevant salt in the acid. Thus, lead nitrate, although soluble in water, has (owing to common ion effect) only slight solubility in concentrated nitric acid, and the corrosion rate is reduced. Similarly, lead chloride is less soluble in moderately concentrated hydrochloric acid than
LEAD A N D LEAD ALLOYS
4:83
it is in water. In concentrated hydrochloric acid, however, the converse is the case, since in high chloride-ion concentrations lead forms soluble complex anions. Inspection of products of corrosion and correlation with thermodynamic data frequently gives an indication of the cause of corrosion. Thus from Figure 4.14 it will be seen that a potential of about 1V (neglecting the influence of other ions which would be unlikely to decrease this value) is required for the formation of lead dioxide in neutral solution. This is somewhat higher than that likely to be generated by galvanic action between dissimilar metals commonly used in civil engineering, and the presence of lead dioxide among the corrosion products is usually taken as an indication that an impressed current is responsible for corrosion. From an inspection of the more common compounds of lead, it will be seen that, in many environments, the corrosion product will be relatively insoluble (Table 4.14). Often, however, compact protective films are prevented from forming on the surface of the metal. The nature of the film is influenced by the mode of crystallisation, and in the case of the lower oxides for example, frequently little protection is afforded. Lead dioxide often forms a good adherent film, especially when it is produced from a sulphate film or other adherent compounds during anodic oxidation. When it is formed away from the surface by chemical reaction it gives no protection. It is a strong oxidising agent and, unlike the lower oxides, is not affected by most acids. Concentrated hydrochloric acid gradually dissolves it to form hexachloroplumbic acid, and with alkalis, plumbates are formed. Table 4.14 Compounds of lead Solubility at
Product
25°C
colour
Acetate Bromide Carbonate Basic carbonate Chloride Chromate Dioxide
soluble (55 g/100 ml) 5 . 7 x 10-6 3.3 x 1 0 - l ~ as above I x 10-4 1 . 8 x IO-^ insoluble
Fluoride Formate Hydroxide Iodide Monoxide Nitrate Phosphate Sulphate Sulphide Sulphite Triplumbic-tetroxide
3 . 7 x 10-8 soluble (1.6g/lOOml) 4 x 10-15 1.4 X insoluble soluble (60 g/IOO ml)
white white white white white orange black or dark brown white white white yellow yellow-red white white white black white red
Compounds of lead
Formula
1 x 10-8 3 . 4 x 10-28 insoluble insoluble
With sparingly soluble salts of lead, the compactness of the deposits may be strongly influenced by the concentration of the relevant anion. Very low concentrations frequently resulting in imperfect coatings.
4:84
LEAD A N D LEAD ALLOYS
Atmospheric Corrosion Lead is used for roofing, gutters, flashings, downspouts, etc. and exhibits excellent resistance to air (dry or humid). The sequence of patina formation is orthorhombic PbO -+ basic lead carbonate normal lead carbonate normal lead sulphite normal lead sulphatea. The oxide is initially converted to plumbonacrite (6PbC0, .3Pb(OH), .PbO) and hydrocerrusite (ZPbCO, .Pb(OH)2)4'. While these have an extremely low solubility, they can produce a white flocculant 'run-off' in wet weather, which can stain surrounding surfaces in the very early stages of e x p o ~ u r e ~ ' *In ~ ~marine *~~. environments, the initial film reacts with sodium chloride when wet to produce basic lead chloride and sodium hydroxide. This may result in corrosion of adjacent materials such as a l ~ m i n i u m ~The ~ - ~lead ~ . patina stabilises, but takes approximately twice as long as in other atmospheric environments to do so. A common treatment for new lead is a resin-based patination oil which suppresses the formation of basic carbonates allowing the slow controlled growth of a strongly adherent normal carbonate patina from the outset 42.43.15 . Galvanic corrosion is not normally significant because the corrosion films formed are electrically insulating, although an isolated instance of severe galvanic corrosion of lead in contact with stainless steel in the presence of lime mortar has been reported. Severe corrosion can be caused by organic acid fumes such as acetic or formic acids. These can be liberated by new wood, especially oak, and also by varnishes, glues, urea formaldehyde, plastics, fabrics and drying-oil paints, which can liberate fumes for a considerable time after application4'. +
-+
+
Distilled Water
In distilled water free from dissolved gases, corrosion is slight though significant. The rate is increased by the presence of oxygen. With oxygen together with very small concentrations of carbon dioxide, very rapid corrosion takes place, with basic carbonates forming a white turbidity. At moderate CO, concentrations, a degree of passivation of the lead surface occurs, but corrosion is still significant. At high CO, contents, corrosion is increased due to the formation of soluble b i c a r b ~ n a t e Lead ~ ~ . is therefore not suitable for distilled water containers. It is, however, used for steam heating coils, but if the condensate is not recycled without access to air, rapid failure is likely. Condensation corrosion is also a common cause of failure in lead-work on buildings. Trapped water is evaporated from and condensed on the underside of the lead during thermal cycling in the environment. This repeated condensation causes the production of lead oxide and lead hydroxide which is soluble and migrates away from the surface, leaving it unpassivated. Subsequent reaction with CO, in the atmosphere produces copious quantities of basic lead carbonate, resulting in blistering, perforation, and finally disintegration of the lead4,' . Adequate ventilation and adherence to codes of practice are essential to prevent this'28. Water can be admitted through cracks caused by thermal fatigue, which is a consequence
LEAD A N D LEAD ALLOYS
4:85
of overfixing the lead, using sheets which are too large, or of using lead containing insufficient copper. Natural Waters
Because of the long life of lead pipework, water may be conveyed through existing lead pipes for some years to come. This may not be hazardous if the waters contain sufficient carbonate, sulphate or silicate (see Section 2.3), and are alkaline. The presence of ‘aggressive carbonic acid’49or organic acids will render relatively hard waters plumbosolvent. Waters from peaty moorland frequently contain quinic acid from the roots of bilberry and heather which increases attack, as d o aggressive agents such as nitrates and carbon dioxide in stagnant water. Rain water is also frequently plumbosolvent. Treatments given to water include deacidification with milk of lime, whiting, or by limestone bed, and removal of organic material with alkalis and aluminium sulphate. These treatments will encourage the formation of a protective carbonate ~ c a l e ~Zinc ~ ’ ~orthophosphate ~. treatment is also reported to control the dissolution of lead from pipes”. The permitted lead content of tap water in the EC is currently 5 0 ~ g / whilst l ~ ~ in the USA it is 15 ~ g / l ~ ~ . Lead usually has excellent resistance to seawater owing to the formation of a passive film of basic carbonate and carbonate-chloride double which should be compared with its behaviour in solutions of alkali chlorides (see salts p. 4:87).
Underground Corrosion Stray-current Corrosion
Stray currents are a source of damage to buried metal structures (see Section 10.5) and lead pipes and cable sheaths are particularly susceptible to it. Although lead can corrode under cathodic (alkaline) conditions, it is generally the anodic sites on the pipe or cable sheath which corrode. Lead is considered to be endangered if the current density is more than 25 mA m-z56.This is influenced by the conductivity of the soil, which is largely determined by the moisture content, but may be affected by salting of roads in winter. The limit of corrosion may be considered as 100 metres from the current source 56. Non-metallic links in pipework may break electrical continuity. This will produce more numerous corrosion sites, but they are frequently less intense. The surface will normally be covered with a mostly whitish corrosion deposit associated with either a smooth pitted surface or a more general rough etched a p p e a ~ a n c e ~ The ~ . corrosion product may comprise oxides, carbonates, hydroxides and chlorides 58. Glassy watery crystals containing PbCl, .Pb(OH), and PbCI2.6Pb0.2H20 have been identified. The use of lead pipes for earths for alternating currents has also resulted in serious c o r r ~ s i o n No ~ ~ protective ~~~. coating is fully effective, but some give good Electrolytic corrosion may also occur on the inside of cable sheaths by the passage of current from the cable sheath to the
4:86
LEAD AND LEAD ALLOYS
Electrochemical Corrosion
Corrosion cells are established by inhomogeneity of the lead69or its environment, although severe corrosion due to metal composition is not common. 'Geological cells' formed between soils which differ in water content, degree of aeration, or the presence of various chemicals or bacteria can give rise to the passage of large corrosion currents at an e.m.f. of up to 1.5 V. Extensive long-term tests have been conducted on lead in soils7'-73~76*7s. The worst combination of soils is wet clay and cinders. The carbon in the cinder acts as an efficient cathode and severe anodic corrosion takes place in the clay environment. Moisture held in the clay permits the passage of relatively high currents. Anodic corrosion can occur when cables are in contact with dissimilar metals such as steel support racks or copper bonding ribbon. A new (clean) section of cable may also become anodic to an old (passivated) cable and can corrode. Soils of high permeability are less aggressive since water tends to be mobile, so reducing concentration cells 77 and frequently drains readily to allow free movement of oxygen, thus reducing the effect of aeration cells. Both mechanisms give rise to pitting corrosion. Where oxygen circulates freely, a stable patina is often formed which is similar to that formed in air. Sandy soils tend to be among the best. Very large grained soils are normally good for the reasons given above, but under certain conditions severe localised pitting can be caused due to aeration cells79.Clays and silts tend to be worst. Cables are often laid in sand or crushed chalk. Sulphates, silicates, carbonates, colloids and certain organic compounds" act as inhibitors if evenly distributed, and sodium silicate has been used as such in certain media. Nitrates tend to promote corrosion, especially in acid soil waters, due to cathodic de-polarisation and to the formation of soluble nitrates. Alkaline soils can cause serious corrosion with the formation of alkali plumbites which decompose to give (red) lead monoxide. Organic acids and carbon dioxide from rotting vegetable matter or manure also have a strong corrosive action. This is probably the explanation of 'phenol corrosion', which is not caused by phenol", but thought to be caused by decomposition of jute or hessian in applied protective layers 82-'5. Calcium hydroxide leached from incompletely cured concrete causes serious corrosion of lead (see Section 9.3). This is because carbon dioxide reacts with the lime solution to form calcium carbonate, which is practically insoluble. Carbonate ions are therefore not available to form a passive film on the surface of the leads6. Typically, thick layers of PbO are formed, which may show seasonal rings of litharge (tetragonal PbO) and massicot (orthorhombic PbO)'7-88. To prevent undergraund corrosion, lead is frequently protected with coatings of tar, bitumen, resin, etc., which are only effective if they completely insulate the metal from corrosive agents and stray currents. No coating is fully effective, but some give good p r o t e ~ t i o n ~ ~. Th " ~e' ~ ~ , ~ ~ ~ most successful method used is cathodic protectionM which for impressed currents, if correctly applied, can protect indefinitely (see Chapter IO). It is effective at a potential of E" = -0-8V65or about 0 -1 V more negative than
LEAD A N D LEAD ALLOYS
4:87
its equilibrium potential in the soil in question". Both impressed currents67 and sacrificial anodes have been usedM-'*.An excessively negative potential can increase the pH of the environment, thus causing corrosion. Caustic soda has also been observed from electrolysis of de-icing salt. Patches of conductive lead sulphide can be formed on lead in the presence of sewage. This can result in the flow of a large corrosion currentg9. Sulphate-reducing bacteria in soils can produce metal sulphides and H,S, which results in the formation of deep pits containing a black mass of lead sulphidew. Other micro-organisms may also be involved in the corrosion of lead in S O ~ I ~ . ~ ' . Cables are frequently laid in ducting for protection, but are still susceptible to corrosion by aeration cells set up between the cables and the duct walls, and to attack by corrosive solutions, especially from concrete ducts. They are also prone to corrosion by organic acids from wooden ducting, and to galvanic corrosion with iron supports. Damage by insects and animals may also occur92.
Chemicals Corrosion data reported as weight losses can be misleading because of the high density of lead; volume losses or yearly penetration figures are to be preferred for this metal. It should also be remembered that in chemical applications the thickness of lead used is usually greater than that of other metals, and higher corrosion rates, by themselves, are therefore not so serious. Since lead is protected by relatively thick films of corrosion products, short-term tests can be misleading, as once the film has formed there will be a significant decrease in the corrosion rate. Several sources of corrosion data are available, which should be consulted for specific information on corrosion re~istance~~'~'.
Gases Lead will resist chlorine up to about 100°C97,is used for dry bromine at lower temperatures 98 and is fairly resistant to fluorine94.Hydrofluoric acid does not passivate lead, so lead should not be used in this environment. Lead is very resistant to sulphur dioxide and fairly resistant to sulphur trioxide, wet or dry, over a wide temperature rangey4.
Acids Mineral Acids
Sulphuric acid is frequently made, stored and conveyed in lead. The corrosion resistance is excellent (see Figure 4.15) provided that the sulphate film is not broken in non-passivating conditions. Rupture of the film may be caused by erosion by high velocity liquids and gases containing acid spray.
4:88
LEAD AND LEAD ALLOYS
In such an environment an inner lining of acid-resistant brick is often used. Thermal cycling may also disrupt the film. Acid of more than 85% concentration tends to dissolve the lead sulphate film, although lead has been used in cold quiescent conditions with concentrations of over 90%. Nitrosylsulphuric acid, and nitrosyl chloride formed as a result of chloride in the water, can cause corrosion in sulphuric acid and lead-chamber plants. Alloying is not generally beneficial in this instancew and some elements (such as copper) can increase the corrosion rate. Nitric acid readily attacks lead if dilute and the metal should not be used for handling nitrate or nitrite radicals except at extreme dilutions and preferably with a passivating reagent such as a sulphate, which will confer some protection. An example of this is the wash water from cellulose nitrate units. Corrosion decreases to a minimum at 6570% HNO, and lead has been used for storage of nitric acid in the cold at this c o n ~ e n t r a t i o n ~Resis~.~~~. tance to a mixture of 98.85% H2S04and nitric acid of 1 *50-1 e 5 2 S.G. can be excellent lo’.
O F
OC
SULPHURIC ACID
(%I
Fig. 4.15 Corrosion of lead by sulphuric acid as a function of temperature. Concentrations below 50% are not shown because resistance of lead is very good even at temperatures including boiling (after Fontana IOo)
Hydrochloric acid should be regarded as aggressive to lead and its use cannot generally be recommended, although a satisfactory life has been obtained with acid of up to 30% concentration at ambient temperature and 20% concentration at 100°C. Antimonial lead is markedly more r e ~ i s t a n t ~ ~Resis.~’. tance of lead to corrosion by HCI is presumably due to the formation of a protective film of lead chloride which is only slightly soluble at these concentrations combined with the rate-limiting effect due to the high hydrogen
LEAD AND LEAD ALLOYS
4:89
evolution overpotential of lead. With mixed hydrochloric and hydrofluoric acids for pickling steel, the behaviour of lead is uncertain. The life can be increased, however, by adding the hydrofluoric acid first to passivate the surface. The presence of aluminium fluoride can prevent the formation of a protective film and severe corrosion may result IO2.
Phosphoric acid and chromic acid Lead has good corrosion resistance to these acids. Its resistance to chrome-plating solutions will be discussed later.
Organic Acids
Acetic acid Lead is attacked by most weak organic acids which produce water soluble lead salts, in the presence of air or organic oxidants. Lead is resistant to cold glacial acetic acid and is used for making storage vesselsY3. Aqueous acetic acid, solutions containing acetates, and acetic acid vapour rapidly corrode lead. The lead oxide protective film is dissolved, yielding salts which are carbonated in the presence of CO, and water to form basic lead carbonate, which in these circumstance does not form a passive film. In the absence of oxygen, corrosion in dilute solutions (0-01Mor less) is slightIo3. Formic acid behaves in a manner similar to acetic acid. Lead is resistant to oxalic, tartaric and fatty acids only in the absence of oxygen. Dilute (0.1~-0.001~) acetic, propionic, butyric, succinic and lactic acids all corrode lead to about the same extent. Pyruvic acid appears to inhibit corrosion after a short period of attack. In most cases the corrosion products are x PbCO,, y Pb(OH),, the ratio of x to y being 2:1, and corrosion is intergranular IO4. Lead in building can be corroded by organic acids from new wood, decaying wood and lichens (see Sections 9.3 and 18.10). This is a common phenomenon with run-off from lichens which grow on tiles and slates. Where this occurs, a sacrificial strip of lead has been advocated ' 2 6 .
Fuel oil Organic acids are thought to be responsible for corrosion of lead by fuel oil. Formerly, mercaptans were held to be the cause, but now it is believed that naphthenic acids are responsible. Lubricating oils Bearings d o not normally fail due to corrosion, but where this has occurred it has been associated with the acidity of white oils, the peroxide content and the presence of air. Peroxides are the controlling factor, but corrosion is reduced in the absence of air. The corrosion product consists of a basic lead salt of two or more organic acids IO5 (see Section 2.1 1). Alkalis
Lead is not particularly resistant to alkalis, but in some cases the corrosive action of sodium hydroxide and potassium hydroxide can be tolerated (KOH to 50% and up to 6 0 ° C NaOH to 30% and 25"C, 10% and 90°C)93.The rapid attack of lime solutions is discussed earlier (also see Section 19.3).
4:90
LEAD AND LEAD ALLOYS
Salts
Lead is not generally attacked rapidly by salt solutions (especially the salts of the acids to which it is resistant). The action of nitrates and salts such as potassium and sodium chloride may be rapid. In potassium chloride the corrosion rate increases with concentration to a maximum in 0.05~solution, decreases with a higher concentration, and increases again in 2~ solution. Only loosely adherent deposits are formed. In potassium bromide adherent deposits are formed, and the corrosion rate increases with concentration. The attack in potassium iodide is slow in concentrations up to 0 . 1 ~ but in concentrated solutions rapid attack occurs, probably owing to the formation of soluble KPbI,. In dilute potassium nitrate solutions ( 0 . 0 0 1 ~ and below) the corrosion product is yellow and is probably a mixture of Pb(OH), and PbO, which is poorly adherent. At higher concentrations the corrosion product is more adherent and corrosion is somewhat reduced IO6. Details of the corrosion behaviour of lead in various solutions of salts are given in Figure 4.16.
4
8
I
I
I
12
16
20
TIME
(DAYS)
Fig. 4.16 Relationship between weight loss or weight gain and time for lead immersed in various environments (selected from Reference 107). LEGEND f 0.25 N NaF; pH 6 . 3 u 0.5 N (NH,),SO,; pH 2 . 9 g 0.5 N CH3COONH4 + CH,COOH; pH 4 . 6 b 0.5 N Na2S0,; pH 5 . 4 h 0.04 N Ca(OH),; pH 13 c 0.01 N NaOH; pH 12.8 0.5 N CH,COONa; pH 16 0.01 N Ba(OH),; pH 11.9 0 . 5 N Ba(OH),; pH > 13 d 0.5 N NaCI; pH 4 . 9 e 0 . 5 N CH,COONH, + NaOH; pH 8 . 9
LEAD A N D LEAD ALLOYS
4:91
Lead Anodes Anodes for electroplating and for electrolysis of brine are frequently made of lead and lead alloys. Despite the formation of a passive film of lead dioxide (see anodic behaviour), there is generally a very slow continued corrosion which leads to thickening of the PbO, film. The tensional forces produced can cause growth of the anode. The film may also crack, releasing PbO, particles. Alloying elements are frequently added for strength or to stabilise the film. Rolled or extruded structures are generally more resistant to corrosion than cast. In seawater, lead anodes with 1 or 2% silver may be used for cathodic protection of at current densities of up to 120 A m-,* ' I 2 . Lead with 6% antimony and 1% silver has also been recommended. It is thought that silver might provide small stable nucleation sites for PbO, formation 1 3 - ' I ' in a manner similar to the Pb/Pt b i - e l e ~ t r o d e ~ (see ~ ' ~ 'Section ' 11.3), which is serviceable at 250 A m-,. A lead, 1% Ag, 0.5% Bi or 0.5% Te alloy with a platinum micro-electrode will perform well at 500 A m-,. In environments containing sulphuric acid, the introduction of cobalt ions into solution reduces the corrosion rate of pure lead markedly'15. There is disagreement over the effectiveness of cobalt with antimonial alloys 1 1 % 114. I18 . It is important to keep a PbO, electrode well above the PbO,/PbSO, potential or rapid corrosion will occur''6. Anodes for the electrowinning of copper have traditionally been made of antimonial lead, but new high purity processes have necessitated a change to lead-calciumtin alloy^'^^.^''. The less porous nature of the PbO, layer reduces the amount of lead transferred to the cathode to 2-3 ppm. In the electrolytic recovery of zinc, the traditional anode has been made of an arsenical hypoeutectic lead-silver alloy known as Tainton lead. Lead-thallium has been reported to show good resistance"', and interest in the excellent behaviour of lead-calcium-silver Izo has recently been revived Other alloys used are lead-tin and lead-silver-tin. Anodes made by sintering powdered lead, alone and with additions of cobalt and of silver have been reported to have good corrosion resistance'22. More exotic approaches involving embedded catalytic particles and catalytic particles in a semiconducting polymer coating have also been suggested. Chrome plating anodes, tanks and pipelines are normally made of lead containing 6% antimony despite occasional and sometimes spectacular failures. The electrolyte is usually chromic and sulphuric acids, with fluosilicic acid in the case of mixed catalyst baths. Tanks are sometimes treated anodically when new to produce a PbO, coating. If this is subsequently damaged, severe local corrosion can occur. Flame treatment of the surface of tank linings has been found to be beneficial, and treatment of anodes has been advocated. Rolled or extruded anodes are generally preferred. Corrosion is controlled by the sulphate concentration of the bath, with maximum corrosion at 4-9g/l depending upon temperature. It has been found that 0 5 g/l of magnesium fluosilicate supresses corrosion without affecting the plating process12'. With high efficiency baths, all alloys suffer rapid and severe pitting corrosion when no current is passing. Relatively little attack occurs with BS 334 type A, Type B2, type C (8%) and 7% Sn/Pb in mixed catalyst baths, whereas with simple sulphate-catalyst baths, type B2 and
4:92
LEAD A N D LEAD ALLOYS
tin-lead are attacked at room temperature and at 40°C, and types A and C are attacked at room temperature only.
Battery Corrosion The complex nature of the lead-acid battery is dealt with in several excellent I1,14,20,45, 131 . Lead acid batteries typically consist of lead alloy supports which carry an electrochemically active mass, the composition of which differs between positive and negative plates, and with the state of charge of the battery. Failure normally occurs in the positive grids of a battery. The main cause of failure is loss of contact between the grid and the active mass due to 'grid growth' which is caused by the change in volume of the active material during the chargeldischarge cycle, and by corrosion of the metal surface, which can be accelerated by stress. The process of grid growth is restricted by utilising a sufficiently strong grid. This is achieved by a combination of appropriate alloy composition and physical characteristics of the grid such as dimensions of the grid wires. Grid growth is rapidly accelerated in the event of intergranular corrosion occurring. The corrosion rate is greatest close to the reversible PbO,/PbSO, potential as a result of a solid state reaction between PbO, and the underlying lead surfacel16. This corresponds to the rest or open circuit condition. Passivation at the metal/active mass interface, or of the active mass itself can also lead to failure. Detrimental changes in the morphology of the active mass and microstructural changes in the grid material can also occur. Traditionally battery grids have been made from lead with 6-14% antimony with a small amount of arsenic. While alloys in the region of 5-6070 antimony are still used in some industrial, deep discharge and traction applications, high antimony contents have been largely replaced in automotive batteries by complex low antimony or lead-calcium-(tin) alloys. The reduction in antimony content has been made possible by the introduction of nucleants. It is increasingly common to find different alloys used in the positive and negative grids. Also expanded grids from wrought alloys are now widely used. High antimony alloys exhibit high strength, good castability and give good deep cycling performance. The latter requires that the active mass has good adhesion to the metal, is structurally stable during cycling and does not passivate. Recent work has confirmed that antimony reduces shedding of active material '34, produces a surface film of greater porosity which becomes more porous during cycling, that it promotes stability of the active mass, and has shown that PbSO, is more reluctant to nucleate on antimonial lead132.Although corrosion rates may appear quite high, attack is normally of a general nature which allows a satisfactory service life. This is because the eutectic is preferentially corroded, which reduces intergranular corrosion. Antimony reduces the oxygen overpotential on the positive grid. SbS+ions can migrate from the positive grid to the negative and be reduced to metallic a n t i m ~ n y ' ~ 'This . reduces the hydrogen overpotential, leading to excessive gassing, thus consuming water from the electrolyte, reducing charge efficiency and liberating stibine. During overcharge,
LEAD A N D LEAD ALLOYS
4:93
antimony increases the rate of formation of the inner corrosion layer on the positive grid. Low antimony alloys typically contain less than 3% antimony, with some alloys containing as little as 0.6%. The most commonly used alloys are 1.3-1.8% Sb. They always contain As to assist hardening, and a nucleating agent such as Se or S with Cu. These are necessary because the coarse dendritic structure is prone to porosity and hot cracking during casting. The addition of nucleating agents gives a fine grained structure with good corrosion resistance. Tin is often added to increase fluidity in casting alloys. The reduced antimony content allows the production of low-maintenance batteries which require the addition of water infrequently in the second half of their service life; corrosion is also reduced. There is still enough antimony present, however, to prevent premature failure by passivation of the positive grids. Lead-calcium-(tin) alloys are used in maintenance-free automotive starting lighting and ignition (SLI) batteries, in stationary batteries 140,14* and have been suggested for applications such as negative grids in some traction batteriesIa. It is essential that a correct calcium content and a suitable calcium-tin ratio is used. In the binary lead-calcium alloys, a fine grained structure with serrated grain boundaries is produced by a discontinuous precipitation reaction. It is thought that the serrated nature of the grain boundaries reduces the severity of intergranular corrosion. The addition of tin changes the nature of the precipitation reactions to give two areas of stability. One is with high calcium-low tin and the other is in the region below 1.8% tin and less than 0.07% calcium. Outside these areas, a secondary precipitation reaction occurs which eventually gives a structure which is very susceptible to corrosion136;this is in the form of both deep penetrating corrosion and general corrosion. The secondary precipitation reaction takes place at the grain boundaries and can be initiated by a brief high temperature excursion after a considerable service life. Failure can subsequently occur in weeks. In the regions of stability, these alloys are very stable and exhibit high corrosion resistance. The behaviour of wrought lead-calcium-(tin) alloys is completely different but with the appropriate composition and processing conditions the final structure is extremely corrosion resistant I4l. Lead-calcium alloys containing tin are generally less corrosion resistant, but less prone to passivation. The formation of semiconducting SnO, in the film on the positive grid tends to reduce passivation caused by deep discharge. It is common, therefore, to use the ternary alloy for positive grids and binary for the negative, which saves on the considerable cost of tin. Despite the inferior deep-discharge performance and reduced stability of active mass structure, acceptable life can now be achieved in SLI batteries, and use in other areas is increasing. Batteries made from these alloys have a much reduced rate of self-discharge compared with antimonial alloys, thus giving a longer shelf-life, and maintain a high discharge voltage throughout their life. P.C. FROST E. LITTAUER H. C. WESSON
4:94
LEAD AND LEAD ALLOYS
REFERENCES 1. Hine, F., Ogata, Y., Yasuda, M. ‘Consumption of Lead-silver Alloy Anodes in Sulfuric Acid, B. Electrochem., 4, 61-65 (1988) 2. Sumitoma Metal Industries Ltd Japan Pat. 5928598 (1984) 3. Prengaman, R. E., ‘Structure Control of Non-Antimonial Lead Alloys via Alloy Addi-
4.
5. 6.
7. 8. 9. 10. 11. 12. 13.
14. 15.
tions, Heat Treatment and Cold Working, Pb80, Ed. Proc. 7th Inr. Lead Conf., Madrid, Lead Development Association, London (1983) Kawabata, R., Miyase S. and Tagaya, M. ‘Effects of Alloying Elements on Flash Points of Lead‘, Trans. J.I.M., 5, 85 (1964) Rutter, J. W. and Aust, K. T., Truns AIME, 218, 682 (1960) Heubner, U. and Reinert, M., ‘Effect of Small Silver Contents on the Characteristics of Lead and its Alloys’, Pb80, Seventh Internutionul Leud Conference, Lead Development Association, London Lingane, J. I., Amer. Chem. SOC.,60, 724 (1938) Landolt-Bornstein, Physikulisch-Chemische Tabellen ond Ergunzungsbunde, Berlin, Springer 1923, 1927, 1931, 1935, 1936 Bullock, K. R., ‘Electrochemical and Spectroscopic Method of Characterising Lead Corrosion Films’, J. Electrounul. Chem., 222, 347-366 (1987) Von Fraunhofer, J. A. Anti-Corros. Nov. (1968) and Dec. (1968) Kuhn, A. T. ed. The Electrochemistry of Leud, Academic Press, London (1979) Pavlov, D., in Procs. of the Symposium on Advances in Leud-Acid Butteries, Bullock K. R. and Pavlov, D., (Eds), Electrochem. SOC. 110, (1984) Shreir, L. L. and Hayfield, P. C. S., in Cathodic Protection, Theory und Pructice, Ashworth, V. and Booker, C. J. L. Ellis Horwood, Chichester, 108 (1986) Dasoyan, M. A. and Aguf, I. A., Current Theory of Leud Acid Butteries, Technology Ltd, Stonehouse, with ILZRO Inc. N.Y. 46 (1979) Bullock, K. R., Trischan, G.M. and Burrow, R. G., ‘Photoelectrochemical and Microprobe Laser Raman Studies of Lead Corrosion in Sulphuric Acid’, J. Electrochem. SOC.,
130, 1283 (1983) 16. Bullock, K. R. and Butler, M. A., ‘Corrosion of Lead in Sulphuric Acid at High Potentials’, J. Electrochem. Soc., 133, 1085 (1986) 17. Bullock, K. R. J. Electrochem. SOC.127, 662 (1980) 18. Caulder, S. M. and Simon A. C. in Refs. 11, 43. 19. Dawson, J. L. in Ref. 11, pp. 309 20. Pavlov, D. in McNicol, B. D. and Rand, D. A. J. eds, Power Sources for Electric Vehicles, Elsevier, Amsterdam 142 (1984) 21. Burbank, J.. Simon A. C. and Willihnganz E. in Tobias C. W. ed., Advunces in Electrochemistry and Electrochemicul Engineering, 8, Wiley Interscience, New York 157 (1971) 22. Thirsk, H. R. and Wynne-Jones, W. F. K. Truns. Inst. Metul Finish, 29, 35 (1957) 23. Jones, P., Thirsk, H. R. and Wynne-Jones, W. F. K. Truns. Furuduy SOC., 52, 1003 (1956) 24. Littauer, E. L. and Shreir, L. L. Proceedings of the First International Congress on Metullic Corrosion, London (1961), Butterworths, London 374 (1%2); Shreir L. L. Corrosion, 17, 118t (1961) 25. Pavlov, D. and Rogachev, T. ‘Dependence of the Phase Composition of The Anodic
26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.
Layer o n Oxygen Evolution and Anodic Corrosion of Lead Electrode in Lead Dioxide Potential Region’, Electrochim. Acta., 23, 1237 (1978) Bullock, K. R. ‘Electrochemical and Spectroscopic Methods of Characterising Lead Corrosion Films, J. Electroanul. Chem., 222, 347 (1987) Pavlov, D. and Popova, R. Electrochim-Acto, 15, 1483 (1970) Ruetschi, P. J. Electrochem. SOC., 120, 331 (1973) Dini, J. W. and Helm, J. R. Metul Finish, 67, (8/1969) Beck, F. in Ref. 11, pp. 70 Hampson, N. A. in Refs. 11, 30 Jones, P. Lind, R. and Wynne-Jones, W. F. K. Trans. Faraday SOC.,50,972 (1954) Littauer, E. L., P h D Thesis University of London, (1961). Ives, D. G. and Smith, F. R. Truns. Furuduy Soc., 63, 217 (1%7) Smith, F. R. Disc. Furuduy SOC., 56, 113 (1973) Hayes, M. and Kuhn, A. T. in Ref. 11, pp. 207
LEAD AND LEAD ALLOYS
4:95
37. Salzberg, H. W. J. Electrochem. SOC., 100, 588 (1953) 38. Delahay, P., Pourbaix, M. and Van Rysselbergh, P. J. Electrochem. Soc., 98, 57 (1951) 39. Carr, J. P . and Hampson, N. A. ‘The Lead Dioxide Electrode’, Chem. Rev., 72, 6, 679-703 (1972) 40. Tranter, G. C . ‘Patination of Lead: An Infra-red Spectroscopic Study’, Br. Corros. J . 11, 4, 222 (1 976) 41. Olby, J. K. J . Inorg, Nucl. Chem., 28, 2507 (1966) 42. Hill, R. H., Frost, P. C. and Smith, R. ‘Corrosion of Aluminium in Contact with Lead in Atmospheric Environments’, in: Pb80, Ed. Proc. 7th Int. Lead, Conf., Lead Development Association, London 194-203 (1983) 43. Hill, R. H., Frost, P. C. and Smith, R. ‘Various Aspects of Weathering and Corrosion of Lead in Building Applications’, in: PB83 Ed. Proc. 8th Int. Lead Conf., Lead Development Association, London, 103 (1985) 44. Boffardi, B. P. and Sherbondi, A. M. ‘Control of Lead Corrosion by Chemical Treatment, Paper 445, Corrosion ’91 NACE Conf., Cinncinnati, Mar. 11-12 (1991) 45. Brown, H. E. Lead Oxide Properties and Applicafions, ILZRO. New York 194-234 (1985) 46. US Govt. ‘Safe Drinking Water Act’, (1974), Amendment (June 1991) 47. Brill, R. H. Ed. Science and Archaeology, MIT Press 91-99 (1971) 48. Hofmann, W. Lead and Lead Alloys Properties and Technology, Springer-Verlag, English Trans. by Lead Development Association, London 302 (1970) 49. Wesson, H. C. Corrosion Prevention and Control, 6, 9, 12 (1959) 50. Evans, U. R. An Introduction to Mefallic Corrosion, 2nd Ed. Arnold, London (1963) 51. Heap, J. H. J . SOC. Chem. Ind., 32, 771 (1913) 52. Hawkes, C. A. Chem. Ind., 264, (1944) 53. Miles, G. J. SOC. Chem Ind., 67, 10 (1948) 54. E.E.C. Directive 80/778/EEC, o n the Quality of Running Water Intended for Human Consumption, July 1980. 55. Beccaria A. M. et a/., ‘Corrosion of Lead in Sea Water’, Br. Corros. 1. 17, 2, 87 (1982) 56. Haehnel, 0. ETZ 60, 713 (1939) 57. Ref. 48, p. 315 58. Glander, F. and Glander, W. Z. Metallk, 44, 97 (1953) 59. Amy, L. and Moujnos, C . Rev. Gen. Elect. 66, 187 (1957) 60. Reiner, S. Z. Mefallkde, 30, 277 (1938) 61. Borel, 1. Bull. Ass. Suisse Electr., 28, 54 (1937) 62. Gosden, J. H. Chem. and Ind., 1069 (1956) 63. Radley, W. G. Electr. Engng. 57, 168 (1938) 64. Uhlig, H. H. The Corrosion Handbook, 5th Ed., New York/London (1955) 65. Hornung, R. Techn. Mitt. P T T . , 31, 265, 318 (1953) 66. Compton, K. G. Corrosion, (Houston), 12, 37 (1956) 67. Doyle, E. J. Corrosion, (Houston), 11, 17 (1955) 68. Robinson, H. A. and Featherly, R. L. Corrosion, (Houston), 3, 349 (1947) 69. Ref. 48, p. 313 70. Beccaria, A. M. et a/., ‘Investigation on Lead Corrosion Products in Sea Water and In Neutral Saline Solutions’, Werksfofle und Korros., 33, 416-420 (1982) 71. Romanoff, M. Underground Corrosion, NBS 579, National Bureau of Standards, April 227 (1957) 72. Robson, W. W. and Taylor, A. R. Some Experiments in The Mechanism of Corrosion of Lead Pipes in Soils, Report MM/19/54, Associated Lead Manufacturers Ltd. (1954) 73. Denison, I. A. and Romanoff, M. J. J. Res. Nut. Bur. Stand, 44,259 (1950) 74. Beavers, J . A., Koch, G. H. and Berry, W. E. Corrosion of Metals in Marine Environments, MCIC Report MCIC-86-50 (1982) 75. Cook, A. R. and Smith, R. ‘Atmospheric Corrosion of Lead and Its Alloys’, in Ailor, W. H. Afmospheric Corrosion, Wiley Interscience, New York (1982) 76. Logan, K. R. Underground Corrosion, Nat. Bur.-Stand, Circular C450 (1945) 77. Haase, L. W. Werkst. U.Korros. 2, 90 (1951) 78. Burns, R. M. Bell Sysf. Tech. J . , 15, 603 (1936) 79, Burns, R. M. and Salley, W. J. Indust. Eng. Chem., 22, 293 (1930) 80. Ref. 48, p. 310 81. Ref. 48, p. 293 82. Radley, W . G. and Richards, C. E. J . Inst. Elect. Engng. 35, 685 (1939)
4:96
LEAD AND LEAD ALLOYS
Senez, J. C. and Pichinoty, F. F . Corros. Anticorros, 5, 203 (1957) Cole, E. L. and Davies, R. L. Chem. and Ind. (Rev.) 39, 1030 (1956) Bonde, G. and Lunn, B. Ingen. Intern. Edit. 2 , 103 (1958) Ref. 48, p. 294 Brtif, B. S. and Siftar, J. Corrosion et Anticorrosion, 6 , 342 (1958) Wolff, E. F. and Bonilla, C . F. Trans. Electrochem. SOC., 79, 307 (1941) Schmelling, E. L. and Roschenbleck, B. Werkst. U. Korrosion, Mannheim, 9, 529 (1958) 90. Reinitz, B. B. Corrosion, (Houston), 9, 425 (1953) 91. Wolzogen-Kuhr, C. V. Water, 40, 281 (1956) 92. Ref. 48, p. 317. 93. Lead for Corrosion Resistant Applications, Lead Industries Association Inc., New York 83. 84. 85. 86. 87. 88. 89.
94. 95. 96. 97. 98. 99. 100.
101. 102. 103. 104. 105. 106. 107. 108.
109.
( 1974) Corrosion of Lead, Lead Development Association, London (1971) Corrosion Resistance of Lead and Lead Alloys, Chem. Eng. Feb (1953) Rabald, E. Resisrance of Lead to Corrosives,from Corrosion Guide, (1951) Lead in Modern Induslry, Lead Industries Association, New York (1952) Frost, P. C. The Corrosion of Lead and Lead Alloys in Bromine, Report MM/4/8 1, Lead Industries Group Ltd., (1981) Wickert, K. Korros-Metallsch. Beih. 20, 147 (1944) Fontana, M. G. Industr. Engng. Chem., 43, 9, lO5A (1951) Ref. 48, p. 291 Camuil, J. Sci. Industr. Res. 20, 114 (1944) Turnbull, D. and Frey, D. R. J. fhys. Colloid Chem., 51, 681 (1947) Coles, E. L., Gibson, J . G. and Hinde. R. M. J . Appl. Chem. 8 , 341 (1958) Wilson, B. S. and Garner, F. H. J. Inst. Petrol., 37, 225 (1951) Vaivads, A. and Liepina, L. Latv. P.S.R. Zinat. Akad. Veslis. 8 , 119 (1954) Katz, W. Metalloberjlache, 11, A.161 (1953) Heubner, U. and Reinert, M. Development of Improved Lead Materials for Chemical Plant, Pb8O 7th Int. Lead Conf., Madrid, Lead Development Association, London 204-214 (1983) Heubner, U. and Reinert, M. Eflect of Small Silver Contents on the Characteristics of Lead and its Alloys, Pb8O 7th Int. Lead Conf. Madrid, Lead Development Association, London 130-137 (1983)
110. Barnard, K. N., Christie, G. L. and Gage, D. G. ‘Service Experience with Lead Silver Alloy Anodes in Cathodic Protection of Ships’, Corrosion, 15, 11, 581-586 (1959) 1 1 I . Peplow. D. B. and Shreir, L. L. ‘Lead/Platinum Electrodes for Marine Applications’, Corr. Tech. Apr. (1984) 112. Morgan, J. H. ‘Lead Alloy Anode for Cathodic Protection’, Corr. Tech. 10/12. 348-352 (1958) 113. Koch, D. F. A. Electrochimica Acta, 1, 32 (1959). 114. Eggett, G. and Naden, D. ‘Developments in Anodes for Pure Copper Electrowinning from Solvent Extraction Produced Electrolytes’, Hydrometallurgy, Elsevier, Amsterdam, 1, 123-137 (1975) 115. Lander, J. J. J . Electrochem. SOC.,99, 467 (1954) 116. Lander, J. J. J . Electrochem. Soc., 103, 1 (1956) 117. Prengaman, R. D. Wrought Lead-Calcium-Tin Anodes for Electrowinning, AIME Conference, Los Angeles, 28/2/84. 118. Gendron, A. S., Ettel, V. A. and Abe, S. ‘Effect of Cobalt Added to Electrolyteon Corrosion Rate of Pb-Sb Anodes in Copper Electrowinning’, Canad. Met. Quart., 14, 1, 59-61 ( 1975) 119. Ref. 48, p. 298. 120. Hanley, H. R., Clayton, C. Y. and Walsh, D. F. Trans. A I M E , Yearbook 91, 275 (1930) 121. UK Pat. App. GB 2149424A (1983) 122. Kir’ yakov, G. Z. and Dunaev, Yu. D. Izvest. Akad. Nauk, Kazakh. S . S . R . Ser. Khim, 2, 32 (1957) 123. UK Pat. App. GB 2085031A (1980) 124. UK Pat. App. GB 2096643A (1981) 125. Carter, V. E. and Campbell, H. S. J. Met. Finish., 8 , 103 (1962) 126. Corrosion of Lead Roofing, Interim Report for Ecclesiastical Architects; and Surveyors’ Association (1986) 127. Joyce, S. J. Thesis, Brighton Polytechnic (1983)
LEAD AND LEAD ALLOYS
4:97
128. Murdoch, R. ‘The Lead Sheet Manual’, Vols 1,2, and 3, Lead Sheet Association, London, (1990, 1992, 1993) 129. Heubner, U. et al., Metallgesellschaft ‘Lead Handbook’ (1983) 130. Ref. 48, pp. 347 131. Sharpe, T. F. in Bard, A. J . Encyclopedia of Elecfrochemisfry of the Elements, 1, 235 (1974) 132. Webster, S., Mitchell, P. J., Hampson, N. A. and Dyson, J. I. ‘The Cycle Life of Various Lead Alloys in 5M H2S0,, J . Electrochem. SOC., 1, 133 (1986) 133. Peters, K. and Young, N. R. ‘Some Aspects of Corrosion in Lead Acid Batteries, I. Chem. E . Symposium Series, 98, 185 (1980) 134. Gibson, I. K., Peters, K. and Wilson, F. in Thompson, J . Ed., Power Sources, 8, Academic Press, London 565 (1980) 135. Dawson, J. L., Wilkinson, J . and Gillibrand, M. I. in Collins, D. H. Ed., Power Sources, 3, Oriel Press, Newcastle-upon-Tyne, 1-9 (1970) 136. Prengaman, R. D. ‘Structural Control of Non-Antimonial Lead Alloys Via Alloy Additions, Heat Treatment and Cold Working’, Pb80 Ed., froc. 7fh Int. Lead Conf., Lead Development Association, London 34 (1983) 137. Bystrom, J. A r k . Kemi, Miner01 Geol., 20A, 11, (1945) 138. Katz, T. Ann. Chim., (Paris), 5 , 5 (1950) 139. Barnes S. C. and Mathieson, R. T. in Simon, A. C. Batteries (2), Collins, D. H. Ed., Pergamon Press, New York 41 (1%5) 140. Ref. 20 p. 212 141. St. Joe Minerals Corporation, U.K. Patent 1 338 823 (1970) 142. Prengaman, R. D. Improvements in Alloys, Oxides ond Expanders for Leod Botferies, Lead Development Association, London 3 (1984)
4.4
Magnesium and Magnesium Alloys
Magnesium is a divalent metal and is silvery white in appearance. It is the eighth most abundant element and sixth most abundant metal. The atomic weight is 24.32 and the specific gravity of the pure metal 1 -738 at 20°C. The structure is close packed hexagonal. The melting point is 650°C and the boiling point 1 107°C. The specific heat at 20°C is 1 *030kJ/kg "C and the thermal conductivity at 20°C is 157.5 W/m"C; the electrochemical equivalent is 0*126mg/C. The standard electrode potential E ; , Z + , ~=, -2.37 V, but in 3% sodium chloride the electrode potential is -1-63 V (vs S.C.E.), i.e. - 1 -38 V (VSS.H.E.). For engineering purposes magnesium is rarely used in the unalloyed condition. Small percentages of aluminium, zinc, etc. as indicated in Table 4.15 are added to improve mechanical and other properties. Magnesium is itself used for alloying with other metals. Applications of magnesium alloys are many and varied but their light weight - about two-thirds that of the aluminium alloys - and high strengthto-weight ratio have made them of particular interest to the aircraft and guided-weapons industries. In the transport industry too, their light weight is attractive. Other features for which they are noted are their high stiffnessto-weight ratio, their great ease of machinability, good casting qualities and high damping capacity. In the nuclear engineering field special magnesium-base alloys are extensively used as canning materials for uranium in gas-cooled reactors. The compositions of the more common magnesium-rich alloys used in Great Britain are given in Table 4.15. Similar alloys are in use in the USA and elsewhere and their American designation together with the equivalent British alloys are given in Table 4.16. Many of the casting alloys are given various simple heat treatments to improve their properties, while the wrought alloys can be obtained in a number of tempers. As with other electronegative metals, the absence of serious corrosion of these alloys in ordinary industrial atmospheres is largely a result of the formation of protective films which inhibit further attack. Similarly, when serious corrosion does occur, or when it occurs after a period of successful use, it can usually be traced to a change in conditions of such a nature that protective films already formed have suffered dissolution or break-down. No alloying ingredients are known which effect any substantial improvement 4:98
Table 4.15
Nominal composition of magnesium-rich alloys % Composition (remainder magnesium)
Designation
Casting aloys
Z5Z RZS TZ6 MSR A MSR B ZRE 1 ZT1 MTZ AS
AZ91 AZ91X C AZG ZE63 Wrought alloys ZW3 1
zw
ZW6 ZTY AM503 AZ3 1 AZM AZ855 ZM21 ZM61
AI
Zn
Zr
Th
-
4,5 4.0 5.5
0.7 0.7 0.7 0.6 0-6 0.6 0.7
-
-
-
-
-
2.2 2.2
8.0 9.5 9.4 7.5-9.5 6.0
0.5 0.5
-
-
-
-
-
3.0 6.0 8.0
-
-
0.4 0.3-1.5 3.0 6.0 3.0 1.3 5.5 0.5
-
1.0 1.0
0-4 2.0 6.0
0.7
-
0.6 0.6 0-6 0.6 0.6 -
-
-
1.8
3.0 3.0 -
-
Rare earth metah
Fractionated rare earth metals
Be
Mn
-
-
-
1.2
-
-
2.7 -
1.7 2.5 -
-
-
-
-
2.5
-
-
-
0.75 -
-
-
-
-
-
-
-
-
-
-
-
-
-
0.0015
-
-
0.3 0.3 0.3 0.15 (min) 0.3 -
Ag
-- 50 -
2.5 2.5
-
-
-
-
-
-
-
-
-
1.5
-
0.3 0.3 0.3 1.0 I .o
-
-
I
>z
0
-- 50 -- Ez
-
-
5
2
E
- F5 v)
?!
3
4: 100
MAGNESIUM AND MAGNESIUM ALLOYS Table 4.16
Equivalent British and American designations for magnesium alloys
Wrought alloys British designation ZW6 AM503 AZM AZ855 ZM21 ZM6 I
A mericon designation (AsTM) ZK60A MIA AZ61A AZ8OA ZM21 ZM61
Cast alloys British designation ZSZ RZ5 TZ6 MSR ZREl ZT 1 MTZ A8 A291 ZE63
American designation (ASTM) ZK5lA ZE4 I A ZH62A QE22A EZ33A HZ32A HK31A AZ8 I A AZ91 ZE63
in the general corrosion behaviour of the magnesium alloys; though manganese is usually regarded as beneficial this is probably because it offsets the deleterious effects of iron and other cathodic metals which may be present rather than because it makes any positive improvement in the resistance of the magnesium itself. All metals added to produce the various alloys are intended simply for the improvement of the physical and mechanical properties and not for any effect they may have on the corrosion behaviour. On the other hand, none of these additions has any very marked deleterious effect on the corrosion resistance of the alloys. Since corrosion resistance depends on film formation, it follows that the behaviour of the alloys will vary considerably with the medium to which they are exposed. The corrosion of the metal is governed largely by the solubility and other characteristics of the film. Thus, magnesium fluoride is very insoluble in hydrofluoric acid and as a consequence magnesium does not dissolve in this acid. Initial attack forms a film of magnesium fluoride and even though the film is not impervious to other corrosive influences it effectively seals the metal against further reaction. In dilute aqueous hydrofluoric acid, attack may take place, and if so it will be of a pitting type similar to that which takes place in tap water. In fact the corrosion is due to the water itself and not to the acid. Magnesium sulphate on the other hand is readily soluble in dilute sulphuric acid and no protective film is formed when magnesium reacts with this acid; attack is rapid and continuous with evolution of hydrogen. It should be noted, however, that magnesium sulphate is only slightly soluble in concentrated sulphuric acid. When, therefore, magnesium is immersed in strong sulphuric acid, initial attack produces a film of magnesium sulphate which quickly saturates the acid at the interface, and the reaction is reduced to a vanishingly low rate. As long as water is excluded no further attack takes place. In considering the corrosion behaviour of magnesium alloys, therefore, it is of the utmost importance to know the nature of the medium to which the metal is to be exposed. In general, atmospheric attack in damp conditions is largely superficial; aqueous solutions bring about attack which varies not only with the solute but with the volume, movement and temperature
Standard electrode potentials (S.H.E.)
Table 4.17
Metal
Magnesium A I umi ni um Zinc Iron
Potential
V -2.37
-1.66 -0.76
-0.44
Table 4.18 Steady-state potential of several metals in 1 M solutions of different electrolytes (vs 0.I
N
Calomel, E, = 0.336 V)
Potential (V) Metal
Magnesium Aluminium Zinc Iron
Sodium chloride
Sodium sulphate
Sodium chromate
Hydrochloric acid
Nitric acid
Sodium hydroxide
Ammonium hydroxide
Calcium hydroxide (sat'd)
Barium hydroxide (sat'd)
-1.72 -0.86
-1.75 -0.50
-O.% -0.71 -0.67 -0.16
-1.68
-1.49 -0.49 -1.06 -0.58
-1.47 -1.50 -1.51 -1.22
-1.43 -0.80 -1.50 -0.18
-0.95 -1.54 -1.40 -0.30
-0.88 -1.53 -1.49 -0.25
-1.15
-1.19
-0.72
-0.76
-0.80 -1.14 -0.66
4: 102
MAGNESIUM AND MAGNESIUM ALLOYS
of the liquid. Many organic liquids are quite inert to magnesium, but some of those which contain reactive polar groups, as might be expected, are reactive in some degree towards the metal. Galvanic corrosion of magnesium, Le. the enhanced corrosion to which the anodic member of a pair of metals in contact is subject to when both are in contact with a common electrolyte, is of considerable practical importance, since magnesium is anodic to all other structural metals in most electrolytes. The standard electrode potential of magnesium is given, along with the potentials of other metals, in Table 4.17 and the steady-state potentials of magnesium in various solutions are listed in Table 4.18'.
Corrosion of Magnesium Alloys in Atmospheric Conditions In clean, dry atmospheres (with r.h. below about 60%) uncontaminated magnesium alloys retain a lustrous surface almost indefinitely. If the atmosphere is clean but not dry and the humidity approaches 100% a scattered pattern of corrosion spots appears after a period, but considerable areas of unaffected surface remain for a very long time. (This effect is probably attributable to galvanic corrosion on a micro-scale, each spot representing corrosion of the magnesium adjoining a cathodic particle contained in the surface as an impurity.) It is quite otherwise if the surface has been contaminated either by corrosive dusts or by cathodic particles introduced by abrasive treatments of various kinds. For example, if a piece of magnesium which has been shot blasted is exposed to damp conditions, the whole surface rapidly becomes covered with a greyish layer of corrosion product. In drier atmospheres the layer may resemble a patina and develop very slowly. In many ways the corrosion of magnesium alloys in normal atmospheric conditions is a close approximation to the initial formation of rust on mild Table 4.19 Corrosion rates of magnesium-rich alloys (g m - * d - ' ) in three different environments Type of tesi Alloy
A8 AZ91
z5z ZREl RZ5
TZ6 High-purity A8
Immersion for 30 days in 3% NaCl solution 3.5-133.0 av. 72.7 1.3-144.0 av. 52.6 5.5-9.3 av. 8 . 0 20.5-32.7 av. 26.6 12.0-40.2 av. of 26 tests 32.6 26.8-67.6 av. of 19 tests 53.4 3.0
Sea-water spray 3 times per day for 6 months 1 5-2 5 av. 2 . 0 0'5-2.3 av. 1 . 7 0'7-0.9 av. 0 . 8 1.2-2.0 av. 1.6
Atmospheric exposure for 2 years 0.22-0 26 av. 0.24 0.19-0.26 av. 0.23 0.34-0.35 0.34 0.36-0.37 av. 0.36
1.1-1.6
av. of 12 tests 1 . 3 0.9-1.2 av. of 4 tests 1 . 1 0.4
0.40-0' 44 av. of 4 tests 0.41 0.28
MAGNESIUM A N D MAGNESIUM ALLOYS
4: 103
steel when similarly exposed and is just as superficial. Study of the crosssections of a number of test bars of different alloys which were exposed to the weather for three years at Clifton Junction, Manchester, showed that the loss of section was remarkably small and the deterioration of mechanical properties was not greater than could be attributed to this and to the roughening of the surface which had occurred. Corrosion rates in normal industrial atmospheres measured as loss of weight over a period are extremely uniform among the various alloys. Table 4.19, last column, gives the corrosion rates (in g m-’ d-’) for a number of alloys determined at Clifton Junction in recent years. The highest value recorded (0.4 g m-’ d-’) is equivalent to a rate of penetration of 0.076 mm/y, which is appreciably less than that of mild steel. Composition of Corrosion Product
The compound produced by the interaction of magnesium with moist air appears to be essentially magnesium hydroxide (S, = Magnesium oxide reacts slowly with water to form hydroxide so that films of oxide formed at high temperatures, during hot-forming operations for example, eventually become hydrated. The final composition of the corrosion product undoubtedly depends upon the nature of the atmosphere and the gases it contains. Thus the natural air-formed corrosion product in industrial areas always consists largely of carbonate and sulphate in addition to hydroxide. In marine atmospheres magnesium chloride is formed and eventually oxychloride by reaction with magnesium hydroxide formed at the same time. Since the chloride is hygroscopic, moisture is attracted and the corrosive effect is hence much worse than that of water alone. Marine A tmospheres
If by a ‘marine atmosphere’ one means the conditions at a site within a few metres of high-water mark in an otherwise clean country atmosphere, then corrosion of unprotected magnesium alloys is remarkably small. The presence of actual liquid spray is exceptional, and during periods of high wind, when such spray is generally produced, the humidity is usually low and evaporation of the droplets is rapid. This explains why crystals of salt may be found on exposed metal yet little evidence of severe corrosion may be seen. Salt solution is corrosive but dry salt particles are almost without effect. Two other factors which operate in such regions are: (a) the washing effect of clean rain followed by rapid evaporation; and (b) the scouring by blown sand particles. Unpainted magnesium items exposed near the sea coast invariably exhibit a dull, slightly rough surface, with no corrosion product evident. This appearance is in sharp contrast to that of metal which has been exposed to conditions of persistent dampness; in such cases the corrosion product, considerably more voluminous than the metal from which it is formed, is retained on the surface as a greyish white powder. The glossy surface of painted specimens and the polish of newly prepared bare metal specimens exposed near the sea coast is also quickly lost. Corrosion rates of
4: 104
MAGNESIUM AND MAGNESIUM ALLOYS
painted specimens which have been exposed so long that the paint has perished are often higher than those of bare metal. The cause of this is probably the retention of water by the sponge-like texture of the deteriorated paint*. In a similar way corrosion may be most in evidence at junctions and narrow gaps. This is due not so much to true ‘crevice corrosion’ as to the retention of moisture. Surface moisture evaporates quickly, and the incipient corrosion to which it gives rise quickly comes to an end. In a narrow recess, however, trapped moisture, unable to evaporate, continues the corrosion process over long periods. True crevice corrosion on the other hand is a phenomenon which owes its nature to the development of anodic areas within the crevice, caused by exclusion of oxygen for example, and does not appear to occur with the magnesium alloys. In conditions where sea-water spray may be deposited regularly on magnesium articles with no alleviating mechanism for its removal, or where breaking waves may drench the components, the effect is quite different. Corrosion of bare metal will be heavy and will be intensified at junctions with other more noble metals. Unless magnesium alloys can be adequately protected in such combinations it is better to avoid their use. This matter is dealt with under the section on protection. In an experiment similar to that referred to on p. 4.100, tensile test bars were exposed at Clifton Junction, Manchester, for six months, during which time they were sprayed three times daily with sea-water. Whereas exposure to industrial atmosphere alone had little effect, bars of the same alloys were much more heavily attacked by sea-water spray.
High-purity Alloys and Galvanic Corrosion In addition to the alloying ingredients which are added, certain other metals are usually present in small amounts. In the alloys which contain aluminium, for example, iron usually amounts to about 0.02-0-05%.By special techniques and care in melting this can be reduced to about one-tenth of the above figure. Many workers have shown that such high-purity alloys have a markedly better resistance to salt water than those of normal purity, but their behaviour towards industrial atmospheres is not greatly different. Furthermore, the practical value of the higher resistance to corrosion is largely offset when components are used in electrical contact with other more cathodic metals. The effect of a steel bolt for example, even when it has been zinc or cadmium plated, is much greater at the point of contact than that of the excess of local cathodes in the impure alloys. Galvanic corrosion at joints with other metals therefore is not markedly less in the case of the high-purity alloys. Nevertheless, such alloys have their place, and when they can be used without other metal attachments provide better intrinsic resistance to corrosion by sea-water than the alloys of normal purity. Alloys containing zirconium as a grain-refining agent have the iron content automatically reduced to about 0.004% by settling out of impurities during the alloying procedure. Private communication from A . P. Fenn, Esq., Birmingham Aluminium Casting (1903) C o . Ltd.
MAGNESIUM A N D MAGNESIUM ALLOYS
4: 105
Reference to Table 4.19 will show that greatly superior corrosion rates of the high-purity alloys are only in evidence in the more severe conditions of test by immersion in salt water, and that in less drastic conditions, and especially in industrial atmospheric exposure, there is little to choose between the alloys. Figure 4.17 illustrates the corrosion occurring on high-purity AZ3 1 and ZW3 in contact with steel bolts. Tested alone in sea-water, the corrosion rate of the former is much the lower. It is evident from the illustration, however, that the governing factor in galvanic corrosion is the type of electrolyte present rather than the composition of the alloy.
Nature of the Corroded Surface When corrosion occurs on a smooth machined magnesium alloy surface, this surface is roughened by the chemical action, and after the initial attack the degree of roughness does not change appreciably. In the usual industrial atmospheric conditions the attack is uniform, but in immersed conditions, including corrosion under pools of condensate, attack may be, and usually is, irregular; some areas become anodic to other areas and, as corrosion proceeds at the former, a pitted effect results. Even in atmospheric attack the roughening is really a microscopic form of pitting. There is a noticeable difference between the appearance of the aluminiumcontaining magnesium-rich alloys on the one hand and the zindzirconiumcontaining magnesium alloys on the other. In the former the microscopic pits in the surface which has been exposed to the weather tend to be narrow and relatively deep, while in the latter they are wider and tend to overlap, leading to a slightly wavy appearance. The unequal attack which occurs in tap water, condensate and other mild electrolytes may lead to perforations of thin-gauge sheet and even to deep pitting of castings. In stronger electrolytes the effect is variable. In chloride solutions such as sea-water, attack on the metal usually results in the pitting of some areas only, but where the metal surface has been rendered reactive, as by shot blasting, attack may be so rapid that uniform dissolution over the whole surface may occur. In either case magnesium-base alloys are not usually suitable for use in aqueous liquids since they are not intrinsically resistant to these electrolytes.
Methods of Corrosion Testing In considering the corrosion of magnesium and its alloys it is important to examine the methods available for assessing corrosion tendencies and particularly those known as accelerated tests. Tests carried out by immersion in salt water or by spraying specimens regularly with sea-water are worthless as a means of determining the resistance of magnesium alloys under any other than the particular test conditions. Extrapolation to less corrosive conditions is not valid and even the assessment of the value of protective measures by such means is hardly possible. The reason is to be found in the fact that corrosion behaviour is directly related to the formation of insoluble
(a)
(b)
(C)
(d)
(e)
Fig. 4.17 Samples of high-purity AZ31 (upper photographs) and ZW3 (lower photographs) magnesium-base alloys, fitted with mild-steel nuts and bolts and exposed to a variety of corrosion conditions. (a) 4 - 5 hours' immersion in 3% salt soh. (b) 180 days' immersion in distilled water, (c) 4 days' immersion in borehole water, ( d ) 180 days' in humidity cabinet sea-water spray and (e) 180 days' atmospheric exposure
MAGNESIUM AND MAGNESIUM ALLOYS
4: 107
films. In chloride solutions there is no stable insoluble film formed from the solution itself and no previously formed film (by chemical reaction) is impermeable to the chloride ion. Even existing protective films are penetrated relatively easily by chloride ions, and organic films of paint or varnish are subjected to osmosis and to swelling in conditions which are quite unlike normal experience. Except for the specific purpose of determining the behaviour of the materials in dilute chloride solutions, accelerated methods of test of this nature are inadmissible and the results misleading. Corrosion in the atmosphere is usually a continuous, relatively gentle, but persistent, process. Since it is continuous the time scale cannot be shortened and to attempt to obtain results merely by increasing the severity of the conditions is illogical. Such tests can only give information about the behaviour in the chosen conditions. The only true indication of behaviour is obtained by exposing samples to the conditions to be experienced, if these are known or can be assessed. If the severity is of an intermittent nature then it is usually permissible to accelerate the time basis by omitting or reducing the intervals of less severe exposure. Even in this case, however, the effects of recovery by drying out, etc. should not be overlooked. In particular, the samples exposed should take account of the effects of joints with both similar and dissimilar materials; acid vapours from wood and plastics and the electric stress introduced by coupling to other metals in the presence of an electrolyte can vitiate completely any deductions concerning the protective value of paints based on tests on isolated pieces of metal.
Intergranular Corrosion (Sections I .3and 1.71 True intergranular corrosion of magnesium alloys does not occur for the reason that the grain-boundary constituent is invariably cathodic to the grain body. It follows, therefore, that corrosion will be principally concentrated on the grains and the grain-boundary constituent will not only be more resistant to attack but will in some measure receive cathodic protection from the corrosion of the neighbouring grain. Examples are occasionally quoted which purport to show attack at the grain boundaries, but this is not intergranular corrosion properly so called; indeed it is the opposite and might better be called granular attack, for it is the grain and not the boundary which is preferentially attacked. Because the grain boundary is cathodic to the grain proper, attack is concentrated on the area of the anodic grain adjoining the boundary until eventually the grain may be undercut and fall out of the matrix. The important difference from true intergranular corrosion is quite clearly that attack can proceed only grain by grain and cannot make its way through the body of the material following a grain-boundary path. As usually cleaned for microscopical examination, a corroded specimen has invariably lost the delicate tracery of intergranular material in the cleaning process and thus may present something of the appearance associated with intergranular attack. If special steps are taken, however, it is possible to mount a corroded specimen of magnesium alloy with the grain boundaries still intact, showing where some grains have entirely disappeared while others are in the process of dissolution round the edges where the cathodic
4: 108
MAGNESIUM AND MAGNESIUM ALLOYS
Fig. 4.18 Corrodedgrain in ZREl showing the grain boundaries still intact. Attack occurs on the periphery of the grains and thus is not intergranular
effect of the grain-boundary material is strongest. This is illustrated in Fig. 4.18 which shows a sample of corroded ZREl with the intergranular constituent still intact in many areas and with one grain almost etched free from the containing network.
Protection of Magnesium Alloys The proneness or otherwise to corrosion is essentially the same in all the magnesium-base alloys and it is important to note that the requirements of protection therefore do not vary for the magnesium alloy under consideration. If conditions are such that any one of the alloys can be used satisfactorily without protection, then any other of the alloys can be so used. On the other hand, if a given protective scheme is found necessary for a particular alloy, then the same protective scheme will be found necessary (and will be equally effective) with any other magnesium-base alloy. It will be realised that since the tendency of magnesium to corrode is governed by the nature of the environment to which it is exposed the degree of protection necessary is also controlled by the same factor. The methods of protection available are of two basic kinds: chemical or electrochemical treatments which oxidise the metal and produce a film which is more stable than the metal itself, and coating methods which rely
MAGNESIUM A N D MAGNESIUM ALLOYS
4: 109
upon the application of some extraneous material to provide a more or less impervious coating and thus restrict access of corrosive influences. Generally the two methods are complementary and for best effects, or for protection under the more drastic conditions, they are used in conjunction with each other. The first class referred to above includes various simple treatments in which the metal is dipped in acidic or near-neutral solutions, usually containing chromate in the hexavalent condition, which has a strong passivating action on many reactive metals. The resulting chemical film which is formed on the magnesium-rich alloys consists essentially of magnesium oxide but also contains chromium compounds. This class of treatment also includes methods of anodic oxidation achieved either by galvanic or electrical means. In general the films resulting from such electrical methods are thicker and often harder than those produced by the simple immersion methods, and usually form more effective barriers against corrosion. Both types, however, should be regarded as suitable for withstanding normal atmospheric conditions for relatively short periods only, since they are all porous or at least permeable to water to some degree. They are, in fact, to be considered as foundation treatments for coating with more impermeable organic materials such as paint and enamel. (See also Section 15.3.) The second class of protective measures includes all those processes generally known as painting, as well as ‘temporary’ protective treatments such as greasing and oiling. It is conventional to apply paint films over one or other of the chromate-containing oxide films (produced in one of the immersion-type baths), but paints can be applied equally well to the electrolytically developed coatings. It is not usually wise to apply paint directly to the bare metal. The reason for this is twofold. In the first place, a chromated surface, especially a freshly chromated surface, is not as likely to have deteriorated in storage as a bare metal surface and consequently will provide a better basis for paint. Secondly, and more important, the natural surface of magnesium alloys in contact with damp air is alkaline, because of the presence of the naturally formed oxide and hydroxide, and this may lead to the rapid deterioration of paint films. This is especially true of the oil-based and some of the synthetic air-drying paints which are sensitive to alkali. In all cases a priming coat of paint containing some form of hexavalent chromium such as zinc or strontium chromate should be used. Table 4.20 lists a number of the better-known processes for producing protective films on magnesium alloys by chemical and electrochemical processes. High Temperature Stoved Epoxy Resins
In recent years use has been made of the strong adhesion, toughness and water impermeability of some of the epoxy resins to secure greatly improved surface protection of magnesium alloys. By this means it has been possible to employ these alloys even in situations where they are drenched repeatedly with sea-water. Not all the epoxy resins are equally efficacious and all have to be stoved at a relatively high temperature (180-220°C) in order to develop the requisite
Table 4.20
D.T.D
process
Spec.
United States Spec.
The more usual cleaning and chromating processes for magnesium-base alloys* purpose
Composirion (‘70 by wt.)
Method of use
Resulting appearance
Suitable container
Clean and bright but may be loose black smut on Mg-Zr-Zn alloys
Glass, polythene, rubber and earthenware
Metal removal
Clean and bright but may be loose black smut on Mg-Zr-Zn alloys
Glass, polythene. rubber and earthenware
Metal removal
Smooth white matt
Rubber, ebonite and polythene Steel
Notes
A ..
c. e
0
Pwkling and cleaning
Nitric acid pickle
91 IC. Sec. 2.1.1
MIL-M-3171C. Sec. 3.2.4
Cleaning rough castings
5-10% conc. nitric
Sulphuric acid pickle
91 IC, Sec. 2.1.1
MIL-M-3171C. Sec. 3.2.4
Cleaning rough castings
2-5% conc. sulphuric acid in water
Fluoride anodising treatment Caustic soda clean
Chromic acid bath
911C. Sec. 2. I .3
-
911C. Sec. 2.1.4
MIL-M-3171C. Sec. Type VII. 3.9 MIL-M-317lC. Sec. 3.2.3
MIL-M-3171C. Sec. 3.4.1
Super cleaning for maximum corrosion Cleaning finished parts or greasy components generally Removal of paint
Removal of old chromate and fluoride films; also removal of corrosion product and oxide films
acid in water
10-30% ammonium bifluoride in water 2-5% caustic soda in water
10-15% chromic
an hydride (Cr03) in water
Repeated brief dips until clean, followed by thorough rinse, preferably with hose Repeated brief dips until clean, followed by thorough rinse, preferably with hose Application of 120 V a.c. for 5-30 min Immerse and boil for 15-30 min; wash thoroughly
Immerse and boil for 15 s to 30 min or as necessary; wash thoroughly
Clean, showing no water ‘break’ when wetted with cold water
Clean grey appearance
I
0
z
E
C
I
Glass and steel
No dimensional change M.E.L.B. Pat. 721 445
No dimensional change Proprietary brands of alkaline cleaners may be used provided these do not attack metal Soda ash with washing soda and soap are suitable for preparing alkaline cleaning baths No dimensional change unless much impurity is present For small delicate parts 0.1-0~5% silver chromate may be added to reduce attack caused by impurities such as chloride in corrosion product
> z
0
I
Q
z rn
g Fr
s v1
Chromaling
Acid chromate bath (I.G. bath) (chrome pickle)
911C. App. I I , Bath iv
MIL-M-3171C. Type 1. Sec. 3.5
Protection in storage Paint foundation Repair of chromate film
15% sodium or
Hot half-hour bath (R.A.E. bath) (black bath)
911C. App. 11, Bath
MIL-M-3171C. Sec. 6.4.1
Good paint foundation Protection in storage
3% ammonium sulphate. 1.5% potassium or sodium dichromate and 0.5-0.75% 0.880 ammonia solution 10% sodium or potassium dichromate. 5% magnesium sulphate and 5 % manganese sulphate
Chromemanganese bath (M.E.L. black bath)
... 111
911C. App. I I , Bath V
MIL-M-3171C. sec. 6.4.1
Good paint foundation Protection in storage
potassium dichromate and 20-25% nitric acid (s.g. 1.42)
Immerse for IO s to 2 min. drain for 5 s, wash in cold or warm water
Golden bronze often with iridescent colours
Immerse in bath and simmer for 30 min; wash
Usually black; light brown on D.T.D. I I S
Glass, earthenware, slate, aluminium and stainless steel Glass, steel and aluminium
Metal removal Cannot be used on parts to fine tolerances Useful for rough castings No dimensional change Used chiefly for finished work
I n 2-
z Immerse in bath for 2 h at 20°C; proportionately less if bath is heated
Usually black or dark brown; light brown on D.T.D. 118
Glass, steel and aluminium
No dimensional change Used chiefly for finished work where heating may be undesirable where inserts are present
Special surface treatments
H.A.E. process
-
Dow 17 process
-
~~
MIL-M-45202B Corrosion and (ORD) abrasion resistance
MIL-M-45202B Corrosion and abrasion resistance
~
Table reproduced by
permission
of Magnesium Eleklron Lld
12% potassium hydroxide, 1% aluminium (high purity) 3.5% trisodium phosphate (crystals), 3.5% potassium fluoride (anhydrous) and 2.2% potassium manganate 24% ammonium bifluoride, 10% sodium dichromate and 8.6% orthophosphoric acid (85%)
Application of up to 90 V ax. for up to 2 h
Dark brown, ceramic-like coating, brittle
Glass, rubber and steel
Increase in dimensions of 0,025 to 0.050 mm
Application of up to 110 V a.c. at 50 to 500 A h 2 ; temp. 70-80°C; time 2-30 min
Grey-green to dark green dependent on thickness
Steel, rubber and vinylbased materials
Increase in dimensions of 0.005 to 0.038 mm
5I
> z
0
50
3z
3
?? c. I I
4:112
MAGNESIUM A N D MAGNESIUM ALLOYS
adhesion to the metal and the necessary water resistance in the resin. Furthermore, a technique of baking the metal before application of the resin, followed by dipping of the still warm metal, is essential in order to fill subcutaneous blemishes. In the absence of such a procedure trouble can still arise from ‘pockets’ which retain moisture and which are only bridged by a normally applied resin film.
The Effect of Surface Finishing on the Corrosion Behaviour of Magnesium Alloys It is probable that all corrosion of magnesium alloy surfaces exposed to a damp atmosphere, or still more immersed in an electrolyte, is largely galvanic in origin and much influenced by the presence of exposed cathodic particles. Some of these are present in the alloys as unavoidable impurities, and nearly all foreign metallic particles not in solution are cathodic to magnesium. Various methods of chemical pickling or etching as well as mechanical means of metal treatment may remove a proportion of these, but unfortunately they will usually expose a further number in the lower layers. Furthermore, some methods of mechanical abrasion may increase the number of foreign cathodic particles in the surface by entrapment, and even pickling solutions are not exempt from causing the deposition of more noble metals in solution by displacement. Partly used pickling baths which have been in use for some time, in particular, become enriched in cations of other metals and may redeposit these metals by displacement. In practice, pickling baths based on nitric acid are less likely to give rise to this effect though they may not be able to remove foreign particles already in situ. Chief among the processes which bring about harmful effects on the corrosion resistance of magnesium alloy surfaces are shot and grit blasting and the use of emery cloths and papers. Various blasting operations are used for the removal of adhering foundry sand from sand castings but it should be recognised that they lower very considerably the natural resistance of the surface and reduce its ability to form protective films. Metal particles either of the abrasive itself or of material scoured from the equipment, are lodged in the magnesium surfaces by blasting operations, while splinters of emery (which form quite effective cathodes) are picked up by the use of such materials. On the other hand, glass papers are usually harmless; the splinters of glass are non-conductive and therefore incapable of acting as cathodes. Despite the fact that silicon carbide is a conductor, the use of Carborundum paper and belts does not usually lead to any serious deterioration in the corrosion resistance of magnesium alloys. By the use of many commercial abrasive processes, the corrosion resistance of magnesium alloys can be reduced to such an extent that samples of metal that may lie quiescent in salt water for many hours will, after shot blasting, evolve hydrogen vigorously, and the corrosion rate, as measured by loss of weight, will be found to have increased many hundred-fold. The effect in normal atmospheres is naturally much less, yet the activation of the surface is an added hazard and is the opposite of passivation which is essential if later-applied paint finishes are to have proper durability.
MAGNESIUM A N D MAGNESIUM ALLOYS
4:113
The use of chromating baths and acid pickles is powerless to remove all the evil effects of such treatments, but one of the electrolytic processes, namely that of Fluoride Anodising at a high voltage in a solution of ammonium bifluoride, is very effective in removing cathodic foreign metals. In this process the magnesium surface itself is quickly converted to insoluble and non-conductive magnesium fluoride and this reaction thereafter terminates. The current is thereupon automatically directed to and concentrated on the local metal cathodes which are conductive, and these are either dissolved or dislodged from the surface. Carbon, in the form of graphite resulting from the use of die lubricants in forming and in pressure diecasting, may also be an active cathode. It is not dissolved by the electrolysis but the film is undercut and insulated from the metal surface by a layer of magnesium fluoride. In this condition it is less harmful than when in direct contact with the metal; furthermore, it can more readily be removed by treatment in chromic acid or in hot caustic soda solution, processes which, in the absence of prior fluoride anodising, are not completely effective.
Recent Developments The detrimental effects of ‘heavy metal’ impurities and surface contamination on the corrosion performance of magnesium alloys have been described (4.101). Speciality high purity alloys, with extremely good corrosion resistance, were developed for use in the nuclear industry and this concept has recently been applied to the most commercially used magnesium alloyAZ91. Quantitative studies2v3have determined the threshold levels for Fe, Ni and Cu impurities in this alloy system below which a 50 to 100 fold improvement in salt fog corrosion resistance is obtained. The attainment and control of these low impurity levels for both high pressure die and sand castings has been d e m o n ~ t r a t e d ”resulting ~ in the ASTM designation of the high purity alloys AZ91D and AZ91E respectively. Die cast AZ91D components have passed the stringent corrosion tests and field trials of several automotive manufacturers and items, such as grilles, clutch housings, aircleaners, valve covers and wheels have already given several years trouble free service. AZ91E sand castings have recently been specified for aerospace applications as replacements for existing ‘normal purity’ AZ91C components. The high purity concept is being extended to the die casting alloy AM60B’ and to other alloys in the Mg-A1 system. In the Mg-Zr system alloys containing yttrium and rare earth additions have been developed 8* ’, WE54 (Mg-5% Y, 4% RE-Zr) WE43 (Mg-4% Y, 3% RE-Zr). These alloys are available in both wrought and sand cast forms and possess high strength at both ambient and elevated temperatures. They also exhibit a high level of corrosion resistance. Fe and to a lesser extent Ni impurities are naturally controlled” to low levels in the presence of Zr. In the absence of Zn or other active” alloying constituent good corrosion resistance, similar to AZ 91 E, is obtained. Corrosion rates of 0.1-0.2 mg/cmZ/day under ASTMBl17 salt fog conditions are typical for WE54 and the high purity AZ91 alloys which are comparable with those of some aluminium alloys This considerable ’s6
’.
4: 114
MAGNESIUM A N D MAGNESIUM ALLOYS
improvement in corrosion resistance does not however protect against galvanic corrosion. In corrosive environments the standard techniques to reduce or eliminate the effect of galvanic couples must still be employed l 2 * l 3 . Rapid solidification technology has been applied to several magnesium alloy systemsI4 and extruded material of some of these systems have exhibited excellent corrosion resistance. Fluxless melting Is techniques, employing protective atmospheres of air, carbon dioxide and sulphur hexafluoride (SF,), are now being used by many foundries. Flux inclusions in castings, particularly pressure die castings, have in the past contributed to magnesium's poor corrosion reputation. By employing fluxless techniques the risk of deleterious flux inclusions, due to improper melt handling, is avoided. Chromate conversion coatings I 3 are still the most widely used pretreatments prior to painting. With the increased emphasis of the hazards associated with hexavalent chromium, several chromate free treatments ' 3 ~ 1 6 ~ 1 7 have been used on magnesium. These treatments are not as effective as chromating and consequently should be restricted for use in mild environments only. The NH35 chromate treatment 1 8 , with its significantly reduced chromium addition, has been developed for use on high purity AZ91 pressure die castings. Epoxy based primer systems remain the best suited for the corrosion protection of magnesium. Cathodic epoxy electrophoretic paints6, chromate inhibited epoxy-polyamide primers l 9 and high temperature stoving epoxy sealers" are used to provide protection up to 180°C. For higher temperature applications up to 300"C, epoxy silicone or polyimide" based systems can be used. The following checklist is given as a general guide" to minimise the corrosion of magnesium components in service: 1. Design - good design to minimise exposed dissimilar metal couples, radius sharp edges and avoid water traps. Allow for protection of mating faces. 2. Specify good quality castings, forgings and extrusions. 3. Select protective scheme to suit operational environment - for new applications err on the side of overprotection until performance experience has been obtained. 4. Ensure a clean metal surface free from cathodic contaminants. 5 . Apply good quality conversion coatings. 6. Ensure correct organic protection scheme application as soon as possible after conversion coating. 7. Observe 'wet assembly' procedures on exposed galvanic couples. 8. Inspect and maintain protection. K. G. ADAMSON D.S. TAWIL REFERENCES I . Mears, R . B. and Brown, C. D., Corrosion, 1, 113 (1945); Hanawalt, J . D., Nelson, C. E. and Peloubet, J . A., Trans. Amer. Inst. Min. (Metall.) Engrs., 147, 275 (1942) 2. Reichek, Clark, Hillis, Confrolling the Salt WaferCorrosion Performance of Magnesium AZ91 Alloy, SAE Paper 850417
MAGNESlUM AND MAGNESIUM ALLOYS
4: 115
3. Hillis, The meets of Heavy Metal Contamination on Magnesium Corrosion Performance, SAE Paper 830523 4. Clark, AZ91E Magnesium Sand, Casting Alloy. The Standard for Excellent Corrosion Performance, Proc. IMA World Magnesium Conference, Los Angeles, June (1983) 5 . Kaumle, Toemmeraas, Bolstad; The Second Generation Magnesium Road Wheel, SAE Paper 850420 6. Product Design and Development for Magnesium Die Castings, Dow Chemical Co. publication 7. Hillis, Reichek; High Purity AM60 Magnesium Alloy, SAE Detroit Congress Feb. 26 ( 1986) 8. Unsworth; ‘Developments in Magnesium Alloys for Casting Applications’, Metals and Materials, 83-86, February (1988) 9. King, Fowler, Lyon; Light-weight Alloys for Aerospace Applications 11, Proc. TMS Meeting, New Orleans, USA, Feb 17-21, 1991, pp. 423-437 10. Emley; Principles of Magnesium Technology, Pergamon Press, 176-190, 685 (1966) 11. Hanawalt, Nelson, Peloubet; ‘Corrosion Studies of Magnesium and its Alloys’, Trans Am. Inst. Mining Met. Eng. 147, 273-299 (1942) 12. Hawke; Galvanic Corrosion of Magnesium, 14th International Die Casting Congress, Toronto, May (1987) 13. ASMMetals Handbook, 9th Ed, 13, 740-754 (1987) 14. Das, Chang, Raybould; ‘High Performance Magnesium Alloys by Rapid Solidification Processing’, Light Metal Age, Dec. 5-8 1986 15. ‘Use of Air/CO,/SF, Mixtures for the Improved Protection of Molten MagnesiumCouling’, Proc. IMA World Magnesium Conference, Oslo, June (1979) 16. Corrosion and Protection of Magnesium, AMAX Magnesium Publication, (1984) 17. Magnesium: Designing Around Corrosion, Dow Chemical Co. Publication, (1982) 18. The NH35 Chromating of Magnesium Pressure Die Castings, Norsk Hydro Publication, (1985) 19. Robinson; ‘Evaluation of Various Magnesium Finishing Systems’, Proc. IMA World Magnesium Conference, New York (1985) 20. Clear Baking Resin for Surface Sealing Magnesium, U.K. Specification DTD5562, HMSO 21. Improved Protection of Magnesium Alloys Against Synthetic Aviation Lubrications at Elevated Temperatures, Rendu, Tawil; SAE Paper 880869 22. Surface Treatments for Magnesium Alloys in Aerospace and Defence, Magnesium Elektron Publication
BIBLIOGRAPHY
Metals Handbook, 1, 8th edn, American Society for Metals, Chicago (1961) Pearlstein, F. and Teilell, L., ‘Corrosion and Corrosion Prevention of Light-Metal Alloys’ Paper No. 114, Corrosion, 73, Anaheim, March (1973) Emley, E. F., Principles of Magnesium Technology, Pergamon Press, Chapter xx (1966) Adamson, K. G., King, J. F. and Unsworth, W., Evaluation of the Dow I7 Treotment for Magnesium Alloys, Ministry of Defence D.Mat. Report No. 192, February (1973) King, J. F., Adamson, K. G. and Unsworth, W., Impregnation ofAnodicFilms for the Protection of Magnesium Alloys, Ministry of Defence D.Mat. Report No. 193, February (1973) Adamson, K. G., King, J. F. and Unsworth, W., Evaluation of High Temperature Resistant Coatings for the Protection of Magnesium Alloys, Ministry of Detence D.Mat. Report No. 196, July (1973)
4.5
Nickel and Nickel Alloys
Physical and Mechanical Properties Composition of Metal and Alloys
Commercially pure nickel has good mechanical properties and good resistance to many corrosive environments and therefore finds application where this combination of properties is required. Of more importance, however, is the fact that nickel forms a wide range of alloys having desirable engineering and corrosion-resistant properties. With regard to corrosion resistance to aqueous solutions, among the most important of these alloying elements are Cr, Fe, Cu, Mo and Si. Since the range of corrosion-resistant nickel alloys includes some that owe their corrosion resistance to passivity and others that are resistant because they are sufficiently noble not to displace hydrogen from acidic solutions, the corrosive environments in which nickel alloys can be successfully used are very varied, embracing acids, salts and alkalis (both oxidising and non-oxidising in character) sea-water, natural waters and the atmosphere and combinations of these encountered industrially. In addition to nickel alloys, nickel also forms an important alloying element in stainless steels and in cast irons, in both of which it confers additional corrosion resistance and improved mechanical and engineering properties, and in Fe-Ni alloys for obtaining controlled physical and magnetic properties (see Chapter 3). With non-ferrous metals nickel also forms important types of alloys, especially with copper, i.e. cupro-nickels and nickel silvers; these are dealt with in Section 4.2. Nickel is also widely used as an electrodeposited underlay to chromium on ‘chromium-plated’ articles, reinforcing the protection against corrosion provided by the thin chromium surface layer. Additionally the production of articles of complex shape to close dimensional tolerances in nickel by electroforming - a high-speed electrodeposition process - has attracted considerable interest. Electrodeposition of nickel and the properties of electrodeposited coatings containing nickel are dealt with in greater detail in Section 14.7.
The nominal compositions of commercially pure wrought nickel and the main types of modern corrosion-resistant nickel alloys are given in Table 4.21; some of these supersede earlier variants no longer in production. Applications of nickel alloys are not confined to those where corrosion resistance to aqueous solutions is a prime requirement, and the complete 4 : I16
NICKEL A N D NICKEL ALLOYS
4 : 117
range of nickel alloys that are available commercially for other specialised uses, notably those involving service at high temperatures, is therefore much greater than indicated by Table 4.21. The corrosion and oxidation resistance of nickel alloys at elevated temperature, is described in Section 7.5. In general, the alloys listed in Table 4.21 are confined to those in which nickel is the principal alloying element, but it should be noted that highly alloyed stainless steels containing 20-30% Cr, and 20-30% Ni with additions of molybdenum and copper have some features in common with the Ni-Cr-Fe-Mo-Cu alloys given in the table. In addition to the alloys in Table 4.21, Ni-Sn and Ni-Ti alloys also possess useful corrosion resistance. Ni-Sn alloys are extremely brittle and, because of this, are used only as electrodeposited coatings. Ni-Ti alloys over a wide range of compositions have been studied, of which perhaps the intermetallic compound NiTi (55 a06Ni-44-94Ti) has attracted the most interest. Structural Features and Physical and Mechanical Properties
Nickel normally crystallises in the f.c.c. structure; it undergoes a magnetic transformation at 357°C and is ferromagnetic below that temperature. In all the alloys shown in Table 4.21 the f.c.c. (austenitic) structure is substantially retained, and in consequence most of the alloys possess the combination of properties required of materials for widespread industrial acceptability, i.e. tensile strength, ductility, impact strength, hardness, hot and cold workability, machinability and fabrication. Table 4.22 gives the physical properties for nickel and a range of nickel alloys; Table 4.23 shows the mechanical properties. The data given in these tables are those published by manufacturers. It is seen that, compared with nickel, the alloys have considerably lower thermal conductivity and much higher electrical resistivity. As with nickel, some of the alloys undergo magnetic trans,formation; e.g. the Ni-Cu alloy 400 has a transformation temperature close to 0°C. The mechanical properties in Table 4.23 are generally those of wrought material in the annealed condition those of materials in other conditions and of cast alloys may differ appreciably. In all cases alloying considerably increases the proof stress and tensile strength. The elongation values of wrought alloys are generally only slightly below those of nickel. The hardness values of wrought alloys are generally below 200 H, for annealed material. As with stainless steels, some nickel alloys have a propensity to form intergranular precipitates of carbides and intermetallic phases during heat treatment and sometimes during welding. The presence of such intergranular precipitates may render the materials susceptible to intergranular attack in certain corrosive environments. To minimise this possibility, the content of carbon and, in some cases, other alloying elements is carefully controlled. The subject of intergranular corrosion of specific nickel alloys and of methods of avoiding it is dealt with in greater detail later in this chapter.
Table 4.21 Type and designation Ni 200 20 1 Ni-Cr-Fe 600 600L 601
690 800 800L Ni-Cr-Fe-Mo 718 H-9M Ni-Cr-Fe-Mo-Cu G3 G 30 825 825 h Mo 925 20 28 Ni-Cr-Mo C 276 c 4 c 22 625 Ni-Mo B2 Ni-Cu
400 K 500 Bal.
+ Co
C
Ni
0.08 0.01
99.6 99.6
0.08 0.025( -) 0.05 0.03 0.05 0.03(-)
Bal. Bal. Bal. Bal . 32.5 32.0
0-03 0*03(-) 0.007 0.03( -) 0.03 0.025( -) 0.02 0.02( -)
Nominal compositions of corrosion-resistant nickel alloys Fe
Cr
Nominal composition (weight per cent) Mo Cu Ti AI
0.2 0.2
0.1 0-1
0.2
23.0 30.0 21.0 21
8.0 8.0 14.1 9-5
Bal. Bal.
0.4
52.5 Bal .
19.0 22
Bal. 19
3.05 9
Bal. Bal. 42.0 42.5 42.0 37.2 31.0
22.2 29.5 21.5 21.0 21.0 20.0 27.0
19-5
7.0
15.0
5.0 3.0
0.015( -) 0-01(-) 0.05
Bal. Bal. Bal. Bal.
15.5 16.0 22.0 21.5
3.0(-) 3 2.5
16.0 15.5 13.0 9.0
0.01(-)
Bal .
1.0(-)
2.0(-)
28.0
0.2 0.1
Bal . Bal .
0.015( -)
15.5 15.5
Bal . Bal. Bal . Bal . Bal.
Nb
W
Other
??
03
0.4(-) I .4
0.5
0.4 0.4
1 .o
0.6
Ti:C I S ( + ) Zr 0.02(-)
5.0
2 2.0 1-7 2.2 2.2 2.2 3.5 1.2
6-2 3.0 2.5 3.5
0.4 0.4
0.25 X 0.7 X 0.9 0.8 2.1
m r
z
U
zn Ti:C 30(+) Nb:C 8( + )
0.2(-)
x
>
0.75 2.5
0.I 0.2 0.3
zn
N 0.05
x
m r
:: s r
v)
0.005
- substantially the balance of the alloy composition. although
5.5
1.2 1.0 ( - ) maximum
3.7
V 0.35(-)
3.0
V 0.35(-)
0.7(-) 0.2
0.2
31.5 29.5 0.6 (+) minimum
2.7
3.6 X
Xincl.Ta.
other elements such as dmxidants and impurities in small amounts arc included In the balance
Table 4.22
Type and designation Ni 200 20 I Ni-Cr-Fe 600 600L 601 690 800 800 L Ni-Cr-Fe-Mo 718 H-9M Ni-Cr-Fe-Mo-Cu G3 G 30 825 825 h Mo 925 20 28 Ni-Cr-Mo C 276 c 4 c 22 625 Ni-Mo B2 Ni-Cu 400
K 500
Physical properties of corrosion-resistant nickel alloys
Melting range ("C)
ckgm-')
Specific heat (J kg-l K - 1 )
1435- 1445 1435- 1445
8.89 x 10' 8.89
456 456
1370-1425
8.42 8.45 8.05 8.14 8.02 8.0
46 1 460 448 450 460
1300- 1370 1355-1385
1370-1400 1310-1 365 1370-1425 1330-1370 1325-1370 1355-1400 1290-1 350
1300-1 350 1315-1350
Thermal conductivity (W m-' K - I )
Electrical resistivity (Q m)
Modulus of elasticity (G Pa)
13.3 x 13.3
74.9 79.2
0.09 x 10-6 0.08
214 207
550
13.3 14.0 13.75 14.5 14.2 15.9
14.9 14.8 11.2 13.9 11.7 11.5
1.03 I .05 1.22 1.15 0.99 0.97
214 214 206.5 210 196 200
8.2
430
14.2
11.4
1.24
204
8.30 8.22 8.14 8.3 8.14 8.05 8.0
453
14.6 12.8 13.9
10.0 10.2 11.1 12
199 202 198 200 20 1 195 195
Density
Mean coeficient
of thermal expansion
W-9
435 500 442
15.0
11.7 10.8
1.13 1.16 1 .OO 1.10 1.17 1.03 0.99
8.89 8.64 8.69 8.44
427 406 414 410
11.2 10.8 12.4 12.8
9.4 10.1 10. I 9.8
I .30 1.25 1.14 1.29
205 21 1 206 208
9.22
373
10.3
11.1
I .37
217
8.83 8.46
419 419
14.1 13.7
21.8 17.5
0.51 0.62
179 I79
441 500
15.0
13.2 14.9
4: 120 Table 4.23
NICKEL A N D NICKEL ALLOYS Typical mechanical properties of corrosion-resistant nickel alloys 0.2% proof
Type and designation Ni 200 20 1 Ni-Cr-Fe 600 600L 601
690 800 800 L Ni-Cr-Fe-Mo 718 H-9M Ni-Cr-Fe-Mo-Cu G3 G 30
825 825 h Mo 925 20 28
Ni-Cr-Mo C 276 c 4
c 22 625 Ni-Mo 82
Ni-Cu 400 K 500
Form of material
stress (M Pa)
Tensile Strength (M Pa)
Elongation Hardness (070) (HV)
Annealed sheet Annealed
157 103
450 403
44 50
Annealed sheet Annealed Annealed sheet Annealed Annealed Annealed
269 180 min. 292 300 min. 249 180 rnin.
629 550 min. 675 600 min. 592 450 min.
42 30 min. 46 45 min. 30 min. 35 min.
Solution annealed, precipitation hardened Annealed sheet
1035 min
1240 min. 12 min.
372
730
57
31 1
692
58
324
689
56
317 240 min. 356 240 min. 220 min.
672 550 min. 769 550 min. 500 min.
42 25 min. 49 30 min. 35 min.
176 220 max. 225 max.
355
792
61
I92
42 1 407
80 1 800
54 51
200 205
414 min.
827 min.
30 min.
247 max.
Annealed sheet and plate
412
894
61
215
Annealed Annealed sheet
216 275 min.
542 620 min.
51.5 25 min.
170 max.
Solution heattreated plate Solution heattreated plate Annealed sheet Annealed Annealed rounds Annealed Solution heattreated sheet Solution heattreated sheet Annealed sheet Solution heattreated sheet Annealed sheet
100 max.
180 max. I51 max. 179
I72
Methods of Fabrication
Nickel and wrought nickel alloys may be fabricated by welding or, less commonly, by brazing or silver soldering. In order to minimise the deleterious effects that may result from integranular precipitation, either low-heat-input welding procedures employing flux-coated electrodes, or the MIG, TIC or plasma arc procedures, are recommended. Thick sections may be welded using the submerged arc process and a relatively restricted heat input. Oxyacetylene welding is rarely used because of the high heat input and the danger of carbon transfer into the metal.
NICKEL AND NICKEL ALLOYS
4: 121
Corrosion Behaviour in Aqueous Environments Theoretical Considerations
Nickel occupies an intermediate position in the electrochemical series; EGiz+,Ni = -0.227 V, so that it is more noble than Zn and Fe but less noble than Sn, P b and Cu. Figure 4.21 shows a revised potential-pH equilibrium (Pourbaix) diagram for the Ni-H,O system at 25OC'. The existence of the higher anhydrous oxides N i 3 0 4 ,Ni203and NiO, shown in an earlier diagram' appears doubtful in aqueous systems in the absence of positive identification of such species. It is seen that: 1. Nickel is thermodynamically stable in neutral and moderately alkaline solutions although not in acidic or strongly alkaline solutions. 2. The metal would be expected to dissolve in acidic solutions forming Ni'+ ions with liberation of H,. 3. The metal should be capable of passivation by forming a surface layer of Ni(OH), and perhaps NiO (see later) of nickel in neutral and moderately alkaline solutions. 4. The metal may be unstable in strongly alkaline solutions, dissolving to form Ni(0H); ions. 5 . In strongly oxidising neutral and alkaline conditions passivation should be possible through formation of a film of NiOOH.
On the basis of these data, nickel is considered to be a slightly noble metal, although in practice, as will be seen below, it is considerably more corrosion resistant in both acidic and alkaline solutions than would be predicted from Fig. 4.19. Several complications are involved in the calculation of potential-pH equilibrium diagrams for temperatures other than 25°C3-4v5,including the fact that the pH scale itself varies with temperature; thus, diagrams in which the pH scale refers to the temperature for which the equilibria are calculated are probably preferable for most purposes '. The most notable consequence of increasing temperature on the equilibria appears to be a widening of the pH range within which the hydroxide Ni(OH), is thermodynamically stable.
Anodic Behaviour of Nickel
Many investigators have studied the anodic behaviour of nickel. A complete discussion of the reactions occurring during anodic dissolution and passivation of the metal is outside the scope of this chapter, which is confined to a brief summary of the main features of practical significance. Anodic E-i curves for nickel obtained by potentiostatic, potentiokinetic or, in earlier days, galvanostatic techniques, have been published by many workers. Unfortunately, good agreement is not always found between data from different sources. The principal reasons for the discrepancies appear or in the soluto lie in the nature and amount of impurities in the meta16*7.8 tion9*", both of which may have a profound effect on the shape of the curve, and in variations in experimental procedure"-".
4 : 122
NICKEL AND NICKEL ALLOYS
3 -6
2
\ \
NiOOH
'\
1
L
Lu' 0
-1
-2
-6
0
7
14
PH Fig. 4.19 Potential-pH equilibrium diagram for the Ni-H20 system at 25°C (after Silverman' )
Figure 4.20 shows a curve for nickel in 0.5 M H2S0,'* which illustrates the main features of the anodic behaviour of the metal that are of interest with regard to its corrosion resistance. It is seen that in acidic solution nickel is capable of passivation and that the extent of the passive range (DE) is con5 The passivation of nickel in acidic solution is a feature siderable, ~ 0 . V. not predicted by the potential-pH equilibrium diagram (see Section 2.1) and is one reason why, in practice, the corrosion resistance of the metal in acidic solutions is better than that indicated from consideration of thermodynamic equilibria. A second, perhaps more important, reason lies in the fact that in the active range ( A B C ) the anodic overpotential is substantial because the exchange current density for nickel dissolution is small (Table 21.17). This, coupled with the fact that in the electrochemical series nickel is only moderately negative with respect to hydrogen, ELi2+= -0*227V, equilibrium, means that in practice the rate of dissolution of nickel in acidic solutions is slow in the absence of oxidants more powerful than H + or of substances capable of making the anodic reaction kinetically easy. The anodic dissolution current density of nickel in the active state as a function
4: 123
NICKEL AND NICKEL ALLOYS
of potential does, however, depend in a critical manner on the rate at which the measurements are made 13* "* l9 and on pH13. To explain this, Sat0 and Okamoto " proposed that in acidic solution anodic dissolution of nickel is catalysed by OH- and proceeds by way of the following reaction sequence: NiOH(ads.) + eNi + OHNiOH (ads. ) NIOH+ + eNiOH' Ni2+ + OHthe overall rate of reaction being controlled by the concentration of OHions. Burstein and Wright '' consider that the first stage in the sequence, Le. formation of NiO(ads.) to form a pre-passive layer is the rate determining step. This mechanism appears to provide a basis for explaining the sluggish anodic dissolution of nickel in acidic solution and also to account in part for the variations in the anodic behaviour reported from different sources. In solutions containing high concentration of Ni2+ and SO:- Vilche and Arvia2' consider that dissolution of Ni to Ni2+and formation of Ni(OH), are competing processes in the pre-passive region. The anodic dissolution of nickel is also dependent on the amount of cold work in the meta1'9v2',and in the active region the anodic current density of cold worked material at a given potential is up to one order of magnitude greater than that of annealed material. +
+
+
0
I
I
1
2
EH(VI
Fig. 4.20 Potentiostatic &log i curve for nickel, anodically polarised in 0.05 M H2S04 saturated with N, at 25'C (after Sato and Okamoto")
At high potentials in acidic solution nickel becomes transpassive ( E F ) , and in this region corrosion occurs preferentially at grain boundaries", as with stainless steels. In the passive and transpassive states anodic dissolution results in the formation of Ni2+ ions in solution". At still higher potentials nickel exhibits secondary passivity (FG), and although the anodic current is several orders of magnitude greater than in the passive region (DE) it is not localised at grain boundaries2'. At potentials above the range of secondary passivity the anodic current density increases and dissolution proceeds through an oxide film, probably NiOOH and is accompanied by evolution of 0,. In this region grain boundaries are preferentially attacked again". The corrosion behaviour of nickel in acidic solutions in the regions of transpassivity, secondary passivity and beyond, is of limited practical significance, since these potentials are beyond the range of the redox potentials of most aqueous solutions.
4 : 124
NICKEL A N D NICKEL ALLOYS
The influence of temperature on the anodic behaviour of nickel has been studied 3 , 8 , and in acidic and neutral solutions the active-passive transition is not observed at temperatures greater than about 100°C (Fig. 4.21).
0 1 E, (V) At test temperature
,
+
Fig. 4.21 Effect of temperature on the anodic behaviour of nickel in 0.025 M H2SO4 0.025 M K,S04 (pH 1.3) de-aerated with H,. The curves were determined potentiokinetically at a scan rate of 2 V/h and proceeding from negative to positive (after Cowan and Staehle3)
As with most other metals, the anodic behaviour of nickel is influenced by the composition of the solution in which measurements are made, particularly if the solution is acidic. Acidic solutions containing C1- ions22~29 or certain sulphur compounds in particular have a pronounced influence both in increasing the rate of anodic dissolution in the active range and in preventing passivation, and in stimulating localised corrosion3’. Thiourea and some of its derivatives have a complex effect, acting either as anodic stimulators or inhibitors, depending on their concentration3’. In alkaline solutions, except possibly in high concentrations at elevated temperatures, nickel is normally passive. Passivity of Nickel
In many aqueous solutions nickel has the ability to become passive over a wide range of pH values. The mechanism of passivation of nickel and the properties of passive nickel have been studied extensively -perhaps more widely than for any other element, except possibly iron. In recent years the use of optical and surface analytical techniques has done much to clarify the ~ i t u a t i o n ~Early ” ~ ~ . studies on the passivation of nickel were stimulated by the use of nickel anodes in alkaline batteries and in consequence were conducted in the main in alkaline media. More recently, however, attention has been directed to the passivation of nickel in acidic and neutral as well as alkaline solutions. Most authorities nowadays accept the view that passivity of nickel, as of most other metals, is due to the formation of a film of oxide or hydrated
NICKEL A N D NICKEL ALLOYS
4 : 125
oxide. Ellipsometric measurements, both in alkaline solution33and in acidic solution34, support the existence of surface oxide films on passive nickel several nanometres thick, although impedance measurements 3s suggest that, in acidic solutions at least, the passive layer is electrically complex and is not an ideal dielectric. In acidic solutions the film has been reported to be hydrated nickel oxyhydroxide, NiOy(OH)2-zy.MH,O in which y is greater in the passive film than in the pre-passive film formed in the active r e g i o P . In neutral solutions films consisting of Ni03’ and Ni(OH),” possibly with some Ni03’ have been described. In alkaline solutions Ni(OH), has been r e p ~ r t e d ~ * ’ ~ ~ . In alkaline solutions, galvanostatic measurementsa suggest that passivation of nickel is due to formation of a monolayer of Ni(OH),. This probably forms by a solid state process involving nucleation and growth, according to the general model for such growth proposed by Armstrong, Harrison and Thirsk. In some alkaline conditions, particularly concentrated solutions at higher temperatures, thicker films are undoubtedly formed. As indicated when discussing anodic behaviour the mechanism of film formation is complex, involving adsorption of O H - ions to form a prepassive layer followed by either dissolution or film formation as alternative processes. In certain concentrated acidic solutions, e.g. H z S 0 4 ,nickel, whilst not truly passive, may exhibit ‘pseudo-passivity’ owing to crystallisation of a layer of nickel salt (in conc. H 2 S 0 4 probably P-NiS04.6H20) on the surface 4 ’ . Influence of Alloying on Anodic Behaviour of Nickel
During recent years a considerable amount of information has been published on the anodic behaviour of nickel alloys. The data include studies both of binary alloy systems in which nickel forms the major alloying component and of more complex commercially produced nickel alloys. The data are sufficiently numerous to permit a rational and fairly complete interpretation of many of the corrosion-resistant properties of nickel alloys on the basis of their anodic behaviour. Potential/anodic current density curves illustrating the influence of binary alloying additions to nickel are shown as follows: Cr, Fig. 4.22; Fe, Fig.4.23; Cu, Fig.4.26; Mo, Fig.4.28 (curve for Alloy B); Si, Fig.4.29; Sn, Fig. 4.30; Ti, Fig. 4.3 1; Al, Fig. 4.32; and Mn, Fig. 4.33. The deductions that may be drawn from the data about the influence of these alloying elements on the anodic behaviour of nickel are summarised in Table 4.24. It should be noted that the data refer mostly to the behaviour of the alloys in H2S04.Passivity is, however, influenced by the composition of the solution as well as that of the metal and for this reason the influence of alloying additions may be different in solutions containing other ions. In particular, C1- and other similarly aggressive ions have a large influence and may prevent passivation, either completely or partially. If passivity cannot be maintained over the entire surface of the metal, pitting develops, and this is considered later. Broadly speaking the binary alloying additions fall into two categories: (1) those that improve passivity of Ni, viz. Cr, Si, Sn, Ti, AI and (2) those
4: 126
NICKEL AND NICKEL ALLOYS
Fig. 4.22 Effect of chromium content on the anodic behaviour of Ni-Cr alloys in 0.5 M H,SO, (de-aerated with H2) at 25°C; the potential was increased incrementally by 0.025 V every 3 min (after Hodge and Wilde42) Oscillationswithin max. c.d. range
0
1
E" (V)
Fig. 4.23 E-log i relationship for the anodic behaviour of Ni-Fe alloys in 0.5 M H2S0, (de-aerated with H,) at 25°C (after Economy, et a/.43;see also References 44-46
Table 4.24
Alloy addition
Fig.
(W
no
Influence of alloying on anodic behaviour of nickel
Influence on anodic behaviour Active region
Max.c.d. prior to pmivation, icri,
Potential of passivation, E,,
c.d. in passive region, i,
Passive region
Considerable decrease Little effect
Potential range increased
More noble, n passivation above =50%
Large increas
Potential range reduced, eliminated above =50%
0-40Cr
4.22
Potential range reduced
Large decrease
Less noble
0-70 at. % Fe
4.23
Potential range increased
More noble
0-7001
4.26
Potential range increased
Little effect on magnitude, but potential range of max. c.d. increases. Oscillations often observed within max. c.d. range Increase
Potential range reduced
cu
cu
N o passivation
N o passivation
No passivation
N o passivation
Decrease
Less noble
Large decrease
Potential range increased
4.31
Potential range moved to more noble potentials Potential range moved to less noble potentials Potential range moved to much less noble potentials Disappears
0-IOA1
4.32
Potential range increased
0-62 at % Mn
4.33
Potential range increased to much more noble Dotentials
0-28MO
4.28
0-16.S
4.29
Electrodeposited Sn-35Ni
4.30
0-1OOTi
Potential range much increased
Much less noble Much less noble?
Large decrease
Decrease
Less noble
Large increase
More noble
0-7% AI considerable decrease Large increase
Potential range much increased Potential range increased Potential range greatly decreased
4: 128
NICKEL A N D NICKEL ALLOYS
Fig. 4.24 Anodic behaviour of Alloy 600 in 0 . 5 M H,SO, (de-aerated with N 2 ) at 24°C containing different concentrations of CI- ions (after Piron, et a / . 4 7 )
that ennoble Ni, viz. Cu and Mo. Iron and manganese do not belong in either category. Although Ni-Fe alloys can be passivated, their passivity is less than that of nickel and they are also less noble than nickel. In the presence of chromium, however, iron has a considerably beneficial influence on passivity, as may be seen by comparing the curve for the Ni-lSCr-8Fe Alloy 600 in Fig.4.24 with the curves for binary Ni-Cr and Ni-Fe alloys in Figs. 4.22 and 4.23 respectively. Alloying elements which enhance the passivity of nickel are expected to improve the corrosion resistance to oxidising media, in particular acidic solutions containing oxidants. Generally, this is found to be so in practice, although it should be noted that strongly oxidising acids, e.g. HNO, and H2Cr0,, or other acidic solutions containing powerful oxidants may render such alloys transpassive, in which condition the corrosion resistance may be impaired. In less oxidising media, particularly in acidic solutions where hydrogen evolution is the cathodic process, not all alloying elements which improve passivity are beneficial, although some are. In these circumstances chromium, silicon and probably aluminium are unhelpful and might be expected to confer little benefit, because the passivation potential, although displaced to slightly more negative values, is not displaced sufficiently to permit passivity to develop in hydrogen-evolving acidic solutions. In contrast, alloying additions of titanium and tin (in the electrodeposited Sn-35Ni
NICKEL AND NICKEL ALLOYS
4 : 129
Alloy F
(Ni-22(r-20 Fe-6.5 Mo)
(Ni-22 (r-20 Fe-6.5 Mo-2 Cu)
Fig. 4.25 Anodic behaviour of Alloy F and Alloy G in boiling 10% H2S04 de-aerated with H, (the potential was increased incrementally every 3 min;. after Leonard48; see also Reference 49)
alloy) are undoubtedly beneficial because the active/passive transition is displaced to sufficiently negative potentials that enable passivity to be maintained in non-oxidising acidic solutions. In fact silicon is also often beneficial, especially in H,SO,. In dilute H,S04, Ni-Si alloys containing about 10% Si do not passivate spontaneously, but the rate of anodic dissolution rapidly falls to a low value owing to the formation of a silicon-rich surface layer. In concentrated H2S04such Ni-Si alloys are passive, whilst in H,SO, of intermediate concentration the corrosion behaviour is complex, being governed by the nature of the cathodic process, which changes as corrosion proceeds 53. The alloying elements molybdenum and copper do not, by themselves, enhance passivity of nickel in acid solutions, but instead ennoble the metal. This means that, in practice, these alloying elements confer benefit in precisely those circumstances where chromium does not, viz. hydrogen-evolving acidic solutions, by reducing the rate of anodic dissolution. In more oxidising media the anodic activity increases, and, since binary Ni-Mo and Ni-Cu alloys do not passivate in acidic solutions, they are generally unsuitable in such media. Relatively small amounts of molybdenum in Ni-Cr-Fe alloys, as in stainless steels, render passivation much easier and it may be seen from Fig. 4.25
4: 130
NICKEL AND NICKEL ALLOYS
O
-
7
6 0
1
1
E..IV)
Fig. 4.26 Anodic behaviour of Ni-Cu alloys in 0.5 M H,SO, (de-aerated with N2) at 25°C; the curve was determined potentiokinetically at 0 . 4 V/h for the 7 8 . 3 and 49.9% Ni alloys and at 3 V/h for the 30.4% Ni alloy proceeding from more positive to more negative (after Osterwald and Uhlig5')
that the further addition of 2% Cu enhances the effect. The major effect is to reduce the maximum current density prior to passivation, icri,, , although the current density in the passive range, i,, is also reduced. Potentiall anodic current density curves of Ni-Cr-Fe-Mo and Ni-Cr-Fe-Mo-Cu alloys plotted in the conventional way d o not show these effects clearly, but they may be illustrated by employing fast scan rates49or elevated temperatures (see Fig. 4.25). Because of the effect on icri,., Ni-Cr-Fe-Mo and Ni-Cr-Fe-Mo-Cu alloys have good corrosion resistance to acidic solutions both in oxidising conditions and when corrosion is accompanied by hydrogen evolution. The addition of chromium to Ni-Mo alloys containing about 15% Mo confers passivity, as may be seen by comparing the curves in Fig. 4.28 for Alloy B (Ni-28Mo), Alloy N (Ni-16-5Mo-7Cr) and Alloy C (Ni-16Mo15 *5Cr).Chromium, however, displaces the active region in these alloys to more negative potentials, so that whilst the chromium-containing alloys are more corrosion resistant than the chromium-free alloy in oxidising acidic media, they are less resistant in most hydrogen-evolving acidic solutions. An interesting illustration of the effect that quite small alloying additions may sometimes have on anodic behaviour is seen in Fig. 4.275' from a comparison of the Ni-30Cu alloy Alloy 400 with its age-hardening variant Alloy K500, which contains 2.7% A1 and 0.6% Ti. The presence of these elements in the latter alloy is responsible for a well-defined passive region, whereas the former alloy shows only a slight tendency to passivate in acidic
4: 131
NICKEL AND NICKEL ALLOYS
O
m
0 Fig. 4.27 Anodic behaviour of Ni-Cu alloys in 10% H,SO, at ambient temperature. 1, Ni; 2, Alloy 400; 3, Alloy K500 solution treated; 4, Alloy K500 aged (after Flint and Barker )
''
solutions. Furthermore, a clear distinction may be seen between the passivity of the age-hardening alloy in the solution-treated condition, where aluminium and titanium are substantially in solid solution, and in the aged condition, where the alloy is strengthened by precipitates of Ni,Al and Ni,Ti. In addition to the removal of most of the aluminium and titanium from solid solution, precipitation also increases the effective copper content of the matrix. Both of these effects may be responsible for the reduction in passivity of the aged material. Another indication of the influence of precipitated phases on anodic behaviour may be seen in the curve for Alloy C in Fig. 4.28, where the small peak in the middle of the passive range is probably attributable to anodic dissolution of an intermetallic phase ( p ) and M,C carbide5'. The influence of minor alloying elements and the effect of formation of other phases on the anodic behaviour of nickel alloys are thus not negligible and should not be ignored. Pitting (Section 1.61
Pitting of nickel and nickel alloys, as of other metals and alloys, occurs when passivity breaks down at local points on the surface exposed to the corrosive environment, at which points anodic dissolution then proceeds whilst the
4: 132
NICKEL AND NICKEL ALLOYS
EH (V)
Fig. 4.28 Anodic behaviour of Alloys B, C and N in boiling 10% H 2 S 0 4 de-aerated with H,; the potential was increased incrementally (after Leonard 4 8 )
Fig. 4.29 Anodic behaviour of Ni-Si alloys in 25% H,S04 (de-aerated with N2) at ambient temperature (after Barker and Evans ”)
NICKEL A N D NICKEL ALLOYS
4 : 133
Fig. 4.30 Suggested anodic behaviour of electrodeposited Sn-3SNi alloy; 1 , 'observed' curve; 2a, H2 evolution; 2b, H2 oxidation; 3, 'true' anodic curve (after Clarke and Elbournes4)
E" ( V I
Fig. 4.31 Anodic behaviour of Ni-Ti alloys in HCI + 3.5% NaCl(pH I), de-aerated with argon, at 22.2OC; the potential was increased by 0.02 V every minute (after Sedriks, el
major part of the surface remains passive. Since most (sometimes all) of the cathodic reaction accompanying corrosion is distributed over the passive surface, it follows that the fewer the number of sites of breakdown the more intense is the anodic dissolution at each site, Le. the fewer the pits the faster they grow, at least in the early stages. Pitting of nickel has been shown to develop preferentially near structural features in the metal, such as grain boundaries, and also at imperfections in the surface, such as scratchesS9.
4: 134
NICKEL AND NlCKEL ALLOYS
0 OAl
-.
Fig. 4.32 Anodic behaviour of Ni-AI alloys in 0.5 M H2S04, de-aerated with H,, at 22OC; the potential was increased by 0.01 or 0.02 V every 3 min in the active range and by 0.04 V in the passive range (after Crow, et a/.56)
Electropolishing appears to be helpful in reducing the tendency of pits to develop at surface imperfections, but not necessarily at sites associated with structural features of the metal. In practice, pitting of nickel and nickel alloys may be encountered if the corrosive environment contains chloride or other aggressive ions and is more liable to develop in acidic than in neutral or alkaline ~ o l u t i o n s ~In~acidic . solutions containing high concentrations of chloride, however, passivity is likely to break down completely and corrosion to proceed more or less uniformly over the surface. For this reason nickel and those nickel alloys which rely on passivity for their corrosion resistance are not resistant to HCI. Figure 4.34 illustrates, by means of potential/anodic current density curves, the influence of pH and C1- ions on the pitting of nickel2*. The tendency to pit is associated with the potential at which a sudden increase in anodic current density is observed within the normally passive range (EBon Curve 1 in Fig. 4.34). It can be seen that in neutral 0.05 M Na,S04 containing 0.02 M C1- (Curve 1) EBhas a value of approximately 0.4 V E,, . When pitting develops, the solution in the pits becomes acidic owing to hydrolysis of the corrosion product (see Section 1.6) and when this occurs the anodic current density increases by at least two orders of magnitude and tends to follow the curve obtained in 0.05 M H,SO, + 0.02M NaCl (Curve 2). Comparison of Curves 2 and 3 illustrates the influence of C1- ions on the pitting process. Owing to the hydrolysis reaction, pit development is an autocatalytic process and often there is an induction period before pit growth attains
NICKEL AND NICKEL ALLOYS
17 at
4: 135
YO Mn
Fig. 4.33 Anodic behaviour of Ni-Mn alloys in 0 . 5 M H2S04saturated with H, at 20°C (after Horton et aLS7)
observable proportions. In some circumstances, e.g. neutral and alkaline solutions, the induction period may be very long in practice. As with other passive metals and alloys, development of pitting in nickel may be inhibited in flowing solutionsm. Figure 4.35 illustrates the effect of temperature on the rate of development of pitting, measured as a corrosion current in an acidic solution containing C1-; it is seen that quite small increments in temperature have large effects. The influence of temperature is of considerable significance when metals and alloys act as heat transfer surfaces and are hotter than the corrosive environment with which they are in contact. In these circumstances,
4: 136
NICKEL AND NICKEL ALLOYS
-3 I 1
0
E"(W
+ +
Fig. 4.34 Influence of pH and CI- ions on the anodic behaviour of nickel in SO:C1- ion solutions at 20°C (potentiokinetic polarisation at 0.05 V/rnin). I , 0.05 M Na2S04 0.02 M NaCl; 2, 0 . 0 5 ~ H2S04 0 . 0 2 ~NaCI; 3, 0.05 M H2S04 0.05 M NaCl (after Szklarska-Srnialowska22)
+
+
Time (min)
Fig. 4.35 Influence of temperature on breakdown of passivity of nickel in H,S04 solution (pH 0.4) containing 0.05 M C1- (after Gressrnann")
+ Na,SO,
NICKEL AND NICKEL ALLOYS
4: 137
deep pointed pits may develop rather than the shallower rounded pits usually found when there is no thermal gradient. A possible explanation is that anodic dissolution becomes concentrated at the base of the growing pit in preference to its sides under the influence of the thermal gradient in the metal. Figure 4.36 shows the influence of pH on the breakdown potential of nickel in alkaline solutions containing C1- ions, and it is apparent that the breakdown potential becomes more positive as the pH increases, i.e. breakdown is unlikely unless the solution has a very high redox potential.
1.
>
LU’
-
0-
11
12
13
14
PH
Fig. 4.36 Influence of pH and CI- ion on the breakdown potential of commercial nickel in alkaline solutions (0.001-5 M NaOH) de-aerated with N, (after Postlethwaite6’)
Alloying nickel with other elements has a marked influence on the susceptibility to pitting. Figure 4.37 shows the variation of the breakdown potential with chromium concentration for binary Ni-Cr alloys63,and it is seen that breakdown becomes significantly less probable as the chromium increases above 10%. Alloying with iron in addition to chromium yields a further improvement, as may be seen from Fig. 4.24, which shows that the Ni-lSCr-8Fe alloy Alloy 600 exhibits little tendency to breakdown even in an acidic solution containing 1% NaCI. In practice, Ni-Cr-Fe alloys exhibit a high degree of pitting resistance and, as with stainless steels, the addition of a few per cent molybdenum improves their resistance even further. Nickel alloys which rely on nobility for their corrosion resistance, viz. Ni-Cu and Ni-Mo alloys in acidic solution, do not usually pit in these circumstances. It should be noted, however, that the Ni-Cu alloy Monel400 normally forms a protective oxide film in neutral and alkaline solutions, and this is of particular significance with regard to its corrosion resistance to
4: 138
NICKEL AND NICKEL ALLOYS
I 0
I
1
1
10
20
3(
‘I. Cr Fig. 4.37
Influence of the chromium content of Ni-Cr alloys on the breakdown potential in 0.I M NaCl at 25°C de-aerated with N, (after Horvath and Uhlig63)
sea-water. In circumstances where the supply of 0, is insufficient to maintain the film in good repair, as in stagnant conditions, pitting may develop. Crevice Corrison (Section 1.6)
In recent years crevice corrosion has received increased attention owing to the serious hazards that develop if this type of localised attack is overlooked or ignored. Crevice corrosion can be an especially serious problem with passive metals and alloys because breakdown of passivity in the deoxygenated solution that develops in crevices leads to anodic dissolution. Of the nickel alloys those containing molybdenum and, to a lesser extent, copper offer the best-resistance to this form of attack. Ni-Cr-Mo alloys are among the most resistant of metallic materials to crevice corrosion, although their resistance may be impaired if intergranular precipitates of molybdenum-rich M,C carbide are allowed to formM (see Intergranular Corrosion). In cast materials at least, solution heat-treatment of the 625 type of alloy is beneficial and if such a heat-treatment is given, reduction of the niobium content of the alloy may be cost-effective6s. Intergranular Corrosion (Sections I . 3 and 1.7) As with most other metal and alloys systems, nickel and certain of its alloys may suffer intergranular corrosion in some circumstances. In practice, intergranular corrosion of nickel alloys is usually confined to the vicinity of welds as a result of the effects produced by the welding operation on the structure of the material in those regions. Alloys that are subjected to other similarly unfavourable thermal treatments may also become susceptible. The compositions of most commercial nickel alloys that are marketed today are,
NICKEL AND NICKEL ALLOYS
4: 139
however, carefully controlled to minimise the possibility of intergranular corrosion developing in welded material during service. Intergranular corrosion of nickel and its alloys is nearly always associated with grain boundary precipitates. In certain commercial grades of nickel, which contain carbon as an impurity, lengthy exposure t o high temperatures may result in the formation of a grain boundary film of graphite which in some circumstances renders the material susceptible to intergranular corrosion on subsequent exposure in an environment to which the material is otherwise well suited, viz. caustic alkalis; with nickel this form of corrosion may be intensified by stress in the metal. For these reasons, the low-carbon grade of commercial nickel, Nickel 201, is, in practice, preferred where this form of attack is a possibility. With material of higher carbon content the possibility of intergranular corrosion developing to a serious extent may be minimised by applying a stress-relieving heat treatment after fabrication. The presence of other elements in nickel, notably sulphur, may also render the metal liable to intergranular penetration and embrittlement . The types of chromium-containing nickel alloys that owe their corrosion resistance to passivity, viz. Ni-Cr-Fe, Ni-Cr-Fe-Mo and Ni-Cr-Fe-Mo-Cu alloys, may become susceptible to intergranular corrosion in circumstances broadly similar to those that produce susceptibility in stainless steel^^'^^. In these materials, preferential attack by the corrosive environment occurs at zones immediately adjacent to grain boundaries at which precipitates of the chromium-rich carbides M2,C6 or possibly M, C, have formed, the attack being concentrated on the chromium-depleted zones adjacent to the precipitate, since these zones cannot become p a s ~ i v a t e d ~ " *As ~ ~with *~~. stainless steels, the appropriate preventative measures are to minimise carbide formation by controlling the carbon content of the material to levels as low as practicable-nowadays 0.02% C max. is attainable- to increase the chromium content and to add elements such as titanium and niobium to form carbides more stable than M,, C, with the residual carbon and thus prevent chromium-depletion. It should be noted, however, that owing to the higher activity of carbon in nickel-rich alloys than in stainless steels, a greater proportion of stabilising element such as titanium is needed in the former materials than in the latter7'. Intergranular corrosion of stainless steels and Ni-Cr-Fe alloys has been observed to occur in the absence of grain boundary carbide precipitates in the alloy during laboratory tests in highly oxidising acidic solutions such as H N 0 3 containing chromates or and is associated with segregation of P and Si to grain boundaries. A review of intergranular corrosion of alloys in the Fe-Ni-Cr system, including stainless steels and nickel alloys, is available,,. Another type of nickel alloy with which problems of intergranular corrosion may be encountered is that based on Ni-Cr-Mo containing about 15% Cr and 15% Mo. In this type of alloy the nature of the grain boundary precipitation responsible for the phenomenon is more complex than in Ni-Cr-Fe alloys, and the precipitates that may form during unfavourable heat treatment are not confined to carbides but include at least one intermetallic phase in addition. The phenomenon has been extensively studied in recent years The grain boundary precipitates responsible are molybdenum-rich M,C carbide and non-stoichiometric intermetallic p s8*64974-79.
4: 140
NICKEL A N D NICKEL ALLOYS
phase (Ni, Fe, Co), (W,Mo, Cr),S*. Depending on the nature of the corrosive environment attack in this type of alloy may be either at depleted zones adjacent to grain boundaries or on the grain boundary precipitates themselves. Thus two different mechanisms of intergranular corrosion operate in this type of alloy, one involving attack on the depleted regions being observed in HC1 (and perhaps other hydrogen-evolving acidic solutions), the other, in which the precipitates themselves are preferentially attacked, being observed in more highly oxidising acidic media. An observation of significance made some years ago was that limiting the silicon content of this type of alloy to very low levels reduced the tendency for formation of the intermetallic phase during welding7' and this led to the introduction of improved commercial alloys of the C276. More recently a composition possessing even greater thermal stability, Le. C4, has been developed, in which iron and tungsten present in the earlier alloys have been largely replaced by nickel 79 and further alloys have been introduced with higher Cr and lower Mo contents, e.g. C22 and 625 (see Table4.21). Ni-Mo alloys containing about 28% Mo are a third category of nickel alloy liable to intergranular corrosion in the welded condition. In these alloys preferential corrosion may develop at zones adjacent to welds exposed to HCIl and other hydrogen-evolving acids in which this type of alloy is used. Corrosion is preferentially concentrated on molybdenum-depleted zones adjacent to grain boundaries in which molybdenum-rich M,C carbide has precipitated. The susceptibility of this type of alloy to intergranular corrosion is reduced by controlling the carbon and iron content to levels as low as is practicable and also by addition of about 2% VB0or 3.5-5% W". Niobium may also be a beneficial addition''. but titanium and zirconium accelerate intergranular corrosion of this type of alloy". Bimetallic Corrosion
Owing to their intermediate position in the galvanic series, nickel and nickel alloys may stimulate corrosion of metals less noble to themselves when in bimetallic contact and thus receive cathodic protection or suffer intensified corrosion from contact with more noble metals and graphite. In general, in mild environments such as unpolluted atmospheric conditions, nickel and nickel alloys are compatible with a fairly wide range of other metals and alloys, but in strong electrolytes such as sea-water and marine atmospheres the range of compatible couples is less. Table 4.25 gives guidance in very general terms, but should not be assumed to apply in every circumstance, since other factors may influence the issue. The relative surface areas of the two metals in contact plays a large part in determining whether bimetallic corrosion is serious or not, and the combination of a small area of the more negative (less noble) metal or alloy in contact with a large area of the more noble material is usually the most dangerous situation (see Section 1.7). Protection of the less noble metal by painting or other means, if properly carried out, is usually effective in minimising bimetallic corrosion. In aggressive environments nickel and the different types of nickel alloy are not necessarily wholly compatible one with another.
NICKEL AND NICKEL ALLOYS
4: 141
Corrosion in Natural Environments The A tmosphere
Nickel and nickel alloys possess a high degree of resistance to corrosion when exposed to the atmosphere, much higher than carbon and low-alloy steels, although not as high as stainless steels. Corrosion by the atmosphere is, therefore, rarely if ever a factor limiting the life of nickel and nickel alloy structures when exposed to that environment. Table 4.25 Bimetallic corrosion effects of nickel. and nickel alloys (General guidance only; other factors, including relative surface areas, often exert an important influence) Corrosive environment Most atmospheric conditions except marine atmospheres
Sea-water and marine atmospheres
Corrosion of nickel or nickel alloy is stimulated by bimetallic contact with:
Bimetallic contact with nickel or nickel alloy has little or uncertain influence*:
Au
Rh Pd C (graphite) Ti Cu and Cu alloys Stainless steels Cr plate C steel? AI and AI alloyst Mg and Ntg alloyst
Pb Sn Soft solders Cd Zn Galvanised steel AI clad Carbon steelf AI and AI alloysf Mg and Mg alloysf
Ag and Ag brazing
Austenitic cast iron5 Low-alloy steels Cast iron (ungraphitised) Wrought iron Carbon steel Cd AI and AI alloys Zn Mg and Mg alloys
Pt
C (graphite) Graphitised cast iron Au Pt
alloys Cu and Cu alloys Pb Sn Soft solders Other Ni alloys
Bimetallic contact with nickel or nickel alloy stimulates corrosion of:
Little effect in most atmospheres. except marine. Effects in sea-water and marine atmospheres depend on surface area relationships. t I f properly painted. 1 If unpainted or improperly painted. 8 Contact with small area or Alloy 400 has little eRect.
The appearance of bright nickel is, however, impaired by exposure to moist, polluted atmospheres owing to the phenomenon known as fogging. Vernon showed more than 60 years ago that for 'fogging' of nickel to occur, a high humidity-greater than about 70% r.h.-and the presence of SO, were both necessary"'. Fogging is due to the catalytic oxidation of SO, in polluted atmospheres by the nickel surface and subsequent corrosion of the nickel by the liquid film of HISO, dhus formed on the surface, the corrosion product - a basic nickel sulphate- being responsible for fogging. In the early stages the film can be readily removed by wiping with a cloth, but once the surface has become fogged the bright appearance cannot be restored merely by wiping and mild abrasion is needed. Some nickel alloys, viz. the
4: 142
NICKEL AND NICKEL ALLOYS a/Kure b4ach.N.C. (24m lot)
Jewark, N J.
2
4
6
Time (years)
Fig. 4.38 Atmospheric corrosion of nickel and nickel alloys during exposure tests at sites in the USA. 1, Nickel 200; 2, Alloy 600;3, Alloy 800; 4, Alloy 825; 5, Alloy 400 (after van Rooyen and Copson")
Ni-30 Cu Alloy 400 also undergo fogging, but alloys containing 15% Cr or more d o not exhibit this phenomenon. Fogging is prevented by a very thin film of chromium deposited on the surface-a fact which forms the basis for the bright appearance of decorative chromium-nickel plate (see Sections 13.7 and 13.8). Nickel and nickel alloys do not form thick layers of corrosion products when freely exposed to outdoor atmospheres in circumstances where the surface is periodically washed by rain, but such deposits may form on sheltered surfaces. Quantitative data on the rate of loss of metal and of pitting of nickel and nickel alloys exposed to outdoor atmospheres are available83-86.Figure 4.38 shows results obtained at three sites in the USA over a 7 year period8' and Fig. 4.39 gives results from a 10 year test at Birminghama6. In both series of tests, Ni-Cr-Fe alloys gave lower weight losses than nickel itself or Ni-Cu alloys and the American results bring out the
4: 143
NICKEL AND NICKEL ALLOYS
Upper surface
1000-
- i .
E E
v
N
E
0
v
u
I 0
100-
0 .-
0
10-
1.1
Lower surface
Fig. 4.39 Atmospheric corrosion of nickel and nickel alloys at Birmingham, England, during exposure tests of 10 year duration. I , Nickel 200; 2, Alloy 600; 3, Alloy DS (Fe-37Ni18Cr-2Si); 4, Alloy 400; 5 , Alloy K500; 6, Ni-28Mo (after Evanss6)
point that over long periods the corrosion rate of Ni-Cr-Fe alloys in the atmosphere declines to a very low value whilst that of nickel and Ni-Cu alloys remains approximately linear. Comparison of the data in Figs. 4.38 and 4.39 shows that corrosion at the UK site was several times greater than that at the most aggressive American site. This has also been observed with stainless steels in a test programme where a direct comparison was made between identical test samples”. Fresh Water
Nickel and nickel alloys are normally resistant to fresh water and natural waters at temperatures up to normal boiling point; there may, however, sometimes be a risk of pitting in waters of high acidity or high salinity in stagnant conditions. In flowing conditions, oxygen dissolved in the water is normally sufficient to maintain passivity. Aerobic bacteria appear to have little influence, but corrosion may become severe in the presence of bacteriainduced decay products.88Steam condensates containing 0, and CO, may, however, be aggressive towards nickel and Ni-Cu alloys, in which circumstances Ni-Cr-Fe alloys are more resistant. Sea-water
Nickel and nickel alloys possess good resistance to sea-water in conditions where the protective properties of the passive film are fully maintained. As pointed out above, Ni-30 Cu Alloy 400, in contrast to its behaviour in acidic solution, normally forms a protective film in neutral and alkaline environments, including sea-water; this alloy and its age hardening variant
4: 144
NICKEL A N D NICKEL ALLOYS
Alloy KSOO is widely used in sea-water. A particularly valuable feature of the behaviour of nickel and its alloys in sea-water is the ability of the protective surface film to remain in good repair in highly turbulent and erosive conditions. Because of this the alloys are used extensively in pumps and valves and other similar equipment in contact with sea-water flowing at high velocity. The protective film on nickel, Ni-Cu and Ni-Cr alloys is normally kept in good repair providing the effective sea-water velocity is greater than approximately 2 m/sS9 and in these circumstances overall corrosion rates are normally of the order of 0.01 mm/y. In sea-water flowing at slower velocities and more especially in stagnant conditions, pitting and crevice corrosion may develop, particularly beneath deposits and marine growths at the surface of the metal. Some data for the Ni-30 Cu Alloy 400w are shown in Fig.4.40; the corrosion was mostly pitting .
%J
N
'E
-e
01
Y
0)
6 05.ul P 0
0
Fig. 4.40 Corrosion of Monel 400 in sea-water at Port Hueneme Harbour, Cal., USA (after Brouillette 90)
Ni-Cr-Fe alloys are liable to suffer more intensive pitting than Ni-Cu alloys and nickel itself in low velocity sea-waters9,but the addition of a few per cent of molybdenum to Ni-Cr-Fe alloys greatly improves the resistance to pitting and crevice corrosion. Table 4.26 shows some data reported by Niederberger, Ferrara and Plummer" which illustrate the magnitude of the improvement. It will be seen from Table 4.26 that Ni-Cr-Mo alloys possess the best resistance to corrosion and pitting in sea-water. A specimen of NiCr-Mo alloy has been reported to be immune to corrosion in sea-water over a 10 year period, having suffered no weight loss and no pitting". With cast material annealing may improve resistance to crevice corrosion if niobium is also present in the alloy65.When immersed in deep sea-water nickel and nickel alloys undergo less corrosion than in shallow conditions9'.
4: 145
NICKEL AND NICKEL ALLOYS Table 4.26
Corrosion of nickel alloys in quiet and in slow moving sea-water (after Niederberger, e f a i 9 ' ) Quiet sea- water
Alloy composition
w t , loss (')
Ni-35Cu Ni-30Cu-3AI Ni-16Cr-7Fe Ni-35Cr-2Fe Ni-47Fe20Cr Ni-27Mo Ni-l6Mo7Cr-4Fe Ni-30Fe21Cr-3Mo Ni-22Cr9Mo-2Fe Ni-2OCr5Mo-6Fe Ni-16Cr16MO-4W
20.40 19.50 11.85 9.32
Slow moving sea- water
Range of pit depth (mm) Exposed
Crevice
0-0.5
0-75-0.8
WI. loss (')
Range of pit depth (mm) Exposed
Crevice
3.25 0.15-0.7
21.10 24.40 12.55 7.62
0.25-0.38 0.6-0.68 0.8-I '0 0.5-0'7 3.25 0.73-3.25 3.0 0'38-1 '65
3.63 0-33-0.35
24.50 54.80
3.63 0.25-0.35
3.63 0.2-0.4
Nil
0-0.05
I .45
Nil
0.08-0.I
0.25
0-0'03
0-0.23
0.20
Nil
Nil
Nil
0.25
Nil
Nil
0.20
Nil
0.35-0.63
0.15
0-0'05
Nil
Nil
Nil
Nil
0.10
Nil
Nil
0.55-0.65 3.25 3.0
0.55-0.58
42.80
3.4 0.5-0.53
0.50
15.72
0.08-0.I 5 0.03-0.65
Conditions of exposure: Quiet sea-water -suspended from raft. Slow moving sea-water -velocity 0.3-0.6 m/s. Location: Harbor Island, Wrightsville. Beach. N.C.. USA. Duration of test: 2 years; panel size 12 x 3 in (0.305 x 0.076 m). crevice area I x I in (0,025 x 0.025 m).
Underground
As other cheaper materials usually give satisfactory performance, nickel and nickel alloys are not normally required for applications involving resistance to corrosion underground. Data on their behaviour in these circumstances are therefore sparse; in particular, whether micro-organisms responsible for the accelerated corrosion of ferrous and other metals in certain anaerobic soils have any influence on nickel and its alloys, is uncertain.
Corrosion in Chemical Environments Acids
The wide range of corrosion-resistant nickel alloys that are produced commercially is capable in practice of handling most types of acid. Since the nickel-alloy range includes some that are corrosion resistant by virtue of their relative nobility and others that owe their resistance to passivity, alloys suitable both for hydrogen-evolving acids and for more oxidising acids are available. Table 4.27 contains a summary of data mainly derived from laboratory corrosion tests to illustrate the behaviour of individual alloys in some common mineral and organic acids.
4: 146
NICKEL AND NICKEL ALLOYS
The data in Table 4.27 refer to solutions of pure acids; in practice the presence of impurities often has a large influence and modifies the corrosion resistance to a greater or lesser extent. Oxygen from the air stimulates corrosion of alloys of the relatively noble type, including nickel itself, Ni-Cu and Ni-Mo alloys, but may be helpful in maintaining passivity of the other type, viz. alloys containing 15% Cr or more. Other oxidants such as Fe3+ or Cu2+ (which may sometimes in practice be present through corrosion of ferrous and copper-base alloys also in contact with the acidic environment) usually have a similar influence. The presence of halide ions, especially C1and F-, in H2S04, H,P04 and HNO, is usually highly detrimental to the corrosion resistance of both the noble and passive types of alloys. It should be noted, however, that Alloy 690 possesses sufficient resistance to HNO, HF mixed acids to be practically useful and that a Co-20Cr-1 SW-1ONi alloy is similarly resistant to certain HNO, -HCI mixtures. In addition to impurities, other factors such as fluid flow and heat transfer often exert an important influence in practice. Fluid flow accentuates the effects of impurities by increasing their rate of transport to the corroding surface and may in some cases hinder the formation of (or even remove) protective films, e.g. nickel in HF. In conditions of heat transfer the rate of corrosion is more likely to be governed by the effective temperature of the metal surface than by that of the solution. When the metal is hotter than the acidic solution corrosion is likely to be greater than that experienced by a similar combination under isothermal conditions. The increase in corrosion that may arise through the heat transfer effect can be particularly serious with any metal or alloy that owes its corrosion resistance to passivity, since it appears that passivity breaks down rather suddenly above a critical temperature, which, however, in turn depends on the composition and concentration of the acid. If the breakdown of passivity is only partial, pitting may develop or corrosion may become localised at hot spots; if, however, passivity fails completely, more or less uniform corrosion is likely to occur. Table 4.27 provides a basis for selecting the nickel alloy type likely to be suitable for service in particular acids. The Ni-Cr-Fe-Mo-Cu and Ni-Cr-Fe-Mo alloys, both wrought and cast, are the types most often selected for H 2 S 0 4 ,and they possess the additional advantage that their resistance is not greatly affected by the presence of SO2. Cast Ni-Si alloys containing 9% Si or more and, preferably, alloying additions of copper, titanium and m ~ l y b d e n u r n ~are * ~also ’ used for H2S04.Most nickel alloys have good resistance to pure H,PO,, but the presence of halide ion impurities reduces the resistance in the higher concentration range. Alloy 690 possesses good resistance to HNO, and is one of the few metals able to withstand the combination of HNO, plus HF; it is not, however, resistant to mixtures of HNO, and HCI. Among metallic materials Alloy B2 is one of the most suitable for handling HCl, particularly in the absence of air and other oxidants; in oxidising conditions Ni-Cr-Mo alloys are usually more suitable. Nickel itself and Alloy 400 both possess good resistance to HF; in practice Alloy 400 is used for aqueous and anhydrous HF, but precautions are necessary against stress-corrosion cracking. In practice, nickel and nickel alloys have good resistance to most organic acids.
NICKEL AND NICKEL ALLOYS
4: 147
Alkalis
Nickel and its alloys are among the most resistant metallic materials to caustic alkalis. Nickel itself possesses outstanding resistance to NaOH and KOH and is used to contain these substances over the entire concentration and temperature ranges that are of practical interest, viz. 0-100% and up to 350°C. At the higher concentrations and temperatures KOH is significantly more corrosive than NaOH towards nickel and in these circumstances the metal is sometimes cathodically protected. At temperatures above 300"C, low-carbon nickel (0.02% C) is preferred to avoid the possibility of intergranular attack developing after long exposure; if material of higher carbon is employed it should be annealed after fabrication and before exposure to caustic alkalis to prevent stress-assisted intergranular corrosion. The corrosion rate of nickel in sodium hydroxide is adversely affected by heat transfer by small amounts of oxidisable alkaline sulphur-containing salts, e.g. Na,SO,, Na,S,O,, Na,S and, at high temperatures, by alkaline oxidising agents, viz. NaCIO, and Na,O, . In the former circumstance Alloy 600 is more resistant than nickel, but not in the latter. When Alloy 600 is used for service in caustic alkalis, it should be stress relieved after fabrication to minimise the possibility of stress-corrosion cracking. Other nickel alloys, notably Alloy 600, also possess good resistance to caustic alkalis. Salts
Nickel and nickel alloys generally possess good corrosion resistance to acidic, neutral and alkaline salts, including halides, that are not oxidising in character. Oxidising salts are usually corrosive towards Ni, Ni-Cu and Ni-Mo alloys, but not to Ni-Cr and Ni-Cr-Fe-Mo-Cu alloys unless they contain appreciable quantities of both oxidiser and halide ions, e.g. FeCI, , CuCI,, NaOCI. Ni-Cr-Mo alloys are among the few metallic materials that are resistant to oxidising halide salts. Wet and Dry Gases
Nickel and its alloys are usually resistant to dry gases, including NH, , SO,, F, , CI,, HCI and H F even at high temperatures, and are often the preferred materials for handling such gases. Nickel and Alloy 600 are used in service at elevated temperatures with dry C1, , HCI, F, and HF. Alloy 600 and certain other nickel alloys are resistant to dry SO, and dry NH, . When moist or under dew-point conditions these gases are in many instances appreciably more corrosive towards nickel and most nickel alloys, with some exceptions. Ni-Cr-Mo alloys do, however, possess good resistance to condensates containing SO, and C1 at temperatures well in excess of 1OO"C, and also to solutions containing NH, and its salts. Ni-Cr-Mo alloys are among the most resistant metallic materials to moist halogens.
Table 4.27 A No v
H,SOa
Nickel 200 and 201
A , O-2Oqo('). RT
Alloy 400
A , 0-85%('), 30°C
Corrosion resistance of nickel and nickel alloys to acidic solutions
Hi PO, A , 0-85%('), RT
HNO,
U
HCI
HF
B, 0-70%('), RT
A , 0-10%(')
B, 0-50% ( I ) , 7OoC
A , 0-90%('), A , O-90%('), 100°C
U
A, O-8%"),
RT
A , O-%%('), 65°C A , 0-50%, BP
Alloy C276 c4 c22
A , &%Yo, 65OC E , O-IOVo, BP
Alloy G
B , 0-10%. BP
A , 0-85qoC'),65°C A , 0-50%, BP A , 0-85%, 65°C A , 0-50%, BP
A , 0-30%. BP
U
A , 0-70%, RT A , 0-30%, 65OC
A , 0-37%('), RT
B, 0-25%('), 65°C B, 0-2'70, BP
A , O-37%, RT A , 0-2%. 65°C B, 30-7070, 65°C B, 2-37%, 65°C B, 0-IO%, BP A , 0-1%, BP
Acetic, E , O-IOOVo, BP
RT Formic, A , 0-90%('), RT
B, 0-60%('), RT A , >90%('),RT
E, O-60%('), 95°C B, 25-85%('), 30°C Alloy B2
Other acids
A , 0-45%('), RT
RT B, 5-45%, RT A , 0-5%,
Acetic, A , 0-100%('), RT B, 0-40%(*), RT E , 65-100%(2),RT Acetic, A , 0-IOOVo, BP Formic, A , O-WVo, RT, BP B, 0-60%, 65'C Acetic, A , 0-100%, BP Formic, A , 0-90%, 65°C B, 0-90'70, 65OC
A , 0-45'70, RT
E , 3O-85%, BP (continued opposite)
z0 x
rn r
> z
0
2
0
x
m
r
F
6
s
Table 4.27
Alloy Alloy 825
Alloy 600
HZ
A , O-5%, 80°C B, 0 4 5 % . BP
HNO,
H2 p04
A , 0-20%, BP
B, 40-80%, 100°C
B, 20-80%, BP A , O-8O%, 75'C
A , 0-70%. RT(3)
A , 0-80%; RT(')
(continued)
A , 0-30%, BP
B 38-70%, 60°C
B , 30-70%, BP A , 0-70%, 75°C
The data show corrosion resistance as a funclion of acid composition, concentration taken as a firm indication of performance in SeWice. A = < 0.1 mm/year B = 0.1-0.5 mm/year S = no data. but often suitable in service U = unsuitable
A , 0-270, R T
B, 2-15%, RT(3)
Other acids
HF
HCI
S
x m
Acetic, A , 0-100%, BP Formic, A , 0-100%, BP Oxalic, A , 0-10%. BP B, 10-50%. BP
>
Acetic, A , 0-100%, R T
x m
and temperature. Since the data are mostly derived from laboratory corrosion tests in pure solutions. they should not be Vn = concentration w/w R T = room temperature BP = boiling point
zn
( I ) = air-free solutions: aeration increases corrosion (3) = saturated with air (3) = not resistant at high temperatures
r
z W
z0
r
F
0' 2
4: 150
NICKEL AND NICKEL ALLOYS
Organic Compounds
Nickel and nickel alloys are resistant to many organic compounds and are often suitable for handling organic acids, alcohols and halogenated hydrocarbons. It should be borne in mind, however, that halogenated organic compounds may undergo hydrolysis in the presence of water or steam and release appreciable quantities of the corresponding halogen hydracid, and this will often dictate the choice of the alloy. Detailed information should be sought concerning the suitability of alloys for particular circumstances. Water and Steam at High Temperatures
The corrosion rates of nickel and nickel alloys in pure water and steam at elevated temperatures are generally extremely low, typically of the order of 1 pm/year. The metal and its alloys are therefore often selected for service in these environments in circumstances where contamination of the water by metal ions is to be avoided. It should be noted, however, that the possibility of stress corrosion may need to be taken into account in certain circumstances (see below). Additionally where phosphate water treatment has been used in PWR secondary heat exchangers, severe localised corrosion has occurred when alkaline phosphates have been permitted to a c c ~ m u l a t e ~ * * ~ . Conjoint Action of Stress and Corrosion (Chapter 8)
As with alloys of other metals, nickel alloys may suffer stress-corrosion cracking in certain corrosive environments, although the number of alloy environment combinations in which nickel alloys have been reported to undergo cracking is relatively small. In addition, intergranular attack due to grain boundary precipitates may be intensified by tensile stress in the metal in certain environments and develop into cracking. Table 4.28 lists the major circumstances in which stress corrosion or stress-assisted corrosion of nickel and its alloys have been recorded in service and also shows the preventive and remedial measures that have been adopted, usually with success, in each case. With regard to stress-corrosion cracking in the Ni-Cr-Fe system, including both nickel-base alloys and stainless steels, a vast number of papers has been published. A detailed review of work published before 1969 is available% and the authors have since published additional data9'. The susceptibility of nickel alloys, principally Alloys 600 and 800 to stresscorrosion in water-cooled nuclear reactor heat-exchanger circuits has received much attentjon. The influence of both metallurgical variables (e.g. alloy composition, heat-treatment) and water chemistry (additives, inhibitors) have been extensively studied and reviewed.99"0z In recent years several new Ni-Cr-Fe-Mo and Ni-Cr-Fe-Mo-Cu have been introduced with improved resistance to sulphide stress cracking in sour oil and gas environments.'03-'@'
NICKEL AND NICKEL ALLOYS
Table 4.28 Alloy type Nickel 200 Alloy 600 Alloy 400 Ni-Cr-Fe and Ni-CrFe-Mo-Cu (Ni approx. 1 .25 mm/y. The material is ordinarily considered unsuitable. $See text. p. 5.54. $Not recommended. See pp. 5.47 and 5.53. qSee p. 5.46.
t A.
the existence of special circumstances, usually localised in character, which can give rise to corrosion even with an apparently innocuous medium. Thus, the presence of bimetallic couples, the existence of deep crevices, the presence of abrasive particles in a liquid stream, the incidence of local tensile stress, or the application of reciprocating stresses, are all features which demand consideration when a specific material of construction is to be selected.
Resistance to erosion Titanium has outstanding resistance to erosion resulting from the presence of abrasive particles entrained in cooling water and in process In one practical trial as a steam condenser tube, under circumstances known to result in rapid erosion of conventional condenser-tube alloys, titanium was virtually unmarked after more than 15 years service. As a turbine-blade material subject to impingement by water droplets moving at very high speeds, titanium has been shown to be superior to the conventional Fe-l3Cr, Fe-lSCr-SNi +Ti and to Monel. Under such circumstances, the harder the blade the better resistance it offers to erosion,
TITANIUM A N D ZIRCONIUM
5:45
and titanium alloys of the Ti-6A1-4V type give even better service than commercially pure titanium 18, particularly for large turbine generators.
Resistance to crevice corrosion Titanium is more resistant to crevice corrosion than most conventional metals and alloys, particularly where differential aeration is involved, e.g. it is very resistant to crevice attack in sea water at normal temperatures. This form of corrosion becomes more severe when acidity develops in a crevice and this is more prone to occur under conditions of heat t r a n ~ f e r ~ ’ *.’ ’Under * ~ ~ these circumstances, especially in the presence of halide, even titanium may suffer attack, and the metal should not be employed in strong aqueous halides at temperatures in excess of 130°C. This limiting temperature can be raised to 180°C by use of the or by coating with noble metals86. (See also Sections Ti-0- 15Pd all~y’’-*~ 1.4 and 1.6.) Some crevice attack upon titanium can also occur in the presence of gaseous chlorine gas at temperatures below loO°C, but this is mainly confined to crevices formed between titanium and organic sealing compounds. Here again, the Ti-O.15Pd alloy is less prone to attack. Titanium in contact with other metals In most environments the potentials of passive titanium, Monel and stainless steel, are similar, so that galvanic effects are not likely to occur when these metals are connected. On the other hand, titanium usually functions as an efficient cathode, and thus while contact with dissimilar metals is not likely to lead to any significant attack upon titanium, there may well be adverse galvanic effects upon the other metal. The extent and degree of such galvanic attack will depend upon the relative areas of the titanium and the other metal; where the area of the second metal is small in relation to that of titanium severe corrosion of the former will occur, while less corrosion will be evident where the proportions are reversed”. Metals such as stainless steel, which, like titanium, polarise easily, are much less affected in these circumstances than copper-base alloys and mild steel. In acid solutions, the behaviour of titanium/dissimilar-metal couples may differ from that just described, and on occasion titanium may be anodic to stainless steel and even to aluminium 24. In chemical-plant environments, therefore, it is usual to take the precaution of insulating titanium from adjacent components constructed from other metals. Resistance to stress-corrosion cracking Commercially pure titanium is very resistant to stress-corrosion cracking in those aqueous environments that usually constitute a hazard for this form of failure, and with one or two exceptions, detailed below, the hazard only becomes significant when titanium is alloyed, for example, with aluminium. This latter aspect is discussed in Section 8.5 under titanium alloys. For commercially pure titanium, the specific environments to be avoided are pure methanol and red, fuming nitric acid25-28v65, although in both environments the presence of 2% of water will inhibit cracking. On the other hand, the presence of either bromine or iodine in methanol aggravates the effect. When it does occur, stress-corrosion cracking of commercially pure titanium is usually intergranular in habit. Resistance to fatigue and corrosion fatigue The resistance of titanium to fracture by fatigue, induced by imposition of rapidly reversing stresses,
5:46
TITANIUM AND ZIRCONIUM
compares favourably with that of the more conventional metals and alloys. Commercially pure titanium has a definite fatigue limit, in air, at about half its tensile strength, and at this figure fracture may take place at between lo7 and lo8 reversals. In this respect the commercially pure metal resembles steel rather than the non-ferrous alloys. Reversed stresses at a figure below the limit indicated are not likely to result in fatigue failure, irrespective of the number of reversals applied. For many metals, the presence of corrosive environments coupled with reversing stress results in fracture by corrosion fatigue at a stress level well below that of the normal fatigue limit. While, given the appropriate environment, titanium is not immune to this effect, its generally good resistance to corrosion renders corrosion fatigue a comparatively rare event. Thus, the fatigue limit for titanium wetted with sea water is very similar to the figure obtained in air29.It is, therefore, not surprising that valve springs and valve plates of titanium alloys, used in gas compressors, give better performance than the conventional alloy steels. High-temperaturebehaviour Commerciallypure titanium is an established material for use at the moderately elevated temperatures attained in aircraft exhaust shrouds and firewalls, but neither titanium nor titanium alloys are suitable for use at really high temperatures. The tensile strength of commercially pure titanium shows a steady fall with increase in temperature, the tensile strength at 350°C being approximately half that at room temperature. The creep strength is improved by suitable alloying and an alloy containing 8% Al, 4.5% Sn, 4% Zr and smaller quantities of Nb, Mo, Si has a high creep strength at temperatures up to 600°C. Above about 600°C penetration of oxygen and nitrogen occurs. It has already been indicated that the presence of these elements renders the titanium brittle, and this feature must be taken into account in considering the use of titanium at elevated temperatures. Titanium has nevertheless been successfully employed as an autoclave lining in steam atmospheres at a temperature of 400°C and a pressure of 10MPa.
Examples of the Use of Titanium in Chemical Plant As with all fairly expensive materials of construction, economy in the use of titanium can often be achieved by using it in the form of thin linings upon a thicker load-bearing support. The soundest and most economical method of achieving this is by explosively bonding thin titanium sheet to thick steel plate. This technique, coupled with that of welding the duplex plate into large reaction vessels, is well establi~hed’”~’.~. The decision as to whether to use solid titanium or to employ explosively clad steel, depends upon the size and wall thickness of the construction and is largely decided upon economic grounds. The reader is advised to consult the specialist suppliers and fabricators before deciding which form to employ. For some chemical plant applications the iron content of the titanium employed can influence its behaviour, e.g. in some strengths of nitric acid, and in chlorine dioxide, preferential weld attack may occur if the iron
5:47
TITANIUM AND ZIRCONIUM
content of the titanium is above a certain critical level. This effect is only encountered in a few specific environmentsM, but where these are involved, it is recommended that titanium with an iron content of less than 0.05% is specified. There are also occasions, particularly in hydrogen-containing atmospheres, when surface contamination of the titanium with iron can result in localised corrosion and embrittlement. This effect can be countered by avoidance of undue contamination with iron during fabrication, by postfabrication cleaning and by post-fabrication anodising ‘6*67. It should be emphasized, however, that in general use in the marine and chemical industries discussed below, iron levels up to 0.2% do not adversely affect corrosion resistance. Examples of the use of titanium in the chemical industry are briefly summarised below. For more detailed treatment the reader is advised to consult Reference 32. Titanium is being employed in the bleaching industry36where the good corrosion resistance of the material makes it particularly suitable for equipment in both textile and paper pulp bleaching processes. In the dye-stuffs industry, the inertness of titanium eliminates any products of corrosion which might cause discoloration of the products. A similar situation can also exist in areas like the plastics, pharmaceuticals, and food-stuffs i n d ~ s t r i e s . ~ ~ Equipment lined with titanium has been employed for organic reaction vessels used in contact with nitric acid at elevated temperatures and press u r e ~ in ~ ~a , large rotary ammonium chloride dryer3’, for emulsion pans holding photographic solutions3s, and for general service in corrosive liquors. Shell and tube heat exchangers have been used to handle hydrochloric acid containing free chlorine37and chromic acid3*, while titanium pumps have been found to be useful for organic chlorides containing hydrochloric acid and free chlorinea. Important applications for titanium have been developed in processes involving acetic acid, malic acid, amines, urea, terephthalic acid, vinyl acetate, and ethylene dichloride. Some of these represent large scale use of the material in the form of pipework, heat exchangers, pumps, valves, and vessels of solid, loose lined, or explosion clad construction. In many of these the requirement for titanium is because of corrosion problems arising from the organic chemicals in the process, the use of seawater or polluted cooling waters, or from complex aggressive catalysts in the reaction. Titanium is the only one of the more common structural metals which is not attacked by wet chlorine gas and it is thus widely used as a heat exchange material for cooling the gas after the electrolysis stage. Preheating of sodium chloride brine is carried out in titanium plate heat exchangers, while titanium butterfly valves, demisters, and precipitators handle the chlorine gas produced in the cell. The most important use of titanium in chlorine production is as anodes in place of graphite in the electrolytic process. This is covered in more detail later. The resistance of titanium in nitric acid is good at most concentrations and ~ ’ ~ .tubular heat exchangers are used in at temperatures up to b ~ i l i n g - ’ ~Thus ammonium nitrate production for preheating the acid prior to its introduction into the reactor via titanium sparge pipes. In explosives manufacture, concentrated nitric acid is cooled in titanium coils and titanium tanks are
’.
5:48
TITANIUM AND ZIRCONIUM
used in the reprocessing of spent nuclear fuel elements by dissolution in nitric acid. The excellent corrosion resistance of titanium in sea water has led to one of the largest present and potential future uses for the material outside the aerospace industries. To all intents and purposes, commercially pure titanium is completely unattacked by seawater at ambient and moderately elevated temperatures. This has resulted in titanium becoming firmly established as a heat exchange material for power station condensers, for desalination plant, and in on-shore and off-shore oil installation^'^ where seawater or other polluted waters are used as the cooling medium. Titanium fittings are replacing stainless steel for racing yachtsg0 and the material is also being used for many applications in Naval vessels, particularly minesweepers where its nonmagnetic properties are an advantage. A rapidly growing use in the medical field” is for surgical implants as either bone plates and screws, joint replacements, or for the repair of cranial injuries. Here, titanium and its alloys have the advantages of complete compatibility with body fluids, low density, and low modulus. Applications also exist in dentistry. Titanium impellers have been used in pumps employed for the conveyance of corrosive and erosive ore slurries, for organic chlorides containing hydrochloric acid and free chlorinem, for handling moist chlorine gas, and in the wood-pulp and the textile-bleaching industry, particularly with sodium h y p ~ c h l o r i t e ~ ~ . In the electroplating industry, the use of titanium as hooks4’ and as heating and cooling coils for temperature control of certain acidic liquors has improved the control of plating Perhaps the most significant advance has been in the nickel plating industry where solid nickel anodes have been largely replaced by titanium baskets holding nickel shot and nickel shapes. In this development, the titanium itself is anodically passivated, but at the same time the passive film allows electron transfer to occur between contacting surfaces of titanium and nickel so that the latter is anodically dissolved. Anodic passivation also allows titanium to be employed as a jig for aluminium anodising baths43,because the protective anodic film formed on titanium allows passage of electronic current to the metal contact while virtually suppressing flow of ionic current through the anodically-formed surface film. This aspect is discussed in more detail in relation to special applications. In the field of electrowinning and electrorefining of metals, titanium has an advantage as a cathode, upon which copper particularly can be deposited with finely balanced adhesion that allows the electrodeposited metal to strip easily when required. Titanium anodes are also being employed as a replacement for lead or graphite in the production of electrolytic manganese dioxide. In the field of nuclear energy, titanium has been used for processing of fuel and for elements, where this demands use of nitric acid or aqua control-rod mechanism, in which the short half-life of irradiated titanium is of advantage. Environmental considerations in recent years have dictated that sulphur bearing compounds are removed from the exhaust gases of coal burning power stations in order to reduce the incidence of ‘acid rain’. The flue gas
TITANIUM A N D ZIRCONIUM
5:49
desulphurisation process (FGD), while removing the sulphur, also changes the character of the waste gases and makes them more corrosive to the materials from which power station chimneys are normally constructed. Considerable service experience has now demonstrated that titanium is resistant to the conditions and this represents one of the most promising uses of the material for the future.
Special Applications Anodic Passivation
It has already been indicated that titanium is not particularly resistant to corrosion in hot, strong acids of the type that usually generate hydrogen upon reaction with metals - acids such as sulphuric or hydrochloric. In contact with such acids, corroding titanium assumes a negative electrical potential (approximately -0.7 V, S.C.E.). If this negative potential is artificially raised by slow increments, a critical potential level may be attained, at which corrosion dramatically ceases and the metal acquires a protective film. The potential level at which this occurs usually lies between -0.5 V and -0.2 V (S.C.E.), and it is evident that at this potential level the metal/electrolyte interface has attained a thermodynamic state conducive to the formation of a stable, insoluble titanium dioxideM. This protective surface film has been shown to consist mainly of anatase, a tetragonal form of the oxide, and if the potential is further raised, the film thickens, giving a series of interference tints until it reaches a dark purple colour at a maximum limiting thickness of about 2 x lo-’ m. Once this film has become established, there is very little further passage of current into the electrolyte and corrosion virtually ceases as long as the ‘protective’ potential is applied. The most obvious means of attaining this potential is by application of an anodic direct current from an external source, but the same effect can also be attained to some extent by coupling the titanium to a more noble element such as carbon, or one of the platinum group metals. The latter method is, however, of limited application because it is dependent upon the potential level attained by the noble element, and this may not be sufficiently high to provide a mixed potential which is above the critical value for film formation (see Sections 1.4 and 1.5). Nevertheless, Stern and his a s s ~ c i a t e s ~have ’ - ~ ~shown that the addition of 0.2% palladium to titanium produces a discrete dispersion of palladium particles at the surface, which permits the combination to offer an adequate resistance to corrosion in 5 % boiling sulphuric and hydrochloric acids. Cotton 3*50, 5’ has shown that application of a d.c. potential of about 2 V between titanium and a suitable cathode can prevent corrosion in a wide range of strong non-oxidising acids at concentrations and temperatures which present considerable handling difficulties with most metallic materials of c ~ n s t r u c t i o n ~ ’ ~ * ~ . One full-scale practical application of this principle of anodic passivation is found in titanium heat exchangers handling 8% sulphuric acid containing hydrogen sulphide and carbon disulphide employed in viscose rayon processing 9.52 . It is conservatively estimated that each of these anodically passivated units performs the duty previously undertaken by three graphite heat exchangers. In doing this they require a current of only 1 - 5 A supplied at 15 v.
5:50
TITANIUM AND ZIRCONIUM
Use as Anodes As indicated above, when a positive direct current is impressed upon a piece of titanium immersed in an electrolyte, the consequent rise in potential induces the formation of a protective surface film, which is resistant to passage of any further appreciable quantity of current into the electrolyte. The upper potential limit that can be attained without breakdown of the surface film will depend upon the nature of the electrolyte. Thus, in strong sulphuric acid the metal/oxide system will sustain voltages of between 80 and lOOV before a spark-type dielectric rupture ensues, while in sodium chloride solutions or in sea water film rupture takes place when the voltage across the oxide film reaches a value of about 12 to 14 V. Above the critical voltage, anodic dissolution takes place at weak spots in the surface film and appreciable current passes into the electrolyte, presumably by an initial mechanism involving the formation of soluble titanium ions. Thus titanium by itself cannot function as an efficient anode for the passage of positive direct current into an electrolyte. The surface film of oxide formed upon the titanium has, however, a most useful property: while it will not pass positive direct current into an electrolyte (more correctly, while it will not accept electrons from negatively charged ions in solution), it will accept electrons from, or pass positive current to, another metal pressed on to it. Hence a piece of titanium which has pressed on to its surface a small piece of platinum will pass positive direct current into brine and into many electrolytes, at a high current density, via the platinum, without undue potential rise, and without breakdown of the supporting t i t a n i ~ m ~ ~ ’ ~ ~ . Platinised titanium anodes (titanium carrying a thin surface film of platinum, of the order of 0.0025mm thick) have proved successful in cathodic-protection systems employing impressed-current techniques, as electrodes for electrodialysis of brackish water, and in many applications where established anode materials suffer significant corrosion. Platinumcoated titanium anodes can operate without breakdown at very high current densities, of the order of 5000A/m2, in sea water, as although the very thin platinum coating may be porous the underlying titanium exposed at the pores will become anodically passivated ”. In aqueous chloride where it is necessary to use platinised titanium anodes coated over only part of their surface, e.g. titanium rod tipped with a thin platinum film, it may be necessary to limit the applied voltage to 12 V. The development of platinised titanium has been extended to include the replacement of platinum by deposits of other forms of corrosion-resistant conducting surfaces, such as platinum-iridium and ruthenium oxide. Apart from their corrosion resistance, these surfaces have the ability of operating electrochemically at lower overvoltages than plated platinum or graphite. Thus, for a range of electrochemical cells used in the chlor-alkali industry for the production of chlorine and sodium chlorate, etc. there is a significant advantage in using them compared with graphite”, and they are now the preferred choice for such applications. By a mechanism similar to that discussed in relation to platinum coating, titanium can function as a conducting jig to support aluminium components and assemblies in conventional anodising baths. In this application the exposed titanium acquires the insulating film, but allows current to pass to the aluminium at the points of contact56.
TITANIUM AND ZIRCONIUM
5:51
Pyrophoric Tendency If titanium is exposed to certain vigorously oxidising environments, oxidation does not cease at the surface, and a rapid exothermic reaction in depth ensues. The fundamental reason for this remarkable change in the character of the oxidation is not known with any certainty, but it is significant that in almost every instance the presence of a small quantity of water, sometimes in trace amounts, prevents this rapid oxidation in depth. Investigation into the effect has been mainly devoted to reactions with red fuming nitric acid". It seems that in red fuming nitric acid a preliminary reaction results in the formation of a surface deposit of finely divided metallic titanium; ignition or pyrophoricity can then be initiated by any slight impact or friction. The tendency to pyrophoricity increases as the nitrogen dioxide content of the nitric acid rises from zero to maximum solubility at about 20070, but decreases as the water content rises, the effect being nearly completely stifled at about 2% water. Other media in which titanium is subject to pyrophoricity are anhydrous liquid or gaseous chlorine58, liquid bromine, hot gaseous fluorine, or oxygen-enriched atmospheres at moderately low pressures.
Titanium Alloys Mechanical properties of various titanium alloys are given in Table 5.16. In general the corrosion behaviour of those titanium alloys developed for the aircraft industry is very similar to that of unalloyed titaniums9. The addition of some alloying elements may increase resistance to one medium, but decrease it to othersb0. Additions of zirconium confer a significant increase in corrosion resistance, particularly in sulphuric and hydrochloric acids 59*61. At alloying additions of the order of 50% Zr, however, there can be a significant diminution in resistance to oxidation" and the welding of titanium to zirconium is not advisable, because within the welded zone the proportion of titanium to zirconium will almost inevitably fall within the sensitive composition range. The addition of 0.2% palladium to titanium decreases the corrosion rate in boiling 5% sulphuric acid by a factor of 500, and in boiling 5% hydrochloric acid by a factor of 1500, in relation to the rates obtained with unalloyed titanium. The addition of palladium in these quantities thus provides an adequate measure of resistance to relatively weak concentrations of the acids mentioned48. From the corrosion-resistance aspect, one of the most effective additions to titanium is that of molybdenum. According to Yoshida and his colleague~~~-", the addition of 15% Mo produces an alloy fully resistant to virtually all concentrations of sulphuric and hydrochloric acid at room temperatures, while with 30% Mo, the alloy is resistant to all strengths of boiling sulphuric acid up to a concentration of 40% by weight, and to 10% boiling hydrochloric acid. The stress-corrosion cracking hazard for titanium alloys containing aluminium is significantly higher than that obtaining for commercially pure titanium, and in addition to stress-corrosion cracking in methanol and red
Table 5.16
..
VI
Mechanical properties of some titanium alloys
VI h)
Stress f o r Nominal composition in weight % ond chorocteristics
Ductile medium strength Ti-2.5Cu alloy, weldable and age hardened BSTA 21-24, 52-55, 58
0.2% proof stress (min) (MW
Tensile strength (MPa)
Elongation (min) (‘o)
Young’s modulus
(To of T.S.)
Bend radius on 2 mm
Density
(typicol) (GPa)
21
4.56
S, B, W , E
4.56
S, B, W,E
4.51
S
limit
Annealed
400
540-770
16
105-120
60-65
Solution treated and aged
525
650-880
10
105-120
60-65
170
330-420
25
105- I20
50
Small additions of Pd giving improved resistance to nonoxidisine, acids
It
(g’cm3)
‘ ‘ I % toto‘ plostic stroin in 100 h (MPa)
Production range
> z
W
Medium strength Ti-6A1-4V alloy BS TA10-13, 28, 56
Sheet
900
960- 1270
8
105-120
55-60
Rod
830
900-1 160
8
105-120
55-60
960
1 100-1280
9
110-130
50-60
1095
1250-1420
8
110-130
40-50
970
1110-1340
8
105-1 10
55-60
High strength Ti-4AI-4Mo-2Sn 0.5Si alloy. Creep resistant up to 400°C BSTA 45-51, 57 Very high strength Ti-4AI-4Sn-4Mo-0.5Si BSTA 38-42
alloy
High strength Ti- 1 1Sn-5Zr-2.25A1-1 Mo alloy, creep resistant up to 450°C BSTA 18-20, 25-27
51
4.42
S
4.42
B, W, E
4.62
Table 5.16
(continued)
Medium strength Ti-6Al-5Zr0.5Mo-0.2Si alloy, weldable and creep resistant up to 520°C BSTA 43. 44
Room
850
990- 1 140
6
125
50
4.45
520°C
480
620-780
9
125
50
4.45
Medium strength Ti-5.5A1-3.5Sn3Zr- 1Nb-O.3Mo0.3Si alloy weldable and creep resistant up to 550°C
Room
820
950 min
10
120
50
4.51
B, E
300
B, E B, E
2
-I
? L
300
B, E
540°C
460
590 min
12
120
50
4.51
Medium strength Ti-6Al-7Nb alloy for surgical implant applications
800
900-1200
10
105
55-60
4.52
B
Medium strength Ti-5.8AI-4Sn3.52-0.7Nb-0.5 Mo-O.35Si-O.MC alloy, weldable and creep resistant up to 600°C
Room
910
1030
6
120
60
4.51
B
600°C
450
585
9
High strength Ti-1SMo-3Nb3A1-0.25i alloy, oxidation resistant
965
1035-1350
4
96
4.92
St
High strength Ti-3 * 5AI8V-6Cr-4Mo-4Zr alloy, deep hardenable and corrosion resistant
1180
1250
11
106
4.82
B, E
-~
=!
Table 5.17 Physical properties of unalloyed zirconium ~
2
~~
Atomic Atomic number weight
Crystar structure &low
865°C
4o
9 1 . 2 c.p. hex at 25°C
Above 865°C
Thermal neutron Electrical Temperature specific Thermal Standard Melting Density Thermal electrode absorption cross-section. point at 20" C conductivity resistivity coefficient heat expansion Reactor grade (oc, (g/cm3) (W/m (at 20°C) of resistivity (J,goc) per oc potential (aQ/cm) ("C) (") Microscopic Macroscopic
E >
-
E
b.c.c. at 900°C
a =3'23?A a = 3.61A c = 5.15A
2
z 1845
6.490
22
39.7
44
0.276
5.89
10-6
-1.53
0.180 bardatom
0.08mm
8 5
E
TITANIUM AND ZIRCONIUM
5:55
fuming nitric acid, cracking has been observed in salt solution, in hot solid sodium chloride and in uninhibited chlorinated hydrocarbons. Because of the importance of these alloys to the aircraft industry there has been considerable laboratory investigation of the effect and the reader is advised to consult References 65 and 66 and the literature for a comprehensive treatment of the subject. (see also Section 8 . 5 ) Viewed in perspective, evidence of failure in service has been rare and the practical hazard is certainly very much lower than would appear from the results of laboratory tests. In chlorinated hydrocarbons the effect can be controlled by the addition of inhibitors, and, for example, the appropriate commercial degreasants containing these inhibitors are specified in a British detence standard*.
ZIRCONIUM The growth of nuclear engineering with its specialised demands for materials having a low neutron absorption coupled with adequate strength and corrosion resistance at elevated temperatures, has necessitated the production of zirconium in relatively large commercial quantities. This specific demand has resulted in development of specially purified zirconium, and certain zirconium alloys, for use in particular types of nuclear reactor. In its natural state, zirconium is associated with hafnium, and for use in nuclear reactors it is essential to separate the two because hafnium readily absorbs neutrons. This situation gives rise to bulk production of two forms of raw zirconium metal, a hafnium free reactor grade and a commercially pure hafnium bearing quality (ASTM designations R60001 and R60702 respectively). A number of different zirconium alloys are also commercially available including one containing tin, iron, chromium, and nickel additions (R60802) and a similar material (R60804)but without the nickel. Both of these are used in water cooled nuclear reactors. A zirconium 2+ 070 niobium alloy (R60901) provides a heat treatment capability, while in the chemical industry a similar alloy (R60705) offers good corrosion resistance and better strength than commercially pure zirconium. Generally, for the chemical engineer not particularly associated with atomic energy, unalloyed zirconium containing hafnium is an appropriate choice for those occasions which require the special corrosion resistant properties exhibited by the metal.
Physical and Mechanical Properties The physical properties of unalloyed zirconium are recorded in Table 5.17. Mechanical properties of these grades of zirconium depend to a large extent upon the purity of zirconium sponge used for melting. Hardness and tensile strength increase rapidly, with rise in impurity content, notably oxygen, nitrogen and iron. Typical mechanical properties of chemical grades of zirconium are listed in Table 5.18. * The Cleuning und Prepurulion of MeIulSurfuces,
Defence Standard 03-2/1 (1970). obtainable from the Ministry of Defence, First Avenue House, High Holborn, London, W.C.1.
5:56
TITANIUM AND ZIRCONIUM Mechanical properties of chemical grades of zirconium
Table 5.18 ASTM designation
0.2% proof stress
Tensile strength (MPa)
Elongation
(MPa)
R 60702
207 (rnin)
379 (min)
16 (rnin)
5t
R 69705
379 (rnin)
552 (rnin)
16 (rnin)
3t
(%)
Bend radius ~~
Table 5.19 gives the physical properties of Zr-Sn-Cr-Ni alloy. Table 5.19 Physical properties of Zr-Sn-Cr-Ni alloy Alloy
nominal composition
Density at
Electrical
(g/cm2)
at 21OC ( p cm) ~
6.57
74
20"c
(W) Zr-1.5 Sn0.1 Cr-0.12 Fe -0.05 Ni 1 barn =
Thermal neutron absorption cross-section
Of linear thermal expansion
("C) 20-7000c 6.5 x
25-6000c
Microscopic
Macroscopic
-
0.22-0.24 barn*/ atom
-
Cm2.
The mechanical properties of the alloys will vary slightly according to the purity of sponge, and also with heat treatment. Table 5.20 Minimum mechanical properties of nuclear grade zirconium alloys ASTM designation
Condition
Direction of 0.2% proof stress Tensile strength Elongation (MPa) (MPal (Yo) test
R 6ooo1
Annealed
Transverse
207
296
18
R 60802
Annealed
Transverse
303
386
25
R 60804
Annealed
Transverse
303
386
25
R 60901
Annealed
Transverse
344
448
20
R 60901
Cold worked
Transverse
385
510
15
Behaviour of Commercially Pure Zirconium in Aqueous Environments Zirconium, like titanium, depends upon the integrity of a surface film, usually of oxide, for its corrosion resistance, but there are differences in behaviour between the two metals when they are exposed to aggressive aqueous environments. In general, zirconium does not equal titanium in resistance to certain oxidising media, but it is superior in non-oxidising acids, and in caustic alkalis. The presence of certain impurities in zirconium influences the corro-
TITANIUM AND ZIRCONIUM
5:57
sion behaviour, and while small amounts of hafnium are not deleterious, carbon in amounts greater than 0.06% lessens resistance to hot concentrated The contrast in hydrochloric acid by a factor of several hundreds68969. behaviour between titanium and zirconium in a wide range of media is illustrated in detail in Table5.15. To summarise, zirconium performs well in nitric acid at all concentrations up to 70% and temperatures up to 200"C6', but it will react pyrophorically in a fashion similar to titanium in concentrated nitric acid containing free nitrogen dioxide. If there are appreciable amounts of hydrochloric acid present together with nitric acid, there may be severe attack, and, in contrast to titanium, zirconium is not resistant to aqua regia containing three parts nitric to one part hydrochloric acid. Towards chromic acid, zirconium is resistant at least up to a strength of 50% at a temperature of 90°C. In saturated chlorine water the corrosion rate of zirconium is virtually nil, but unlike titanium it is attacked in moist gaseous chlorine and not in dry chlorine at room temperature". In solutions of metal chlorides, behaviour appears to depend upon whether the chloride solution tends to be oxidising or reducing, and in general zirconium is not as resistant as titanium. Thus it is not resistant to boiling ferric or cupric chlorides at strengths greater than 10070,but it is resistant to mercuric, stannic, manganous, nickel, ammonium, zinc, magnesium, barium and sodium chlorides, and to sea water. Behaviour in aluminium chloride is worth noting, for zirconium is resistant to boiling 25% aluminium chloride, while titanium is attacked. Both metals corrode in boiling 62% calcium chloride. In resistance to hydrochloric and sulphuric-acids zirconium shows a significant advantage over titanium6'. With pure hydrochloric acid at 100°C the corrosion rate is negligible up to the constant boiling strength, i.e. 20% w/w at atmospheric pressure, but at 200°C under pressure there is appreciable attack at acid strengths greater than 18% by weight. The presence of traces of copper and iron in the hydrochloric acid can result in a significantly increased rate of attack, and, for example, in boiling 20% acid 1 OOO p.p.m. of iron or copper raises the rate of attack from less than 0.0075 mm/y to the barely acceptable level of 0 . 5 mm/y. In sulphuric acid where traces of metal ions do not appear to be unduly troublesome, there is no appreciable corrosion up to 66% w/w at boiling point; the rate of attack, however, increases rapidly in boiling 70% acid, and at 200°C under pressure there is significant uniform corrosion at about 40% w/w. The presence of chlorine in sulphuric acid can seriously increase the rate of corrosion. With phosphoric acid the performance of zirconium is again distinctly superior to that of titanium, for while, in general, use of titanium is limited to strengths less than 30% w/w, for zirconium there is no appreciable corrosion at room temperature up to 80% strength. As temperature rises there is an inflection in the corrosion-rate curve, an unacceptable rate being reached in boiling acid at 50% strength. As temperature rises beyond this, the corrosion rate again decreasesm, until at 200"C, under pressure, there is again negligible attack in 80% acid. Neither titanium nor zirconium is recommended for use in hydrofluoric acid. Zirconium is also resistant to attack in a wide range of organic acids, one useful difference from titanium being that it is not corroded in boiling
5:58
TITANIUM A N D ZIRCONIUM
deaerated formic acid at concentrations of 25% and upwards, in which titanium exhibits borderline passivity. In strong chlorinated organic acids, however, there may be some attack at elevated temperatures. It is in its behaviour to caustic alkalis that zirconium shows itself to be superior to those other elements of Groups IV and V whose resistance to corrosion results primarily from an ability to form surface films. Thus, in contrast to tantalum, niobium and titanium, zirconium is virtually completely resistant to concentrated caustic solutions at high temperatures, and it is only slightly attacked in fused alkalis. Resistance to liquid sodium is good. Zirconium is thus an excellent material of construction for sections of chemical plant demanding alternate contact with hot strong acids and hot strong alkalis -a unique and valuable attribute. Because of its good performance in mineral acids, there is little need or incentive to invoke anodic passivation techniques for zirconium. The metal can be anodised in sulphuric acid, but, again in contrast to the behaviour of titanium, it does not form a stable anodic film in chloride solutions, and even in neutral sodium chloride, zirconium rapidly corrodes if an anodic potential of 2 V is applied.
Applications in Industry The chemical industry now provides a major area for the use of zirconium equipment. The material is employed in the form of heat exchangers, stripper columns, reactor vessels, pumps, valves, and piping for a wide variety of chemical processes. These include hydrogen peroxide production, rayon manufacture, and the handling of phosphoric and sulphuric acids and ethyl benzene. Gas scrubbers, pickling tanks, resin plants, and coal gasification reactors are some of the applications where the good corrosion resistance of zirconium towards organic acids is utilised. A particularly useful attribute is the ability of the material to withstand environments with alternating acidity and alkalinity.
Special Applications It has already been indicated that the principal use for zirconium is in the field of nuclear engineering. The very nature of this application demands the lowest possible corrosion rate, and this has necessitated a great deal of investigation into the oxidation rate of zirconium, when exposed to hot water, steam and carbon dioxide. When zirconium oxidises in these environments at elevated temperatures the reaction kinetics follow a law which can be formulated as w = Kt" where w = weight gain, t = time, and K and n are constants at a constant temperature. Initially n has a value of between f and +, and the rate of oxidation decreases with time. However, when a certain thickness in the surface film is attained, the value of n may change and become equal to or greater than unity. The corrosion rate will then become constant or will
5:59
TITANIUM AND ZIRCONIUM
increase. This type of behaviour, which can occur with several metals or alloys, has been called ‘breakaway’ corrosion. Within the period at which the value of n remains below unity, the monoclinic oxide film produced on zirconium is hard, glossy, adherent and usually black or dark coloured. When the kinetic change takes place the character of the film changes, and continued oxidation may lead to heavy surface spalling Unalloyed zirconium produced from Kroll sponge quickly reaches the breakaway point when exposed to steam or hot water at reactor temperatures. Early investigation in the United States established that this behaviour resulted from the almost inevitable presence of nitrogen, but that the deleterious effect could be countered by an addition of tin7’, and the alloy knownasZircaloy2, containingabout 1.5% Sn, 0.1% Fe, 0.1% Cr, 0.05% Ni was developed for use in water-cooled reactors. Even with this alloy, metallurgical treatment during fabrication is known to affect performance, and a rigorous scheme of corrosion testing is e m p l ~ y e d ~to~ ,ensure ’ ~ that the semi-fabricated material and finished product conform to a high degree of corrosion resistance. This test involves the autoclaving of carefully prepared coupons for fourteen days in pure steam at a temperature of 400°C and a pressure of 10MPa. At the conclusion of the test, satisfactory material has a weight gain of 28 f 10mg/dm2, and is covered with a glossy black lustrous film. Defective material manifests itself by high weight gains (up to as much as 100mg/dm2) and the appearance in the surface film of white corrosion product. Most of the considerable volume of published work on the behaviour of zirconium relates to its use in nuclear reactors in contact with water or steam, e.g. in pressurised steam the control of oxidation by use of boric acid has been rep~rted’~. The reader is advised to consult the reviews on this important aspect of the subject cited under References 76 and 77. It should be noted that swarf from a zirconium-titanium alloy containing approximately 50% by weight of each element is prone to pyrophoricity in air. It has also been reported6’ that when zirconium is welded to titanium, the welded zone is much more sensitive to corrosion than either of the parent metals. If, therefore, it is proposed to use any construction in which zirconium is welded to titanium, caution should be observed in the machining of welds, and the corrosion behaviour of the weld should be checked by prior testing in the environment with which the construction will be employed. The pyrophoric tendency of zirconium in contact with red fuming nitric acid has already been mentioned. There is some evidence that the increase in corrosion recorded when zirconium is exposed to hydrochloric acid at 200OC under pressure results from intergranular penetration”. Finally, perhaps, it should be pointed out that because the behaviour of zirconium is often adversely influenced by the presence of impurities in corrosive environments, corrosion testing prior to use should be carried out in actual plant liquors rather than in purer synthetic solutions. J. B. COTTON B. H. HANSON
.
5:60
TITANIUM AND ZIRCONIUM
REFERENCES Evans, U. R., The Corrosion and Oxidation of Metals, Arnold, London, 39-48 (1960) Nakayama, Castings Research Laboratory Report, No. 5 , Waseda University, 57-59 (1956) Cotton, J. B., Werkst. u . Korrosion, Weinheim, 2 No. 3, 152 (1960) Adamson, G. M., Jr. et ai., Proceedings of the Second U.N.International Conferenceon the Peaceful Uses of Atomic Energy, Paper P/1 993, United Nations, Geneva (1958) 5 . Jackson, J. D., Mat. Prot., 4 No. I , 30-33 (1965) 6. Millaway, E. E. and Kleinman, M. H., Corrosion, 23, 88-97 (1967) 7. Wullner, R. L., Mat. Prot., 4 No. 1, 55-56 (1965) 8. Mueller, W. A., J. Electrochem. SOC.,107, 157 (1960) 9. Cotton, J. B., Chemical Engineering Progress, 66 No. 10, 57-62 (1970) 10. CorrosionResistanceof Titanium, IMI Titanium Ltd., P.O. Box 704, Witton, Birmingham 11. Pourbaix, M., Rapport No.21, Centre Belge d’gtude de la Corrosion, Brussels (1953) 12. Fischer, W. R., Werkst. u Korrosion, Weinheim, 10, 243 (1959) 13. Schlain, D. and Smetko, J. S., J. Electrochem. SOC.,99, 417 (1952) 14. T.M.L. Report No. 57, Titanium Metallurgical Laboratory, Battelle Memorial, 116-153,Oct. 29 (1956) 15. Stern, M. and Wissenberg, H., J. Electrochem. SOC.,106, 754 (1959) 16. Thomas, N. T. and Nobe, K., J. Electrochem. SOC., 116, 1 748 (1969) 17. Cotton, J. B. and Bradley, H., Chem. and Ind. (Rev.), 643 (1958) 18. Cotton, J. B. and Downing, B. P., Trans. Inst. Mar. Engrs., 69,311 (1957) 19. Feige, N. G. and Kane, R. L., Metals Engr. Quart., 7, 27-29 (1967) 20. Sims, M. H., Power, 112,890-96 (1968) 21. Greiss, J. C., Corrosion, 24, 96-109 (1968) 22. France, W. D. and Greene, N. D., Corrosion, 24, 247-51 (1968) 23. Takamura, A., Corrosion, 23, 306-13 (1967) 24. Schlain, D., US Bureau of Mines, Report No. 4 965, April (1953) 25. Sedriks, A. J., A.S.M. Trans. Quart., 61, 625-27 (1968) 26. Harey, E. G. and Wearmouth, W. R., Corrosion, 25, 87-91 (1969) 27. Sedriks, A. J., Corrosion, 25, 325-28 (1969) 28. Rittenhouse, J. B. etal., Trans. Amer. SOC.Metals., 51, 871, 895 (1959) 29. Inglis, N. P., Chem. and 2nd. (Rev.), 180 (1957) 30. Engineering, Jan. 6 (1967) 31. Obrig, H. and Ehle, J. C., Chem. Process. Engrg., 50 (1969) 32. Hanson, B. H., The Chemical Engineer, 276-79, April (1978) 33. Barron, L. J., Light Metal Age, 14 Nos. 3 and 4, 16 (1956) 34. Industr. Engng. Chem., 50, 934 (1958) 35. Connolly, B. J., Chem. Proc. Engng., 39, 247 (1958) 36. Bomberger, H. B., Industr. Engng. Chem., 49, 1658 (1957) 37. Carmichael, M. L., Battelle Techn. Rev., 5 No. 12, 9 (1956) 38. Steel, 143 No. 26, 62 (1958) 39. lshii, Y. and Hoskino, Y., Chem. Engng., Tokyo, 21, 559 (1957) 40. Frazer, G. T. et al., Mat. and Meth., 43, 112 (1956) 41. Light Metal Age, 17,27, Oct. (1959) 42. Corrosion, 15, 82 (1959) 43. Hames, W. T., Aircraft Prod., Lond., 20, 369 (1958) 44. Savolainen, J. E. and Dlanco, R. E., Chem. Engng. Prog., 53, 78F (1957) 45. Peterson, C. L. et al., Industr. Engng. Chem., 51, 32 (1959) 46. Schmets, J. and Pourbaix, M., Proceedings of the 6th Meeting of the International Committee for Electrochemical Thermodynamics and Kinetics, Poitiers, 1954, Butterworths, London (1955) 47. Stern, M. and Wissenberg. H., J. Electrochem. SOC., 106,755 (1959) 48. Stem, M. and Wissenberg, H.,J. Electrochem. SOC.,106,759 (1959) 49. Stem, M. and Bishop, C. R., Amer. SOC.Met., Preprint No. 165 (1959) 50. Cotton, J. B., Chem. and Ind. (Rev.), 68 (1958) 51. Cotton, J. B., Werkst. u Korrosion, 11, March 3 (1960) 52. Evans, L. S., Hayfield, P. C. S. and Morris, M. C., Proc. 4th Intern. Congress on Metallic Corrosion 53. Cotton, J. B., Chem. and Ind. (Rev.), 492 (1958) 54. Cotton, J. B., Platinum Metals Rev., 2 , 45 (1958) 1. 2. 3. 4.
TITANIUM AND ZIRCONIUM
5:61
55. Shreir, L. L., Plafinum Mefals Rev., 4, I5 (1960) 56. Jones, J. C., Prod. Finish., Lond., 12 No. 12, 81 (1959) 57. Rittenhouse, J. B. et al., Trans. Amer. SOC. Metals, 51, 871, 895 (1959) 58. Millaway, E. E. and Kleinman. M. H., Corrosion, 23, 88-97 (1967) 59. Golden, L. B. et a/., Trans. Amer. SOC.Mefals, 51, 871, 895 (1959) 60. Schlain, D. and Kenahan, C. B., Corrosion, 14, 405t (1958) 61. Andreeva, V. V. and Gluklova, J . A p p f . Chem., 11, 390 (1961) 62. Cotton, J. B., Chem. and Ind., 357-358 (1962) 63. Yoshida, S. e t a l . , J . Gout. Mech. Lab., Tokyo, 10, 2-21 (1956) 64. Yoshida, S. etal., J . Jap. Insf. Metals, 21 No. 3, 183 (1957) 65. S.T.P. 397, A.S.T.M., 1916 Race St., Philadelphia, USA 66. Jackson, J. D. and Boyd, W. K., The Science, Technology and Application of Titanium, Pergamon, 267-281 (1966) 67. Imperial Chemical Industries, Ltd., Brit. Pat. 1 187 771 (15.4.70) 68. Kuhn, W. E., More Zirconium Facts, 1 No. 2, Carborundum Metals Company, 4 (1957) 69. Kuhn, W. E., Chem. Engng., 156 (1960) 70. Gegner, P. J. and Wilson, W. L., Corrosion, 15, 341t and 350t (1959) 71. Thomas, D. E., Proceedings of the First UN Conference on Atomic Energy, Geneva, 1955, 9, Paper P/537, 407, United Nations, Geneva 72. Cotton, J. B. and Gallant, P. E., Proceedings ofthe Firsf International Congress on Mefallic Corrosion, London, April, 1961, Butterworths, London, 458 (1962) 73. Kass, S., Corrosion, 16, 137 (1960) 74. O'Driscoll, W. G., Tyzack, C. and Raine, T., Proceedings of the Second Conference on the Peaceful Uses of Atomic Energy, Geneva, 1958.5, Paper P. 1 450, Geneva, 75 (1958) 75. Britton, C. F., J. Nuc. Mat., 15 No. 4, 263-277 (1965) 76. Coleman, C. E. and Hardie, D., J . Less Common Metals, 11, 168-85 (1966) 77. Rosa, C. J., J. Less Common Metals, 16, 173-201 (1968) 78. Kuhn, W. E., Corrosion, 16, 141t (1960) 79. Cotton, J. B., Localised Corrosion 676 NACE International Corrosion Conference series NACE-3 (eds R. W. Stahle, B. F. Brown, J. Kruger and A. Agrawal) (1974) 80. Schutz, R. W., Grauman, J. S., and Hall, J. A., 5th International Conference on Titanium, Munich, 2617-24 (1984) 81. Hayfield, P. C. S. and Hanson, B. H. Chemical Processing, 16 (5). 52 May (1970) 82. Coulter, M. 0..Mudern Chlur-Alkali Technology, S.C.I. 5.71 (eds J. H. Collinsand J. H. Entwhistle) (1980) 83. Cotton, J. B. and Scholes, I. R., Trans. Insf. Mar. Eng., 84, 16. 538 84. Satoh et al., 5th International Conference on Titanium, (eds G. Lutjering, U. Zwicker and W. Bunk) Munich, 1165-71 (1984) 85. Kobayashi et a / . ,4th International Conference on Titanium, (eds H. Kimura and 0. Izumi) Kyoto, 2613-22 (1980) 86. Fukuzuka et a/., 4th International Conference on Titanium, (eds H.Kimura and 0. Izumi) Kyoto, 2631-38 (1980) 87. Hanson, B. H., 2nd International Conference on Titanium, (eds R. I. Jaffee and H. M. Burte) Boston, 2419-29 (1972) 88. Gehring, G. A. and Kyle, R. J . Paper 60, Corrosion (1982) 89. Brettle, J., Metals and Materials, 442-51 Oct. (1972) 90. Hanson, B. H., Seahorse. 97, 46-7 Nov./Dec. (1986) 91. Hanson, B. H., Materials and Design, Vol vii, No. 6, 301-7 Nov./Dec. (1986)
5.5 Tantalum
General Tantalum is one of the most versatile corrosion-resistant metals. Its corrosion behaviour can be compared with that of glass in most environments. This behaviour is attributed to the stable passive film of Ta,O, produced on the surface during exposure. The pure metal has a very high melting point (2996OC) and is blue-grey and like lead in appearance. It has a density of about twice that of carbon steel (16.6 g/cm3) and a similar thermal conductivity. It is one of the refractory metals and suitable for high temperature application under protective conditions. It can be readily cold worked, but hot working, however, must be avoided as the metal reacts with gases such as oxygen, nitrogen and carbon dioxide with resultant embrittlement. It can be machined, although care is necessary to obtain a good surface finish. The high strength, good workability and excellent corrosion resistance permit the use of very thin walled components, a commonly employed thickness in chemical plant being about 0.3 mm. These properties, coupled with the metal's ability to promote bubble-type vapour formation on the surface when heating liquids, and dropwise condensation when condensing vapours, make the metal an ideal constructional material for heat-transfer equipment for use with strong acids. The absence of corrosion, coupled with the fact that scale and other deposits appear to be dislocated by thermal cycling, result in a finish on tantalum heating surfaces that is as good as the original, even after 20 or 30 years in service, and also ensure that good heat-transfer properties are maintained throughout the life of the equipment. The use of tantalum for process equipment also ensures freedom from contaminations of the product. The mechanical properties of tantalum are dependent on the previous history of the material and the manufacturer should be consulted if these properties are likely to be critical. The physical and some typical mechanical properties are listed in Tables 5.21 and 5.22. The effect of the temperature on the strength and elongation of tantalum sheet in vacuum is shown in Figs. 5.8 and 5.9.
5:62
Table 5.21
Physical properties of tantalum
~~
Thermal
Melting
Density
I
Boiling
'
point
(g/cm ')
("C)
("C)
Thermal neutron absorption
Electrical
Linear coefl. of expansion ("(3 . .
cross section
Thermal conductivity
Specific heat2
Electrical resistivity
(W/cm"C)
(Jk" Cl
(uWcm)
5.44 at 20°C 7.52 at 1106°C
0.142 at 0°C 0.161 at 1227°C
12.43 at 20°C 54.8 at I O00"C
'.
Temperature coeff. of resistivity
("C)
(badatom)
16.6
2 996
5 425
21
20-500°C 6.6 x lop6 20-1 500°C: 9.0 x
3.82 x lo-' at 0-100"C 3 x io-' at 0-1 0 0 0 ~
2z
4
FC Table 5.22 Mechanical properties of tantalum Modulus
Poisson's
of
ratio
Yield stress
(MN/m
2,
UTS2*
(MN/m *)
("C)
(GN/m')
186 at 2OoC -. ~. 151 at 1 O00"C ~
Stress Ductile lo Recrystallisation relieving Hardness' (VPN) Stability2 brittle transition rempercrrure~ temperature2 (oc) temperature4 Annealed Hard worked (To creep ratelh, min.)
0.35
179-1 060 a t 27°C and 44-310 at 500°C
689-1 034 a t 20°C and 103-138 at 1 O00"C
None detected down t o -196
("C) 1 050-1 500
900
80-100
180
0.113 at 750°C. 96 G N / m
*
.. 8
VI
5:64
TANTALUM
-ae
I
z
0
2 0
sz
W
.
m -
E
Z
-2
I
8 z
600-
E
5M)-
W
cn
ULTIMATE STRENGTH
LOO-
W
i cn
300-
W
200-
z
c
1KI
200
Fig. 5.8
600
800 TEMPERATURE (K)
LOO
I
1000
1200
I
I
I
ILOO
Effect of temperature on the tensile strength and elongation of tantalum 400
. z
-
m -
E
E. 300
-
c
I I-
o z
2200-
t;; w I4
2 100
ti
-
3
0'
I
1
1
500
1000
1500
TEMPERATURE f
I
2000
"c)
Fig. 5.9 Effect of temperature on the ultimate strength of tantalum
Methods of Fabrication The high melting point and reactivity of tantalum with the permanent gases at high temperatures prevents conventional consolidation by melting and casting in air. The metal is in fact consolidated by vacuum sinter-
TANTALUM
5:65
ing, vacuum-arc melting and electron-beam melting of powder compacts. Vacuum sintering yields metal of fine grain, whereas electron-beam melting yields softer coarse-grained metal which requires cold forging prior to rolling. Metal produced by all three techniques will absorb considerable cold work before annealing is necessary.
Rolling and swaging Vacuum-sintered bar can be cold rolled, and reductions up to 90070 between anneals are possible. Arc-cast and electron-beammelted material is generally forged at room temperature prior to rolling and swaging. Drawing Tantalum has a tendency to gall and is normally anodised to provide a surface which will carry a drawing lubricant. Seamless tube is produced by cupping followed by drawing or by hollow shells. Spinning Tantalum can be formed by all conventional spinning techniques, provided a lubricant such as tallow is employed, and can be spun into configurations which cannot be produced by other forming methods. Machining Tantalum is readily machined using high-speed-steel tools, provided a lubricant such as trichloroethane is employed. Blanking and cutting Tantalum can be blanked, cut and sheared using similar equipment and techniques to those used for austenitic stainless steel. Joining Tantalum can be joined by riveting, brazing and welding; however, due to the good properties of welded joints the former techniques are seldom used. Welding Because of the reactivity of the hot metal with the permanent gases, conventional welding techniques cannot be used. In general the practical methods are restricted to tungsten-electrode inert-gas (TIG), resistance, electron-beam (EB) and plasma-arc welding. To ensure statisfactory TIG welds, welding should be done in an inert-gas-filled chamber. Material thinner than 0.5 mm cannot readily be TIG welded and resistance welding has to be used. Spot welding can be carried out in air and under water. EB welding gives a contamination-free narrow weld and heat-affected zone, irrespective of material thickness, and plasma-arc welding has been used in 0.05-1 -0mm sheet and gives a weld with similar properties to EB welds.
Economical Considerations The relatively high cost of tantalum has been a limiting factor in its use. Fabrication techniques, in which thin linings of tantalum are used, result in equipment at a much lower cost than an all-tantalum construction. The long life and reliability of tantalum equipment in severe-corEosion applications often more than offsets its higher initial costs. Therefore, a new situation has been created for utilising the benefits of tantalum products. When tantalum is properly applied, it can often be justified not only on a field replacement basis but also on initial installation.
5:66
TANTALUM
Corrosion Resistance Tantalum’s corrosion resistance is due to the presence of a thin continous surface film of tantalum pentoxide (Ta205).Thus the metal is passive and approaches the inertness of gold and platinum in a large number of very aggressive environments. The metal itself in the active state lies below zinc in the thermodynamic nobility table presented by Pourbaix’. In the passive state, its oxide film, however, puts it just below rhodium and above gold in the Pourbaix practical nobility table. The oxide film adheres well and appears to be free from porosity. At elevated temperatures a suboxide layer develops between the metal and the upper oxide film (Ta205)interface. This suboxide layer is not stable at temperatures higher than 425°C. When heated above this temperature only the stable pentoxide exists and the internal stress set up by the metal during oxide conversion causes the protective oxide film to flake and spall. Owing to this phenomenon, high temperature application of tantalum is limited in atmospheric environments under oxidative conditions. Available reports indicate that tantalum is an effective passive metal in most of the chemical environments, at ambient temperature and up to about 100°C. There are only a few environments in which tantalum corrodes in a rate higher than lmm/y, at temperatures up to about 100°C.
Fluorine and Fluoride Environments Tantalum is severely attacked at ambient temperatures and up to about 100°C in aqueous atmospheric environments in the presence of fluorine and hydrofluoric acids. Flourine, hydrofluoric acid and fluoride salt solutions represent typical aggressive environments in which tantalum corrodes at ambient temperatures. Under exposure to these environments the protective Ta,OS oxide film is attacked and the metal is transformed from a passive to an active state. The corrosion mechanism of tantalum in these environments is mainly based on dissolution reactions to give fluoro complexes. The composition depends markedly on the conditions. The existence of oxidizing agents such as sulphur trioxide or peroxides in aqueous fluoride environments enhance the corrosion rate of tantalum owing to rapid formation of oxofluoro complexes.
Hydrogen Embrittlement Tantalum has a high solubility for hydrogen, forming two internal hydrides, but the exact mechanism of their formation is not precisely known. There is evidence that embrittlement can occur at temperatures below 370°C. Clauss and Forestier6in fact reported that embrittlement can occur when tantalum is deformed in contact with hydrogen at room temperature. Examination of the literature indicates that one of the few defects in the resistance of tantalum to corrosion in aqueous media lies in its susceptibility to hydrogen embrittlement. Although it is inert in concentrated hydrochloric
TANTALUM
5:67
at temperatures as high as 11O"C, some reaction occurs at appreciably higher temperatures and sufficient hydrogen may be absorbed to cause embrittlement. Since it becomes cathodic in galvanic cell circuits with virtually all constructional metals, it must be electrically insulated from other metals with which it could come into contact in a common electrolyte, in order to prevent hydrogen discharge and entry into the metal. Anodising the tantalum, or addition of selected oxidising agents to the en~ironment'~ are proposed to reduce hydrogen embrittlement .
Reactions with Gases: Hydrogen, Nitrogen, Oxygen Tantalum and tantalum alloys react with hydrogen, nitrogen and oxygen at temperatures above 300°C. Hydrogen is dissolved in the metallic matrix The above 35OoC8and evolved at higher temperatures of about 800°C9*L0. dissolved hydrogen embrittles the tantalum and its alloys. This effect can be used to prepare tantalum powder. The reaction with small amounts of nitrogen results in an increased hardness, tensile strength and electrical resistivity. Tantalum is embrittled by higher amounts of nitrogen. The reaction takes place at temperature above 4OOoC8.Nitrides among other phases form at the surface, but at higher temperatures these decompose and all the nitrogen is liberated at 2100°C". Generally, the most important reaction is that of tantalum with oxygen, since it tends to form oxides when heated in air. Reaction starts above 300°C and becomes rapid above 600°C'9. The scale is not adherent, and if the oxidised material is heated above 1OOO"C oxygen will diffuse into the bulk of the material and embrittle it. At 1200°C catastrophic oxidation attack takes place at a rate of about 150 mm/hI3. Oxygen is not driven off by heating alone, but in vacuum above 2300°C it is removed as a suboxide. The first step of the conversion mechanism of tantalum into oxide was shown to occur by the nucleation and growth of small plates along the ( 1001 planes of the BCC meta121*22. The presence of a few atomic percent of oxygen in tantalum increases electrical resistivity, hardness, tensile strength, and modulus of elasticity, but decreases elongation and reduction of area, magnetic susceptibility, and corrosion resistance to HF". The main protective method against atmospheric catastrophic attack is surface coatings of silicides, and a l u m i n i d e ~ ~ ~ .
Atmospheric Conditions Tantalum has a high resistance to general outdoor atmospheres. Tantalum and the Ta-1OW alloy are virtually immune to sea water at ambient conditions and tantalum is only tarnished in oxygenated sea water at 26°C.
Acid Media Tantalum is practically inert to nitric acid at all concentrations and temperatures. The corrosion rate in 70% acid at 270°C is about 0.1 mm/y. It also
5:68
TANTALUM
Hydrochloric acid (%)
Fig. 5.10 Corrosion of tantalum by hydrochloric acid"
resists fuming nitric acids up to at least 150°C and hydrochloric acid at all concentrations up to 190°C though above 25% the corrosion rate rises rapidly and, in addition, the entry of hydrogen caused embrittlement (Fig. 5.10). Tantalum is completely inert to hydrochloric acid mixtures even in the presence of sulphuric acid and its salts in all proportions and concentrations up to boiling point. It is not corroded by phosphoric acid at concentrations up to 85070 and temperatures up to 200"C, provided flouride ions, often found in commercial acid, do not exceed 5 p.p.m. It is practically inert to perchloric acid, chromic acid, hypochlorous acid, hydrobromic acid, hydriodic acid and most organic acids provided they do not contain flourides, flourine or free sulphur trioxide. One exception to flouride attack appears to be in certain chromium plating baths in which fluoride is used as the catalyst, the corrosion rate in 40% CrO, plus 0.5% F at 55-60°C being 0-0005 mm/y. It is completely inert to 98% sulphuric acid to at least 160°C and to even higher temperatures at lower concentrations. Practically, it may be used to 200°C in all concentrations and to 225-250°C at concentrations between 80% and 90%. Fuming sulphuric acid containing sulphur trioxide attacks tantalum at room temperature as do hydrofluoric and fluorosilicic acids. Specific information is given in Figs. 5.10, 5.11 and Table 5.23.
Alkali Media Sodium hydroxide (NaOH) and potassium hydroxide (KOH) solutions do not dissolve tantalum, but tend to destroy the metal by formation of successive layers of surface scale. The rate of the destruction increases with concentration and temperature. Damage to tantalum equipment has been experienced unexpectedly when strong alkaline solutions are used during cleaning and maintenance.
5:69
TANTALUM
Sulphuric acld I%)
Fig. 5.11
Corrosion of tantalum by sulphuric acid I*
Table 5.23 Corrosion by miscellaneous acids 14* '" " Acid
Chromic
Concentration
PJO) All concentrations 10-50
Phosphoric (air free)
36.5 70 70 10-85 96 96
Temperature ("C)
Corrosion rate (mm/y)
100 Boiling 90 25 100 50-250 215-220 225-230
1.3 0.9-1 '25
l o w 9s - I ) can also cause disruption of the corrosion scale, leading to enhanced metal loss and corrosionassisted cracking of the substrate 14'. Laboratory exposures show parabolic
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
7143
rates with and without CI additions to the gas. There appears to be a direct effect of HCI on the scale integrity. Low CI (0.2% HCI) produces porous and discontinuous Fe,O,, whilst high CI (O.SCr0 HCI), causes complete disintegration of the Fe,O, and an irregular and porous magnetite scale. It was originally thought that the direct relationship between C1 content and corrosion rate was due to such a direct influence of HCI on scale integrity. It is now known, however, that HCI aids the release of potassium from the coal, which increases the overall corrosion rate .,41 Fluidised-bed systems produce higher combustion intensities at lower temperatures than combustion of pulverised fuel in conventional fossil-fuelfired boilers. The mineral matter for corrosion does not form fused salts and is not expected to release corrosive species. Fluidised bed combustors can, therefore, burn lower grade, cheaper fuel in smaller plant with better pollution control than traditional boilers .,41 Minchener et ~ 1 . ' - report that the bubble phase of atmospheric fluidised bed combustion has a p O , in the range 2 x l o - ' to 2 x IO-,. Combustion in the dense phase is sub-stoichiometric, with the PO, as low as lO-I3, and SO, and SO, present in the range 500-5 OOO ppm. Low Cr-Mo steels show heavy scaling in these conditions, whereas 9-129'0 Cr steels show good resistance to sulphidation up to 650°C. Roberts et al. 145, however, report that for pressurised fluidised-bed combustion, ferritic steels at or below 9% Cr show heavy general corrosion above 540-560OC. Chemical Environments
The oil industry frequently uses stainless steels or exotic bonded alloys for the processing of crude oil in the temperature range 200-600°C. These materials are very expensive and there is a strong economic incentive for finding cheaper alloys which are resistant to H,S and some gaseous organic sulphides arising from the S content of the crude oilla. Metal sulphides show the same type of predominant defects as metal oxides, Le. cations in Fe,, - x ) S , Cro +,,)S3. The defect concentration in most sulphides is much higher than those in the corresponding oxides, but the defect mobilities are only slightly higher. Thus the higher diffusivities and growth rates are determined by the higher defect concentrations. Cr and AI only slightly reduce the corrosion rate, and much higher AI is needed than that required for oxidation protection. Very protective scales are only formed at a S pressure lower than that for formation of the base-metal sulphide.'41 Mrowec et ~ 7 1 . ' ~ examined ' the resistance to high-temperature corrosion of Fe alloys with Cr contents between 0.35 and 74 at% Cr in 101 kPa S vapour. They found that the corrosion was parabolic, irrespective of the temperature or alloy composition, and noted that sulphidation takes place at a rate five orders of magnitude greater than oxidation at equivalent temperatures. At less than 2% Cr, the alloys formed Fe,, -$growing by outward diffusion of Fe ions, with traces of FeCr,S, near the metal core. Narita and N i ~ h i d a ' ,examined ~ the sulphidation of low Cr-Fe alloys at 700-900°C in 101 kPa of pure S. They found that the addition of small quantities of Cr significantly decreased the corrosion rate due to the formation of
7:44
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
FeCr2S4in the inner reaches of the scale. The scale comprised an outer FeS layer, with an inner layer of FeS, FeCr,S, and Cr,S, in varying amounts depending upon the Cr content. The corners of specimens corroded more rapidly than flat faces due to breakaway conditions. At low Cr contents the rate was increased, but above 4-6% Cr the parabolic rate constant decreased. Above 7.4% Cr an intermediate layer, containing FeCr,S, and varying amounts of Cr3S4, the proportion of the latter increasing with increasing Cr content, formed between the inner and outer layers. In view of this potentially rapid degradation of Cr-containing steels by high-temperature sulphidation in petrochemical and coal gasification reactors, AI is much used in the Fe alloys for these application^"^-^^^. A1 in Fe reduces the sulphidation rate in S, vapour by up to two orders of magnitude ' j 4 due to the high thermodynamic stability of aluminium sulphide relative to iron sulphide, the low rate of sulphidation of AI compared with pure Fe and the large PB ratio of Al,S, (3.7). The addition of 5% A1 in Fe in 101 kPa S, vapour between 500°C and 700°C resulted in the rate of reaction decreasing by a factor of ten. Paralinear kinetics were observed, with the inner layer of a duplex FeS scale containing a finely dispersed AI,S, phase which acts as a diffusion barrier to Fe2+ migration. Increasing the temperature to 800°C resulted in a rapid take off of the corrosion rate, with catastrophic corrosion rates above 800°C due t o the large volume of A12S, causing an increase in scale porosity. Condit et ul.''' examined the sulphidation of several Fe-Cr-A1 alloys under a variety of sulphidising conditions. They noted that, in the early stages of sulphidation, a thin compact inner layer forms which is high in Cr and AI. Subsequently, a thicker microcrystalline outer layer forms with a uniform Fe, Cr and A1 composition. Formation of the outer compact layer was favoured by increasing pS2 and decreasing temperature, with the layer forming much more rapidly in H,S than in pure S,. The sublayer disappeared more or less rapidly dependent upon alloy composition. The authors propose three stages for scale development. First a thin compact layer forms due to the penetration of S into the alloy with preferential formation of sulphide from those metals with the highest affinity for S. Fe,, - $ also forms due to the abundance of Fe in the alloy. The outer layer then dissociates to release sulphur which dissolves in grain boundaries of the alloy to form Cr and AI sulphides. The Fe released by this dissociation sulphidises again at the interface between the two layers. The volume increase associated with the conversion of metal to sulphide generates mechanical stress which causes the outer layer to break up and permits permeation of S. This initiates a second stage, where growth of the scale is linear and comprises a porous outer layer, with FeS, Cr,S, and AI,S, evenly distributed, possibly as FeCr2S4, FeAI,S, and FeCr,,,Al,, -x,S4. S then diffuses through the pores to the scale/metal interface. The third stage comprises the formation of an outer compact layer of Fe,, - $and continued thickening of the inner layer. Addition of Cr to Fe-A1 alloys aids the formation of Cr sulphides and AI,S, which together markedly reduce the sulphidation rate146.Between 2% and 5 % C r then, more than 3 % A l is required to obtain protection. At 9% Cr, however, only 1To A1 is needed to give protection since the Cr is sufficiently active to lower the S potential seen by AI (secondary gettering). Thus
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
7:45
all Fe-Cr-AI alloys initially form FeS, Cr sulphides and A12S3.The Cr and AI are then exposed to a much lower pS2at the scale/metal interface and Cr sulphides and AI,S, grow preferentially if their activity is high enough. Ultimately, a protective layer of AI,S, or AI,S, Cr sulphides develops at the scale/metal interface and the reaction rate decreases substantially. Karlson et found, from on-site experience of cement-producing plant, that corrosion of Fe surfaces may occur in gases containing O,, SO, and alkali chlorides such as NaCl and KCI between 300°C and 500°C. They reported that the corrosion rates may be extraordinarily high (5-10 mm/ month) implying liquid-phase corrosion. Laboratory simulation of the plant conditions demonstrated the need for both SO, and the alkali chloride in the environment. The principle corrosion reaction was found to be:
+
2Fe,03
+ 12[K, Na]CI(s, I) + 12S02(g) + 9 0 2 ( g )
-+
4[K, Na13Fe(S04)3(s,1) + 6C1,(g) A thermodynamic evaluation of this equation indicated that the reaction could proceed with SO2 levels as low as 100ppm.
co/co, Failures of mild steel components in Magnox reactors in the UK and Italy after approximately 5 years of operation alerted the world to the potential for breakaway oxidation of low-alloy steels in C O / C O , environments". The CO,, 1Vo CO, 300 vppm CH,, 250 vppm H,O, 100 vppm H, environment used in CAGRs was selected on the need to minimise oxidation of the graphite reactor core and deposition of C from the coolant gas157.Corrosion rate tests of 15 000-20000 h, in the limited range of conditions anticipated by the designers, showed that the maximum reduction in corrosion rate of ferritic steels in CO, at 600°C is realised at around the 9% Cr level'58-'@'. Therefore 9CrlMo steel was chosen for the evaporator and primary superheater sections of the CAGR"'. However, in the late 1960s, Taylor (reported in Reference 158) identified evidence for a significant change in the corrosion mechanism for 9% Cr steels at around 550°C. This change could lead to rate inversion with increasing temperature in steels containing 0.7-0.8% Si, or breakaway in 0.4-0.5% Si steels. Because of their importance to the nuclear power generation industry, these observations initiated a vast amount of research into the oxidation of low-alloy steels in CO/CO, environments. It is now clear that low-alloy steels exhibit three types of behaviour when exposed to CO/C02, i.e. protective, transitional and linear-breakaway (Fig. 7.14), with the time to breakaway and the breakaway rate being of crucial importance in determining component life. For mild and low-alloy steels in CO, the first scale to form is a compact coarse columnar layer of Fe,04'6'-'63.Growth of this layer is controlled by outward grain boundary diffusion of Fe ions 17,1639161. The inward countercurrent of vacancies is initially annihilated at the metal surface, but eventually vacancy condensation at the scale/metal interface gives decohesion, the scale develops microporosity, an inner layer grows within the space Then created by the departing metal ions and a duplex scale forms'63-165.
7:46
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
I
/
/
Breakaway
w = at
Protective
w
=
atb
Time Fig. 7.14 Schematic diagram of stages of low-alloy steel oxidation in CO/COI
for Cr-containing steels, the protective scale comprises an outer, coarse columnar-grained magnetite layer and an inner, slightly porous, Cr-rich spinel of fine (0.1 pm) equiaxed grains6'* . As with low-alloy steels in other environments, the Cr is not mobile in the scale, but is oxidised in situ 163.166 . Studies have shown that the M,O, spinel nucleates at asperities on the surface and duplex growth is also known to be favoured in the vicinity of inclusions and specimen corners 163. Once initiated, the inner layer grows by inward diffusion of 0, probably as COz, down microfissures and micropores 113*167, in both the lateral and vertical directions, until a complete layer is obtained la. The growth of this layer subsequently follows the parabolic rate law. During this protective stage, decreasing the water and CO content of the gas appears to decrease the rate constant of Cr-containing steels157but has little effect on the rate constant of carbon steels 161*168.The rate constant of all steels has been found to decrease with decreasing temperature and increasing Si content IO3- 157*16'* 1699I7O. Ferguson et ai.168 have also reported that increasing the S content of carbon steels reduces the parabolic rate constant. An increase in the Si content of the steel has been reported to give a significant increase in the duration of the protective regime for carbon 169. A similar benefit has also been reported for Cr-containing steel steels 157. 158, 163. 171 . Increases in the duration of the protective regime are also realised with reductions in the CO and H 2 0 contents of the gas168,in the temperatures 158*161*168and, for carbon steels, in the surface roughness or, for 9Cr steels, with increased surface cold work.'51 Robertson and Manning43found that breakaway may also be delayed by some S-containing gases. Following the initial protective period, under certain conditions of temperature, alloy and gas composition the oxidation goes through a transitional stage into breakaway. Several authors have reported that breakaway oxidation is initiated once the scale reaches a critical thickness 157* or weight gain6' and only occurs below an initially protective duplex 1 7 3
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
7:41
Both single-layer protective and duplex-layer protective scales have a PBR of 2.1, whereas breakaway scales show a PBR of approximately 2.7, containing around 23% porosity'63,and graphitic carbon grains (up to 6% by weight) between the oxide grains 16'. During this transitional period, and once full breakaway is established, growth occurs throughout the oxide and close to the scale/metal interface I". Growth below the oxidant surface requires oxidant transport through the scale, but solid-state lattice diffusion is much too slow to account for the rates observedt7.Therefore, a porous scale is required. Mechanisms for pore growth have been postulated by several authors 165*173-176 who have suggested that the transition from single-layer to duplex growth is due to the initiation of pores within the scale. Then breakaway was thought to be caused by increased porosity giving unlimited access of oxidant to metal surface. However Atkinson and Smart47have shown single layer scales to be slightly porous. Robertson and Manning43 have proposed that oxidant access is always available and the type of scale which develops is dependent upon conditions at the scale/metal interface. Breakaway occurs if space is created by continuous scale deformation or creep 17'. The inward penetration of CO, via cracks and micropores gives: '"9
3Fe
+ 4C0, = Fe,O, + 4CO 2CO = 0, + 2C via the Boudouard reaction
at the scale/metal The carbon initially diffuses into the steel but ultimately, the steel may become saturated with C'77.There is no detectable solubility of C in FeO, Fe,O,, MnO or Cr,03178,and C can only permeate through pores or faulty scalesw. Thus, the C then deposits at the scale/metal interface and prevents the formation of a coherent protective layer once it reaches a critical activityt7'. Carbides in the steel have been found to be the preferred sites for breakaway. These are either pre-existing carbides or carbides precipitated by C injection during oxidation 163. Precarburised or graphite-painted steel breaks away rapidly, as do thin foils, due to the smaller C sink a ~ a i l a b l e ' ~ ~ . Pritchard et ai.', reported that the proportion of C in the scale increased with scale thickness and water content of the gas, and was higher in breakaway oxide. For 9% Cr steel, breakaway oxidation is associated with heavy carburisation of the metal and C deposition within the oxide, with preferential breakaway occurring at corners and edges For Fe in CO/CO, at atmospheric pressure, Surman'79found that if 1 0 - ~e
pcopco + pco, e 0.3
then the oxidation is parabolic and very little C deposition occurs. He concluded that magnetite is not sufficiently catalytic to promote the Boudouard reaction unless CO > 10% and moisture is present and surmised that H,O promotes the formation of a Boudouard catalyst. If the CO is greater than 0.4 in the above expression, oxidation and carbon deposition occur simultaneously at a linear rate German and Littlejohnt6'have observed that increasing the Si content of carbon steel reduces the linear rate constant during breakaway and Banks
7:48
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
and Lorimer17' have shown that Cr and Si reduce the creep rate of Fe304, thus reducing the post-breakaway rate of Cr-containing steels. However, Si has been reported as having no effect on the post-breakaway rate on 9Cr s t e e l ~ ' ~ ~Small " ~ ' . changes to the C O and H,O content of the environment and the temperature have no significant effect on the breakaway rate constant, but a large reduction in the CO can cause reversion to protective kineticslS0,as does reducing the CO, pressure to atmospheric ''I. Increasing the CO switches breakaway on again'". L. W. PINDER REFERENCES 1. Holmes, D. R. and Stringer, J. in Corrosion of Steels in CO,, Ed. Holmes, D. R., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 165 (1974) 2. Pilling, N. B. and Bedworth, R. E., J. Inst. Met., 29, 529 (1923) 3. Tammann, G., Z. Anorg. Allgem. Chem., 111, 78 (1920) 4. Gulbransen, E. A. and Ruka, R., Trans. AIME, 188, 1500 (1950) 5. Pinder, L. W., C.E.G.B. Unclassified Report, MID/SSD/80/0050/R, August 1980 6. Howe, C. I., McEnany, B. and Scott, V. D., Corr. Sci., 25 No. 3, 195 (1985) 7. Goursat, A. G. and Smeltzer, W. W., Oxid. Met., 6 No. 2, 101 (1973) 8. Kuroda, K., Labun, P. A., Welsh, G. and Mitchell, T. E., Oxid. Met., 19 Nos 3/4, 117 (1983) 9. Koch, F. and Cohen, J. B., Acta Crystallography, B 25, 275 (1969) 10. Cheetham, A. K., Fender, B. E. G. and Taylor, R. I., J . Phys. C . , V , 2160 (1971) 1 I . Catlow, C . R. A., Mackrodt, W. C., Norgett, M. J. and Stoneham, A. M., Phil Mag A , 40 No. 2, 161 (1979) 12. Chen, W. K. and Peterson, N. L., J. Phys. Chem. Solids, 36, 1097 (1975) 13. Wagner C., Atom Movements, ASM, Cleveland, 153 (1951) 14. Wagner, C., Z. Phys. Chem., B621, 25 (1933) 15. Atkinson, A., Rev. Mod. Phys., 57, 437 (1985) 16. Atkinson, A. and Taylor, R. I., J. f h y s . Chem. Solids, 46, 469 (1985) 17. Atkinson, A. and Taylor, R. I., High Temperature - High Pressure, 14, 571 (1982) 18. Dieckmann, R. and Kohne, M., Eer. Eunsenges Phys. Chem., 87, 495 (1983) 19. Garnaud, G. and Rapp, R. A., Oxid. Met., 11, 193 (1977) 20. Channing, D. A. and Graham, M. J., Corr. Sci., 12,271 (1972) 21. Channing, D. A. Dickerson, S. M. and Graham, M. J., Corr. Sci.. 13, 933 (1973) 22. Francis, R. and Lees, D. G., Corr. Sci., 16, 847 (1976) 23. Rahmel, A,, Werkstofle und Korrosion, 16 No. 10, 837 (1965) 24. Rahmel, A., Korrosion, 18, 41 (1966) 25. Eubanks, K.G . , Moore, D. G . and Pennington, W. A., J. Electrochem. Soc., 109, 382 ( 1962) 26. Svedung, I., Hammar, B., and Vannerberg, N. G., Oxid. Met., 6 No. I, 21 (1973) 27. Caplan, D., Corr. Sci., 6, 509 (1966) 28. von Fraunhofer, J. A. and Pickup, G. A., Corr. Sci., 10, 253 (1970) 29. Janssen, S. and Lehtinen, B., Metallurgie, 7 , 61 (1967) 30. Caplan, D. and Cohen, M., Corr. Sei., 6, 321 (1966) 31. Price, W. R., Corr. Sci., 7, 473 (1967) 32. Pinder, L. W., C.E.G.B. Unclassified Report, MID/SSD/80/0057/R, August 1980 33. Stott, F. H., Mat. Sci. Tech., 5 No. 8, 734 (1989) 34. Atkinson, A., Mat. Sci. Tech., 4 No. 12, 1046 (1988) 35. Dunitz, J. D. and Orgel, L. E., J. Phys. Chem. Solids, 3 , 318 (1957) 36. Azaroff, L. V., J. Appl. Phys., 32 Part 9, 1658 (1969) 37. Cox, M. G. C., MacEnany, B. and Scott, V. D., Phil. Mag., 26, 839 (1972) 38. Hodge, J. D., J. Electrochem. SOC., 125 No 2, 55c (1978) 39. Rahmel, A. Z., Electrochem., 66 No 4, 363 (1962) 40. Moreau, J., Compte Rendu, 236, 85 (1953) 41. Surman, P . L. and Castle, J . E., Corr. Sci., 9, 771 (1969) 42. Atkinson, A. in Oxidation of Metals and Associated Mass Transport, Ed. Dayananda, M. A. et a / . Warrendale, P.,The Metallurgical Society of the AIME, 29 (1987)
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
7 :49
Robertson, J. and Manning, M. I., Mat. Sci. Tech., 4, 1064 (1988) Bruckman, A., Emmerich, R. and Mrowec, S., Oxid. M e t . , 5 No. 2, 137 (1972) Fromhold Jr., A. and Sato, N., Oxid. Met., 16 Nos. 3/4, 203 (1981) Harrison, P. L., Oxid. Met., 22 Nos. 1/2, 35 (1984) Atkinson A. Phil Mag. B, 55, 637 (1987) Whittle, D. P., Evans, D. J., Scully, D. B. and Wood, G. C., A c t a M e f . , 15, 1421 1967) Wagner, C., Corr. Sci., 5, 751 (1965) Runk, R. B. and Kim, H. J., Oxid. Met., 2 No. 3, 285 (1970) Kim, H. J. and Runk, R. B., Oxid. Met., 2 No. 3, 307 (1970) Nosek, E. and Werber, T., Oxid. Met., 25 Nos. 314, 121 (1986) Bohnenkamp, V. J. and Engell, H. J., Arch. Eisenhuettenw., 33, 359 (1962) Caplan, D., Sproule, G. I., Hussey, R. J. and Graham, M. J., Oxid. M e f . , 13, 255 1979) Caplan, D., Sproule, G. I., Hussey, R. J. and Graham, M. J., Oxid. Met., 12, 67 1978) Malik, A. U., Oxid. Met., 25 Nos. 516, 233 (1985) Tomaszewicz, P. and Wallwork, G. R., Oxid. Met., 19 Nos. 5/6, 165 (1983) Tomaszewicz, P. and Wallwork, G. R., in High TemperatureCorrosion, Ed. Rapp, R. A., NACE, Houston (1983) 59. Boggs, W. E., J . Elecfrochem. Soc., 118, 906 (1971) 60. Ahmed, H. A. and Smeltzer, W. W., J . Electrochem. SOC., 133, 212 (1986) 61. Pons, M., Caillet, N. and Galerie, A., Cow. Sci., 22, 239 (1982) 62. Smith, P. J., Beauprie, R. M., Smeltzer, W. W., Stevanovic, D. V. and Thompson, D. A., Oxid. Met. 28 Nos. 5 / 6 , 259 (1987) 63. Ahmed, H. A., Underhill, R. P., Smeltzer, W. W., Brett, M. E. and Graham, M. J., Oxid. Met., 28 Nos. 5/6, 347 (1987) 64. Tomaszewicz, P. and Wallwork, G. R., Oxid. Met., 19 Nos. 3/4, 75 (1983) 65. Wagner, J. B., in Defectsand Transport in Oxides, Ed. Smeltzer, M. S. and Jaffe, R. I.. Plenum Press, New York, 283 (1974) 66. Seybolt, A. U., Trans. AIME, 242, 752 (1%8) 67. Rahmel, A. and Tobolski, J., Werksfofleund Korrosion, 16 No. 8, 662 (1965) 68. Robertson, J. and Manning, M. I., Mat. Sci. Tech., 5, 741 (1989) 69. Darken, L. S., Trans. AIME, 150, 157 (1942) 70. Tuck, C. W., Corr. Sci., 5, 631 (1965) 71. Svedung, I. and Vannenberg, N. G., Corr. Sci., 14, 391 (1974) 72. Wood, G. C., Richardson, J. A., Hobby, M. G. and Banstead, J., Corr. Sci., 11, 659 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.
(1971) 73. Rochet, F., Rigo, S., Froment, M., d'Anterroches, C., Maillot, C., Roulet, H. and Dufour, G., Phil. Mag. B, 55, 309 (1987) 74. Adachi, T and Meier, G. H., Oxid. Met., 21 Nos. 5/6, 347 (1987) 75. Atkinson, A., Corr. Sci., 22, 87 (1982) 76. Logani, R. C. and Smeltzer, W. W., Oxid. Met., 3 No. 3, 279 (1971) 77. Logani, R. C. and Smeltzer, W. W., Oxid. Met., 1 No. 3, 3 (1969) 78. Logani, R. C. and Smeltzer, W. W., Oxid. Met., 3 No. 1, 15 (1971) 79. Jackson, P. R. S. and Wallwork, G. R., Oxid. M e r . , 20 Nos. 1/2, 1 (1983) 80. Donati, and Garaud, Corrosion of Steels in CO,, Ed. Holmes, D. R., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 28 (1974) 81. Dewanckel, B., Leclercq, D. and Dixmier, J., Ibid., 42 82. McAdam, G. and Young, D. J., Oxid. Met., 28 Nos. 3/4, 165 (1987) 83. Nishida, K., Narita, T., Tani, T. and Sasaki, G., Oxid. Met., 14, 65 (1980) 84. Pritchard, A. M., Antill, J. E., Cottell, K. R. J., Peakall, K. A. and Truswell, A. E., in Corrosion of Steels in COz. Ed. Holmes, D. R., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 73 (1974) 85. Menzies, L. A. and Tomlinson, W. J., JISI., 204, 1239 (1958) 86. Menzies, I. A. and Lubkiewicz, J., Oxid. M e t . , 3 No. I , 41 (1971) 87. Dalvi, A. D. and Coates, D. E., Oxid. Met., 5 No. 2, 113, 135 (1972) 88. Yearian, H. J., Randell, E. C. and Longo, T. A., Corrosion, 12, 515 (1956) 89. Douglas, D. L., Gesmundo, F. and de Asmundis, C., Oxid. Met., 25 Nos. 3/4, 235 (1986) 90. Khanna, A. S. and Gnanamoorthy, J. B., Oxid. Met., 23 Nos. 112, 17 (1985) 91. Hossain, M. K., Corr. Sci., 19, 1031 (1979) 92. Caplan, D. and Sproule, G., Oxid. Met., 9, 459 (1975) 93. Rahmel, A,, Jaeger, W. and Becker, K., Arch. Eissenhuttenw., 30, 351 (1959) 94. Wolfe, I., Grabke, H. J . and Schmidt, P., Oxid. Met. 29 Nos. 3/4, 289 (1988)
7:50 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106.
107. 108. 109.
110. I 1I. 112.
113. 114. 115.
116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139.
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
Baxter, D. J . and Natesan, K., Reo. High Temp. Materials, 314, 149 (1983) Schutze, M., Mat. Sci. Tech., 4, 407 (1988) Stringer, J . , Corr. Sci.. 10, 513 (1970) Evans, H. E. and Lobb, R. C., Corr. Sci., 24 No. 3, 209 (1984) Kubaschewski, 0. and Hopkins, B., Oxidation of Metals and Alloys, Butterworth, London (1969) Manning, M. 1. and Metcalfe, E., Proc. Sixth European Congress on Metallic Corrosion, London, 121 (1977) Barbehon, J., Rahmel, A. and Schutze, M., Oxid. Met., 30 Nos. 1/2, 85 (1988) Christl, W., Rahmel, A. and Schutze, M., Oxid. Met., 31 Nos. 1/2, 1 (1989) Whittle, D. P., Oxid. Met., 4 No. 3, 171 (1972) Deadmore, D. L. and Lowel, C . E., Oxid. Mer., I1 No. 2, 91 (1977) Hsueh, C. H. and Evans, A. G., J. Appl. Phys., 54, 6672 (1983) Douglass, D. L., Oxidation of Mefals and Alloys, American Society of Metals, Metals Park, Ohio, 137 (1971) Mitchell, T. E., Voss, D. A. and Butler, E. P.. J . Mat. Sci., 17, 1825 (1982) Manning, M. I., Corr. Sci., 21, 301 (1981) Evans, A. G., Crumley, G. B. and Demaray, R. E., Oxid. Met., 20 Nos. 5/6, 193 (1983) Norin, A., Oxid. Met., 9 No. 3, 259 (1975) Appleby, W. K. and Tylecoate, R. F., Corr. Sci., 10, 325 (1970) Jha, B. B., Raj, B. and Khanna, A. S., Oxid. Met., 26 Nos. 3/4, 213 (1986) Hancock, P. and Hurst, R. C., in Corrosion of Steels in COz, Ed. Holmes, D. R., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 320 (1974) Ward, G . , Hockenhull, B. S. and Hancock, P., Mer. Trans., 5, 1451 (1974) Forrest, J . E. and Bell, P. S., Corrosion and Mechanical Stress at High Temperatures, Applied Science Publishers, London, 339 (1981) Rolls, R. and Nematollahi, M., Oxid. Mer., 20 Nos. 1/2, 19 (1983) Simms, N. J. and Little, J . A., Mat. Sci. Tech., 4, 1133 (1988) Pinder, L. W., C.E.G.B. Unclassified Report SSD/MID/R58/77, November 1977. Wiles, C., PowerGen, private communication. Cory, N. J . and Herrington, T. M., Oxid. Met., 28 Nos. 5/6, 237 (1987) Cory, N. J. and Herrington, T. M., Oxid. Met., 29 Nos. 1/2, 135 (1988) Hauffe, K., Oxid. Mer., 285 (1965) Effertz, P. H. and Miesel, H., Machinenshaden, 55, 14 (1971) Potter, E. C. and Mann, G. M. W, Proc. NACE 2nd International Congress on Metallic Corrosion, New York, 872 and 878 (1963) Hurst, P. and Cowen, H. C., Proc. Conf. Ferritic Steels f o r Fast Reacfor Steam Generators, British Nuclear Energy Society, London, (1977) Mayer, P. and Manolescu, A. V.,High TemperatureCorrosion, Ed. Rapp, R. A,, NACE, Houston, Texas, 368 (1983) Rahmel, A. and Tobolski, J., Corr. Sci., 5, 333 (1965) Kofstad, P., Oxid. Met., 24 Nos. 5/6, 265 (1985) Bruckman, A. and Mrowec, S., Corr. Sci., 7 , 173 (1973) Sheasby, J . S., Boggs, W. E. and Turkdogan, E. T., Met. Sci., 18, 127 (1984) Tuck, C. W., Odgers, M. and Sachs, K., Corr. Sci., 9, 271 (1969) Griskin, A.M., Perkov, V. G., Sentyurev, V. P. and Yaschenko, Ya Y., Thermal Engineering, 16, I2 1 ( 1969) Armitt, J . , Holmes, D. R., Manning, M. I. and Meadowcroft, D. B., The Spalling of Steam Grown Oxides from Superheater and Reheater Tube Steels, EPRI-FP-686,TPS 76-655 Final Report (February 1978). Lux, J. A., American Power Conference, Chicago, Illinois, 29 April to May 1 (1974) Clarke, F. and Morris, C. W., in Corrosion Resistant Materials f o r Coal Combustion Systems, Ed. Meadowcroft, D. B. and Manning, M. I., Applied Science Publishers, London, 47 (1983) Lees, D. J . and Whitehead, M. E., /bid., 63 Latham, E. P., Meadowcroft, D. B. and Pinder, L. W., CRSC-EPRI lnt. Conf. on Chlorine in Coal, Chicago, October, 1989 (proceedings to be published) Gibb, W. H., in Corrosion Resistant Materials f o r Coal Combustion Systems, Ed. Meadowcroft, D. B. and Manning, M. I., Applied Science Publishers, 25 (1983) Brooks, S. and Meadowcroft, D. B., [bid., 105
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
7:51
140. Laxton, J. W., Meadowcroft, D. B., Clarke, F., Flatley, T., King, C. W. and Morris, C. W., C.E.G.B. private communication. 141. Mayer, P. and Manolescu, A. V., in Corrosion Resistant Materials for Coal Combustion Systems, Ed. Meadowcroft, D. B. and Manning, M. I., Applied Science Publishers, London, 87 (1983) 142. Cutler, A. J. B. and Reask, E., Corr. Sci., 21, 789 (1981) 143. Perkins, R. A., in Corrosion Resistant Materials for Coal Combustion Systems, Ed. Meadowcroft, D. B. and Manning, M. I., Applied Science Publishers, London, 219( 1983) 144. Minchener, A. J., Lloyd, D. M. and Stringer, J., Ibid., 299 145. Roberts, A. G., Raven, P., Lane, G. and Stringer, J., Ibid., 323 146. Zelanko, P. D. and Simkovich, G., Oxid. Met., 8 No. 5, 343 (1974) 147. Mrowec, S. and Przybylski, K., Oxid. Met., 23 Nos. 3/4, 107 (1985) 148. Mrowec, S., Walec, T. and Weber, T., Oxid. Mer., 1 No. 1, 93 (1969) 149. Narita, T. and Nishida, K., Oxid. Met., 6 No. 3, 181 (1973) 150. Sutherland, R. B. and Prescott. G. R., Corrosion, 18, 277t (1961) 151. Backensto, E. B., Prior, J. E., Sjooberg, J. W. and Manuel, R. W., Corrosion, 18, 253t (1962) 152. Malinowski, E., Metal, 94 No. 4, (1962) 153. Burns, F. J., Corrosion, 25, 119 (1969) 154. Strafford, K. N. and Manifold, R., Oxid. Met., 1, 229 (1969) 155. Condit, R. H., Hobbins, R. R. and Birchenall, C. E., Oxid. Met., 8 No. 6, 409 (1974) 156. Karlsson, A., Moller, P. J. and Johansen, V., Corr. Sci., 30, 153 (1990) 157. Rowlands, P. C., Garrett, J. C. P., Hicks, F. G., Lister, S. K., Lloyd, B. and Twelves, J. A., in Corrosion of Steels in CO,, Ed. Holmes, D. R., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 247 (1974) 158. Holmes, D. R., Mortimer, D. and Newell, J., Ibid., 151 159. Newell, J. E., Nucl. Energy I n t . , 17, 637 (1972) 160. Taylor, J. W. and Trotsenberg, P. V., in Corrosion of Steels in CO,, Ed. Holmes, D. R:, Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 180 (1974) 161. German, P. A. and Littlejohn, A. C., Ibid., 1 162. Hussey, R. J., Sproule, G. I., Caplan, D. and Graham, M. J., Oxid. Mer., 11, 65 ( 1977) 163. Gibbs, G. B., Pendlebury, R. E. and Wooton, M. R., in Corrosion of Steels in CO,, Ed. Holmes, D. R., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 59 (1974) 164. Cox, M. G. C., McEnaney, B. and Scott, V. D., Ibid., 247 165. Gibbs, G. B., Oxid. Mer., 7, 173 (1973) 166. Harrison, P . L., Dooley, R. B., Lister, S. K., Meadowcroft, D. B., Nolan, P. J., Pendlebury, R. E., Surman, P. L. and Wooton, M. R., in Corrosion of Steels in CO,, Ed. Holmes, R. D., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 220 (1974) 167. Harrison, P. L., Dooley, P. B., Lister, S. K., Meadowcroft, D. B., Nolan, P. J., Pendlebury, R. E., Surman, P. L. and Wooton, M. R., Ibid., 220 168. Ferguson, J. M., Garrett, J. C. P . and Lloyd, B., Ibid., 15 169. Rowlands, P . C., Garrett, J. C. P., Popple, L. A,, Whittaker, A. and Hoaksey, A., Nucl. Energy, 25 No. 5, 267 (1986) 170. Camona, G. A., Imbergamo, M. and Ronchetti, C., in Corrosion of Steels in CO,, ed. Holmes, D. R., Hill, R. B. and Wyatt, L. M., British Nuclear Energy Society, London, 45 (1974) 171. Grandison, N. 0. and Facer, R. I . , Ibid., 208 172. Gleave, C., Calvert, J. M., Lees, D. G. and Rowlands, P . C., Proc. Roy. SOC. [ A 1.379, 409 (1982) 173. Mrowec, S., Corr. Sci., 7, 563 (1967) 174. Gibbs, G. B. and Hales, R., Corr. Sci., 17, 487 (1977) 175. Evans, A. G., Rajdev, D. and Douglass, D. L., Oxid. Met., 4, 151 (1972) 176. Kofstad, P., Oxid. Met., 24, 265 (1985) 177. Banks, P. and Lorimer, G. W., inMaterialsroSupply IheEnergy Demand, Ed. Hawbolt, E. B. and Mitchell, A., American Instituteof Mechanical Engineers, New York, 231 (1982) 178. Wolfe, I. and Grabke, H. J., Solid State Comm., 54, 5 (1985) 179. Surman, P. L., Corr Sci., 13, 825 (1973)
7:52
THE OXIDATION RESISTANCE OF LOW-ALLOY STEELS
180. Surman, P. L. and Brown, A . M., in Corrosion oJSteels in C 0 2 , Ed. Holmes, D. R . , Hill, R . B. and Wyatt, L. M., British Nuclear Energy Society, London, 85 (1974) 181. Goodison, D., Harris, R . J . and Goldenbaum, P., British Joint Corrosion Group Symposium on Metal-Gas Reactions in Atmospheres Containing CO,, London, March, ( 1967) 182. White, J. CEGB Private Communication
7.3 High-temperature Corrosion of Cast Iron
Introduction When cast iron is exposed to high temperatures under oxidising conditions, oxidation of the metal results, with the formation of a surface scale. In addition, the dimensions of the component become distorted. Although such dimensional changes can occur also in inert atmospheres or in vacuum, the evidence available suggests that this ‘growth’ is frequently associated with oxidation, and accordingly it is appropriate to consider it as an aspect of the corrosion of the iron. The composition of the atmosphere to which components at high temperature may be exposed varies very widely, and most work on these aspects has accordingly been carried out in clean air. The aggressiveness of air is considerably enhanced by the presence of trace amounts of other reactive gases such as steam, carbon dioxide and sulphur dioxide; thus the figures subsequently quoted may in fact be appreciably lower than those encountered in specific atmospheres. The data presented should, however, prove an adequate guide to the order of the effect to be expected.
Growth Components designed for high-temperature duty may either remain at a steady high temperature for their entire life, or, as more commonly happens, may undergo cyclic variation between a minimum temperature, often room temperature, and a maximum temperature. The maximum temperature involved may be either above or below the critical temperature range of the iron. This is the range within which the transformation between ferrite or pearlite and austenite occurs and for the majority of unalloyed irons it may be regarded as being 700-850°C. (See Section 20.4.) Conditions of cyclic reheating are more severe than conditions of steady high temperature, and cyclic reheating through the critical range is particularly liable to cause excessive growth of the iron. Generally, the studies which have been carried out suggest that growth of up to 40% by volume can occur within the first 40 h of cyclic reheating to 900°C with a frequency of 1-4 h/cycle, while subsequent cycling produces 7:53
7:54
HIGH-TEMPERATURE CORROSION OF CAST IRON
growth at a rate rarely exceeding 20% increase in volume in 100 h. The rate of growth which develops increases with increase in temperature and possibly also with increase in frequency of cycling. Although the fact that a 60% increase in volume may occur after only 140 h of cyclic reheating suggests that unalloyed iron is totally unsuited for such applications, iron is in fact extensively used under such conditions, e.g. as furnace doors and fire bars. This may be partly because these applications involve lower cycling frequencies than those which cause the very high rates of growth mentioned, but undoubtedly a major factor determining the use of unalloyed cast iron for such duties is its cheapness, which outweighs the superior growth resistance of more highly alloyed and more expensive irons. At temperatures below the critical range, much less growth occurs, rarely exceeding 3vo for 100 h of cyclic reheating. Here too the rate of growth depends on the temperature and the frequency of cycling. At temperatures below 400°C growth becomes negligible for most irons while below 350°C it is negligible for all irons. This threshold is probably related to the marked decrease in strength which occurs when irons are heated above 400"C, which results in the component being more easily distorted by the development of internal volume changes. Clearly, unalloyed irons have a very considerable usefulness up to about 700"C, and even in steam plant, where dimensional stability is important, there is a case for the use of unalloyed iron at temperatures up to 400°C. At temperatures below the critical range, an important cause of growth is graphitisation, Le. the decomposition of the carbide constituent of pearlite to give ferrite and graphite. Unalloyed irons usually contain up to 0.8% combined carbon and complete graphitisation of this can theoretically result in a volume increase of 1 6% '. This value has been confirmed by Gilbert and White' who have shown that ferritising a fully pearlitic iron gives a linear growth of up to 0.7% (Le. about 2.1070 volume increase). Clearly, the rate of growth due to this mechanism will be controlled by the stability of the carbide in the pearlite and this will vary with the composition of the iron. The presence of certain elements, notably silicon, decreases the stability of the carbide, while it is stabilised by the presence of other elements, notably chromium. An iron with a low silicon content and containing some chromium may thus be expected to have good growth resistance, but since excessive carbide stability can lead to a hard, brittle alloy, there is a limit to the benefit which can be derived from such stability. It should be emphasised that unless large amounts of carbide-stabilising elements are present in the iron, all that will be achieved is a slower rate of growth; there will not be a decrease in the total growth possible. Phosphorus appears to have a beneficial effect on the growth rate. At sub-critical temperatures it helps to stabilise the carbide, while at temperatures up to about 900°C the presence of the hard phosphide eutectic network restricts the deformation to which the much more ductile matrix would otherwise be subject. Since the phosphide eutectic melts at about 950°C irons containing appreciable amounts of this constituent should clearly not be exposed to this temperature. Another cause of growth which is of equal importance with graphitisation is the penetration of oxides into the metal along the graphite flakes. This presumably takes place because oxidising gases can be adsorbed on to the
-
7:55
HIGH-TEMPERATURE CORROSION OF CAST IRON
graphite and so allowed access to the metal/graphite interface. Since the oxides are more bulky than the metal from which they are derived, internal stresses are set up and growth results. As might be expected, the amount of growth due to internal oxidation increases as the graphite content increases (Fig. 7.15) and also as the section size increases, since this leads to a coarsening of the graphite. On the other hand, a white iron which contains no graphite is very growth-resistant since it does not readily graphitise, nor is it easily penetrated by oxidation. For similar reasons, nodular graphite irons are resistant to growth.
b
C
0
10
20
30
40
50
60
1
I
Time (weeks 1
Fig. 7.15 Effect of amount and distribution of graphite on growth in air at 500°C. Curve a: Iron 6 (Fe-3.61CIoI.~-1.63Si-0.76Mn-0.094S-0.28P). Curve b: Iron 1 (Fe-3~25CIoIaI1.58Si-0.65Mn-0.107S-0~25P). Curve c: Iron 21 (Fe-3.39CloI,~-1.73Si-0~41Mn-O-O13S0~05P-0~67Ni-O~O7SMg-0~004Ce) (after Gilbert4)
At temperatures above the critical range, the maximum amount of growth due to graphitisation may account for less than 10% of the total growth observed. Undoubtedly a large contribution to the total growth is made by the oxidation of the iron, since the stresses set up in the oxide layers by the differences between the expansion of the oxides and the iron during the alternate heating and cooling cycles generate cracks in the scale, which prevent the reaction from ever becoming self-stifling. The increase in oxidation rate due to the temperature rise does not, however, satisfactorily account for the marked increase in growth rates when the critical range is exceeded. Benedicks and Lofquist’ have given an interpretation of some dilatometer curves produced by Kikuta3which explains the sudden increase in growth at these higher temperatures in terms of the ferrite-austenite transformation.
7:56
HIGH-TEMPERATURE CORROSION OF CAST IRON
This explanation implies that at each complete cycle through the critical range there is a net expansion which is due to the fact that the expansion involved in the austenite to ferrite change does not balance the contraction involved in the ferrite-austenite change. For an iron containing 0.7% combined carbon initially, the net growth per cycle involved may be up to 0.5% by length (1 - 5 % by volume). The net growth per cycle decreases with the number of cycles but the possibility that each cycle, at least initially, can contribute this amount of growth suggests that the mechanism can give rise to very high growth rates. As with sub-critical growth, the duration of each stage of the cycle and the rate of heating and cooling will largely determine the rate of growth achieved, very slow and very rapid cycling being probably the least dangerous in this context. All the remarks so far made have been concerned with conditions of cyclic reheating. When an alloy is held at a steady temperature above the critical range, some growth will arise from graphitisation, partly offset by the contraction involved in the ferrite-austenite transformation, but most of the growth will be due to oxide penetration. Work carried out by Gilbert4on irons maintained at 500°C for 64 weeks (Fig. 7.15) has shown that in ordinary unalloyed flake irons graphitisation and oxidation cause roughly equal amounts of growth, and that as the carbon content increases the effect of oxidation becomes more important and the overall rate of growth increases. Nodular graphite irons grow very slowly under these conditions. Irons designed specifically for good oxidation- and growth-resistance have highly oxidation-resistant matrices, containing either no carbides at all or very stable carbides, and have critical temperatures either below room temperature or above the maximum temperature anticipated. The alloys most commonly used are Silal, Niresist, Nicrosilal and Fe-30Cr. Details of these irons and their properties are given in Table 7.3. The extremely fine graphite structure present in Silal probably makes a major contribution to its good heat resistance. However, when Silal is produced with nodular graphite, its heat-resistance is further enhanced. Two other alloys which have been used for their good oxidation- and growth-resistance are Cralfer (Fe-7A 1-0.750) and Fe- 14 5Si. The production of the former, however, entails considerable difficulties while the latter has poor mechanical properties and poor resistance to thermal shock, with the result that neither is extensively used for this purpose today.
Scaling When an iron is exposed to an oxidising atmosphere, it develops a scale which consists of a series of layers of oxides of varying composition. The thickness of the scale naturally depends on the temperature and the duration of oxidation ( t ) . The scale does not, however, thicken at a uniform rate with time since its very presence reduces the accessibility of the metal surface to tpe oxidising gases. Ideally, the thickness of the scale should increase as ti, but in practice cracks develop in the scale, and these allow the gases to reach the metal surface somewhat more readily than is postulated by this relationship. Cracking will always tend to occur as the film
Table 7.3 Composition
Mechanical properties
Total
-.
...
- -
Silal
2.5
6.0
-
-
'Nicrosilal
2.0
5.0 20.0
Nome
Heat-resisting irons
2.0
n.
.
Fine graphite in silico-ferrite matrix
- Fine graphite in
Critical
Ultimate
154
Nil
280
>920
Nil after 80 x 1 f h at 870°C (White and Elsea')
216
2070
140
800°C) is required a high-carbon cast version of type 310S24 may be used where other requirements allow it. This type of alloy will not be discussed here, but information is given in Reference 8.
Physical Properties Data for a variety of alloys have already been given in Section 3.3,page 3.35.
Mechanical Properties The martensitic steels are used in the hardened and tempered conditions, the tempering temperatures used obviously being in excess of the proposed service temperature. Alloying elements used to improve the creep strength include molybdenum, vanadium, niobium, cobalt and tungsten and these also have the effect of increasing the resistance to softening or tempering so that the proof strengths of the creep-resisting variants are substantially higher than those for the simple Fe-13Cr alloy. 0.2% proof strengths at various temperatures for several martensitic types are shown in Fig. 7.22. Creep and long-term rupture strengths are shown in Figs 7.23 to 7.25 as stress to give 0.1070total plastic strain in 10 OOO h and as stress to give rupture in 10OOO and 100 OOO h plotted as a function of temperature. The ferritic steels rapidly lose strength at elevated temperature as shown in Fig. 7.26 and are of little value for load-bearing applications. The austenitic grades, used mainly in the solution treated (softened) state, have low strength at ambient temperature but maintain strength at elevated temperatures much better than the martensitics and the ferritics. As can be seen from Figs 7.23 to 7.25, creep and rupture strengths are far superior
7:72
HIGH-ALLOY STEELS
6. 7. 8. 9.
1. 4lOS21 2. Fe-12Cr-0.75Mo-O.lC
3. Fe-12Cr-0.8Ni-O.6M0-0.18V-O.lZC 4. Fe-11Cr-0.8Ni-O.6Mo-0.4Nb-0.3V-0.13C
442819 321S31 347S31 316831
5. Fe-10.5Cr-6Co-0.8Mo-O0.45Nb-0.3Ni-0.2V-0.O7C 10, Fe-16.5Cr-11.5Ni-1.5Mo-lNb-O.08C
1000
800 I
N
E E
23 600 u) 0
e
tj 400
200
7 and 8 10 Temperature I
L
I
I
("C) I
I
6
Fig. 7.22 0*2%proof stress values for various steels
above 600°C. Below this temperature proof strength is the limiting factor, but this can be improved somewhat in certain grades by alloying with nitrogen. No more than a brief outline of the mechanical properties can be given here, for detailed information Reference 9 should be consulted. It should be noted that while steels used for creep resisting purposes may conform to the standard specifications, sometimes specially limited composition ranges within these specifications are used in the interests of strength, structural stability or resistance to embrittlement. Fabrication
Forming and fabrication characteristics are described in Section 3.3 on stainless steels. Creep-resisting steels are, of course, intended to resist deformation at elevated temperatures, but in fact the mechanical power required for deformation at the forging temperature is little greater than that required for the stainless steels. Creep-resisting steels often have to be used in thicker sections than is the case with stainless types and this can lead to the need for special techniques for forming and welding.
7:73
HIGH-ALLOY STEELS
1. 2. 3. 4. 5.
410S21 Fe-12Cr-0.75Mo-O.lC Fe-12Cr-0.8Ni-O.6Mo-0.18V-0.12C Fe-11Cr-0.8Ni-0.6Mo-0.4Nb-0.3V-0.13C Fe-10.5Cr-6.OCo-0.8Mo-0.45Nb-0.3Ni-0.2V-O.07C
0. 347831 9. 316S31
400
- 300
n
. 5 E E
ln
g
200
c
v)
100
8
Temperature ("C)
0 400
450
500
550
600
650
700
750
Fig. 7.23 Stresses to give 0.1% total plastic strain at IOOOO h for various steels
High-temperature Corrosion Scale Structure and Oxidation Rates
Since the paper by Pilling and Bedworth" in 1923 much has been written about the mechanism and laws of growth of oxides on metals. These studies have greatly assisted the understanding of high-temperature oxidation, and the mathematical rate 'laws' deduced in some cases make possible useful quantitative predictions. With alloy steels the oxide scales have a complex structure: chromium steels owe much of their oxidation resistance to the presence of chromium oxide in the inner scale layer. Other elements can act in the same way, but it is their chromium content which in the main establishes the oxidation resistance of most heat-resisting steels. In 1929 Pfeil" published a most interesting account of the way layered structures form and the manner in which they influence oxidation rates. From detailed studies of the growth and composition of scales he was able to show clearly how the formation of barrier layers reduced scale formation by hindering outward diffusion of iron through the scale. Naturally, this work had to be largely based on the study of scales of sufficient thickness so that the mechanism of the early stages of oxidation could not be studied in this way. Pfeil analysed the outer, middle and inner layers of scales formed
7:74
HIGH-ALLOY STEELS
1. 410S21 2. Fe-12Cr-0.75Mo-0.1 C 3. Fe-12Cr-O.BNi-O.6Mo-O.18V-0.12C
7. 321831 8. 341831 9. 316331
4. Fe-1 1Cr-0.8Ni-O.6Mo-O.4Nb-0.3V-O.13C 10. F e l l 6.5Cr-11.5N i-l.5Mo-1 Nb-0.08C
100
0
1
1
I
450
500
550
Temperature ("C) 600
650
700
I
750
Fig. 7.24 Stresses to give rupture at 10 OOO h for various steels
on steels containing various alloying elements and thus was able to demonstrate, for example, that the inner layer of scale formed on a 13% Cr steel in air at 1 OoO°C contained 34% Cr203,the middle layer 1.64% and the outer layer 0.89%. The development of the electron-probe microanalyser has given research workers a powerful tool with which to determine composition variations of scale layers and also of underlying metal. Wood and his co-workers have used this instrument to great advantage to help explain the behaviour of iron-chromium alloys "-I4. He found that on oxidation at high temperature (1 1OOOC) protective scales are largely Cr203,containing small proportions of iron oxide. The formation of the Cr,O, causes depletion of the subjacent alloy in chromium, and the chromium content of the alloy at the interface may be as low as 5.3% without apparent transformation of the Cr,O, to a spinel. Failure of the protective Cr,O, scale with time is considered due to scale lifting and cracking followed by rapid oxidation of the depleted alloy. As the catastrophic break-through progresses the content of the inner layer is diluted to a limiting value of 20-25% Cr, while an outer, virtually pure iron oxide layer develops and the depleted subjacent alloy is almost entirely eliminated. These findings largely explain the observed behaviour of chromium-rich steels at temperature in oxidising atmospheres. Oxidation rates increase only relatively little with increasing temperature until, above some temperature, they increase rapidly. The exact value of this 'breakdown'
7:75
HIGH-ALLOY STEELS
400
Z 300 E E
2. 3. 4. 7. 8.
-
Fe-12Cr-0.75M 0-0.1 C Fe-1 2Cr-0.8Ni-O.6Mo-O.18V-0.1 ZC Fe-I 1Cr-0.8Ni-O.GMo-0.4Nb-0.3V-0. 13C 321831 347831 9. 316S31 10. Fe-l6.5C r-1 1.5Ni-1.5Mo-1 N b-0.08C
~
. 5 In Y)
Gi
200 -
100 -
\A
, ("C)I Temperature 450
0
Fig. 7.25
500
2
550
600
650
700
Stresses to give rupture at 1OOOOO h for various steels
I 0
10 78 9
100
300
500 Temperature ( O C 1
Fig. 7.26
700
900
Tensile strength of ferritic stainless steel 442S29
750
7:76
HIGH-ALLOY STEELS
temperature, depends on alloy content but can also be affected by other features external to the alloy. For a detailed review of the mechanism of oxidation of chromium-bearing steels with a comprehensive bibliography Reference 15 should be consulted.
Attack by Gases Flue Gases
There is naturally a desire with corrosion by oxidation to produce in the laboratory quantitative data which can be used for design purposes. This can be a more questionable procedure in the field of oxidation than is even the case with ‘wet’ corrosion since so many features can affect the results obtained. Apart from variations in atmosphere composition, gas velocity, rate of heating or cooling, frequency of thermal cycling, method of sample preparation, sample geometry and time of test can all have marked effects on corrosion rate measured under some circumstances. Most laboratory tests are useful mainly in allowing comparison of alloys and a general assessment of the range of temperatures over which an alloy may be useful, although it must be recognised that, ideally, the behaviour of an alloy selected for service should be further checked under conditions closely simulating service, especially if envisaged service is near the temperature ceiling for the alloy indicated by the laboratory tests. A test procedure which has proved very useful was first described by HatfieldI6. The samples are cylinders 32 x 1 2 - 5 m m in diameter with a standard abraded finish which are supported on open-ended refractory boats in a tubular furnace. In the original test the atmosphere, which was produced by burning towns gas with a 50% excess of air, was passed over the specimens at a standard velocity after first preheating to test temperature over refractory packing in a separate furnace chamber. More latterly, natural gas has been used with suitable modification of air:gas ratio to give
-
200
. hl
E E? 150 i-
1
m .m
2
._m
100
m (II
* P
50
0 1
2
3
4
5
6
7
Number of cycles Fig. 7.27
Total weight gain versus number of 6-h test cycles (20.14% Cr)
7:77
HIGH-ALLOY STEELS
a generally similar test atmosphere. No significant variation in results has been noted. Test times are short, but thermal cycling is incorporated so that any disruptive effects of differential expansion and contraction of metal and oxide may operate. The specimen is heated to temperature for seven 6-h cycles with intermediate cooling to room temperature and weighing, together with any loose scale shed during cooling. Before each test cycle the specimen is lightly brushed and reweighed. Gain in weight versus cycle plots for a number of temperatures determined for a ferritic steel with 20*14(rlo chromium are shown in Fig. 7.27. The behaviour shown is typical of all steels although the temperatures above which rapid oxidation occurs differ. At lower temperatures, the oxidation rate falls with time (cycles) as a protective scale grows, but the gain in weight before the near-protective behaviour is established increases with temperature. Above some critical temperature there is marked progressive oxidation usually with periodic scale shedding. The change from protective to semi-protective behaviour can sometimes occur during the seven cycles of a test (Le. breakaway oxidation). Obviously it could also occur after some longer time, although experience has shown that a temperature 50°C below that at which rapid oxidation appears in this text is a reasonable choice for maximum service temperature. The total gain in weight over the seven 6-h cycles is designated the scaling index and this value is plotted against test temperature for a series of steels of varying chromium content in Fig. 7.28. These were laboratory-produced
200
I
-
r N 0
. --
N'
E 150 m
t .m
m
.61% :r
w
c m .-
-; 100
1.45%
X
25.37% Cr
.35%
W
D
14.97% Cr
.-
m
.-C -
P
m
50
-
400
500
600
700
800
b
30.16% Cr
1
900
I
1000 1100 1200
Test temperature, ("C)
Fig. 7.28
Scaling index versus temperature for several chromium steels
7:78
HIGH-ALLOY STEELS
steels containing about 0.05% carbon, 0.5% silicon and 0.5% manganese, but no other alloying except for the inevitable small amounts of impurities. The sharp change from protective to limited protective behaviour with increasing temperature can be clearly seen. From such plots a simpler method for presenting results can be derived, that is the temperatures at which certain scaling indices could be obtained. The temperatures to give indices of 10, 50 and 100g/m2 are convenient values and are referred to as the SI,,, SI,, and SI,, temperatures. SI temperatures are given for a number of tests, including some carbon and low-alloy types for comparison, in Table 7.8. As well as the types listed in Table 7.7, a selection of creep-resisting grades is included. In addition some of the special stainless steels (see Section 3.3) are also included to demonstrate the effects of some other alloying elements. Table 7.8
C
Si
Oxidation resistance of a number of steels as shown by a short time, cyclic test Mn Cr
Ni
Mo V
Nb
Others
SI ("C)
Grade SI10
0.55
0.24 0.38 0.53 0.10 0.07 0.08 0.28 0.14 0.09 0.14 0.11 0.11 0.07 0.045 0.05 0.05 0.05 0.05 0.075 0.05 0.07 0.05 0.05 0.11 0.07 0.045
0.25 0.26 0.22 3.37 0.19 0.23 0.21 0.31 0.29 0.19 0.42 0.43 0.48 0.48 0.55
0.32 0.42 0.39 1.49 0.55 0.51 0.76 0.40 0.19 0.47 0.48 0.60
0.82 0.55 0.61 0.50 0.48 0.64 0.37 0.31 0.41 0.46 0.86 1.06
0.73 0.89 0.26 0.78 0.37 0.73 1.34 0.85 1.04 0.80 0.76 1.69 1.17 1.49 1.12
0.09 3.12 3.11 8.49 9.04 0.09 12.86 12.80 16.61 12.28 10.89 10.59 11.43 10.70 15.94 13.86 16.01 20.48 28.86 13.12 18.68 18.25 18.28 16.38 22.00 25.20 20.32
0.09 0.34 0.94 0.26 0.36 9.08 0.25 0.50 2.50 0.32 0.58 0.81 2.62 0.60 4.08 5.50 0.25 0.22 1.82 0.09 10.03 8.95 9.95 10.50 14.38 20.24 33.50
< 400 < 500
0.59 0.94 0.20
0.68 0.85 0.62 1.33 0.76
0.28 0.18 0.39 0.13 0.21 0.17 0.33 6.01 Co 0.29 3.26 Cu 1.61 0.37 1.74 Cu 0.16N 4.01AI 0.43 Ti 0.77
2.52 0.25 Ti,0.33Al
SI,,
600 620 600 600 960 640 605 820 800 790 810 805 810 760 810 807 770 860 lo00 1120 1100 880 860 860 820 1060
640 615 950 1010 < 500 695 < 500 620 800 835 790 825 750 850 800 820 800 820 800 815 750 770 800 825 780 818 750 830 850 875 900 1075 900 1160 1060 1125 860 915 820 890 820 890 700 845 980 1070 1030 1090 1180 970 1010 1140
< 500
0.92
SI20
The major beneficial effect on oxidation resistance comes from alloying with chromium, silicon and aluminium. Chromium represents the basic alloying addition for most oxidation-resisting steels and can be accommodated up to about 14% with a martensitic structure and 30% (practically) in a ferritic structure. Suitable alloying with nickel allows austenitic structures also with high chromium contents, 25% chromium being the highest value used currently (3 10S31). While silicon and aluminium both strongly
HIGH-ALLOY STEELS
7:79
complement the beneficial effects of chromium, they are both strong ferrite formers, which limits their use. Aluminium is used in some very resistant steels but these are ferritic and so can only be used at the high temperatures available from an oxidation resistance point of view when stressing is relatively low. Unlike aluminium, silicon is present in small quantities (0-2-0-5 % ) in most commercially produced steels (see Table 7.8) and there is evidence that even such small amounts contribute substantially to the behaviour of chromium steels. In Fig. 7.29, SI,, temperatures are plotted against silicon content for a series of martensitic steels with 10.89-13- 14% chromium. Nickel is much used to control structure. It can have a slightly adverse effect on the oxidation resistance of martensitic steels but is beneficial in the larger amounts relevant to the austenitic types. The other commonly used alloying elements have little effect (at least in the quantities used) although manganese in substantial amounts is somewhat detrimental and molybdenum can be harmful if service conditions are such that the volatile MOO, can attain significant levels in the gases adjacent t o the steel. Rare earths in small quantities can be beneficial, as they can in other alloy systems.
Si (%)
Fig. 7.29 Temperature to givea scaling index of 100 versus silicon content (10.89-13.14% Cr)
Although behaviour does not show any great variation over quite a wide range of oxidising atmospheres, the air:fuel ratio can exert some effect. This was shown by tests on an Fe-18Cr-SNi alloy at 850’C”. The steel was submitted to a test of the type just described, but in atmospheres produced by catalytic burning of 2: 1,4: 1 and 6: 1 air:towns gas (pre-natural gas), the 4: 1 ratio corresponding to stoichiometric combustion. Mean oxidation gains expressed as g/m2 were: 6:l air:gas 4:l air:gas 2:l air:gas
32 12 4
7:80
HIGH-ALLOY STEELS
At temperatures below 850°C this effect of varying air:gas ratio tended to disappear and it was not apparent in tests at 750°C. From the data shown in Table 7.8 it can be seen that 850°C is about the temperature of transition from protective to semi-protective behaviour, so atmospheric effects would be at their greatest. The effect of sulphur from the gas phase is critically dependent on the effectiveness of fuel combustion. With good combustion to the limit of the available oxygen, and even down to 50% air deficiency, no serious effect was found from high-sulphur fuel in tests with 321312 steel up to the usual limit of service temperature at 85OoC, as shown in Table 7.9. With 310324 steel at 1 100°C some effect from high sulphur content was found in 2:l air:gas with effective combustion, but none in 4:1 and 6:1 mixtures. Table 7.9 Comparison of high- and low-sulphur fuel in tests at 850°C with 321S12 steel” Air:gas ratio 6: I
4:I
2: 1
Sulphur in gas *
Oxidation gain in seven 6-h test cycles (g/mZ)
Low
26
High
27
Low
17
High
21
Low
10
High
12
* T h e towns gas contained 460 mg of S/m’. High sulphur addition was made in the form of H,S equivalent 10 0.5% SO, in the 6:l combustion products. ;.e. about three times as much as from ordinary high-sulphur
fuel?.
With poor combustion, on the other hand, very severe ccelers ion of attack, dependent upon the formation of sulphide in the scale, can occur. This destroys the protective action of the scale, and results in sulphide penetration of the metal in advance of oxidation. The effect is illustrated by tests
Fig. 7.30 Unetched section through Fe-25Cr-21Ni after attack in 4:l air:gas 1 100°C with poor combustion, showing sulphide penetration: x 150
+
HzS at
7:81
HIGH-ALLOY STEELS
with 310324 steel in stoichiometric 4: 1 air:gas at 1 100°C, in which the burner was modified to give incomplete combustion. The tests were for two 6-h cycles, and oxidation gains were ( a ) good combustion, high-sulphur fuel, 16 g/mZ; ( b ) bad combustion, low-sulphur fuel, 12 g/m2; ( c ) bad combustion, high-sulphur fuel, 3 19 g/m2. Sulphide penetration into the metal under condition ( c ) is illustrated in Fig. 7.30. Air
Air tends to be less aggressive than the flue gas used for the standard test described earlier, but the useful range of temperatures for each steel is effectively similar. Edwards and Nicholson '* reported some long-term testing of four austenitic grades in air saturated with water (at room temperature) at temperatures of 650°C-875"C for up to 10 OOO h. They make the point, and show convincingly, that in any long-term assessment, metal wastage must not be based on scaling alone but that the effect of subsurface penetration must also be considered (this applies also, of course, to testing in other gases). Thus their values (Fig. 7.31) are compounded from surface loss and subsurface penetration, and their work is especially valuable in that an assessment of the effect of long-term heating on mechanical properties was also made.
U 01
c
u
0.1
01 r L
-0 0.05 u
2
,(d',
0
,
J
50 700 750 800 850 650 700 750 800 850 Temperature ( O C )
Fig. 7.31 Metal wastage of several steels due to oxidation in air (saturated with water at room temperature) for IO OOO h at various temperatures. ( a )Type 302S31, ( b ) type 321S31, ( c ) type 316331 and ( d ) type 310S31. T I ~ = , ~Ssurracc I PPcnclratlon (after Edwards and Nicholson")
+
Nitrogen can be absorbed from air during prolonged heating, but with steel in the unstressed or lightly stressed state the rate is very slow except for temperatures above 1 050°C. Considerable absorption can occur at lower temperatures during creep, however. This fact is presumably due to the exposure of oxide-free surface during creep, and it has been noted that nitrogen absorption is especially marked at cracks. The following nitrogen contents have been reportedI9 for the 0.75 mm surface layers of creep specimens in 34733 1 steel (original nitrogen content 0.053%) after creep failure:
7:82
HIGH-ALLOY STEELS
650°C (life 10 970 h under 108 MN/m2), 0.077% N, 700°C (life 37958 h under 46MN/m2), 0.65%N2 800°C (life 16629h under 1.5 MN/m2), 0.90% N,. Steam
Modern boiler developments involving increased steam temperatures and pressures have made it increasingly important to consider the behaviour of high-alloy steels under conditions typical for superheater tubes and steam pipes. For satisfactory service the steels must, of course, possess adequate mechanical properties, especially creep resistance, but they must also be sufficiently resistant to oxidation to ensure long life. Short-term laboratory tests are of value in yielding comparative data for different steels and, in fact, results generally similar to those for the flue gas test already described are obtained, but prolonged tests approximating more nearly to service conditions are desirable. Rohrig, van Duzer and Fellows'' exposed samples in an experimental superheater fed with steam at 2.6MN/m2 from a power plant. Some 42 materials were tested for periods of up to 16 OOO h, attack being estimated after test by weight loss following descaling. It was concluded that at 593°C attack continues at a high rate on carbon steel, whereas the rate for most alloy steels decreases with time (Table 7.10). Table 7.10
Losses from exposure for 7461 h in steam at 2.62 MN/m2 and 593°C;
Steel Mild steel A.I.S.I. 403 (Fe-12Cr) A.I.S.I. 347 (Fe-1SCr-SNi + Nb) A. I .S.I. 309 (Fe-250- 12Ni) A. I .S. 1. 3 10 (Fe-25Cr-20Ni) Fe-35Ni-15Cr
Calculated penetration in loo00 h (mm) 0.107 0.015 0.002 0.002 0.003 0.002
Data after Rohring, van Duzer and Fellows".
Eberle, Ely and Dillon" tested commercial tubes in a small superheater receiving plant steam at 14MN/m2 and superheating it from 538" to 677°C. Penetration was estimated from scale thickness measurements after 6 950 h and comparison was made between the attack by steam on the inside of the tubes and that by flue gas from pulverised coal firing on the outside (Table 7.11). A collaborative test programme covering low-alloy and high-alloy steels was carried out by the Central Electricity Generating Board and various steelmakers. Samples were exposed in specially constructed chambers held at 566"C, 593°C and 621°C fed with power-station steam at a pressure of 3.45MN/m2 for times of up to 16286 h. In the assessment of the results both metal lost from the surface and subsurface penetratien were measured. The results have been reported by King, Robinson, Howarth and Perry in a C.E.G.B. report. Selected data are shown in Fig, 7.32, in which the broken lines have been obtained by extrapolation of the experimental results.
7:83
HIGH-ALLOY STEELS
Table 7.11 Comparison of internal and external scaling of superheater tubes after 6 950 h in steam at 13.8 MN/m2 and 500-670°C * ~~
Estimated penetration (mm/y) Steam Flue gas
Steel A.I.S.I. A.I.S.I. A.I.S.I. A.I.S.I.
304 (Fe-18Cr-8Ni) 321 (Fe-IKr-INi+Ti) 347 (Fe-18Cr-8Ni+Nb) 318 (Fe-16Cr-13Ni-3Mo+ Nb)
0.038 0.038 0.010 0.013
0.021 0.023 0.023 0.029
After Eberle, Ely and Dillon” and relating to tests in a superheater raising steam from 540-C to 670%
Exposure time Ih) Fig. 7.32 Metal wastage of several steels due to oxidation in steam at various temperatures. ( a ) Mild steel. ( b ) Fe-ZCr-0.25M0, ( c ) Fe-12Cr + M o + V, ( d ) A.I.S.I. 316 and ( e ) Fe-18Cr-I2Ni-1Nb. TtOtal = SrurfaccPpcnctla,ion (after King et a/.)
+
7:84
HIGH-ALLOY STEELS
Other Industrial Gases
All oxidising gases can lead to oxide formation on chromium steels at elevated temperatures and in some instances this can be associated with absorption of some other substance in the steel. Carbonaceous gases are a good example and whereas high-alloy steels successfully resist flue gases even under conditions of considerable air deficiency, reduction of oxygen content eventually leads to conditions under which at a sufficiently high temperature considerable carburisation of the metal occurs. An example is the endothermic gases used as protective atmospheres for other metals which, at elevated temperature, can rapidly cause embrittlement of high-alloy steel. The absorption of nitrogen from air has been mentioned and similar effects can occur with nitrogen under similar circumstances. However, nitriding is much more likely in ammonia or in gases containing ammonia, as indicated by the following figures for the nitrogen contents of the outer 0.25 mm layers of samples of 310S31 steel (initial N content 0.06%) after 250 h in ammonia: 500"C, 0.25% N,; W " C , 0.55% N,; 1 OOO"C, 0.92% N,; 1 050"C, 1.19% N,. Nitriding leads to serious embrittlement. Hydrogen at high pressure and temperatures above 400°C has a considerable adverse effect on carbon steel, dissolving in the steels and combining with carbides to produce methane and so causing fissuring and considerable embrittlement. However, chromium stabilises the carbides and stainless steel may be safely used in hydrogen at dull red heat".
Ash Attack The degree of oxidation in a gaseous environment can be modified greatly by the deposition of even small amounts of certain fuel ashes. The topic of ash corrosion has been reviewed with extensive bibliographyz3, but some consideration of high-alloy steels will be given here. Any substance which can form a low melting point mixture with the normally protective oxide scale (Le. 'flux' the oxide) formed on high-alloy steels, is potentially dangerous. While such substances are not common in fuels the danger should be borne in mind where high-alloy steels are used as containment vessels for high-temperature processes. Sulphates, which form part of the ash from the combustion of many fuels, are not harmful to high-alloy steels, but can become so if reduction to sulphide occurs. This leads to the formation of low melting point oxide-sulphide mixtures and to sulphide penetration of the metal. Such reduction is particularly easy if the sulphate can form a mixture of low melting point with some other substance. Reduction can be brought about by bad combustion, as demonstrated by Sykes and Shirley", and it is obviously important to avoid contact with inefficiently burnt fuels when sulphate deposits may be present. Reduction can also be brought about in atmospheres other than reducing ones and the presence of chlorides or vanadium pentoxide has been shown to be sufficient to initiate the reaction. It has also been shown" that it can be initiated by prior cathodic polarisation in fused sodium sulphate. The effect of even small amounts of chloride on oxidation in the presence of sulphate is illustrated in Fig. 7.33".
7:85
HIGH-ALLOY STEELS
I
E
I
f
x
120
LIlL 100
200
300
LOO
500
600
Duration (h)
Fig. 7.33 Gains in weight due to oxidation of type 347S31 steel in air at 750°C while in contact with Na2S04. Curve A plain, and curve B containing 0 . 3 % NaCl (period X- Y) (after Sykes and Shirley ”)
Not all sulphates are as readily reduced as sodium sulphate, for instance, calcium sulphate does not usually lead to sulphide penetration, although the presence of other substances with calcium sulphate may lead to accelerated oxidation for other reasons. The results for laboratory tests on a series of metals and alloys in sodium sulphate + sodium chloride and calcium sulphate + calcium chloride mixtures are shown in Table 7.1225. In many cases sulphide peneration could be noted with the sodium salts but not with the calcium salts. Table 7.12 Effect of 9O:lO su1phate:chloride mixtures on various metals at 750°C (tests for 6 h in air)* Moreriol Cr Ni Mild steel Fe-2OCr Fe-28Cr-2Ni Fe-22Ni- 14Cr Fe-ISCr-12Ni + Nb Ni-I3Cr Ni-2OCr +Ti + AI * Data after Shirley ‘’.
of rnerol (g/m2) Sodium salts Calcium s o h
LOSS
No mixture 7 20 490 4 3 2 9 2 4
30 60 1430
1120
50
I350
60
I160
40 50
950 470 20 20
2 700 3 700
110 110
7:86
HIGH-ALLOY STEELS
Table 7.13 Corrosion tests in air with specimen half immersed in sodium chloride Test temperature
Weigh1 loss after descaling following a 24 h test (g/m2)
("C)
Steel
550 650 750
430
304
321
100
30 270 660
20 100
320 I050
650
347
310
20
30 190
210 400
750
It has been suggested that corrosion by sulphates can occur by the formation of pyrosulphates which melt at relatively low temperatures. Accelerated corrosion due to the presence of pyrosulphates has been demonstrated for ferrous alloys inchding stainless The r61e of chlorides in the presence of sulphates has already been mentioned, but these can aIso have a serious effect in the absence of other contaminants. The presence of chloride not only leads to considerable acceleration of oxidation rate but can also give substantial subsurface intergranular penetration of the steel. Corrosion test results for several steels are shown in Table 7.13. The attack noted on the calcium sulphate calcium chloride mixtures indicated in Table 7.12 can possibly be attributed solely to the presence of chloride. Very small amounts of chloride are sufficient to cause serious acceleration, as illustrated in Fig. 7.34. Samples of 347831 steel were heated in air to 650°C for 20-h cycles. Between cycles the samples were cooled to room temperature and weighed together with any loose scale,
+
Test A (NaCl lded Test 0 Gain in //Test C A weight due to oxidation
/Jest
,5
/
/
Test 0
__
T l N n. C - - l. nddedl -- -,
Test C lNaCL added) No addition 0
20
10
60
80 100 Time (h)
120
110
160
180
Fig. 7.34 Gains in weight of 347S31 steel in air at 650°C with various periodic additions of sodium chloride. Stepped curves show quantity of NaCI added, broken curves show gains in weight attributable to oxidation
HIGH-ALLOY STEELS
7:87
brushed, weighed again and then dipped in a dilute sodium chloride solution, removed, dried and weighed yet again to give the quantity of salt added to the surface. Gains in weight attributable to chloride and to oxidation are shown; it should be noted that the latter are probably low since no account was taken in this method of assessment of any losses of volatile substances. In tests of this sort, when the addition of chloride is discontinued, the oxidation rate slowly returns to the very low value to be expected in air at this temperature. While most interest has been in effects of chlorides, there is evidence that other halides can have similar effects. The subject of corrosion in steam-raising plant burning chloride-bearing coal was considered in the conference reported in Reference 27. In the case of oil ash, the most serious damage is associated with vanadium compounds. Organic vanadium compounds which cannot be economically removed from the residual oils form vanadium pentoxide in combustion which can have a considerable corrosive effect on heat-resisting steels. Vanadium pentoxide itself has a melting point in the neighbourhood of 660°C and forms compounds which have even lower melting points. Prior to the development of the gas turbine the problem was not of such overriding importance, since components, such as superheater tubes in steam boilers, which had to be thin, were at a low enough temperature to escape serious action. Tube supports in superheater and oil-cracking furnaces fired with residual oil often suffered severe attack, but as they were cast products, relatively inexpensive and replaceable, they were regarded as being of necessity expendable even though losses of metal of thicknesses 2 10 mm/ year from Fe-24Cr-21Ni and Fe-30Cr steel supports at 800-900°C could occur. With the advent of the gas turbine and the possibility of higher steam temperatures an entirely new order of assessment was necessary as the very much smaller rates of attack occurring in the range 650-800°C involved quite prohibitive expense in relation to costly creep-resisting high-alloy steel blades or tubes. No complete solution has emerged, but useful measures to reduce the seriousness of the attack in some cases have been evolved, and the limits within which vanadium-bearing oils may reasonably be used have been defined. The oil ash problem is complicated and intensified by the presence of alkali oxides and sulphates which form a variety of low-melting complexes with vanadium oxide. An idea of the complications occurring in practice can be obtained from the account of extensive field and laboratory tests carried out for a committee of the Council of British Manufacturers of Petroleum Equipment 28, with particular reference to superheater supports and metal temperatures in the range 600-850°C. The field tests were carried out in three boiler installations, one marine and two land-based. Materials tested included Fe-26Cr, Fe-18Cr-8Ni + Nb, Fe-24Cr-22Ni and Fe-32Ni18Cr+Ti, as well as a number of nickel-base alloys. It was concluded that deposits in the superheater zones of land-based boilers tend to have a higher V:Na ratio than is present in the original oil, but that with installations operating at sea, contamination by sea-water may reverse this relation. Deposits of high V:Na ratio accelerated oxidation primarily through scale fluxing, and were most corrosive to steel, while deposits of high sodium sulphate content accelerated oxidation through sulphide action and were most damaging to alloys of high nickel content. A ferritic 26% Cr steel was
7:88
HIGH-ALLOY STEELS
the only one which showed good resistance to deposits containing sodium and vanadium in all proportions; this type of alloy unfortunately has relatively poor stress-carrying capacity at high temperature. Since the main action of vanadium is related to fluxing, much attention has been given to the inhibition of this action by formation, through suitable conditions, of higher-melting compounds; calcium and magnesium compounds are the most generally favoured of such additivesz9. Preliminary washing of the oil to reduce sodium to very low limits has also been advocated 30. The stability of calcium and magnesium vanadates is, unfortunately, not great enough t o prevent their substantial conversion to sulphates by the sulphur oxides normally present in the flue gases, so that with additions of alkaline earth oxides to, say, two or three times the stoichiometric equivalent of the VzO,, stabilisation as vanadates is incomplete, and the improvement only partial. For critical components the only safe procedure is usually either to limit the temperature at which steels are used to a maximum of about 6Oo0C, or to use more expensive distillate-oil fuel free from vanadium. The whole subject of vanadium attack has been reviewed by Sachsl’. An interesting case of ‘ash’ attack is encountered with valves in engines powered by high octane fuels containing lead compounds. These compounds are deposited from the gases as mixtures of lead oxide, sulphate and bromide, and can cause serious scale-fluxing effects with high-alloy valve steels.
Molten Salts Molten salt baths are widely used in heat treatment and for steel carburising. As in the case of ash attack, danger to alloy steel containers arises mainly from enhanced oxidation brought about by scale fluxing. Such oxidation can only proceed where oxidising conditions obtain, so that while alkali chlorides form useful heat-treatment baths for steel, they produce severe attack, even with heat-resisting steel containers, at the surface of the bath, with formation of chromates and ferrites. Austenitic Fe-23Cr-12Ni and Fe-35Ni-15Cr steels are used for cyanide-hardening-bath containers, but conversion of cyanide to carbonate in use brings danger of fluxing attack. Molten alkali hydroxides are particularly dangerous, not only because of scale fluxing, but also because they induce stress corrosion where stress is a serious factor.
Molten Metals Heat-resisting steels have limited uses in contact with molten metals. They are not recommended for use with molten zinc, cadmium, aluminium, antimony or copper, because of excessive attack and embrittlement effects. In brazing and silver soldering, contact between the molten non-ferrous alloy and the steel occurs for only a very limited period of time. With molten lead or tin, limited use of high-alloy steels is possible. In the case of containers for lead baths, it is important to avoid the combination
HIGH-ALLOY STEELS
7:89
of lead oxide and air at the bath surface because of fluxing action, but with mechanical removal of lead oxide and use of carbonaceous coverings, Fe-35Ni-15Cr and Fe-25Cr steels are successfully used. With tin, behaviour depends considerably on temperature, slight action taking place at 300°C and considerable attack at 600°C. The alkali metals have acquired special interest through their suitability for use as heat-exchanging fluids in atomic reactors. They are generally satisfactorily resisted by the heat-resisting steels, although detailed studies have shown effects on prolonged contact at high temperatures. Thus, Brasunas3* describes the leaching of nickel by lithium at 1OOO"C from Fe-18Cr-1ONi steel, with transformation of the nickel-impoverished surface layer to a ferritic structure and ultimate production of subsurface cavities. He also indicates that there is some penetration by sodium and lead at this temperature with precipitation of intermetallic compounds within the steel.
Applications The heat-resisting steels are used for a wide range of general engineering and chemical engineering applications where the corrosion resistance, and in some instances strength, of the lower-alloy steels is inadequate. The martensitic steels, because of their lower oxidation resistance, are normally used for the less onerous conditions, and certain limitations in ease of fabrication generally precludes their use for large structures and containment vessels. Their combination of moderate corrosion resistance plus strength at modest temperatures has led to widespread use as turbine discs and blades, bolts and similar parts. They have also been used on steam plant for the less onerous conditions. The ferritic steels are limited in scope because of lack of hot strength, but the cheaper types such as 430 are used in sheet form for the fabrication of parts such as heat exchangers. The higher chromium varieties are of importance in being much more resistant to sulphur attack than the nickel-bearing types and so are widely used as superheater supports or in sulphide-roasting furnaces, mainly as castings. Strength limitations and brittleness call for care in design. The austenitic steels combine good oxidation resistance with ease of fabrication and thus are most widely used. In addition, while being quite weak at room temperature, they are among the strongest materials in the 550-750°C range and are thus widely used for this purpose. Typical applications are furnace parts, heat exchangers, gas turbine parts, steam superheaters and piping, and chemical plant equipment for containing reactions and products at elevated temperatures. J. E. TRUMAN
REFERENCES 1. 2. 3. 4.
Hatfield, W. H., J.I.S.I., 115, 517 (1927) Aitchison, L., Engineering, 108, 799 (1919) Dickenson, J. H. S., J . I . S . I . , 106, 103 (1922) Farenwald, F. A., Proc. Amer. Soc. Test. Mat., 24. 310 (1924)
7:90 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
NICKEL AND ITS ALLOYS
Johnson, J. B. and Christiansen, S. A., Proc. Amer. SOC.Test. Mat., 24, 383 (1924) Vanick, J. S., Proc. Amer. SOC. Test. Mat., 24, 348 (1924) Hatfield, W. H., J.I.S.I., 115, 483 (1927) Edeleanu, C . and Estruch, B., I.S.I. Special Report No. 86, 220 (1964) High Temperature Properties of Steels, I.S.I. Publication No. 97 (1967) Pilling, N. E. and Bedworth, R. E., J. Inst. Met., 29, 529 (1923) Pfeil. L. B., J.I.S.I., 119, 501 (1929) Wood, G. C. and Whittle, D. P., Corrosion Science, 4 No. 3, 263-269 (1964) Wood, G. C. and Whittle, D. P., Corrosion Science, 4 No. 3, 293-315 (1964) Wood, G. C., Corrosion Science, 2, 255-269 (1962) Wood, G. C., Corrosion Science, 2, 173-192 (1962) Hatfield, W. H., J. Inst. Fuel, 11, 245 (1938) Sykes, C. and Shirley, H. T., I.S.I. Special Report No. 43, 153 (195 I ) Edwards, A. M. and Nicholson, A., I.S.I. Publication No. 117, 149 (1969) Kirkby, H. W. and Truman, R. J., I.S.I. Special Report No. 64,244 (1959) Rohrig, I. A . , van Duzer, R. M. and Fellows, C. H., Truns. Amer. Soc. Mech. Eng., 66,
277 (1944) 21. Eberte, F., Ely, F. A. and Dillon, J. A., Trans. Amer. Soc. Mech. Eng., 76, 665 (1954) 22. Inglis, N. P. and Andrews, W., J.I.S.I., 128, 383 (1933) 23. Hancock, P., Corrosion of Alloys at High Temperatures in Atmospheres Consisting of Fuel Combustion Products and Associated Impurities, H.M.S.O. (1968) 24. Simons, E. L., Browning, G. V. and Liebhafeky, H.A., Corrosion, 11, 50% (1955) 25. Shirley, H. T., J.I.S.I., 182, 144 (1956) 26. Jonakin, J., The Mechanism of Corrosion by Fuel Impurities, 648 and 649, Butterworths, London (1963) 27. Meadowcroft, D. B. and Manning, M. I. (eds), Proc. Conf. on Corrosion Resistanr Materials for Coal Conversion Systems, Applied Science Publishers (1982) 28. British Petroleum Equipment News, 7 No. 4, 54 and 7 No. 5, 48 (1959-1960) Gardner, G. M. and Sanders, D. G., Residual Fuel Oil Ash Corrosion, 29. Buckland, B. 0.. A.S.M.E. Paper A-52-161 Preprint (1952) 30. Buckland, B. 0. and Sanders, D. G., Modified Residual Fuel for Gas Turbines, A.S.M.E.. Paper 54-A-246 Preprint (1954) 31. Sachs, K., Melallurgia, 57, 123, 167, 224 (1958) 32. Brasunas, A. de S., Corrosion, 9, 78 (1953)
7 . 5 Nickel and its Alloys
Oxidation Pure Nickel
Although the oxidation of nickel has been extensively studied it is only recently that the process has been clearly understood. The relative simplicity of the system in which only a single-phase layer of oxide, NiO, forms has encouraged research, and a further simplification is that the expansion coefficients of the oxide and metal are similar, (17.1 and 17.6 x "C-', respectively,) so that the effects of thermal cycles can be largely neglected. Nickel, in comparison with metals such as iron, cobalt and copper, has a relatively good resistance to oxidation at high temperatures. The growth of the oxide generally follows a parabolic law, but deviations are observed depending on the surface preparation, alloy purity and microstructure. Figure 7.35 shows a comparison of the parabolic rate constant for the oxidation of high-purity nickel with the tracer lattice diffusion coefficient of nickel in NiO, and it can be seen that it is only at temperatures in excess of about 1200°C that the activation energies of the two processes become similar (230-250 kJ mol-') I . At lower temperatures the rate of oxidation is increasingly greater than would be predicted by assuming that the process is controlled by bulk diffusion in the oxide lattice, with activation energies being reported in the range 155-170 kJ mol-'. The effect of prior cold-work in the nickel is to increase the oxidation rate, but the observed rate law is usually less than parabolic. Both these observations suggest that the rate of oxidation is controlled by grain boundary diffusion in the oxide; the less than parabolic rate observed in the cold-worked material occurs because the initially fine-grained oxide coarsens during the oxidation process thereby eliminating some short-circuit diffusion paths. Models have been developed to describe the oxidation reaction where the rate is controlled by dual-lattice and grain-boundary diffusion, in which the effective diffusion coefficient is given by, De, = D, + 2 ( D ' S ) / g
(7.10)
where g is the grain size normal to the growth direction, 6 is the grain boundary width, and D ' and D, are the diffusion coefficients of the boundary 7:91
7:92
NICKEL AND ITS ALLOYS
1400
1000
i
I
-8
-
- 10
I
u) N
E
-12
.
T ("C) 800 700 600 500 400 3 00 I I I I I I Growth of NiO Solid points Measured in oxidation 0 Rhines and Connell (1977) A Atkinson et a l ( 1 9 8 2 ) Graham and Cohen ( 1 9 7 2 ) Open points C a l c u l a t e d f r o m Grain Boundary Diffusion of 0 and Grain s i z e measurements
U
Q
Y v
A
0
--14
m
-0
-16
-18
Fig. 7.35
0
Calculated f r o m Lattice Doffusion of N I in N i 0
m 0
I 6
8
10 12 1 0 4 u (K-')
14
I 16
Arrhenius plot of the parabolic rate constant for the oxidation of Ni to NiO (after Atkinson ')
and the lattice, respectively. This model has been used by Atkinson' to calculate parabolic rate constants for nickel where grain boundary diffusion dominates, and there was good agreement between the calculated values and those obtained experimentally, as shown in Fig. 7.35. A single layer of nickel oxide forms during the early stages of growth of the oxide, but as the layer thickens a duplex structure develops; this consists of an inner region of equiaxed, fine-grained crystallites and an outer region of large columnar crystals. The inner layer is generally more pronounced on less pure material, and is believed to be due t o the presence of impurities segregating to the grain boundaries thereby inhibiting grain growth. Tracer diffusion studies have shown that the outer layer grows by movement of nickel vacancies along the grain boundaries, and the inner layer by molecular oxygen penetration along microcracks and fissures which are present in the outer layer due to the build-up of stress in that layer. In the case of nickel oxide, compressive stresses result because of the constraints imposed on the oxide layer by the receding metal. Dilute Nickel Alto ys
The resistance of nickel to oxidation may be modified considerably by alloying, although the rate of oxidation still in general obeys a parabolic rate
7:93
NICKEL AND ITS ALLOYS
law, the rate constant increasing exponentially with temperature. In general the rate constant increases linearly from the value for nickel with increasing additions of a second element, but above a given level, which depends on the solute, the change becomes slower and for some elements the rate constant then decreases. Results obtained by Horn2 are illustrated in Fig. 7.36, which shows that beryllium, silicon and chromium in particular can pro1201
w TI
4
280
-
240
-
200
-
0
2
4
6
IO
Solute metal 1At.V.)
Fig. 7.36 Effect of alloying on the rate constant for oxidation of nickel at 9oo°C2
7:94
NICKEL AND ITS ALLOYS
Table 7.14 Relative specific scaling rate constant for binary nickel alloys at 900°C2 Element
K,’
Be Ca
x IO* 17
216 15.8 102
AI
Si Ti Zr Ce Th Cr Mo W Mn cu Au
78.7 204 275 383 48.2
26.5 66.5 35.5 24.4
7.3
-
K,* = ( K . m C K.,,t.l)/atomic qo solute. where K.,,M IS approximately 20 x 1 0 - 8 .
duce enhanced oxidation resistance. From the linear portions of the rate constant/concentration curves, Horn calculated the difference in rate constant from that of nickel produced by 1 atomic 070 of each of the solutes investigated and obtained the values given in Table 7.14. The concentration at which the curves deviate from linearity is associated with that at which the oxide of the solute begins to form a complex oxide with NiO, e.g. NiCr204, or a discrete second phase. At the lower concentrations the effect had been attributed to lattice distortion due to the solute ion; but it is now generally believed to be associated with either Wagner Hauffe doping of the lattice (Section 1 4 , where there is significant solubility in the oxide lattice, or with modification of the grain boundary structure where segregation is an important effect. It would appear that the effects of impurities at the grain boundary must be either (a) to increase the diffusion rates or (b) to influence the microstructure and increase the number of short-circuit paths. However, theoretical modelling of the grain boundary structure by Duffy and Tasker3 and Table 7.15 ‘Alloying factor of oxidation’ for nickel alloys *
F t for stated addition (Wt.
Element 5
Chromium Manganese Tantalum Molybdenum Copper Niobium Platinum
4 3 3
24 If I 1
10
If 4 2 3 If
-I -3
70)
20
30
-3I
-4I -2I
-6I -2I
2f
9 -
-
-21 2
-I
-3
-
‘2
40
-
50
-SI 1
-
-I
Derived from curves of Reference J. rate of alloy relative to that of pure nickel
t Approximate oxidation
*In the tables and figures K x IW indicates that the actual values given are K x IO-.‘..
NICKEL AND ITS ALLOYS
7:95
T("C)
-f
11001000 900600 I
I
700
500
600
I
-lC
-1
i
-
h
I
Lo N
E
2 -14
G!
0 m
0 J
-16
-18 .
X Denotes Cr lattice
\
diffusion measured in the present study
\
NI (lattice)
\\
\
I
I
I
1
I
\
\\Cr (lattice) I
I
I
Fig. 7.37 Diffusion coefficients for some impurities in NiO grain boundaries compared with the corresponding lattice diffusivities (the grain boundary width is assumed to be 1 nm) (after Atkinson and Taylor4)
experimental measurements of grain boundary diffusion rates4 indicate that impurities often decrease grain boundary diffusion rates (Fig. 7.37). Thus it would appear that the effects of impurities in increasing the oxidation rate of nickel most probably result from a reduction in the oxide grain size with a consequent increase in the number of short-circuit diffusion paths. A semi-quantitative indication of che effects of different elements on the resistance to oxidation of nickel is given in Table 7.15 which lists values for the relative oxidation rate with respect to that of nickel for different concentrations of solute element. These values are approximately valid over quite wide ranges of time and temperature5.
7:96
NICKEL A N D ITS ALLOYS
Nickel- Chromium Alloys
Isothermal Oxidation The alloys based on the nickel-chromium system are of paramount importance in the field of high-temperature alloys. As shown in Tables 7.16 and 7.17, addition of chromium to nickel has a complex effect on the oxidation behaviour; small additions are deleterious, the isothermal oxidation rate increasing with chromium content to a maximum at about 7% chromium. With less than about 9% Cr, internal oxidation occurs and the chromium content of the matrix is sufficiently reduced for the alloys to appear magnetic (Table 7.17), although before oxidation only alloys with less than 7% Cr are magnetic. A progressive improvement in oxidation resistance results from further additions of chromium up to a chromium level reported variously as about 20% and 40-90%, depending on temperature, and these alloys are more resistant than either of the constituent metals although pure crack-free chromium gives oxidation values very similar to those for commercial Ni-20Cr alloys'. Barrett and his colleagues9, and Kosak l o have summarised existing information on the scales formed on nickel-chromium alloys. Up to about 10% Cr, the thick black scale is composed of a double layer, the outer layer being nickel oxide and the inner porous layer a mixture of nickel oxide with small amounts of the spinel NiO - Cr,O,. Internal oxidation causes the formation of a subscale consisting of chromium oxide particles embedded in the nickelrich matrix. At 10-20% Cr the scale is thinner and grey coloured and consists of chromium oxide and spinel with the possible presence of some nickel oxide. At about 25-30% Cr a predominantly chromium oxide scale is Table 7.16 Oxidation data for nickel-chromium alloys6 Chromium
content (To) ~~~~
Temperature ("C)
p 02 (am)
Rate constant ( K p)
900
air
900 900 900 900
air air air air
0.28 4.9 5.8 8.2 0.0
~
0.0 I .97 4.12 5.89 8.0
0.0 0.3 1.0
3.0 10.0
0.0 0.32 0.92 2.0 3.45 5.67 7,64 8.71 11.1 14.9 20.0
~
lo00 lo00 lo00 lo00 lo00 1096
1096 I 096 1 096 1096 1096 1096 10% 1096
(g*m-Js-'x~~z)
~~~
1
3-48 15.0 25.8 28-3 5.55
1 1
5.48 23.6
I
29.7
1
39.6 46.8 58.5 67.8 30.8 3.79 0.35 0.07
1 1 1 1
I 1
I
I096
1 1 1
1 096
I
7:97
NICKEL AND ITS ALLOYS
Results of oxidation tests on nickel-chromium alloys’
Table 7.17
Material composition
(Yo)
Weight gain (gm-’s-Ix 104)
Ni
Cr
100
3
97 96 94 92 91 91 88 86 85 82 80
4 days at 1 038 “C
4 6 8 9 9 12 14 15 18 20
Irregular attack;
8 days af 954 O C
Magnetic Oxide thickness (mm) responset
Mass gain (gm -’s
I
Outer scale Inner zone 6.6 10
0.178’ Thin
0.127’ Nil
W N
-
-
Nil Nil
N N -
0.03
0 4
1 628
780
1743
035
0.3
0.2 N i content in alloy(%)
Fig. 7.48 Effect of composition of alloy and deposit on corrosion, shown as weight loss (g an-*) against plotted points. (Shaded area shows most attack, Le. > O s 5 g ~ m - * ) ' ~ ~
7: 124
NICKEL A N D ITS ALLOYS
Seiersten and Kofstad’” point out that in simple laboratory tests at 650-800°C using sodium vanadate deposits the rate of attack on pure nickel was increased by a factor of about four to five while a plasma-sprayed Ni-Cr-AI-Y coating was even more severely corroded, and the corresponding Fe-Cr-AI-Y coating reacted even more rapidly. However, in dynamic burner rig testing, where continuous replenishment of the deposit occurs, the Fe-Cr-AI-Y coating exhibited corrosion resistance superior to that of Ni-Cr-AI-Y. These results were interpreted in terms of the ability of Ni-base materials to form a stable vanadate [Ni,(VO,),] so that in tests involving only one application of salt an increase in the Na20 concentration of the melt occurred making it more basic and hence less agressive to NiO. The corresponding iron vanadate is less stable. However, in the burner rig where continuous replenishment of salt occurs this limitation does not arise and the overall corrosion rate is controlled by the relative solubilities of the oxides in the molten vanadate. A survey of a broad range of Ni-Cr-Fe-Co alloys immersed in an 80% V,05 + 20% Na,SO, mixture over the temperature range 700-1 100°C revealed a major effect of chromium content in determining corrosion resistance, the level of chromium required increasing with increasing test temperature”. Again at this ratio of vanadium:sodium, attack was largely by the fluxing effect on the protective film of molten vanadates. Results for binary alloys are given in Fig. 7.49. The provision of the high (50-60%) chromium-nickel alloys in usable forms represents perhaps the most promising metallurgical approach to a solution of this particular corrosion problem so far 107-109. 12 0000 NI-30Cr
A NI - LO Cr
‘oooo-
-
NI - 5 0 C r X NI-60Cr
8000-
.
I
E
3 6000$ 0
0
I
3 L 000-
2 000 -
O
6!L
I Temperature I O C 1
Fig. 7.49 Effect of temperature on corrosion of nickel-chromium alloys exposed VZOS 20% NaZS04 salt mixture for 120 h9’
+
IO
an 80%
7 : 125
NICKEL AND ITS ALLOYS
Very recently Nicholls and Stephenson"' confirmed the beneficial effects of chromium additions in a comprehensive series of laboratory tests using synthetic ash deposits in which 75 different alloys were examined. They concluded that nickel- or iron-based material with chromium contents in excess of 25% offered the best resistance to attack in their tests. Cutler et al."' investigated the corrosion resistance of ferritic and austenitic steels in oil-buring power stations. There was a marked superiority of the ferritic materials particularly at gas temperatures of 1 150°C (see Fig. 7.50). The addition of lead compounds to petroleum fuels leads to attack by the combustion products on components, particularly exhaust valves and sparking-plug electrodes, of piston engines. Deposition of lead sulphate results in both sulphur attack of the type described in another section and corrosion more directly associated with lead itself. For the spark-plug electrodes, nickel alloys with manganese and silicon have proved very satisfactory from the corrosion point of view, while the use of a protective nickel-chromium alloy coating applied to steel exhaust valves by a welding technique has been established for many years. For high-performance engines, the valves themselves have been manufactured from a high-strength nickel-chromium-base alloy, but with increased operating temperatures further corrosion resistance has been required and the application of an aluminised coating has been found effective.
Fer r i t ic 11 50°C 800°C Gas temperature
-
c I
-E 3 E m ffl
0
0
c
1-
.-___500
----550
*
6 00
Metal temperature
650
("C)
Metal loss of austenitic and ferritic steels after exposure in oil-burning power stations as a function of metal and gas temperature (after Cutler er a/. " I )
Fig. 7 . 5 0
7: 126
NICKEL AND ITS ALLOYS
Carburisation and Attack by Carbon-containing Gases While carburisation itself is not a normal corrosion process*, in that there is no metal wastage, absorption and diffusion of carbon can lead to significant changes in the mechanical properties of the affected material and in particular to marked embrittlement. Furthermore, initial carburisation can produce an acceleration of the normal oxidation process, a phenomenon that is notable in nickel-chromium alloys. The question of the compatibility of metals and alloys with carbon and carbonaceous gases has assumed considerable importance in connection with the development of the gas-cooled nuclear reactor in which graphite is used as a moderator and a constituent of the fuel element, and carbon dioxide as the coolant. Tests of up to 1 OOO h on a series of metals and nickelcontaining alloys under pressure contact with graphite at 1 010°C showed that only copper was more resistant than nickel to diffusion of carbon and that the high-nickel alloys were superior to those of lower nickel content. The more complex nickel-chromium alloys containing titanium, niobium and aluminium were better than the basic nickel-chromium materials. Tests on a wide range of alloys at temperatures varying from 704 to 927°C have been made by Bernsen et d.”’to determine the temperature limits beyond which engineering materials carburise when held in contact with graphite. Table 7.27 lists the maximum penetrations of the carburised zones; while nickel in general showed no visible evidence of carburisation the associated hardness measurements indicated solution of carbon even at 704°C. At this temperature the chromium-containing alloys showed little tendency to carburisation, but at 816°C carburisation leading to the formation of chromium carbide was rapid. Of the carbon oxide gases, carbon dioxide is the less corrosive, leading normally to mild oxidation only and under certain conditions, e.g. low partial pressures of COz or in the presence of steam, to decarburisation of nickel alloys. Tests on nickel and nickel alloys in carbon dioxide at 1.38 x lo7Pa pressure at 704 and 816°C showed that very little increase in mass occurred in up to 1 OOO h even at the higher temperature (Table 7.28). The scales formed were similar in constitution to those obtained in air and the rate of scaling approximated to the parabolic law. In addition to the general scale formation, nickel exhibited intergranular oxidation while Inconel (Ni-16Cr-7Fe) and Nichrome V (Ni-200) showed localised pitting. Later work’I4 on Nimonic 75 (Ni-20Cr-O-4Ti-O.lC) at pressures of 1 atm or less has shown that oxidation predominates and was generally protective, but at low partial pressure the oxide film was less coherent and uniform, and mass losses due to decarburisation and evaporation of chromium from the specimen surface were recorded at 900-1 0oO”C. In the gas-cooled reactor, reaction.between the coolant and the moderator results in formation of a proportion of carbon monoxide in the atmosphere. This gas can be carburising to nickel-base alloys but the results of tests’” in which CO, was allowed to react with graphite in the furnace indicate that the attack on high-nickel alloys is slight, even at moderately high temperatures and is still mainly due to simple oxidation.
’’*
* See discussion
of ‘corrosion’ in Section I . I .
Table 7.27 Effect of time and temperature on the visible penetration of carburisation in nickel alloys under contact pressure (I-79N/mm2) with graphite"I
816°C
Material type and composition
Nickel
1.08(1)
10 kcal
1
: melting point, element, carbide, resp.
T, M' T : transition point, element, carbide, resp. B, boiling point, element.
Note: 1 kcal ~ 4 . kJ 2
a lower oxygen pressure than the dissociation pressure, the oxide cannot be formed as a pure phase. It also follows from the equations given above that an alloy of a metal will require a higher oxygen pressure to form the pure oxide, and conversely a pure metal can form an oxide solid solution or liquid slag at a lower pressure than that required for pure oxide formation.
THERMODYNAMICS A N D KINETICS OF GAS-METAL SYSTEMS
7: 151
Similar free-energy diagrams, which can be interpreted in exactly the same way, have been constructed for sulphides’, carbides4 and nitrides’ (Figs. 7.56 to 7.58). It is unnecessary to go to the lengths of calculating the oxygen or sulphur potentials of gas phases in order to use these diagrams in certain simple cases. Consider the oxidation of a metal by a hydrogedwater-vapour atmosphere. The reaction involved here is
M
+ H,O
+
MO
+ H,
Therefore = (Ah40 - A h ) - RT ~nP,,o/P,,
-A@
Fig. 7.58 Free energy of formation of nitrides (after Pearson and Ende’) KEY
@
+ 3 kcal suggested accuracies
Note: 1 kcal ~ 4 . k2J
f 10
f
kcal
> 10 kcal
M : melting point, element.
T,
0 : transition point, element, nitride, resp. boiling point, element.
7: 152
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
Thus when the oxidation of pure metal to pure oxide is being considered and the oxygen dissociation pressure of the metal/metal oxide is the same as that of the hydrogenlwater-vapour mixture. It is thus a practical advantage to have a scale around the edge of the diagram showing values of Apo2 (oxygen potential) for various ratios of pH20/pH2. For a mixture of hydrogen and steam at equal partial pressures, the oxygen potential will be equal to the standard free energy of formation of water vapour from hydrogen and oxygen at all temperatures. Therefore, the extrapolation of the standard free-energy line for the system 2H2 0, 2H,O intersects the H2/H20 scale at the right-hand side of the diagram at the ratio 1 / 1 . The point marked H on the left-hand side of the diagrams (Figs. 7.55 and 7.56) is the extrapolation of the same line to the absolute zero, and is thus equal to the standard heat of formation of water vapour from hydrogen and oxygen at 298 K. When the hydrogenlwater vapour ratio is 100/1, the point on the H2/H20 ratio scale representing this ratio is obtained by subtracting the chemical potential for a product molecule (Le. H 2 0 in the reaction H, +O, .+ H20) at an activity of 1/100 from the standard free-energy line and extrapolating the resulting line to meet the scale at the point marked 'l@/l'. It should be observed that the value of the chemical potential for any substance at an activity of 0.01 is the same as that for oxygen at a pressure of atm, and hence can be obtained from the diagram by using the oxygen potential scale. From these examples the construction of this scale is apparent, and, as a corollary, it should be noted that the oxygen potentials of CO/CO2 mixtures can be obtained by joining the point marked C on the left-hand side of the diagram, at the absolute zero of temperature, with the appropriate CO/CO, ratios marked on the scale at the right-hand side of the diagram. A similar scale for pHzS/pHZ, is attached to the sulphur diagram, and one for pcHI/& to the carbon diagram, etc. As examples, it can be seen from Fig. 7.55, by projecting the line which connects 0 and the MnO curve at 1OOO"C to the pol scale, that the dissociation pressure of manganese oxide (MnO) in contact with pure manganese is ~ - ~ ~ apressure t m of oxygen at a temperature of 1 OO0"C. Similarly, it can be shown from the diagram that MnO is reduced to pure manganese in an atmosphere consisting of hydrogen and steam in the proportions 1@:1 above lOOO"C, and in an atmosphere of carbon monoxide/carbon dioxide in proportions 105:1 above 1OOO"C. Referring to Fig. 7.56 it can be shown that the dissociation sulphur pressure, as S, molecules, of a mixture of copper and copper sulphide is 1OP8atmat about 900°C, and sulphide is formed on copper in an atmosphere of H,/H2S in the proportions 103:1at all temperatures below 720°C. Figure 7.59 shows the standard free energies of formation of metal chlorides as a function of temperature6.
+
+
-+
THERMODYNAMICS AND KINETICSOF GAS-METAL SYSTEMS
7: 153
IO
Fig. 7.59 Standard free energies of formation of metal chlorides as a function of temperature (after Villa6). Note: 1 kcal 4 . 2 kJ
Dilute Metallic Solutions The metallurgist is concerned with the formation of homogeneous solutions of small amounts of impurities in metals as well as with the formation of compounds. The limit of solubility of impurities is frequently very small, less than one atomic per cent in concentration, and in these dilute solutions Henry’s law is applicable, i.e. the activity of a dilute solute is proportional to the concentration of solute in the solution. Consider a dilute solution of an element A which has a high vapour pressure in the pure state at the temperature T, the vapour being monatomic, in solution in element B which has an immeasurably low vapour pressure at the same temperature. Then if the pressure of A could be measured unambiguously for a range of dilute solutions it would be found that p A = kx, = k ‘ ( A ) *
(7.13)
*The value of k ’ obviously depends on whether atomic or weight per cent is used for expressing the concentration of A .
7: 154
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
where X, is the mole fraction of A in solution and ( A ) is the atomic, or alternatively the weight per cent of A in solution. If the vapour species of A were di-atomic instead of monatomic, then /?'(A)'
PA2
(7.14)
which is Sievert's law for the solution of diatomic gases in metals. Similarly for a triatomic gas, e.g. SO, (7.15) pso2= kXsXA = k '(S)(O)Z In such a binary solution, the chemical potential of the solute ApA and that of the solvent ApB are related to the integral free energy of formation of the solution, AGS, per mole, containing a mole fraction x, of component A , and X , for component B, by the expression
+
A G S = xAA~A xBA~B Corresponding to the integral heat and entropy of formation of the solution are the partial molar heats AHi and entropies AS, of solution of the components where Ap, = AH,- TAS, AHs = X A N A XBAff, ASs xAASA xBASB
+ +
From the algebraic form of these equations it can be seen that the partial and integral values of a thermodynamic function for a solution are interrelated simply. Figure 7.60 shows the integral value of a function A Z for a binary solution, as a function of x,. At any given mole fraction of each component, the relevant values of the partial properties can be obtained by drawing a tangent to the integral curve at the given composition of the solution; AZ, and AZB are the intercepts of this tangent with the A-rich and B-rich sides respectively. It also follows that AZA is the change in the value of the function Z for the component A when 1 mol of A is transferred from the standard state, usually the pure substance, to an infinite volume of the solution of the given concentration, so that the concentration of each species in the solution remains unchanged during the operation.
A
0.2
0.4 0.6 0.8 MOLE FRACTION OF B
B
Fig. 7.60 Relationship between partial and integral properties
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7 : 155
In the dilute range of concentration the partial heat of solution, MA, of A would be constant and the partial entropy of solution would be given by
ASA = a
- Rlnx,
(7.16)
where a is a constant, and the standard state for A is pure A . At higher concentrations, Henry’s law no longer applies and activities must be substituted for the concentration terms. This statement implies that in the region of solution where Henry’s law is valid, the activity coefficient y of A , defined by = YAXA = yd(A 1 is constant. Further, if we choose pure A as the standard state for A then yA has a constant value whose magnitude depends on the chemical identity of A , whereas if the standard state is the infinitely dilute solution of A then 7; has the constant value of unity in the Henry’s law region. The equilibrium constant for, say, a dilute solution of sulphur in iron in equilibrium with an H,/H,S atmosphere can be simply expressed’ as
K = ( P H J P H ~ S ) x (VoS), when an infinitely dilute solution of sulphur in iron is taken as the standard state. Solutions in Solid Iron
The dilute solutions of elements in solid iron are, at present, the only system for which the thermodynamics has been reasonably well worked out experimentally. The remainder of this section will therefore be devoted to the diagrammatic representation of data for these systems which have been evolved by Richardson4. The heat and entropy of solution of a dilute constituent remain constant when the infinitely dilute solution is taken as the standard state, provided that the solute obeys Henry’s law and that no crystallographic change or change of state of the solvent occurs in the temperature range under consideration8. Thus within a given range of temperature in which the solvent remains unchanged, the partial free energy of solution of the solute may be represented on a free-energy/temperature diagram by a straight line. The intercept of this line with the free-energy ordinate at the absolute zero equals the heat of solution, and the slope gives the partial entropy of solution. However, when the stability of a dilute solution of a substance in iron is being compared with the stabilities of compounds, it is preferable to use the pure substance as standard state, in which case the slope of the free-energy line for the dilute solution of given concentration is given by equation 7.16. At the temperature at which the solvent undergoes a change in crystal structure there will be a discontinuous break in the line and, in the new structure, the free-energy line will have a different intercept at the absolute zero, indicating a change in the heat of solution, and a different slope which indicates a change in the constant a in equation 7.16 for the partial entropy of solution.
7: 156
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
IO' 0-2
D-J
10
D-'
I
3-5
lo-'
* z
t I-
u 4 0 t
1-6
10-2
2 4. u
3-7
0-3
1-0 0-4
0-5
Fig. 7.61 Carbon potentials in iron (after Richardson and Jeffes4). Suggested accuracy of all lines 1 kcal
*
By making use of experimental data for solutions of carbon in iron, Richardson has constructed a free-energy diagram showing the partial free energy of solution of carbon in a,y and 6, and liquid iron (Fig. 7.61) which is similar to the earlier Ellingham diagram9. The figure is divided by the broken lines into areas of constant crystal structure of the iron solvent, and each area is traversed by lines (dotted in Fig. 7.62) showing the partial free energy of solution of carbon, at a concentration shown on each line, as a
THERMODYNAMICS A N D KINETICS OF GAS-METAL SYSTEMS
109
: 157
3
''0
108 I
3-2
107
1-3
106 >
t
1-4
-> c
U 4
z
IO 5
0
m
az a u 1-5
I
IO'
1-6
103
102
-2
1-7 3
1-0
Fig. 7.62 Superposition of Figs. 7.57 and 7.61 (after Richardson and Jeffes4)
KEY
@I @ @
a3
kcal * 3 kcal f 10 kcal f > 10 kcal f 1
suggested accuracies
Note: 1 kcal 5 4 . 2 kJ
M: melting point, element. B: boiling point, element. T, [TI : transition point, element, carbide, resp.
7:158
THERMODYNAMICS AND KlNETlCS OF GAS-METAL SYSTEMS
function of temperature. The intercepts at the absolute zero, marked a,y, 6 , and liquid, give values for the partial molar heats of solution of carbon in each crystallographic form of the solvent. Segregation of Carbides
This diagram may be usefully combined with the standard free-energy diagram for the carbides (Fig. 7.57) to indicate the equilibrium conditions under which carbide particles will segregate from a given solution consisting of an alloying element and carbon in solid iron. Figure 7.62 shows how the diagram for carbon in pure iron and the diagram for the formation of carbides by metals at an activity of 0.01 can be superimposed. The combined diagram can be used to calculate, for example, the temperature below which particles of vanadium carbide can be expected to begin separating from an iron alloy containing vanadium at an activity of 0.01, with respect to pure vanadium as standard state, and carbon at a concentration of 0 - 1 wt%. Since only a small amount of vanadium is present in the alloy, the activity of carbon at this concentration will be the same as that in pure iron to a good approximation. Thus the solution diagram needs no amendment. However, the carbide diagram gives the standard free energies of formation of carbide from the pure substances and the alloy contains vanadium at an activity of 0.01. It is necessary, therefore, to draw a straight line joining that for vanadium carbide at the absolute zero, and spaced a distance equal to 19.15Tlog 0.01 above this line, across the diagram. (The values of this function can be read by joining the cross marked C on the middle left-hand side of the diagram with ‘carbon activity’ = IO-’ on the right-hand side of the diagram.) When this line is drawn, it can be seen that the line for [VI,,
+ c vc +
lies above that for 0.01% C in pure iron at temperatures above 840°C but is below it at temperatures lower than 840°C. Clearly then, carbide particles can begin to form as a separate phase only below 840°C in this alloy. At higher concentrations of vanadium in iron the carbide can form, with 0.01 wtVo of carbon, at higher temperatures, but not significantly higher until a large proportion of vanadium, raising the activity of vanadium by an order of magnitude, is present. Similar diagrams for sulphides and nitrides can be constructed from the data given here and the work of Rosenquist and Dunicz”, and Darken and Curry ‘ I . Effects of Large Amounts of Alloying Elements
The diagrams which have just been described are of only limited value because the presence of an alloying element in solution in the iron influences the thermodynamic behaviour of the solute. Thus it is well known that the solubility of gases in metals at constant pressure is changed by addition of alloying elements”, and since this is only another way of saying that the activity coefficient of the gas atoms in the solution has been changed, we
THERMODYNAMICS A N D KINETICS OF GAS-METAL SYSTEMS
7: 159
might expect this effect to pervade the whole field of alloy-solution thermodynamics. The direction in which a given alloying addition will change the activity coefficient of another dilute solute can be predicted semi-quantitatively along the following lines’j. Let us consider the simple case of a random solution of atoms A, Band S, where S is the dilute solute, in which the energy of binding of a given S atom in the solution may be obtained by assigning a fixed value to the energy of interaction between the A-S and B-S pairs which are formed and the number of A-B pairs which are broken when 1 mol of S is introduced into a large amount of the alloy, under the restriction that the alloy remains at constant composition during the process. It can be shown that
h-’ =X
A
h
+X B M !
-XAMf-’
-X B M t - ’
the solution being regarded as so dilute that S-S pairs are not formed. We may then approximate AHs = RTlny,
in which case In yt-’ = x, In y t
+ x, In y!
AG,,xs RT
--
where y?-’ is the activity coefficient of S in the A-B alloy of atom fraction of A and x, of B, y[ is the activity coefficient of S in solution in pure A, y! is that for S in solution in pure B, and AG,,xs is the excess free energy of mixing of the A-B alloy at this composition. The derivation of this equation has not been attempted here, and the interested reader should consult Reference 13 for further details. It can now be seen that if the A - 8 alloy is ideal, i.e. AG,_,xs = 0, then
X,
Fig. 7.63
Variation of log y s in A-E alloy. Curve (I, A-E solvent ideal; 6 , A-E solvent with AGXS negative; c , A-B solvent with AGXS positive
7 : 160
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
the activity coefficient of S will be decreased by addition of B if B interacts more strongly with S than does A , while it will be increased if B interacts less strongly than A with S. If the excess free energy of mixing of the A-B alloy is positive, or, as an approximation, if the heat of formation of the A-B alloy is positive, then ys will be decreased by the addition of B even when A and B have interactions of equal strength with S, and, conversely, ys will be increased if the heat of mixing is negative in these circumstances. The effects are shown diagrammatically in Fig. 7.63. This simple expression can be used to obtain only a semi-quantitative idea of the effect of an alloying element because the assumptions of randomness and a constant pairwise energy of interaction between atoms are only approximations to the truth in most systems. For quaternary and more complex alloys a suggestion of Chipman and ShermanI4 might be used. Chipman's school have made use of the symbol $ for the rate of change of In y of the dilute solute, C, with small additions '. Thus for the solution of carbon in iron: of alloying elements, A
a In y r Ec" = ___ ax, and it has been suggested that for small additions of several alloying elements to iron, the effect on the activity coefficient of a solute, such as carbon, can be obtained from the expressionLS
Values of $ which have so far been obtained experimentallyi6are shown below.
Alloying element $ = a In yt/ax,
Si 10
Cr -4.3
Mn
Mo
-0.5
-0.8
These apply to liquid iron as the solvent. The Segregation of Carbides from Stainless Steel Containing Small Amounts of Carbon
As an example of the way in which these data could be used, the temperatures at which carbides separate from an 18/8 stainless steel are calculated for carbon contents of 0.1, 0.01, 0-001 and 0-0o01 wt%. These calculations, which of necessity involve several approximations due to our present lack of knowledge, demonstrate the value of the thermodynamic approach to problems involving the precipitation of phases which may have a pronounced effect on the corrosion behaviour of the alloy (see Section 3.3). The steel will be considered to be an ideal ternary solution, and therefore at all temperatures a,, = 0.18, aNi= 0-08and aFe= 0.74. Owing to the y-phase stabilisation of iron by the nickel addition it will be assumed that the steel, at equilibrium, is austenitic at all temperatures, and the thermodynamics of dilute solutions of carbon in y iron only are considered. The effect of nickel on the activity coefficient of carbon will be neglected and the effect of chromium will be taken from the value in the liquid state. From the values quoted above, Ec", = - 4 . 3 at 1 600°C, and assuming that the effect of chromium is simply to change the heat of solution of carbon
7: 161
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
in iron then the point y on the left-hand side of Fig. 7.61 must be depressed by an amount
aAHH, = -4.576 x 1873 x 0.335 = - 12.02 kJ the last term being the change in log yc at 1 600°C when x,, = 0.18. Now in the case of chromium carbide separation from the steel, three possible crystal structures may be taken up, those of Cr,C, (or Cr&), Cr,C, and Cr& It is necessary first to calculate the free energies of formation of the compounds from pure chromium and carbon. The results are:
+ C -, 1/6Cr,,C, 7 / 3 0 + C 1/3Cr,C, 3/2Cr + C 1/2Cr,C,
23/6Cr
AGe = -16380 - 1.54Tcal A
+
AGe = -13900 - 2.05Tcal
B
+
A c e = -1OOOO - 1.39Tcal
C
-0 -1
-2 -3
-4
-5
-6 -7
-8
Fi - 9 Y
410
Y i a -11 -
12
- 13
B'(O.01:
A'
- 14 A'(O.01)
e' - 15 - 16 -17
'
- 18 - 19
-20 TEMPERATURE OC
Fig. 7 . 6 4 Carbide formation free-energy diagram for Fe- 18Cr-8Ni
7: 162
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
Since the chromium activity is 0-18 for the formation of carbide in the steel, each of the standard free-energy lines (A", Bo,C") derived for the carbides must be corrected (moved upwards) for this lower activity (A', B ', C '). Finally, the lines for the formation of these carbides at activities 0.01* [e.g. A' (0*01), B ' (0.Ol)l are shown in Fig. 7.64 and a summary of the results of the calculation is given in Table 7.36. It can be seen from the diagram that a phase consisting of Cr,C, at an activity of 0.01 [B'(O.Ol)] would segregate at 1 OOO, 720, 550 and 440°C from steels containing 0.1, 0.01,0.001 and 0-OOO1% carbon respectively. Phases containing Cr,,C, would be unstable with respect to this segregation at the temperatures stated, but would separate at an activity of 0.01 at temperatures of 710, 540, 440 and 370°C. A Cr,C, phase might appear at 0.01 activity at the highest carbon content and above 830"C, and below 1030°C. Now to complete the solution of the problem one would need to know the solution laws for iron, and a small amount of nickel, in each of these carbide phases, since equilibrium requires that uFcand uNiin the segregated carbide must be 0.74 and 0.08 respectively as well as acr being 0.18. At present nothing is known about these laws except that the metal atoms might well be randomly distributed in the carbide phase, in which case, as an example, acr7c3 = X7Cr7C3
in the ideal case. It is known that Cr,,C, can contain up to 25% iron on the metal atom sites, and Cr,C, up to about 60% iron',. Therefore the minimum activities of Cr,,C, and Cr,C, have been calculated for these phases of maximum iron content using the ideal laws to calculate activities from carbide composition. The free-energy lines which were thus obtained are shown as A and B '(mer.). The picture which emerges from this extremely simplified calculation is that a Cr,C, phase should always precede a Cr,,C, phase in segregation from stainless steel and that the latter should appear at a temperature of 780°C for the carbon content of 0.1 wt%. According to the phase diagram which has gained acceptance for this system, the Cr,C, phase never appears at low carbon contents, and a cubic phase of the Cr,,C, type separates at 900°C for 0.1% [C], and about 500°C for 0.01% [C]. These points, together with one for 0.3% [C] are shown by Table 7.36 Values of A~ICfor chromium carbides at 1 OOOK
Crz3Cs; n = 2316, rn = 1/6
- 17 940
Cr7C3; n = 7/3, rn = 1/3
- 15 950
Ci3C2; n = 3/2, rn = 1/2
- 1 1 400
+ 13 OOO = - 4 940 - 15 950 + 7 900 = - 8 050 - 11 400 + 5 060 = - 17 940
- 6 340
- 4 940 - 1 520 = - 4 460 cal - 8 050 - 3 060 = -11 110 cal -6340 - 4580 = - 10 920 cal
THERMODYNAMICSAND KINETICSOF GAS-METAL SYSTEMS
7: 163
the black dots on the diagram. An agreement between the calculated and measured temperatures and compositions for carbide segregation could thus be achieved only by strong departures from the ideal laws in the carbide phases. Alternatively it is possible that the separation of a Cr,C3 phase has not so far been observed because of rapid transformation in the solid state to the Cr& phase which is stable at lower temperatures. Such a transformation has been observed in the CY Fe-Cr alloys.
Concentrated Ternary Solutions When both solutes are present in large amounts, i.e. greater than about 1 at. Vo of each, no simple theoretical treatment is available to predict their mutual effects on thermodynamic properties. In this case, recourse must be made to the various solutions of the ternary Gibbs-Duhem relation
In order to make any practical use of this equation, a good deal of experimental data are usually required for a ternary system, and it will be found that, at present, such data are seldom available in the literature. The methods of evaluation of such data are fully described in the works of Chipman and Elliott and of Schuhmann”.
Thermodynamic Phase Stability Diagrams Pourbaix’s pioneering workz0 on the graphical presentation of gas-metal equilibria and the concept of stability zones and their boundaries between the various stable compounds lead to the second type of diagrams. Figure 7.65 shows a Pourbaix plot of the log poz of a system against the reciprocal of the absolute temperature for the Zn-0-C systems”. The stability zones under varying conditions of temperature, pressure and atmosphere composition are more completely defined than in the Ellingham diagram. However, the diagrams are considered to be more complex and therefore the object of this presentation is defeated unless the scale is greatly enlarged. Over the years, Pourbaix and his co-workers in the CEBELCOR Institute, founded under his direction, extended these diagrams by including lines for metastable compounds”. Figure 7.66 illustrates such a presentation for the Fe-0 system over the temperature range 830-2200K. Pourbaix used these diagrams as a basis for a discussion of the stability of metallic iron (solid, liquid and vapour phases), the oxides of iron as a function of oxygen pressure and temperature from which he explained the protection of iron at high temperature by immunity and passivation. He also pointed out the
..
4
I - x T 2.0
2.5
+10
L
103
Q\ P
1.o
1.5
0.5
+10
0
0 10
-10
> z -20
-20
-30
-30
-40
-40
-50
-50
0
Q 0 0 I
v) .(
I
2.5 I
I
I
I
I
100
150
200
250
300
Fig. 7.65
I
Y
1.o
1.5
2.0
I
400 500 Temperature "C
1
700
0.5
0
I
I
I
I
1000
2000
5000
x.
Equilibrium in the Zn-0-C system (after Pourbaix2")
f
v)
-0
02
I
Em
m
-
0" a
0"
0 OI
a
&
0 _J
-5
01
-5
w -10
-10
E
> z
0
4-
m
E
z m
g
m -15
-I
-1 5
/7'
-20 - F e 3 0 7 a
=!
c1
-20 Fe m
w
-
:-4
-2 5
Fig. 7.66 Equilibrium diagram for log poZ = f(1/T) for the Fe-0 system (between 830 and 2 200 K ) (after Pourbaix*')
v1 4
Y
f
v1
7 : 166
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
value of these diagrams in the fields of geology, metallurgy, corrosion and catalysis among others. A convenient way of representing the thermodynamic information for a given system is by means of isothermal phase stability diagrams which show the ranges of gas compositions over which a condensed phase can exist either singly or in equilibrium with another condensed phase or phases. Kellog and Basu2, for Pb-S-0 and Ingraham23for the Ni-S-0, Fe-S-0, Cu-S-0 and Co-S-0 systems, pioneered the use of such diagrams by considering the relative stability of condensed phases in these systems. A detailed explanation of the construction of thermodynamic phase stability diagrams may be found in References 22-25. In this section the basic principles of construction and interpretation for the specific situation of gas-metal equilibria will be addressed using a hypothetical system. Construction of Phase Stability Diagrams
The method of construction of this type of diagram will be illustrated using the general case of the three component system metal-sulphur-oxygen (M-S-0) whose values of AGF for the reactions between the various condensed phases are given in Table 7.37 on page 7:191. Assume that at the isothermal temperature of interest the following stable condensed phases (solid or liquid) can be formed: M , MO, MS, MSO,. From the Phase Rule it is clear that the maximum number of condensed phases in contact with each other can be three, in addition to the gaseous phase (SO, and OJ. Following the suggestion of Kellog and Basu2,, the
+4
0
t
N
5: a
-4
[5,
0 I
-8
-1 2
-20
-16
-12
-8
-4
0
Log Po2-
Fig. 7.67 Phase stability diagram for a metal-sulphur-oxygen (M-S-0) system at 1 OOO K. (For the thermodynamic data AG? wo for the various across-boundary reactions, see Table 7.37)
Table 7.37 Data for the construction of thermodynamic phase stability diagram M-S-0 at 1 OOO K
+ fO2PM0
M/MO
M
M/Ms
M
Ms/MO
+ S0,PMs + 0, Ms + f 0 , P M O + so2
MS/Ms04
Ms + 202*MS04
M0/MsO4
MO
*Assumed data of Ace,
+ soz+ io, = MSo4
parallel to Y axis 1
independent of
- 153.2
- 16
NA
-
16
1
+ 229.8
+ 12
-4
-
16
2
-
- 38.3
- 20
-4
-
16
3
-
parallel to Y-axis
- 459.6
- 24
NA
- 12
4
independent of
- -I
- 16.6
-4
+2
- 12
5
3 2
PSS
ps02
for the purpose of illustrating the calculation of the position of the boundary lines and triple points A and B see Fig 7 67
-
9
z
tl
7 : 168
THERMODYNAMICS AND KINETICSOF GAS-METAL SYSTEMS
phase stability diagram may be constructed by plotting the sulphiding potential logpso2along the vertical axis and the oxidising potential logp,, along the horizontal axis as in Fig. 7.67. The position and the slopes of the boundary lines between the areas of stability for the condensed phases of the system are then calculated from the appropriate chemical equations describing the reactions which take place when one condensed stable phase reacts to form the other phase. The only thermodynamic data required are either the standard Gibb’s free energy change at the chosen temperature (AG;) or the equilibrium constant for the reaction at the given temperature ( K , ) . Usually, the most convenient boundary to calculate first is that between the pure metal ( M ) and the metal oxide (MO), i.e. the M / M O boundary since it will be parallel to they-axis. Using the balanced reaction for the formation of the oxide
M+fO,=MO
at IOOOK
and the relationship between AG; and K, gives: AGP,
= -RTlnK,
, = -RTln--~a!fA. aMp02r
For pure M and MO by definition a,,,,= a,,,,, = 1. Converting Inp,, to logp,,, rearranging terms and substituting values for AGPm (Table 7.38), R and T, the following is obtained:
- 153.2 x
103 = 19-15 x 103 x ;iogpoz
or logpo, = - 16 Since pso2does not take part in the reaction, the boundary line between M and MO is independent of logpsoz and so given by a straight vertical line at logp,, = - 16, parallel to the y-axis (line I in Fig. 7.67). It should be noticed that stability areas across the boundary follow the sequence of condensed phases shown in the equation, i.e. on the left-hand side of the boundary pure metal is the stable phase and on the right-hand side the pure metal oxide. To determine the position of the boundary between A4 and MS the following chemical reaction is used:
M+
so,=MS + 0,
For pure M and MS aM= am = 1 Using equation 7.12: AGY,
=
-19-15Tlogh
=
19-15T(logpso, - logp,,)
PSOl
Rearranging:
+
This is the equation of a straight line of the form y = mx c when logpso2 is plotted against logpoZ,where y = logp,,,, m is the slope which here is
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7: 169
unity, x = logp,, and c is a constant which in this case is AGP,/19- 15T. Substituting values for AGO,, = 229.8 x lo3 and logp,, = (where the previous line and the new line intersect) gives: 229.8 x lo3 l0gps02 = 19.15 103 - 1 6 = -4 Thus, the coordinates of the point of intersection ( A in Fig. 7.67) of the two boundary lines M/MS and M / M O are logp,,, = - 4 and logp,, = - 16. These calculated data are now sufficient to draw the boundary line between the stability areas of M and MS. This is constructed by drawing a straight line having a slope of 1 from the point of intersection A . Next, the position of the boundary between MS and MO can be calculated from the reaction:
+
+ + o , = M o +so,
MS Thus,
AGp,
= - 19. 15Tlog, Psoz
-
Po22
which, on rearranging, gives
+
The slope of the line is t and the line is drawn from the point A to the MS/MSO, boundary to be determined next. The boundary between the MS and MSO, stability areas is calculated from the reaction:
MS
+ 20, e MSO,
Thus
or -459 x 103 logPo2 = 19-15 x lo3 x 2
= -12
Since SO, does not take part in the above reaction, the boundary between MS and MSO, stability areas is independent of logp,,, and is given by a vertical line (4) at logp,, = - 12 parallel to the logpso2axis. The intersection point of the MS/MSO, line (4) with that of M O / M s boundary line (3) at point B of Fig. 7.67 completes the stability area of the Ms phase (lines 2, 3 and 4). Finally, the boundary between MO/MSO, is calculated from the reaction: MO
+ SO, + + O,=
MSO,
7 : 170
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
Thus
AGP,
1 = -19.15Tlog1 PS0,P
8,
or
Since this line will start at point B of Fig. 7.67, Le. when the values of = -76-6 kJ, one can calculate the value of logp,,, for the point B by substituting these values in the previous equation:
po, = - 12 and AGP,
Therefore, the calculated coordinates of the triple point for the coexistence of MO, MS and MSO, are logpso, = + 2 and logp,, = - 12 and the slope of the MO/MS04 boundary is - t The straight line from point B having slope - t gives the boundary line (5) between the stability areas of A40 and MSO,. This completes the construction of the phase stability diagram for M-S-0 at 1000K. The stability phase diagrams contain a wealth of information. Using some selected examples from the literature it is intended to show their range of application in the field of corrosion.
.
Control of
Gas Composition for Surface Stability
Many industrial applications of materials at elevated temperature involve their exposure to complex gas mixtures. Usually it is assumed that the main oxidising species control reaction rates26 by forming protective oxides, whereas the formation of sulphides, chlorides etc, which may be solid or liquid, can be detrimental to the performance of the material. In practice, using steel as an example, the partial pressures of oxygen (poz),sulphur (p,,), halogenic gases and the activity of carbon (ac) are controlled by establishing the relevant safe gas equilibria to prevent sulphidation and carburisation of the steel. It is relatively simple to obtain graphically from AGO/ T diagrams and their nomographic scales the ratios of binary equilibrium gas mixtures H,/HCl, Hz/H,O, CO/C02 and CH4/H, in contact with a particular metal or condensed phasez7.However, in multicomponent atmospheres it may be necessary to take advantage of specialised computer data banks and the iterative routines such as MTDATA in use at NPL2* and their facilities for automatically plotting the phase stability diagrams for metals and alloys relevant to the temperature and gaseous conditions of interest. Many such centres are available to outside contract5'. In Fig. 7.68 the oxidising and sulphiding potentials of four different atmospheric environments, i.e. conventional coal combustion (A), fluidised bed combustion (B), conventional coal gasification (C) and coal gasification using nuclear heat (D), are shown on the thermochemical phase stability
7: 171
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
5
0 -5 1
-E
-10
N
m
Q
-15 -I
-20 -25 -30 -30
-1 0
-20
10
0
Log po2 ( a m )
Fig. 7.68 Thermochemical stability diagram for the system Fe-S-0 at 1 OOO K showing the relative corrosion potentials of the atmospheres in conventional coal combustion (A), fluidised bed combustion (B), conventional coal gasification (C) and coal gasificiation using nuclear heat (D) (after Gray and Starr”)
-5
-
~
Cr2S3 + FeS + NzS CrzS3 + FeS
/ FeCrp04
N
0
-I5k
-20
FeZ03
+
Cr203
-30
-2 5
-2 0
Log pO2 (atm)
Fig. 7.69 Thermochemical stability diagram for the 3 10 stainless-steel-S-0 system at 750°C (from Gray and Starr after Natesan and Chopra 30)
*’
7 : 172
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
diagram for the Fe-S-0 system at 1 OOO K2’. From the position of the four areas (A, B, C, D) shown on the diagram, the aggressiveness of the environment can be seen as it increases with decreasing po2 and increasing psz. Thus the corrosivity of the different atmospheres increases from process A to D. However, in such an evaluation of the aggressiveness of atmospheres it is necessary to take into account the differences in the feedstock and other process conditions employed 29. Figure 7.69 and 7.70 show the phase stability diagrams at 1023 K for the 310 stainless-steel-S-0 system and that for the Cr-0-S system relevant to Incoloy 800H alloy, respectively. Comparing these diagrams it is apparent that the boundary between steel AIS1 310 and Cr,O, is at a slightly lower oxygen potential (about - 27.5) than that between Incoloy 800H which has the same boundary at - 2 6 . 5 oxygen potential. However both these diagrams illustrate how convenient it is to obtain the composition of the atmosphere at the isothermal temperature within which each particular phase may be formed. Obviously, this information is particularly useful when assessing the critical gas composition at which the protective Cr,O, oxide can be expected to be stable. However, it has been observed that at low oxygen pontentials the gas compositions must be made more oxidising by about two to three orders of magnitude than those predicted by the equilibrium values; this is possibly because of kinetic effects”. Figure 7.70 gives the additional kinetic phase boundary separating the stability area of Cr,O, and the adjoining stability areas. This observation was confirmed from XPS
-6 -
Thermodynamic
Kinetic boundary
-8 E
--m
Alloy -14
I
-30
I -2 5
I -20
-15
Log p (02) (atm) Fig. 7.70 Phase stability diagram for the Cr-0-S system on Incoloy 800H at I 023 K showing thermodynamic and kinetic boundaries (after Natason ’ I )
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7 173
Boundary shift indicated by XPS
+
-Composition of gas for heat treatment
0
20
10
30
Time (m)
Fig. 7.71 An activity diagram showing the competing formation of sulphides and oxides on chromium. The XPS data (lower) show how sulphide replaced oxide as the surface anion when oxide samples were heated in the gas composition marked on the 0-S diagram, implying that the boundary should be moved. (Reprinted with permission from Pergamon Press; after Huang 33)
work where it was established analytically3’ that there was a definite boundary shift to a higher oxygen potential (Fig. 7.71). Phase Stability Diagrams with a Liquid Phase
There are many examples of these diagrams being used to predict or assess the disruptive effect of a liquid phase at different oxygen potentials on the protective properties of oxides. In Fig. 7.72 the stability of M,O, is shown in equilibrium with liquid Na,SO, as a function of the oxygen activity and the acidity of the liquid salt, at 1 273 K. This stability phase diagram shows that the oxygen potential boundary between basic fluxing and the stable
7 : 174
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
-3.5 0
-E +m
-4
-19.5
-7.5 I
Basic fluxing
I
7
Oxide stable
Acid fluxing
N
ul
~13+
Q
0 0
-,
-a
-1 2
A102-
I -8
-4
0
Fig. 7.72 A thermodynamic phase stability diagram for A l - 0 - S species in the eqilibrium with liquid Na2S04 at 1 O00"C as a function of the oxygen activity and the acidity of the salt (after Stringer 34)
oxide is loga, = -8 approximately, whilst acid fluxing and the stable oxide is loga, = 17 approximately. In recent work3' phase stability diagrams were used to evaluate the effect of molten Na,SO, on the kinetics of corrosion of pure iron between 600°C and 800°C by drawing a series of superimposed stability diagrams for Na-0-S and Fe-0-S at 600"C, 700°C and 800°C and thus to account for the differences in the corrosion behaviour as a function of temperature. Phase Stability Diagrams and the Formation of Volatile Halides
Another problem in high-temperature corrosion can be the effect of the formation of volatile metallic halides which can, in turn, disrupt the integrity of a protective surface oxide. Figure 7.73 shows that in the Ti-0-Cl system at very low oxygen potentials, volatile TiC1, can be formed directly from T i 0 and Ti, whereas from Fig. 7.74 it is clear that in the system U-0-Cl at 450°C the volatile chloride cannot be formed directly from the oxides. Phase Stability Diagrams for Two or More Metals
These isothermal diagrams can be used to consider the phase stability areas for more than one metal in contact with a common atmosphere and thus to assess the condensed phases which can be stable under the prevailing conditions. Figure 7.75 sh0ws.a stability diagram having phase areas for Co-S-0 (solid lines) and for Cu-S-0 system (broken lines). From this diagramz3it can be seen clearly that at 950 K at certain gas mixtures, pure metals C o and
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7: 175
394°C
0
----8 =.Io-lO------
N -
V
a
cn J 0
-24
Arbeitsbereich:
-32 Ti
-40 -1 00
Ti0 ‘- T i 3 0 5
1
I
-80
-60
-40
I -2 0
Fig. 7.73 Composition ranges in the Ti-0-CI system at 394°C (after Knacke’’)
Log p o p (atm)
Fig. 7.74 Composition ranges in the U-0-CI system at 450°C (after K n a ~ k e ’ ~ )
7 : 176
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
+8
+4
/
',
cuo.cuso~
0
0"
v)
a
0)
0 -I
-4
-8 /
co
I
I I
I
-1 2
-1 6
-20
I
I
I
I -8
-1 2
-4
0
Log Po,
Fig. 7.75 Superimposed predominance area diagrams at 950 K for the Co-S-0 system (solid lines) and the Cu-S-0 system (broken lines). Within the area A , Cos04 and CuO are the stable phases (after IngrahamZ3)
0 ,
E
c
-10
-
m
1
N
cn
Q
0,
I
----i
0 A
-20
-
I
I I
I I I
I
1
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7: 177
5
0
5 N
In
a
-10 -2 5
-1 5 Log PO (N m-2)
-5
Fig. 7.77 Thermodynamic stability diagram for the Fe-Ni-Cr system at 1 143 K, assuming metal activities to be unity. ----, phase boundaries involving Fe; ----, phase boundaries involvphase boundaries involving Cr. The location of environments 1 , 2, 3, and 4 are ing Ni; -, indicated by X (after Stott and Smith”)
Log p 0 2 (atm)
Fig. 7.78 Thermodynamic stability diagram for some oxides and sulphides at I OOO K (after
Lions”’)
7 : 178
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
Cu will be unaffected by the atmosphere, whilst at other atmospheric compositions the pure oxides will be stable. Figure 7.76 illustrates a simplified diagram34at 871°C for three metallic elements Cr/Mn/Ni-S-0 in a heatresisting alloy; the range for coal gasification is also included. It is clear that Cr20, is stable in all these atmospheres, but NiS will be stable under these atmospheric conditions above 62OOC in the form of a eutectic liquid with Ni. Thus, an alloy of Cr and Ni may produce either of these phases or their mixtures leading to corrosion problems. Figure 7.77 shows a diagram for the three metals Fe, Cr, Ni as a function of sulphide potential against oxygen potential. This diagram has been used to select atmospheres in the study of high-temperature corrosion in which relatively small changes in oxygen and sulphur have a marked effect on the kinetics of corrosion, scale morphology and scale composition of 34Fe39Ni-27Cr alloy ingots. The atmospheres selected for the study are shown in Fig. 7.77 as X1, X2, X3 and X4. Figure 7.78 shows the stability diagram at 1 OOO K for AI, Ti, Si, Cr and Fe sulphides at oxidising potentials between, logp,,, - 10 and -50 and suphiding potentials, logp,,, between 0 and
log p0, (atm)
Path 1
log p0, (atml
log p0, ( a m )
Path 3
Path 4
Fig. 7.79 Five possible reaction paths on a schematic thermodynamic phase stability diagram, and the corresponding distribution of phases in the reaction systems (after Stringer 34)
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7 :179
-25. Isobaric lines for logp,, ranging from 0 to -40 are also included in the diagram as straight lines3*. The diagram was produced for a study aimed at finding improved materials which would be immune to sulphur corrosion and lead to the increased efficiencyof thermal and nuclear power stations. From Fig. 7.78 it is clear that the oxides are unstable under high sulphur pressure and very low oxygen pressure. It is also clear that the formation of SOzhas to be taken into consideration as the reaction between sulphur and oxygen significantly lowers the oxygen activity; under high sulphur pressure and low SO2pressure only some oxides are stable (AZO,, TiO, and SO,), and the oxides of Fe, Co, Ni and perhaps Cr are decomposed.
gas
I
I I
A
1.
I
2.
I I
A
3.
A A 4
Fig. 7.79 (continued)
A
5.
1
7 : 180
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
Phase Stability Diagrams and the Sequence of Phases in Surface Interaction Layers (Reaction Paths)
A number of author^^^'^ have indicated how the stability diagrams may be used to suggest or explain the possible sequence of phases in surface interaction layers formed during gas-metal interactions. Stringer and Whittle4' suggested the concept of the 'reaction path' on the stability diagrams which enables prediction of the sequence of phases in surface layers formed when the activity of the oxidant follows a certain path through the various stability areas. Figure 7.79 shows five possible paths in a stability diagram for the A-S-0 system. Stringer34,using the reaction path graphical method in the A-S-0 system and the five possible paths shown in Fig. 7.79, accounted for the following sequence of interaction phase layers: ASO,/AO/AS/A, AS04/AS/AO/A, ASO,/AO/A or AS04/AS/A and ASO,/AO/AS,AO/A.
The slopes of the reaction path lines, between the point P (giving oxygen and sulphur activities in the gaseous atmosphere) and Q (initial oxygen and sulphur activities in the metal A), are determined by the relative diffusion rates of the species in the phases. From the diagram it can be seen that small differences in slopes can result in significantly different distributions of phases. Stringer34points out that only the lack of precise knowledge of diffusion coefficients prevents accurate calculation of reaction paths, and therefore these diagrams, at the moment, are more useful for the interpretation of oxygen and sulphur potential from the observed phase distributions than for predictive purposes. It is also clear that small changes in the position of points P and Q can have a significant effect on the phase distribution in the surface layers. From the diagrams it is also seen that, when the metal A is saturated with oxygen and sulphur, and therefore the point Q is located at the corner of the rectangle giving the stability area of the metal A , then the innermost phase layer will consist of a mixed sulphide and oxide layer. C 1
\
\
',
AC
a,
0
A0
Fig. 7.80 A schematic thermodynamic phase stability diagram for the A-C-0 system, showing three reaction paths. Paths 2 and 3 are only possible if gaseous diffusion in pores in the oxide product results in a carbon activity increase through the scale, as shown in Fig. 7.81 (after Stringer 34)
THERMODYNAMICS AND KINETICSOF GAS-METAL SYSTEMS
7: 181
Figures 7.80 and 7.81 illustrate the use of a reaction path graphical method for systems where the surface oxide is porous. Figure 7.80 shows the phase stability diagram for a metal-carbon-oxygen (A-C-0) system. Considering reaction path 1 between P and Q it follows that only the metal oxide A 0 could be formed. However, if the oxide is porous, gaseous molecules of CO, can now penetrate to the metal surface (see Fig. 7.81) and thus, following the reaction CO, A A 0 CO, there must be a gradual increase in the CO partial pressure towards the m e t a l h e t a l oxide interface within the porous oxide. Figure 7.80 shows the effect of the change of the CO/CO, ratio on the carbon and oxygen activities. If the carbon activity rises high enough (see reaction path 2) carburisation may be possible, or even carbon deposition if a, exceeds unity, as shown in path 3 Fig. 7.80.
+
C02 + A + A 0
+
+
+ CO
Fig. 7.81 A sketch illustrating how gaseous diffusion processes in pores within an oxide layer can result in an increase in the CO/COz ratio, and hence the carbon activity, through the layer (after Stringer 34)
Phase Stability Diagrams and the Effect of Temperature Figure. 7.82 shows a three-dimensional phase stability diagram for the Fe-S-0 system between 800 and 1 OOO K. These diagrams are obtained from a knowledge of the variation of A@, for the different reactions which describe the appropriate phase boundary. In general, changes in temperature may have a significant effect on the areas of stability. These may become larger or smaller as the temperature is increased. A detailed description of the method for their production and interpretation may be found elsewhere23s42s43.
Integral Free Energy-Concentration Diagrams As mentioned earlier, this type of diagram may be useful for the quantitative thermodynamic assessment of gas-metal systems which form non-stoichio-
7: 182
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
Fig. 7.82
Predominance volume diagram for the Fe-S-0 system for the temperature range 800-1 OOO K (after Ingraham *3)
metric condensed phases and which are sensitive to relatively small stresses in the condensed phases or small changes in chemical potentialsa, in the gaseous phase and other physical or chemical effects that can be expressed in terms of energy"' (e.g. irradiation, vibrations, gravity, grain size, penetrating liquids, variation of surface tension, magnetic effects& etc.). It will be shown that any effect that slightly alters the relative position of integral free energy-concentration curves may have a drastic effect on the equilibrium and disturb the stability or the composition of the condensed phases. In this sub-section it is intended first to outline the theoretical basis of these diagrams by considering a simple metal-A-gas4 binary system followed by a quantitative treatment of a hypothetical metal M(at. wt. 50) and oxygen binary system. Finally the application of these diagrams will be illustrated using the Ti-C, Fe,,,-Zn,,, and Fe,,,-Zn,,, systems. The key to an efficient use of these diagrams is the understanding of the properties of a common tangent to two or more free energy-concentration curves and in particular the information which may be obtained from the socalled 'tangency rule'. It is therefore intended to develop the tangency rule using a simple isothermal binary system of a pure solid metal A and a pure gas B at 1 atm pressure having two non-stoichiometric solid compounds denoted by phase I and phase 11. Figure 7.83 shows the AG,-concentration diagram for this system. The vertical axis y, represents the isothermal free energy changes (AG,) - which are obtained when one mole of a mixture
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
0
X?
1.0
Mole fraction of gas B
XB
X1
7 :183
Fig. 7.83 The graphical method for obtaining equilibrium compositions from free energy vs. composition curves at a given temperature. Points of contact give equilibrium compositions X I for phase I and X Z for phase I1
having a given composition is formed. The horizontal axis gives the composition of the mixture in mole fraction, of the gas B. On the left-hand side of the scale we have X , = 1 (Le. pure metal) and the concentration of the gas X , = 0, whereas on the right-hand side of the scale X , = 0 and X , = 1 (Le. pure gas B at 1 atm pressure) since X,, X , = 1. The integral free energy changes with concentration for phases I and I1 are shown as AG, and AG,l curves, respectively. Using the diagram it is now possible to predict quantitatively the equilibrium composition of the two condensed phases when in contact with each other, Le. when growing as solid layers on the surface of the metal A . According to G i b b ~in ~ a~ two-component , system any condensed phases at equilibrium will have to satisfy simultaneously two energy conditions, namely (1) the mixture of the phases will acquire the lowest overall free energy, and (2) the chemical potential (or partial molal free energy) of a particular component must be the same in all the phases that are in contact with each other. These two conditions are now sufficient to predict the exact composition of condensed phases at equilibrium with each other. Gibbs’ definition of the chemical potential47of the gaseous component B in a mixture at constant temperature T and pressure p is given by
+
Therefore this partial differential represents mathematically the tangent to any AG-concentration curve. In our case for a common tangent to AGI and AGIl we must have not only a common slope -dYl = - =dY2 &I
&2
but also a common intercept on the AG,,, axis. Thus the equation for a common slope and intercept in Fig. 7.83 must be of the form
7 : 184
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
- y 2 = m ( x , - x 2 ) . It can be shown4' that the intercept of this tangent on the y axis where X , = 1 (pure gas B at 1 atm pressure) gives the for 1 mol of B dissolved in a mixture of AB numerical value of ApB = having a composition of the point of contact of the common tangent to any AG,-concentration curve. Similarly, the intercept of the same tangent on the y-axis at X , = 1 (pure metal) gives the numerical value of ApA = A c , , Le. the chemical potential (or partial molal free energy) of 1 mol of metal A dissolved in the phase having the composition given by the point of contact of the tangent to the AG-concentration curve. In Fig. 7.83 using the common tangent construction the equilibrium compositions of phase I-phase I1 at their boundary are found, from the points of contact, to be respectively X , = X , and X , = X,. Figure 7.84 shows a more complicated isothermal, binary system consisting of metal M and gaseous oxygen 0, at temperature of 1 000K. The free-energy-concentration diagram for the system shows four condensed phases, the first being a solid solution of oxygen in the metal, followed by non-stoichiometric condensed phases of nominal compositions M,O", MO" and MO;. Using the tangency rule, it will be shown that each of these oxides must have a region of homogeneity over a range of composition in which it will be the sole stable phase. In Fig. 7.84 the vertical axis represents the isothermal free-energy change associated with the formation of one mole of the M-0 mixture of a given composition, expressed in mole fractions of the metal (X,) and oxygen ( X o ) , shown along the horizontal axis. When a pure solid metal M is in contact with a gas containing oxygen, at first a solid solution of oxygen in the metal is formed. The Gibbs free energy of mixing (AG,) for the corresponding concentrations of oxygen in the solid solution are shown by the curve a-b-c. Note the section a-b of the curve is the only part which can be determined experimentally, whereas the section b-c, representing a supersaturated solution is either extrapolated from the a-b section or calculated theoretically. The values of the free energies of mixing, producing any possible phase, can now be calculated using computer techniques49 in conjunction with the appropriate thermodynamic data coupled with the relevant phase diagram. It is worth noting that there are already a number of powerful programs" which, in conjunction with stored thermodynamic data, can be used to calculate theoretically these curves for an ever-increasing number of binary, ternary and even quaternary systems. Once the solid solution is produced a surface layer of M,o" oxide phase will be formed, having an excess of the metal. This new phase has a separate AGMZw-concentrationcurve shown in Fig. 7.84 by d-e-f-g. This curve is followed by that of AG, phase shown by h-i-j-k. Finally a layer of MO; will be formed, and its free-energy-composition curve is shown by I-m-n-o. Applying now the tangency rule, by drawing common tangents to neighbouring AG,-concentration curves, the range of stability of th_e oxides is determined. As the chemical potential of the oxygen Apo or AGO is fixed by the oxygen pressure in the atmosphere, the equilibrium composition of the MO; oxide layer exposed to the atmosphere is obtained by drawing a tangent to the AGM0; curve with an intercept on the AGmaxis when X o = 1 equal to the value of the oxygen chemical potential of the atmosphere (Le. in our case = tRTlnp,, where pol is the oxygen pressure in the y,
Ac,
+
50 +wt d
wt g metal M 45 40 g oxygen1.6 3.2
-
35
30
25
20
15
10
5
0
4.8
6.4
a
9.6
11.2
12.8
14.4
16
1000 K
_--
/ I
1-t
h p o = AGO = -34 kJ for oxygen - metal dioxide
-20
--L
-30
1 1
Apo = AGO = -87 kJ at equal oxygen chemical potential between M20 - MO boundary
1-
-70
AGm
-80
I
I
-90 A p o = A G o = -139 kJ
-100
oxygen chemical potential between ss - M 2 0 at equilibrium
-110 -120 -130 -1 40
kJ
..
4
XO -mole
fraction of 0
Pure oxygen at 1 atm
Fig. 7.84 Free energy diagram for a binary system consisting of metal M and gaseous oxygen 0,at a temperature of 1 OOO K
7: 186
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
atmosphere - here 1 atm and pure 02,therefore Ap, = 0). Thus a tangent from the point AGO = 0 to AG,,; gives the equilibrium composition. The mole fraction of oxygen in the MO; can now be read off the composition axis. The point of contact, n, gives the oxygen equilibrium composition X , = 0-7inMnO;. ThemolefractionofthemetalMisX,= 1 - X , = 0 - 3 . Therefore the equilibrium composition of the outermost oxide layer will be Mo,300.,with respect to an atmosphere consisting of oxygen at 1 atm pressure. The common tangent between AG,,; and AGMox gives points of contact for MO; X , = 0-675 and X , = 0.325 and X , = 0.55 and X , = 0.45 for MO" (see Fig. 7.84). Similarly the equilibrium compositions on the Mo" and M20X boundary are found by drawing a common tangent to the AGMox and AGM2,. curves. The points i (on the AG,,. curve) and f (on the AGM,,, curve) give the equilibrium compositions M o . 5 2 s o o . 4 7 , and M o . 6 7 5 0 0 . 3 2 5 for the two oxides at their boundary. Finally the equilibrium compositions at the boundary between the solid solution and the adjacent oxide layer (M,o") are found by drawing the common tangent to AGMzOx and AGss. Points b (on the AGss curve) and point e (on the AG,,,, curve) give the respective equilibrium compositions. Thus, the maximum solubility of oxygen in the metal is found from the point b to be X , = 0.045 and the composition of the MzOxin contact with the saturated solid solution is i b f 0 . 7 3 @ 0 . 2 6 ~ . Closer examination of Fig. 7.84 shows that each of the non-stoichiometric oxides has a region of homogeneity over which the compound is the sole stable phase. It has been observed, from a number of gas-metal systems, that the lower oxides (here M,O" and Mo")usually show a wider region of nonstoichiometric behaviour than the higher oxides (here MO;). Regions of Homogeneity of Non-stoichiometric Oxides in the Surface Interaction Layers and the Effect of Oxygen Pressure on their Range of Stability
In Fig. 7.84 each oxide has two points of contact produced by common tangents. These two points predict the range of composition within which each of the oxides is the sole stable phase. Thus the composition of MO; oxide will vary from an oxygen mole fraction X , = 0.7 on the surface of the oxide exposed to the oxygen atmosphere to X , = 0.675 at the MO;/Mo" boundary. It is also clear from the diagram that as long as the oxygen chemical potential remains between Ap0 = 0 (Le. po2 = 1 atm) and Ap, = -34kJ (Le. pol = 2.8 x 1OV4atm)the outermost surface oxide layer will consist of an MO; oxide phase. However, as the pressure of the oxygen is lowered to between 1 and 2.8 x 10-4atm the equilibrium oxygen contents in the MO; surface layer decrease predictably from X , = 0.7 to X , = 0.675. The exact equilibrium concentration of oxygen in the MO; oxide in contact with the gas phase can be obtained by first calculating the oxygen chemical potential in the atmosphere, using the relationship Apo = tRTtnp,,, and then drawing a tangent from that point on the X , = 1 axis to the AG,,; curve. The point of contact with AG,,; curve will give the composition of the MO; in contact with the atmosphere at the pol. It is also
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
k
AG .------cal/mole
0
0.1
0.2 0.3 0.4
7: 187
A
-----------
0.5 0.6 0.7 0.8 0.9
C
Fig. 7.85 Common tangent method applied to AC:;,',-,, reflT'
C sraphile.
1 TI
The value of the free
energy of formation of stoichiometric TIC (ref.: Ti b.c.c, C graphite) calculated from this equation (after Teyssandier et a/.")
evident from Fig. 7.84 that, once the pressure of oxygen falls below the critical po2 = 2.8 x lO-,atrn, the MO; oxide phase cannot form since no common tangent can be drawn from the new oxygen potential point to AGLo2 curve without intersecting other AG,-concentration curves. It is also worth noting that any change in the oxygen pressure in the atmosphere from po2 = 1 atm t o the vicinity of po, = 2.8 x 1OW4atmwill have no effect on the compositions of the MO", M20X and solid solution phases. and 8 x 10-"atm the outerBetween oxygen pressures of 2.8 x most oxide layer will consist of the Mo" phase. Its exact surface composition can be predicted by using the common tangent in the same manner as described for MO;. From Fig. 7.84 it is clear that, as the oxygen pressure in the atmosphere is reduced the composition of the surface oxide layer will vary in a predictable manner from X, = 0.55 to X, = 0.475. Once the oxygen pressure falls to between 8 x lo-'' and 3 e 0 4 x lO-"atm, the M 2 0 xoxide phase will be the only stable phase whose outermost surface layer composition will change from X, = 0.325 to X, = 0.275. Below an oxygen pressure of 3.04 x 10-"atm no oxide will be formed and the equilibrium solubility of oxygen in the solid solution for a particular oxygen pressure can be predicted once again using the tangency rule. AG,-Concentration Di8gram and the R8nge of Stability of Tic at
1 900 K
Free-energy-concentration diagrams have been used in the study of the thermodynamic influence on the non-stoichiometry of the solid titanium carbide deposited from H2-CH,-TiCl, gas mixtures at 1 900 K s i . The authors show how, from the partial pressure measurements of Ti vapour over a range of
7 : 188
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
solid non-stoichiometric TixC,, - x ) compositions, the free-energy-concentration curves were calculated as a function of their composition. Figure 7.85 shows one of the plots obtained for AGTixc,,- x ) together with two tangents drawn to the free-energy-concentration curve. One tangent from the point AG = 0 (on the left-hand side vertical axis, Le. for Ti,C(, - x ) at equilibrium with C-graphite), and the other tangent from AG = 0 (right-hand side vertical axis, i.e. for Ti,C,, - x ) at equilibrium with solid b.c.c. pure titanium metal). The compositions at the two points of contact obtained from the tangency rule predicted the equilibrium compositions of titanium carbide. On the left-hand side of the diagram, Le. on the C/Ti,C(, )-, boundary, the equilibrium composition was found to be Tio.5iCo..,9 and that on the righthand side, i.e. on the Ti$(, -,/Ti boundary, was found to be Ti0.66co.~~. Thus the thermodynamic range of homogeneity of the f.c.c. Ti,C(, - x ) nonstoichiometric T i c was found to be betweenXTi = 0.51 and XTi= 0.66. In practice these predicted values were found to be correct at the outer boundaries of the deposited titanium carbide. It is interesting to note that, using the thermodynamic data for AGTiXC(, -x) and then applying the common tangent method, computer calculated limits of the range of stability of the non-stoichiometric phase were: XTi= 0-49-0.67, fitting well with the observed limits of XTi= 0-51-0.66.
AG,-Concentration Diagrams and the Effect of Physical and Chemical Factors on the Composition and Stability of Surface Interaction Layers IFe,s)-Zn,,,and Fe,-Zn,v, Systems
There are a number of examples in the literature where, during gas-metal or gas-liquid-metal-surface reactions, certain phases, shown in the relevant equilibrium phase diagrams, do not form. In other cases the composition of these phases may be different from those expected from the normal equilibrium phase diagram. In all these cases neither AGO-T diagrams nor the phase stability diagram proved to be of much use, and therefore attempts have been made to apply the AG,-concentration diagrams to analyse thermodynamically the reason for the differences between the phases obtained under laboratory experimental conditions, such as in the study of equilibrium phase diagrams, and those encountered on a large industrial scale, during which the phases were formed as surface layers. For example, in the Fe,,,-Zn,,, system the protective outermost c-phase layer, which according to the equilibrium phase diagram should be a stable phase up to 530°C, does not form during galvanising above about 495°C resulting in a rapid linear rate of attack of the steel and an unacceptable quality of galvanising. Because of the financial importance of this process to steel producers (about one-third of all the steel produced in the world is subsequently galvanised) a great deal of research has been carried out throughout the world to establish the 'true equilibrium phase boundaries' in the Fe,,,-Zno, system and the critical temperature of stability of the phase. Since the AGO-T diagrams or the phase stability diagrams could not account for these discrepancies in this system, AG,-concentration curves were used for
THERMODYNAMICS A N D KINETICS OF GAS-METAL SYSTEMS
7 : 189
-
0.0
?
1. 1000 0
-
E
.10 .09 .08 .07 .06 .05 .04 .03 .02 .01
0
(a)
0.0
7 - 1000
E
7
eE
8 2000 3000
-
I
7.79% Fe by wt
I
I
I
I
I
I
I
I
.IO .09 .08 .07 .06 .05 .04 .03 .02 .01 0 (b) 0.0
r
1. 1000 0
E 7
5
E
8
2000
3000 .10 .09
-
.08 .07 .06 .05 .04 .03 .02 .01 Mole fraction of iron
0
(C)
Fig. 7.86 Free-energy-concentration curves for Fe(,,-Zntr, at 505°C under ( a ) equilibrium conditions, ( b ) pressure conditions and ( c ) galvanising conditions (after Mackowiak ”)
7 :190
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
both the Fe,,,-Zn,,, and Fe,,,-Zn,,, systems shown in Figs. 7.86 and 7.87, respectively. The AG,-concentration curves for Fe,,,-Zn,,, at 505°C were calculated for the liquid phase q , and for the condensed phases from experimental results described elsewhere52.In the construction of these diagrams it was assumed that the position of the AG,-concentration curve for the liquid 7 phase remained unaffected by pressures and stresses and therefore was the same in all three diagrams Fig. 7.86A. B and C. The AG,-concentration curves for all condensed phases (r,rl,a,, () were then fixed with respect to the AGliquid curve by the constraint imposed by the use of the common tangent rule. For clarity three separate diagrams were produced. Figure 7.86a shows the situation under equilibrium conditions, as during the study of the equilibrium phase diagrams3. Figure 7.86b shows equilibria under compressive pressure54, and Fig. 7 . 8 6 ~shows curves under galvanising conditions where the tensile stress in the (-phase layer has altered the AG,concentration curve for the ( phase in the upward direction. Comparing Fig. 7.86a and b, it is clear that the reason for the dramatic increase in the solubility of iron at 505°C from the normal equilibrium value of 0.15% to 0.31% iron by weight under pressure resulted from the slight upward movement of the AG-concentration curve for the phase, whereas the position of the liquid 7 phase remained the same in both cases. It is also worth noting that the composition of the layer at the (/liquid boundary is 6.97% iron by weight which is higher than that shown in the equilibrium phase diagram (6.18% iron). There is also an increase in the contents of iron
r
r
1 .o
F
2.0 E Y 7
3.0
$, 7 .. II
4.0
15
VJ
7
5.0
6.0 0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0
Mole fraction of iron Fig. 7.87
Construction showing position of free-energy-conservation curves of r, ri, 61, 3; and 7 phases for the Fe, -Zn(,) system at 793 K (after Mackowiack ")
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7 : 191
in both the 6, phase and the phase at the SI/( boundary. Figure 7 . 8 6 ~ shows that under galvanising conditions the AG,-concentration curve for the c-phase (shown by a broken line) is above the common tangent between the AG,,, for the 9 liquid phase and 6, solid phase and therefore the { phase is unstable under these conditions. These three diagrams show clearly how a very small change in the relative position of the AG-concentration curves for the different phases can have a dramatic effect on the composition and stability of the solid phases in the Fe,,,-Zn(,, system. A detailed discussion of these diagrams can be found elsewhere”. In addition, Fig. 7.86 shows that any physical effects in the solid phases, which can be expressed in terms of energy, can be added to the AG,-concentration curves. Thus the curves can be corrected for these effects and the resulting new equilibria from the shift of the AG-concentration curves (Le. change in composition of phase, their stability, etc.) can be predicted simply by the use of the tangency rule. Figure 7.87 shows a AG,-concentration diagram for Fe,,,-Zn,,,. It was constructed from the experimental data shown in Table 7.37. The method of construction is described elsewhere“. Figure 7.87 can now be used, by applying the constraints imposed by the tangency rule, to explain why in Fig. 7.88a and b, where the chemical potentials (shown in the diagram) of zinc vapour varied between 0 and - 1 - 8 1 kJ mol-’, the total interaction surface layer consisted of r, r,,6, and clayers; in Fig. 7 . 8 8 ~at a chemical potential only slightly lower (-2.11 kJ mol-’) only r and I’, layers were present whilst at -2.55 kJ mol-’ only a I’ outermost layer was formed. The micrographs in Fig. 7.88 show clearly how from a knowledge of the AG,-concentration diagrams it is possible to select the exact reaction conditions for the production of tailor-made outermost surface phase layers of the most desired composition and thus of the optimum physical and chemical properties for a given system. In addition it shows that according to thermodynamics, there can be predictable differences in the composition of the same outermost phase layer prepared at the same conditions of temperature but under slightly different vapour pressures. Similar results, to the Fe-Zn system were obtained in the Ti(s)-Al(,,and Ti(s,-Al(o)system where, in the solid-liquid couples some of the expected surface layer phases were not formed, whereas in the solid-vapour system it was possible to obtain all the phasess6and predict from the AG,-concentration curves the compositions at the different layer phase boundaries. In the literature some basic relationships have been derived correlating the physical effects on phase and their influence on the value of AG,-concentration curves4’. These mathematical relationships may be used for ‘correcting’ the AG,-concentration diagrams. Thus Castleman used the Gibbs’ principle (tangency rule) to calculate the equilibrium of the metallic phases growing in AI-U and Ni-AI systems under different hydrostatic pressures. He derived an equation for the change in free energy of a phase on compression. De Boer” considered the thermodynamic effect on the dissolution of solids under a simple pressure systeqn (homogeneous and isotropic). Other research workers4’,60761* attempted to correlate the effect of stresses and strains, using mechanical theories, with the thermodynamic consideration of equilibria. The effect of grain size on the value of molal free energy change was also
Table 7.38 Details of specimen preparation
- A& Sample Temp. Temp. No. Fe, K Zn, K 793 793 793 793 793 793
1 2 3 4 5 6
792 788 781 779 776 770
AT, K
Pznin system, Pa
PoZnat 793K, Pa
294.9 268.5 228 .O 217.9 203.7 176.3
299.9 299.9 299.9 299.9 299.9 299.9
1 5 12 14 17 23
4zn9
kJmol-
I
Phases present in final layer
0.11 0.73 1.81 2.11 2.55 3.50
r :rl:ti1:r r :rl:sI:c r :rl:sl : I r:r, r r
Composition at interfaces of individual phase layers at 739K Composition, at. - %Fe Sample
No. 1 2 3 4 5 6 np
r wr-r/r1 32.1-24.2 31.9-23.6 32.3-25.1 32.8-25.4 33.O-26.5 32.8-27.5
=
not present in total layer
l-1
r/rl-rl/61 22.4-16.1 21.6-15.4 21.8-16.6 22.2-20.7 np np
t
4 rl/al-al/r
S1/c-t/Zn(v)
14.7-12.5 14.8-13.4 14.7-13.6 nP nP nP
11.8-11.5 12.2-12.1 12.9- 12.7 nP nP nP
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7 : 193
Fig. 7.88 Scanning electron micrographs of cross-sections through interaction layers with superimposed Fe and Zn K. line scans across the layers (a) Sample 2 (x 210); (b) sam8Ie 3 ( x 550); (c) sample 4 ( x 760); (d) sample 6 (x 1 OOO) (after Mackowiak and Short )
considered by de Boer who derived an expression for the difference in molal free energy as a function of its crystal sizea. It is beyond this short section to present a full account of all the relationships which may be found in the literature and which may be used to correct the AG,-concentration curves for effects such as vibrations, irradiation, acceleration, capillarity and any others which can be expressed in terms of energy. J. C. B. ALCOCK E. EASTERBROOK J. MACKOWIAK REFERENCES 1. Behaviour of High Temperature Alloys in Aggressive Environments, in Proc. Penen Int. Con&, ed. Kirman, I. ef al., The Metals Society, 1050 (1980) 2. Richardson, F. D. and Jeffes, J. H.E., J.I.S.I., 160, 261 (1948) 3. Richardson, F. D. and Jeffes, J. H.E., J.I.S.Z., 171, 165 (1952) 4. Richardson, F. D. and Jeffes, J. H.E., J.Z.,S.I., 175,33 (1953) 5 . Pearson, J. and Ende, U., J. I.S.I., 175,52 (1953) 6. Villa, H.,‘Thermodynamic Data of the Metallic Chlorides’, J. SOC. Chem. I d . , No. 1 (supplementary issue), S%S18 (1950) 7. Sherman, C. W.,Elvander, H.I. and Chipman, J., J. Mefak, 2, 234 (1950) 8. Lumsden, J:, Thermodynamics ofdlloys, Inst. Metals, London (1952)
7 : 194
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
9. Ellingham, H. J. T., J . SOC. Chem. Ind., Lond., 63, 125 (1944) 10. Rosenquist, T. and Dunicz, B. L., J. Metals, 4, 604 (1952) 11. Darken, L. S. and Curry, R. W., Physical Chemistry of Metals, McGraw Hill, New York (1953) 12. Sieverts, A. and Krumbhaar, H., Ber. Dtsch. Chem. Ges.,43, 893 (1910) 13. Alcock, C. B. and Richardson, F. D., Acta Metallurg., 6, 385 (1958) 14. Chipman, J. and Sherman, C. W.. J. Metals, 4, 597 (1952) 15. Wagner, C., Thermodynamics of Alloys, Addison-Wesley, New York (1952) 16. Chipman, J., J.I.S.I., 180, 97 (1955) 17. Goldschmidt, H., J.I.S.I., 160, 345 (1948) 18. Elliott, J. F. and Chipman, J., Trans. Faraday SOC., 47, 138 (1951) 19. Schuhmann, R., Acta Metallurg., 3, 219 (1955) 20. Pourbaix, M. S. N., Disc. Farad. SOC., 4, 139 and 223 (1947) 21. Pourbaix, M. S. N., The Industrial Use of Thermochemical Data, in SpecialPubl. No. 34, The Chemical Society, London, 55 (1980) 22. Kellogg H. H.and Basu, S. K., Trans. Mat. SOC. AIME, 218, 76 (1960) 23. Ingraham, T. R., in Applications of Fundamental Thermodynamics to Metallurgical Processes, ed. Fitterer, G. R. Gordon & Breach, New York London, 179 (1%7) 24. Alcock, C. B., Principles of Pyrometallurgy, Academic Press, London, New York, 6 (1976) 25. Rao, Y.K., Stoichiometry and Thermodynamics of Metallurgical Processes, Cambridge University Press, London 626 (1985) 26. Rhys-Jones, T. N. (ed.), Surface Stability, Inst. of Met., 113 (1989) 27. As Reference 25, pp 379-381; as Reference 48, pp 170-173 28. Davies, R. H. and Barry, T. I., MTDATA Handbook, N.P.L. (1989) 29. Gray, J. A. and Starr, F., in Proc. Petten Int. Conf., ed. Kirman, I. et al., The Metals Society, 3 (1980) 30. Natason, K. and Chopra, 0. K., First Int. Conf. on Materials for Coal Conversion and Utilisation, Gaithersburg, Maryland, 11 (1976) 31. Natason, K., High Temperature Corrosion, ed. Rapp, R. A., N.A.C.E., Houston, 336 (1983) 32. Castle, J. E., Surface and Interface Analysis, 9, 345 (1986) 33. Huang, T. T. et al., Corr. Sci., 24 167 (1984) 34. Stringer, J., in Proc. Patten Int. Conf., ed. Kirman, I. et al., The Metals Society, 739 (1980). Figure 7.72 after Parkins, R. A. and Voule, S. J., Annual Report to EPRI Project No RP 979-6, 1978 35. Buglia, V. et al., Corr. Sci., 30, 327 (1990) 36. Knacke, 0.:in Metallurgical Chemistry, ed. Kubaschewski, 0.. 1972, 549-559, N.P.L., H.M.S.O., London, 549 (1972) 37. Stott, F. H. and Smith, S. in Proc. Patten Int. Conf., ed. Kirman, I. et al., The Metals Society, 781 (1980) 38. Lions, J. et al., in Proc. Patten Int. Conf., ed. Kirman, I. et al., The Metals Society, 769 ( 1980) 39. Rahmel, A., Corr. Sci., 13, 125 (1973) 40. Rapp, R. A., Proc. Workshop on Materials Problems and Research Opportunities in Coal Conversion, Columbus, Ohio State University, 313 (1974) 41. Stringer, J. and Whittle, D. P., Proc. First Petten Colloquium on Advanced High Temperature Materials, 14, 6 (1977) 42. Rao, Y. K., Stoichiometry and Thermodynamics of Metallurgical Processes, Cambridge University Press, London, 631 (1985) 43. Ingraham, R. R., Trans. Met. SOC. AIME, 236, 1064 (1966) 44. Mackowiak, J. and Short, N. R., Met. Sci., 11, 517 (1977) 45. Mackowiak, J., Report on Sodium/Steel Interactions, Sponsored by the Nuclear Installations Inspectorate of the Health and Safety Executive, Ref 98/CS/129/1976 (1977) 46. Miodownik, A. P., Bulletin of Alloy Phase Diagrams, 2, 406 (1982) 47. Gibbs, J. W., The Scientific Papers, vol 1, Dover Publication, New York, 65 (1961) 48. Mackowiak, J., Physical Chemistryfor Metallurgists, George Allen & Unwin, 185 (1966) 49. Kubaschewski, 0. et al., Gases in Metals, Metals and Metallurgy Trust, ILIFFE Books, London, 18 (1970) 50. Stored data and Software (see Appendix 1) 51. Teyssandier, F. et a/., in SpecialPubl. No. 34, The Chemical Society, London, 301 (1980)
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
7 : 195
52. Brown, W.N.,Ph.D. Thesis, University of London (1977) 53. Serebryakova, I. B. and Smirnov, N.S., Stal, 25, 422 (1965) 54. Mackowiak, J. and Short, N. R., Corr. Sci., 16, 519 (1975)
55. Mackowiak, J., The Industrial Use of Thermochemical Data, in SpecialPubl. No. 34, The Chemical Society, London, 55 (1980) pp 267-279 56. Short, N. R. and Mackowiak, J., J. Less-Comm. Met., 1976,45,301-308. Mackowiak, J. and Shreir, L . L., J . Less-Comm. Met., 1 , 456 (1959) 57. Pascoe, G. and Mackowiak, J., J.I.M., 98, 253 (1970) 58. Castleman, L. S., Acta Met., 8 , 137-146 (1960) 59. de Boer, R. B., PhD Thesis, University of Utrecht, Holland (1975) 60. MacDonald, G. F. I., Am. J. Science, 255, 226-281 (1957) 61. McLellan, A. G., h o c . Roy. SOC.A , 307, 1-13 (1968) 62. Katchalsky, A. and Curran, P. F., Nonequilibrium Thermodynamics in Biophysics, Harvard Univ. Press, Cambridge, Mass., (1967) 63. Kingery, W. D., (ed.) The Technical Press of MII, J. Wiley and Sons, N.Y.and Chapman and Hall, London, 187-194 (1959)
Appendix 1 Some Centres Available to Outside Contracts General Reference: Metallurgical Thermochemical databases Contact C. W. Bale and G. Eriksson Canadian Metallurgical Quarterly Vol 29 NO 2 pp 105-132 (1990) For details of ThermoCalc Database: The division of physical metallurgy KTH S-100 44 Stockholm Sweden: Contact Birgitta Jonsson Tel: +46 8 790 9140 Fax: +46 8 100411 Email: bosse @ matsc.kth.se Datapack (X.25): 24037101046 For details of MTDATA, contact the National Physical Laboratory WL) Queens Road, Teddington T W l l OLW UK Contact Hugh Davies Tel: +44 81 977 3222 ext 6497 Fax: +44 81 943 2155 Email: RHD @ UK.Co.NPL.Newton For details of CHEMSAGE: GTT mbh Kaiserstrasse 100 5120 Herzogenrath 3 Germany Contact Bob Fullerton-Batten Tel: +49 2407 59533 Fax: +49 2407 59661
7 : 196
THERMODYNAMICS AND KINETICS OF GAS-METAL SYSTEMS
In Australia the CSIRO operate a data base in conjunction with the NPL: CSIRO Thermochemistry System PO Box 124 Port Melbourne Victoria 3207 Australia Contact A. G. Turnbull Tel: +61 3 647 0211 Fax: +61 3 647 0395 CSIRONET: (node *MXDIA) For further information on F*A*C*T: The Ecole Polytechnique Box 6079 Station A Montreal Quebec Canada H3C 3A7 Contact A. D. Pelton Tel: + 1-514-340-4770 BITNET: 5799 @ Polytecl For general information in the USA: Alcan Cambridge Technology Center 21 Erie Street Cambridge Mass 02139 USA Contact Larry Kaufman Tel: + 1 617 349 1721 Fax: + 1 617 354 0395 For THERMODATA: The University of Grenoble Domaine Universitaire BP 66 38402 Saint-Martin-D’Heres CEDEX France Contact B. Cheynet Tel: +33 76 427690 Email: EARN/BITNET: BCHEYNET @ FRGREN81
Additional databases: THERMOTECH Surrey Technology Centre 40 Occam Road The Surrey Research Park Guildford GU2 5YH Contact N. Saunders Tel: +44 483 502003 J . MACKOWIAK
8
8.1 8.2 8.3 8.4 8.5
EFFECT OF MECHANICAL FACTORS ON CORROSION
Mechanisms of Stress-corrosion Cracking
813
Stress-corrosion Cracking of Ferritic Steels Stress-corrosion Cracking of Stainless Steels
8:32 8:52 8:84
Stress-corrosion Cracking of High-tensile Steels
Stress-corrosion Cracking of Titanium, Magnesium and Aluminium Alloys 8.6 Corrosion Fatigue 8.7
Fretting Corrosion
8.8 Cavitation Damage 8.9 Outline of Fracture Mechanics 8.10 Stress-corrosion Test Methods
8: 1
8:l15 8:143 8:184
8:197 8:208 8:215
8.1
Mechanisms of Stress-corrosion Cracking
The visible manifestations of stress corrosion are cracks that create the impression of inherent brittleness in the material, since the cracks propagate with little attendant macroscopic plastic deformation. In fact, a metal that fails by stress corrosion is usually found to conform to the normal ductility standards for that material and the combination of circumstances that cause a normally ductile metal to behave in this way are the presence of a specific environment, a tensile stress of sufficient magnitude and, usually, a specific metallurgical requirement in terms of the composition and structure of the alloy. The compositions and structures of the alloys and the properties of the environments involved in the various instances of failure indicated in this section are so widely varying as to suggest that rationalisation of all of these experiences in a single explanation would be difficult if not unreal, i.e. it is probable that a number of different mechanisms are involved. This is not to suggest that some systematisation is not possible and indeed the objective of this section is to show that the evidence may be rationalised in a continuous spectrum of mechanisms', a concept that has the merit of avoiding the dangers inherent in believing that some highly specific conditions need to be fulfilled for stress corrosion to occur because the mechanism of failure is invariable. The usual energy balance approach to fracture (see Section 8.9)-by equating the strain energy released to the energy consumed in creating new surface and in achieving plastic deformation-needs modification where corrosion processes are involved to take account of the chemical energy released, and it is the latter that distinguishes stress corrosion from other modes of fracture not involving environmental interaction:
+
Surface energy change Plastic work done = Change in initial stored energy + Electrochemical energy released (8.1)
Since the surface energy term will usually be negligible by comparison with the plastic work term in the stress corrosion of ductile materials, it may be neglected. The remaining terms may be derived from fracture mechanics and conventional electrochemical conditions and, for the various boundary conditions indicated by West 2, result in 8:3
8:4
MECHANISMS OF STRESS-CORROSION CRACKING
P = K:( 1 - v ’ ) / E
+ (ZFpS/M)q
(8.2) where P is an appropriate plastic work term, K, is the stress-intensity factor, Y is Poisson’s ratio, E is Young’s modulus, z is the valency of the solvated ions, F is Faraday’s constant, p is the density, 6 is the height of the advancing crack front, M is the molecular mass of the metal and q is the anodic overpotential. This assumes that the mechanism of crack advance involves the localised dissolution of metal, but an equivalent expression could be written involving hydrogen-induced cracking. At the threshold stress-intensity K,,,, , i.e. the minimum value of K, for stress-corrosion cracking, equation 8.2 yields.
Klscc= { (El1 - Y’) ( P - ZFpSq/M),,,)f (8.3) Clearly the variables that may influence Klrcc,or the threshold stress for initially plain specimens, and hence the susceptibility to stress-corrosion cracking, are P and q , i.e. ~ l s c c=
( k i (~ k2q)min)’
(8.4)
Lowering of the plastic work term P will result from an increase in the effective yield stress or an increase in the work hardening rate in the crack-tip region, either or which, for constant q , will lower Klsccand hence increase the susceptibility to stress corrosion. An increase in the anodic overpotential qa (the potential of the metal is made more positive) relative to the plastic work term will also increase susceptibility to stress corrosion. The anodic overpotential will be some function of the electrochemical conditions within the crack that control the active to passive transition which determines whether or not cracking occurs, i.e. it will be some function of pH, anion activity, metal composition and electrode potential. The interdependence of these terms upon the structure and composition of the metal, upon the details of the electrochemical conditions at the crack tip in terms of local cell action and film formation, and upon the response of the metal to the presence of stress in creating new metal at the crack tip, make the quantification of the argument extremely difficult and its relation t o the detailed mechanisms of crack propagation virtually impossible, as West’ indicates. However, the recognition of the need for a critical balance between a number of variables if stress corrosion is to occur, and the fact that this balance may be achieved in a number of ways, is important, and not least in relation to the diagnosis and the prevention of stress-corrosion failures.
Stress-corrosion Crack Propagation Models The implication of the foregoing equations, that stress-corrosion cracking will occur if a mechanism exists for concentrating the electrochemical energy release rate at the crack tip or if the environment in some way serves to embrittle the metal, is a convenient introduction to a consideration of the mechanistic models of stress corrosion. In so far as the occurrence of stress corrosion in a susceptible material requires the conjoint action of a tensile stress and a dissolution process, it follows that the boundary conditions within which stress corrosion occurs will be those defined by failure
MECHANISMS OF STRESS-CORROSION CRACKING
8:5
under a stress in the absence of corrosion, and failure by corrosion in the absence of stress. Between these extremes, wherein stress corrosion occurs, it is necessary to consider how corrosion processes may be influenced by the application of stress to a metal and how fracture may be facilitated by corrosion. When stress corrosion involves very localised dissolution, with the geometrical requirements of a crack to be fulfilled, the rate of anodic dissolution may be expressed as a rate of crack propagation i, M
CV=-
ZFP
(8.5)
where i, is the anodic current density and the remaining symbols are as defined earlier. Now i, and hence CV will be dependent upon the nature of the phase being dissolved and also upon the associated cathodic processes that occur elsewhere, and which need to produce a current sufficient to balance that at the crack tip. The chemical natures of the sites for these reactions will therefore be of importance, but how does the imposition of a tensile stress influence this situation to produce stress corrosion? Despic et a/.' have shown that the dissolution rate of iron in acid solution undergoes a marked rise when the strain passes from the elastic to the plastic condition in dynamic straining experiments and that this result is due to the exposure of high-index planes and of edges at slip steps, as well as increasing surface roughness, as plastic deformation occurs. Stresses in excess of the yield stress may therefore produce locally enhanced activity at surfaces where slip steps emerge, Le. i, in equation 8.5 may be increased (by up to an order of magnitude) by stresses promoting plastic strain. However, there is no difficulty in accounting for the observed rates of crack propagation by stress corrosion for most systems in terms of the currents passed at static bare surfaces without invoking arguments involving markedly enhanced currents resulting from the exposure of slip steps. The real problem is in explaining why the corrosion proceeds along a narrow front to retain the geometry of a crack, implying that most of the exposed surfaces, including the crack sides, must remain relatively inactive. The transition from electrochemically active to relatively inactive behaviour that the sides of a crack must undergo as the tip advances and creates more crack can only be achieved if the environment forms a film, and this implies, in relation to equation 8.5, that the conditions for maximum crack growth rate will be met if i, is maintained close to the film-free value, i.e. the metal is in the active state and protective films are not allowed to grow over the crack tip or, if this does occur, that the film is repetitively broken. The function of stress, then, in the realistic conditions of film-forming environments will be essentially to prevent, or to fracture, films forming at the crack tip. In this context, two different circumstances' may now be envisaged whereby cracks can propagate by a dissolution-controlled process. The alloy may exhibit structural features, either as a segregate or precipitate, usually at the grain boundaries, that cause a local galvanic cell to be established, i.e. a pre-existing active path is involved, as originally suggested by Dix4. The precipitate or segregate may act as the anode in the local cell or, by
8:6
MECHANISMS OF STRESS-CORROSION CRACKING
H
Cathodic phase
Anodic phase
Fig. 8.1
Pre-existing active path mechanisms, in which H represents cathodic hydrogen
acting as an efficient cathode, may cause the dissolution to be localised upon the immediately adjacent matrix (Fig. 8.1). The lattice characteristics in the region of a grain boundary are such that equilibrium segregation of solutes or the nucleation and growth of precipitates are favoured reactions, and so grain boundaries in particular are potential sites for chemical heterogeneity. Where such pre-existing active paths are non-existent, or are inoperative, the disruption of a protective surface film to expose bare metal may result in a second mechanism of crack propagation, as originally suggested by Parr and Straub5.Thus stress (or probably more correctly, plastic strain) in the underlying metal may bring about the disruption of the protective film whereby active metal is exposed, in the manner shown schema-
-
Oxide metal
\
\
\
d ef or mat i o n
\ \
Oxide
\ \
\ \
Metal
Ox idat ion
region Rupture
Slip line
.\ .
.
\
\
\
\
'\\
la1
(6I
Fig. 8.2 Strain-generated active path mechanisms. (a) Often referred to as the film rupture model and (b) the slip step dissolution model. In both cases growth is by dissolution: film rupture is the rate controlling step, not the mechanism of crack growth
MECHANISMS OF STRESS-CORROSION CRACKING
8:7
tically in Fig. 8.2. The active path along which the crack propagates is cyclically generated as disruptive strain and film build-up alternate with one another, or propagation is related to the slip characteristics of the underlying metal. Although somewhat less favoured than a decade or so ago, transgranular cracking by a strain-generated active path mechanism remains supported by some workers and a significant body of corroboratory evidence. The localisation of the dissolution in such cases is most often related to slip and particularly that which occurs with some face-centred cubic alloys having a low stacking fault energy or displaying short-range order, where planar groups of dislocations are favoured and cross slip made more difficult6 of dissolu(Section 9.2). The observations of Swann and his co-worker~’-~ tion associated with planar dislocation arrays in transmission electron microscopy (TEM) foils led to the suggestion that arrays of fine corrosion tunnels form which subsequently interconnect by the tearing of the remaining ligaments between the tunnels (Fig. 8.3). Stress
Stress Slip plane ,’trace
Fig. 8.3 Schematic representation of the stress corrosion cracking mechanism of the pit (after Pickering and Swann’). (a) Tubular pits initiated at solute-rich slip step. The pits may, but need not necessarily, follow the slip plane once they are initiated. (b) Ductile tearing along a plane containing the tubular pits. The stress is increased across the plane because of the reduced cross section and the stress raising effect
At about the same time that Swann was developing the tunnelling model, Nielsen lo was examining, by TEM, the corrosion products removed from stress-corrosion cracks in austenitic stainless steels after exposure to chloride solutions. In general, these take the form of fans showing lamella markings, suggesting that crack growth was discontinuous. It may, of course, be argued that such oxide films are formed after the crack tip has advanced and that they are simply replicating the stress-corrosion fracture surface. However, when Nielsen exposed samples of the steel to MgCl, solution for only a few minutes oxide-filled corrosion tunnels developed, which were joined by lateral tunnels that increased in number on moving towards the original surface of the specimen, to produce a corrosion-product fan. Somewhat similar effects have been observed ” in a copper alloy exposed to ammonia vapour and again suggest that localised dissolution processes cannot be ruled out as a contributing factor in the growth of transgranular stress-corrosion cracks in some systems.
8:8
MECHANISMS OF STRESS-CORROSION CRACKING
The objection that is most frequently raised to such an essentially dissolution-related crack growth model is that it does not appear likely to lead to the matching and interlocking fracture surfaces often observed in the transgranular cracking of face-centred cubic metals I * . This has resulted in consideration of the possibility that embrittling films formed at the exposed surfaces of metals may play a critical role in stress-corrosion cracking. Edeleanu and Forty3 observed that the cracking of a-brass single crystals exposed to an ammoniacal solution occurred discontinuously, with short bursts of extremely rapid cracking followed by relatively, long rest periods. It was suggested that truly brittle fracture was associated with the bursts while the rate-controlling periods of non-propagation were concerned with the corrosive processes that established the conditions for further crack bursts. The model requires that a-brass can support a freerunning cleavage crack, albeit over short distances of the order of a few microns, and this presented a major difficulty. Thus, while cleavage in a body-centred cubic metal, such as a-iron, has been observed on a microscale, to reflect its well-known tendency for macroscopic cleavage in appropriate conditions, such cleavage of face-centred cubic metals, as in a-brass, has not been demonstrated in like manner. However, relatively recent atomic modelling studies 14.15 indicate the theoretical possibility of short-range cleavage of ductile metals from an initiating surface film of appropriate characteristics. Embrittlement of a metal from corrosive reactions, especially whereby hydrogen enters the metal, has often been invoked in the context of stress corrosion. The opposite boundary condition referred to earlier as limiting the regime in which stress corrosion occurs was that of purely mechanical fracture in the absence of corrosive processes. The energy balance of equation 8. l indicates that, with negligible contribution from dissolution, crack extension will be facilitated by a reduction in t h t surface energy required to form crack faces or a reduction in the plastic work term by embrittlement of the metal in the crack-tip region. If the environment provides species that are adsorbed at the crack tip to reduce the effective bond strength, then the surface energy is effectively lowered, alternatively the species may diffuse into the metal forming a brittle phase, e.g. a hydride, at the crack tip, or interactions may occur at some region in advance of the crack tip where the stress and/or strain conditions are particularly appropriate for the nucleation of a crack (Fig. 8.4). In the latter case hydrogen is usually regarded as the only species that can diffuse with sufficient speed to account for observed crack propagation rates, and in the present context, hydrogen embrittlement, with the hydrogen derived from corrosion reactions, is considered as a particular instance of stress corrosion. Whilst surface-energy lowering has been suggested I6 as a single mechanism that explains all instances of stress-corrosion cracking, it has particular difficulties in explaining the phenomenon in the more ductile metals. Thus, whilst stress-corrosion cracks propagate without marked macroscopic plastic deformation there is ample evidence to show that localised plastic deformation occurs at the crack tip, and in such circumstances, as indicated in relation to equation 8.1, the surface energy term is negligible in relation to the plastic work term (5 J/m2 as opposed to 5 kJ/m2), and so any reduction in surface energy by adsorption will have a negligible effect upon the fracture stress. Moreover, in some
8:9
MECHANISMS OF STRESS-CORROSION CRACKING
Fibrous
u H
Cleavage
3 Crack
Fig. 8.4 Mechanisms involving embrittlement of the metal. (a) Crack-tip adsorption, (b) hydrogen adsorption, (c) decohesion by hydrogen influx to dilated lattice and (d) crack extension due to brittle hydride particle forming at crack tip
instances of hydrogen-related fracture of metals evidence of the fracture mechanism involving enhanced local plasticity due to the presence of hydrogen 17.18 has become apparent in recent years and, for those cases at least, the mechanism of crack growth is hardly consistent with an approach based on equation 8.1. In summary then, given the appropriate balance between electrochemical activity and inactivity, localised corrosion may be distributed in a number of ways and result from a number of different mechanisms in promoting stress corrosion. If the structure and composition of the alloy are such that almost continuous paths of segregate or precipitate exist, usually at the grain boundaries, and which are electrochemically different from the matrix, then a latent susceptibility to intergranular corrosion may be activated by the prescence of stress. Where pre-existing active paths are inoperative, the strain may generate active paths by rupturing an otherwise protective film and, possibly, activating dissolution at emerging slip steps or initiating micro-cleavage which continues from the film in which it initiates into the underlying metal for some distance before arresting. The more crucial role
8: 10
MECHANISMS OF STRESS-CORROSION CRACKING
of stress or strain in the latter case is continued to those alloys that are inherently lacking somewhat in ductility and have a propensity towards brittle fracture, which may be facilitated by a reduction in the energy required for fracture as the result of either adsorption of species or the formation of brittle phases at the crack tip, or of hydrogen in advance of the latter. The suggestion that these different mechanisms of stress corrosion should be considered as occurring within a continuous spectrum, with a gradual transition from one to the others as the dominance of corrosive processes is replaced by stress or strain, leads readily to the notion that alloy composition and structure, electrochemistry and stress may interact in a variety of ways, and that the transformation from one mechanism to another may result from a change in either alloy characteristics or environmental conditions. On the other hand, there are some who consider that all instances of environment sensitive cracking can be explained by a single mechanism of which Galvele", with a surface mobility mechanism, is most recent.
'
The Stress-corrosion Spectrum Stress corrosion has been the subject of a number of extensive reviews '9-23 resulting from major conferences in recent years, and these, together with the following sections of this volume, avoid any need for a general review of the data. Instead, consideration will be given to some of the implications of the various ideas already referred to in relation to stress corrosion in a variety of systems. Pre-existing Active Paths
Where cracking is associated with pre-existing active paths, structurally sensitive attack, such as intergranular corrosion, may be expected to be observed upon unstressed specimens, at least in the earlier stages of exposure before secondary reactions such as those that lead to film formation, obscure the metal structure. In Section 8.2 it is indicated that samples of steel polished as for metallographic examination suffer grain boundary attack when immersed for short times under environmental conditions that would lead to cracking of stressed specimens, and that whilst such grain-boundary corrosion does not penetrate to great depths in the absence of stress, it is possible to disintegrate a piece of unstressed mild steel by intergranular corrosion in a boiling nitrate solution with applied anodic current. The observation that carbon steels contain pre-existing susceptible paths for corrosion, the structural distribution of which is related to the paths followed by stresscorrosion cracks, is also apparent in the aluminium-base alloys that undergo stress-corrosion cracking (Section 8.5). The electrochemical properties of the segregates or precipitates, relative to their associated matrixes, that are involved in these instances of intergranular attack have been the subject of a number of studies following ~~ the early work of Dix' along these lines. Thus, Doig and E d i n g t ~ nused microelectrodes to measure the localised corrosion potentials at grain boundaries in AI-Mg and AI-Cu alloys, and their results correlate well with the cracking propensities of those alloys. The effects of ageing time in these
MECHANISMS OF STRESS-CORROSION CRACKING
8:11
results reflect their effects upon cracking susceptibility. Similarly, heat treatments designed to change the distribution of chemical heterogeneity of grain boundary regions in ferritic and stainless steels and in nickel-base alloys alter the cracking propensities of the latter in ways that support the suggestion that intergranular stress corrosion in these relatively lowstrength ductile materials is related to a latent susceptibility to intergranular corrosion. Of course, an inherent susceptibility to intergranular corrosion is not the only requirement for susceptibility towards intergranular stress corrosion, since for the latter to occur it is necessary for the former to be sustained or enhanced by the application of stress, and there are instances of alloys that are susceptible to intergranular corrosion but not apparently to stress corrosion. The role of stress may be critical in some cases where the material shows a tendency towards intergranular fracture in the absence of corrosive influences, e.g. in some of the high-strengh aluminium alloys, or where the structure of the alloy determines whether or not deformation is localised to sustain relatively bare metal in the grain-boundary crack-tip region. In this latter respect it is noteworthy that the a-brasses, which can be caused to fail in an intergranular manner under constant strain conditions in ammoniacal solutions at pH 7.3, show an increased tendency for transgranular cracking under slow-strain-rate conditions2s. Since the same material, at constant strain, can be made to fail in a transgranular manner by changing the pH of the ammoniacal solution or by small changes in alloy composition, such results simply serve to underline the delicate balance between the factors that promote a particular mechanism of cracking and of the dangers in attempting to rationalise all observations into a single mechanism. Moreover, even within those systems that exhibit intergranular stress corrosion the part played by the response of the metal to the application of stress may be expected to be variable, with an increasing tendency towards a different mechanism, most often resulting in transgranular fracture as the propensity towards intergranular corrosion is reduced and the roles of stress and/or strain become more important. Strain-generated Active Paths
Many corrosion-resistant alloys owe their electrochemical inactivity to a relatively inert film that forms on the exposed surfaces of the metal, so that the relatively active metal is effectively separated from the environment. If the protective film is disrupted for any reason, such as by plastic strain in the underlying metal, the exposed metal is attacked until the protective film reforms, when further reaction is stifled until the film is again disrupted. With such a mechanism it is claimedz6 that the rate of cracking will be dependent upon the rate of film growth, although the physical characteristics of the film, i.e. its thickness, the extent to which it shows plastic or brittle behaviour and the magnitude of the internal stresses that it contains as a result of its mode of deposition, are also likely to be important. Ellipso~ metric studies of the rates of film growth on a-brasses in 1 5 aqueous ammonia have shown the growth rate to increase with zinc content of the brass, temperature and applied potential, parameters that also increase the stress-corrosion crack-propagation rate, thereby providing support for a
8: 12
MECHANISMS OF STRESS-CORROSION CRACKING
film rupture mechanism. The question of whether this would result in intergranular or transgranular cracking is controversial. It is sometimes arguedz7that plastic strain tends to concentrate at grain boundaries, forming dislocation networks in which copious sources exist for plastic flow and hence for promoting intergranular cracking. On the other hand, transgranular cracking will be favoured by planar slip and will therefore be facilitated at low stresses when extensive plastic strain will not result in dislocation networks that would block planar slip. Whilst instances may be cited of transgranular cracking occurring at low stresses and intergranular at high stresses, there is also a considerable amount of evidence to the contrary. The effectsz5of increasing strain rate, and hence of stress, have already been mentioned as resulting in an increased tendency for transgranular cracking, and the effects of increasing amounts of cold work upon the cracking of carbon steels (Section 8.2) are at variance with the expected effects. Preferential oxidation may occur along grain boundaries in the absence of stress, possibly because of equilibrium solute enrichment in such regions, and even where the rupturing of films growing along boundaries is an important part of the stress-corrosion process, as possibly with the a-brassesZ8,these are, within the context of the earlier definition, examples of corrosion along a pre-existing active path. The intergranular stress corrosion of the a-brasses therefore constitute a convenient bridge between the pre-existing and strain-generated active path mechanisms. Where transgranular stress corrosion results from dissolution following repetitive film rupturing it is to be expected that the deformation characteristics of the metal will be important, although these may well be influenced by the presence of the film. It would appear that the height of the slip step formed at a surface must be greater than the thickness of the inactive film if bare metal is to be exposed, and this implies that deformation associated with high slip steps is likely to be more effective in promoting stress-corrosion than when fine slip is operative. However, it should not be assumed that crack initiation will be avoided if only fine slip steps form, since the initiation of cracks in a-brass, for example, in a variety of environments is most often at grain or twin boundaries, despite the subsequent propagation being tran~granular’~.This is probably because the film overlying grain boundaries has different properties from that overlying grain surfaces facilitating crack initiation at the grain boundaries, although if the slip steps are large enough transgranular initiation will also occur. With the face-centred cubic metals, high slip steps are the result of cross-slip being difficult, leading to planar arrays of dislocations, because of low stacking-fault energy or the presence of short range order in the alloy. Swann6has shown a relationship between stacking-fault energy and stress-corrosion susceptibility for copperbase alloys and for austenitic stainless steels, indicating also the tendency for transgranular cracking to dominate the more readily planar arrays of dislocations form. However, whilst the effects of change in alloy composition upon stresscorrosion cracking susceptibility in the present context may be partly due to their effect upon stacking-fault energy, this does not constitute a complete explanation, since alloying may have significant effects upon electrochemical parameters. The effect of the zinc content of brasses upon their filming characteristics has already been mentioned, while in more recent
MECHANISMS OF STRESS-CORROSION CRACKING
8: 13
work Sieradzki et have shown that the tendency for the dezincification of a-brasses correlates well with the effect of zinc content of the brass upon transgranular cracking. The effect of nickel additions to carbon steels upon cracking in boiling 42% MgCI, is equally illuminating. Small additions, of the order of a few per cent, have little effect upon the cracking of ferritic steel in boiling nitrate solution and in the absence of nickel such steels will not fail in boiling 42% MgCI,. However, the addition of only 1To Ni will induce a susceptibility to cracking in MgCI, which follows an increasingly transgranular path as the nickel content is increased, becoming fractographically indistinguishable from the austenitic steel at about 6% Ni. The structure and mechanical behaviour of ferritic steels are not significantly changed by additions of only 1% Ni, yet the change in cracking susceptibility is dramatic and it is difficult to escape the conclusion that this results primarily from changes in electrochemical behaviour 3 i . This is not meant to imply that the mechanical behaviour of alloys as reflected in their response to the application of stress is not important in transgranular stress-corrosion, but merely that the relative importance of different parameters can vary from one system to another. The importance of deformation, and hence mechanical behaviour, in transgranular cracking is most crucial in the slip dissolution model. Static dislocations do not usually show evidence of significant chemical activity, unless associated with chemical heterogeneity resulting from solute segregation, but moving dislocations have been suggested as promoting electrochemical activity relevant to stress-corrosion cracking. Hoar and West 32 showed that the currents associated with straining electrodes may be very much greater than those observed at static surfaces, a difference suggested as being the result of yieldassisted depolarisation. The association of such observations with the formation of tunnels from the crack tip and the tearing of the ligaments between the tunnels to produce crack advancement '0*33 is an obvious extension, but Staehle34considers neither to be a vital part of the mechanism since they are not relevant in every case of cracking in Fe-Cr-Ni steels. The essential step in the slip dissolution model is that a relatively inactive film is broken by emerging dislocations and a local transient dissolution process ensues. The difference therefore between this and the film rupture model as it is sometimes invoked is largely concerned with the differing emphasis placed upon the acts of film rupture and subsequent metal dissolution as the controlling process. It has already been mentioned, virtually as an extension of the film rupture model, that a crack initiated in a brittle film may progress into the normally ductile substrate for a small but appreciable distance before being arrested by plastic deformation in the matrix. The attraction of a microcleavage-based mechanism for transgranular stress-corrosion cracking in a number of systems derives from fractographic observations and the emission of discrete acoustic events and electrochemical current transients accompanying crack growth Is. Thus, stress-corrosion fracture surfaces are characterised by flat, parallel facets separated by steps, opposite fracture surfaces being matching and interlocking. Arrest markings are sometimes observed, suggesting that crack growth is discontinuous, as observed in the experiments of Edeleanu and Fortyi3. Moreover, there is a strong correlation between peak amplitude acoustic emission events and electrochemical
8: 14
MECHANISMS OF STRESS-CORROSION CRACKING
current transient peaks during the transgranular cracking of a-brass exposed to NaNO, solution”. Of course, it may be argued that such observations are not unequivocal demonstrations of crack growth by fast cleavage. Thus, arrest markings make no comment upon the processes occurring between successive markings, which simply indicate that the crack stopped. If, as is likely, the crack stops because of plastic deformation and crack yawning, the acoustic emissions and electrochemical current transients could be a consequence of such deformation. There is a need for measurements aimed at measuring possible cleavage events more directly. Moreover, the expression used in the analytical modelling of cleavage initiated by films appears to involve dislocation-crack interactions which are only likely to be valid under small-scale yielding conditions. Yet the initiation of stress-corrosion cracks in a-brass exposed to NaNO, solution is associated with the onset of yielding and continues with general yielding3’. The latter leads to extensive branching, which seems more likely to be related to shear strains being very effective in producing crack growth than to any dynamic effects, not least because crack branching in cleavage-type fracture usually only occurs at very high crack velocities. The debate between the protagonists of the dissolution and the microcleavage mechanisms for transgranular cracking of the more ductile alloys continues, but the concept of localised embrittiement being involved in stress-corrosion cracking in some systems is not in doubt.
’’
Embrittlement of the Metal in the Crack-tip Region
The literature reports many so-called critical experiments that purport to show the operation of a surface energy lowering mechanism of stress corrosion, but the results are frequently equally explicable in terms of some other mechanism. The effect of grain size upon stress-corrosion cracking susceptibility is a typical example, it having frequently been reported that coarsegrained material is more susceptible to cracking than fine-grained material, detailed analysis showing a Petch type of relationship between the grain diameter I, and the stress ui, to initiate a stress-corrosion crack, i.e. ui = a,,
+ kl-7
(8.6)
where a, and k are constants, of which k may be related to the surface energy associated with the formation of new surfaces by fracture through
(8.7) where G is the modulus of rigidity and the other symbols are as defined earlier. Measurement of the dependence of some stress-corrosion fracture stress on grain size therefore allows a surface energy to be obtained, and ~ , the apparent surface energies so determined since Coleman, et u I . ~found to be appreciably less than the energy values derived in other circumstances, they concluded that the surface energy associated with crack formation is reduced by the adsorption of some atom or ion species in the stress-corrosion medium. There is, however, an alternative explanation of the grain size
MECHANISMS OF STRESS-CORROSION CRACKING
8: 15
dependence of the stress-corrosion behaviour of alloys, and this is concerned with the plastic-flow characteristic of materials as they are influenced by grain size. Thus a relationship of the form of equation 8.6, where ai is the flow stress at constant strain and the grain size term arises from the resistance to the formation of a slip band at a grain boundary, can be shown to be relevant to the plastic behaviour of metals, and it follows that the grain-size dependence of stress-corrosion cracking may simply reflect the fact that the latter is related to plastic flow in the material. Results such as those shown in Fig. 8.18 in Section 8.2, indicating similar slopes for the stress-corrosion fracture stress and flow stress plots against grain size, suggest that the effect of grain size in stress-corrosion cracking is as likely to be related to plastic flow effects as it is t o surface energy lowering. Similar results are available in relation to the cracking of a-brass in NaNO, solutions3'. The specificity of environments that promote stress-corrosion cracking has been adduced in support of a crack-tip adsorption modelf6,but such observations do not appear 37 to discount a dissolution mechanism of crack propagation any more than they support an adsorption mechanism. Yet, almost paradoxically, it is from observations on environmental aspects of stress corrosion of high-strength steels that the strongest evidence in support of environmental-induced brittleness in the crack-tip region derives. The solution requirements for cracking in high-strength steels are not highly specific, Le. failure will occur in a wide range of aqueous and non-aqueous solutions, unlike the situation in relation to the failure of the low-strength ductile alloys, and the common denominator in these environments is hydrogen. The implication is simply that the environment should provide a source of hydrogen, but that species in solution that facilitate the ingress of hydrogen into the metal will enhance cracking, whilst species that lead to the discharge of gaseous hydrogen at the steel surface will retard cracking. In the former category are arsenious salts, which promote hydrogen adsorption and entry, whilst platinum additions to the system may be expected to facilitate hydrogen discharge. Similarly the effect of increasing cathodic current densities applied in stress corrosion tests may be expected to enhance cracking if hydrogen adsorption is involved in the failure mechanisms. The effects of sodium arsenate and chloroplatinic acid additions to a sodium chloride solution upon the cracking propensity of an 18% Ni maraging steel at various applied cathodic current densities conform with expectations if hydrogen adsorption is the controlling factor in the cracking process3'. Other observations, such as those involving measurement of the solution pH and the electrode potential at the tip of a propagating stress corrosion crack in a high-strength steel in showing that the conditions there are conducive to hydrogen entry into the steel, are also sometimes adduced in support of a hydrogen-embrittlement mechanism (see Section 1.6 and 8.4). It is worth mentioning, however, that the demonstration of the existence of acid conditions at the crack tip does not exclude the possibility that some crack extension, however small, results from dissolution, which is also likely to be facilitated by the low pH environment at the crack tip. Indeed the production of hydrogen by cathodic reaction requires a balancing anodic reaction, which may occur at the crack tip and result in advancement of the latter. Whilst discussion continues on the details of hydrogen generation, adsorption and diffusion, and the relative contributions of these to the overall
8: 16
MECHANISMS OF STRESS-CORROSION CRACKING
physical mechanism of hydrogen embrittlement, some aspects of the latter have begun to crystallise as the result of experiments conducted in gaseous hydrogen environments 39. The demonstration that sub-atmospheric pressures of hydrogen gas can readily result in the propagation of cracks in high-strength steels indicates that the mechanism is not likely to involve the diffusion of hydrogen through the metal to voids where a disruptive pressure of gas is generated. This suggests either that hydrogen lowers the surface energy by adsorption or that it accumulates within a few atomic distances from the crack tip, in response to the lowering of its chemical potential by the elastic stress, thereby lowering the cohesive force of the lattice. Oriani3’ prefers the latter explanation because it is the only one that is consistent with the observations of the effect upon crack propagation of small changes in hydrogen gas pressure and the substitution of deuterium for hydrogen. A sufficient reduction in the hydrogen gas pressure surrounding a specimen containing a propagating crack at a given stress intensity caused the crack to stop propagating, but a subsequent increase in pressure, of about 1.6 kN/m2 from 22 kN/mZ, was sufficient to restart the crack and with a delay time so short that the extra hydrogen entering the lattice as the result of the increased pressure could have diffused no more than a few atom spacings. A similarly rapid response of the crack velocity to small changes in applied cathodic current to a maraging steel immersed in sodium chloride solution has been observed. The effect of deuterium in reducing the response to embrittlement appears not to be related to the difference in transport kinetics of the two isotopes but to their solubilities in the dilated lattice just beyond the crack tip. This again is in agreement with a decohesion model. An alternative, or possibly an additional, model for hydrogen-induced failure that has received recent support is that based upon the idea originated by BeachamW, that hydrogen lowers the work for fracture by enhancing localised slip. As O r i a r ~points i ~ ~ out, while at first this may seem contradictory, since enhanced plastic deformation would be expected to increase the work of fracture, if the enhanced deformation is directly useful to crack propagation, the difficulty disappears. The most striking illustration of localised decohesion in heavily defined regions at crack tips is due to BirnbaumI7, working with nickel foils strained within a high voltage TEM. It appears that hydrogen causes both localised slip and enhanced decohesion, which receives support from the theoretical modelling of Daw and Baskes41, showing that the same phenomena that decrease the resistance force for decohesion also decrease the force for shear separation. Although this model is not yet even semiquantified, and some of the experimental observations ambiguous, Oriani 39 regards the decohesion and localised slip models as complementary, rather than competitive. Nor are these the only models that have acquired strong support, since the formation of brittle hydride phases in the crack-tip region in appropriate metals receives support from the observations of some workers. Thus, Scully and Powell4* have developed earlier observations on the formation of a hydrides in a-Ti alloys to explain the stress-corrosion cracking of such materials, involving cleavage of the hydride as an important step in the cracking process. Pugh and his co-workers have extended these observations on the importance of hydride formation in the cracking of Ti
MECHANISMS OF STRESS-CORROSION CRACKING
8: 17
alloys and have shown that the fracture planes correspond to the habit planes of the hydride, as well as showing that Mg-Al alloys may form hydrides43.There are othersa who believe that stress-corrosion cracking in Ti alloys results from dissolution, but that is not consistent with the effects of immersion in methanolic solutions of HCl prior to straining or the recovery from such exposure in subsequent slow strain-rate tests at very low strain rates4’, which are more readily ascribed to the redistribution of hydrogen.
Environmental Aspects of Stress-corrosion Cracking It has often been stated that the environmental requirements for stresscorrosion cracking are highly specific, but the list of environments identified as causing cracking in various alloys continues to grow with time and the concept of solution specificity is not so narrow as it was even a decade ago. Nevertheless, it is clear that cracking environments are specific, in the sense that not all possible environments promote cracking, and the electrochemistry of stress corrosion is essentially concerned with explaining this specificity. In very general terms, it is clear that potent solutions will need to promote a critical balance between activity and passivity, since a highly active condition will result in general corrosion or pitting, whilst a completely passive condition cannot, by definition, lead to stress corrosion. Whilst the relative inactivity of all exposed surfaces except the crack tip may be derived from a noble film in the cases of alloys containing sufficiently noble elements, for the great majority of engineering alloys inactivity at exposed surfaces is the result of the presence of oxide films overlaying metal surfaces. It is not surprising therefore to find that the alloys of high inherent corrosion resistance (such as those based upon aluminium or titanium, or the austenitic stainless steels, that readily develop protective films) require an aggressive ion, such as a halide, to promote stress-corrosion cracking. On the other hand, to crack the metals of low inherent corrosion resistance, such as carbon steels or magnesium-base alloys, requires the presence of an environment that is itself partially passivating. Thus, the carbon steels can be made to fail in solutions of anodic inhibitors, such as hydroxides and carbonates, and the cracking of magnesium-base alloys is achieved with an appropriate mixture of CrOf and C1- ions, but not with either of these species alone. The transition from electrochemically active to relatively inactive behaviour that the sides of a crack must undergo as the tip advances by dissolution and creates more crack may be expected to be reflected in the current response of an initially bare surface exposed to the appropriate environment, since dissolution will be associated with the passage of relatively high anodic current densities, but with the passage of time this current will decay if filming occurs. Very rapid rates of current decay are unlikely to permit much dissolution and are not likely therefore to be indicative of conditions conducive to cracking, whilst very slow rates of decay will be more likely to be indicative of pitting than cracking. Intermediate rates of current decay will be those likely to be associated with cracking, and such results are indeed observed&, but it is not yet possible to predict quantitatively what constitutes a critical rate of decay, although this would be
8: 18
MECHANISMS OF STRESS-CORROSION CRACKING
expected to be potential dependent according to the competition between the solvation and filming processes. A more convenient way of anticipating the range of potentials in which stress-corrosion cracking is likely to occur is available through potentiodynamic polarisation curves. If the potential of an initially film-free surface is rapidly (approximately 1 V/min) changed over an appropriate range, then the currents passed at the surface will indicate ranges of potential in which relatively high anodic activity is likely. The rapid sweep of the potential range has the object of minimising film formation, so that the currents observed relate to relatively film-free or thin-film conditions. If the experiment is now repeated, but with a slow rate of potential change (approximately lOmV/min) so that time is allowed for filming to occur, comparison of the two curves will indicate any ranges of potential within which high anodic activity in the film-free condition reduces to insignificant activity when the time requirements for film formation are met, and this will indicate the range of potentials within which stress corrosion is likely. Figure 8.5 shows schematic polarisation curves determined under such conditions and indicates the various domains of behaviour expected. The technique correctly anticipates the stress-corrosion cracking of carbon steels in a number of totally different environments4. Of course, it is only applicable in those cases where air-formed oxide films can be reductively dissolved so that bare surfaces are created before the potential sweeps; in other cases straining or scraping electrodes must be used to remove the oxide film and the current response of the bared metal then observed potentiostatically at different
/
Passivity
corrosion
--
Cathod i t protection -ve Fig. 8.5
Current density
+ ve
Potentiodynamic polarisation curves and the expected domains of electrochemical behaviour
MECHANISMS OF STRESS-CORROSION CRACKING
8: 19
potentials. However, these different techniques give broadly the same results in any given system. It is now well established that stress-corrosion cracking only occurs over particular ranges of potential for a given metal-environment combination. Such potential dependence must be related to specific reactions whereby the environmental requirements for cracking are met. Probably the simplest situation in this respect arises with hydrogen-induced cracking, where the hydrogen derives directly from the bulk environment to which the metal is exposed, and in which circumstances the conditions for cracking would be predicted to be met where the potential is below that for hydrogen discharge at the relevant pH. The highest potentials at which hydrogen-induced cracking is observed in various ferritic steels exposed to different solutions lie just below the calculated equilibrium potential for hydrogen discharge as a function of pH,’, so there is reasonable agreement between the predicted and observed behaviours. Where crack growth is by dissolution associated with filming reactions to retain crack geometry, the potential dependence of cracking should reflect those requirements, again with some pH dependence because of the influences of that quantity upon the potentials at which the various reactions are possible. Where the necessary thermodynamic data are available for the species involved in a particular system, it should be possible to calculate the limits of the cracking domain. This has been done for the cracking of lowstrength ferritic steel exposed to phosphate solutions and the agreement between the observed and calculated boundaries of the cracking domain is reasonable4*. For that system, as with ferritic steels in other environments, the upper boundary of the cracking domain is met when the stable phase becomes y-Fe,O,, i.e. at potentials where only the latter forms, cracking does not occur. While the potentials and pH values at which that phase can form will depend upon the phases formed within the cracking domain, it is interesting to consider the location of the potential-pH domains for cracking in various systems involving different ferritic steels in a range of environments at temperatures between 20 and 288°C. Figure 8.6 shows the various cracking domains together with the calculated equilibrium potentials for reactions between Fe,O, and Fe,O, and between Fe,O, and Fe and for hydrogen discharge, all at 90°C as representing an average temperature for the various systems involved49. Clearly each cracking domain is associated with the calculated Fe,O,/Fe,O, line and indeed in all of these systems, Fe,O, is observed to form under conditions where cracking occurs, although it is frequently associated with other phases, e.g. FeCO, in the case of cracking by carbonate-bicarbonate solutions and Fe, (PO,), for cracking by phosphate solutions. Moreover, for most of the systems shown in Fig. 8.6 only ductile failures occur in slow strain-rate tests carried out at potentials high enough to form Fe,O, alone. While it is clear that the anions exert a significant influence upon the location of the cracking domains, the importance of Fe,O, formation within the cracking ranges and Fe,O, formation under conditions associated with ductile fracture appear well established, but the reasons for such less so. The exceptions in Fig. 8.6 to only ductile failure occurring at potentials high enough to form Fe,O, involve nitrates and high temperature water. In both of those systems cracks grow from pits, and within the pit-crack
8:20
MECHANISMS OF STRESS-CORROSION CRACKING
1.01
~
'-2
4
6
8
PH
10
12
I
Fig. 8.6 Potential and pH ranges for the stress-corrosion cracking of ferritic steels in various environments, together with the pH-dependent equilibrium potentials for reactions involving Fe -, Fe304, H -, H + and Fe304 .+Fe203 (after Congieton et 0 1 . ~ ~ )
enclaves Fe, 0, forms, despite the external surfaces being covered with Fe,O, films. The initiation of stress-corrosion cracks from pits has been observed in a variety of systems and is usually taken as indicative of the local environment within the pit being potent and different from that of the surrounding bulk environment. Where cracking does extend from pits there is usually reasonable correlation of the onset of cracking with the pitting potential. The perturbation of the electrochemical conditions within pits has inevitably led to similar considerations being given to the conditions within crack enclaves and since the early pioneering work of Brown" the subject has attracted much a t t e n t i ~ n " . ~While ~. there can be no doubt of the existence and importance of localised changes in environment composition and potential within crack enclaves in some systems, it is equally clear that such changes are negligible in other systems. This may be expected to be so where the solution is effectively buffered, the solubility of the solvated species very low and, where the cathodic reaction occurs outside the crack, there is negligible current flow through the crack sides. Such conditions appear to hold for the cracking of ferritic steels in carbonate-bicarbonate and in concentrated hydroxide ~ o l u t i o n s ~ ~ ~ ~ ~ . If significant potential changes exist along cracks then it may be expected that the potential range over which cracking is observed will be a function of whether pre-cracked or initially plain specimens are employed for determining the potential range in which cracking is observed. For a ferritic steel in a carbonate-bicarbonate solution, there are no significant differences in the potential range for cracking for either type of specimen, but this is not
MECHANISMS OF STRESS-CORROSION CRACKING
8:21
so for other systems. Thus, with a maraging steel exposed to NaCl solution at initial pH values of 6 or 11, initially smooth specimens failed in two regimes of potential separated by a region, some 300 mV in extent, in which cracking did not occur 38. However, pre-cracked specimens did display environment-sensitive crack growth over the whole range of potentials, as indeed did smooth specimens that were pitted before exposure to those conditions that did not promote cracking in unpitted specimens. This cracking of a maraging steel in two regimes of potential separated by a range of potentials in which cracking did not occur for initially plain specimens is suggestive of cracking by two different mechanisms above and below the range of immune potentials. This has been observed in a number of different systems and has often been interpreted as indicating dissolutionrelated cracking at the higher potentials and hydrogen-related cracking at the lower potentials. However, the necessity for low potentials to discharge hydrogen has often been queried, more especially where localised acidification of the environment can occur in pits or cracks, thereby raising the potential for hydrogen discharge. This is almost certainly the case with some systems, but it is as well to remember that it cannot be so in all systems for the reasons mentioned earlier, that solution composition changes do not invariably occur in cracks. If the mechanism by which stress-corrosion cracks propagate involves dissolution at the crack tip, then crack velocities may be expected to be related through Faraday’s law to the current density at the crack tip according to equation 8.5. Taking the effective current density as the largest difference between fast and slow sweep rate polarisation curves, or the maximum current densities observed in scraping or straining electrode experiments at appropriate potentials, Fig. 8.7 shows a plot of these current densities against observed crack velocities for a variety of stress-corrosion systems, the line shown being that calculated from equation 8.5. Clearly, for a calculation of this type the agreement between observed and calculated crack velocities is very reasonable, especially since the current density measurements do not take account of the structural dependence of the cracking. An implication of the results shown in Fig. 8.7 is that, for that data, the time during which the crack tip was relatively inactive due to the presence of a film must have been a small proportion of the total time, since otherwise the experimental points would fall well below the calculated line, which assumes continuous dissolution. This is because the crack velocity data in Fig. 8.7 are mostly from slow strain-rate tests which, if conducted at an appropriate strain rate, will prevent filming at the crack-tip. If however the strain rate is less than such values then the crack tips will be inactive for times dependent upon the frequency of film rupture. In such circumstances equation 8.5 needs modification to
where Q is the anodic charge (or charge density) passed, cf is the strain to rupture the film and i,, is the crack-tip strain-rate. An expression is available4’ for the crack-tip strain-rate in slow strain rate tests and is of the form
8:22
MECHANISMS OF STRESS-CORROSION CRACKING
C steel in NO; C steel in NOj A C steet in OH1d3 . A C steel in OHFerritic N i - steel in MgCl2 4 C steel in CG7HCO; = l e - 8 (type 3 0 L ) in MgCL2 0 C steel in COlCO2lH20 x AI-7 Mg in NaCl 0 Brass in NH; 10-4 0
0
0
. E -In
c x V
0
a 8
Y u
z
V
lo-‘
lo4
10-~
V
10-1 1 Current density on bare surface (A/cm2)
10
Fig. 8.7 Observed crack velocities and current densities associated with ‘bare’ surfaces. The line is that calculated from equation 8.5 (after Reference 20)
75
(8.9)
N
where N is the number of cracks along the gauge length and iapp is the applied strain rate. (The constants in equation 8.9 will depend upon the material involved and test specimen size.) Because the first term in equation (8.9) dominates at high &, and , the second term at low iaPp, there is little i.e. the crack growth contributes little effect of crack velocity at high kpp; to the crack-tip strain-rate. However, at low k,,, the stress-corrosion crack growth maintains i,, at values that are appreciably higher than would be obtained if the crack growth had been ignored. Various worker^^'-^* have used equation 8.8, or some modified version thereof, to compare observed with calculated crack velocities as a function of strain rate, but Fig8.8 shows results” from tests on a ferritic steel exposed to a carbonate-bicarbonate solution. The calculated lines move nearer to the experimental data as the number of cracks in equation 8.9 is increased, while the numbers of cracks observed varied with the applied strain rate, being about 100 for iaPp10-6s-’, but larger at slower iap, and smaller at higher kPp.
-
MECHANISMS OF STRESS-CORROSION CRACKING
4
8:23
Applled strain rate
--
Crack tip strain rate
3
LOG. STRAIN RATE
lsec
Fig. 8.8 Comparison of calculated and experimental crack velocities as a function of strain rate for a ferritic steel exposed to 1 N Na,CO, + 1 N NaHCO, at -650 mV(SCE) and 75’C (after Parkins”)
While equation 8.8 gives reasonable predictions of the crack velocities for small specimens involving relatively small cracks, there is a further factor that must be taken into account with larger specimens and, more importantly, real engineering structures. This concerns the phenomenon of crack merging or coalescence, a matter that is only beginning to be recognised as important even though it is often apparent from the inspection of service failures. The latter almost invariably involve the multiple initiation of cracks, probably over a relatively long period of time, as in laboratory tests. It is probable that most stress-corrosion cracks cease to grow, especially under realistic loading conditions, after relatively small amounts of propagation, possibly because of work hardening in the crack-tip region and a reduction in the crack-tip strain rate. However, with continued crack initiation, some new cracks may form sufficiently near inactive cracks to reactivate the latter. Obviously this requires that the interacting cracks are sufficiently close together for their respective stress fields to interact. Small merged cracks may later cease to propagate, but with continuing nucleation of new cracks they may later be reactivated, these processes continuing until eventually some cracks will reach a size where the stress-intensity factor for relatively rapid crack growth, Klscc,is reached and the crack velocity will approach that given by equation 8.5. Figure 8.9 indicates schematically these changek in crack velocity with time, which can be quantified in a simple fashion” to compare predictions with observed behaviour. Such comparison with service behaviour indicates the importance of crack coalescence, in the absence of which lifetimes would be markedly greater than sometimes experienced and, with a containing vessel, a leak rather than a rupture would more often occur.
8:24
MECHANISMS OF STRESS-CORROSION CRACKING Fast fracture
t
\
\\
More crack COalesceme
I
3racks initiate - some cease to Prwagafe I
-xditions lor SC(
1
non-exrstent
.
h,tlaton and coaiescence continue
I
,
4verage CV reduces due to increase in crack number Increased work hardemng
I
I
Some cracks coalesce
I I
mlntalnec by stran rate generated by crack growth
'
reached
1
I
1 1 I I
TIME
Fig. 8.9 Schematic illustration of the effect of time of exposure upon stress-corrosion crack velocity
The Function of Stress If crack propagation occurs by dissolution at an active crack tip, with the crack sides rendered inactive by filming, the maintenance of film-free conditions may be dependent not only upon the electrochemical conditions but also upon the rate at which metal is exposed at the crack tip by plastic strain. Thus, it may not be stress, per se, but the strain rate that it produces, that is important, as indicated in equation (8.8). Clearly, at sufficiently high strain rates a ductile fracture may be propagated faster than the electrochemical reactions can occur whereby a stress-corrosion crack is propagated, but as the strain rate is decreased so will stress-corrosion crack propagation be facilitated. However, further decreases in strain rate will eventually result in a situation where the rate at which new surface is created by straining does not exceed the rate at which the surface is rendered inactive and hence stress corrosion may effectively cease. The implications of a significant role for strain rate are wider than the obvious one that stress corrosion should only occur over a restricted range of strain rates. Thus, in constant load tests, since cracks will continue to propagate only if their rate of advancement is sufficient to maintain the cracktip strain rate above the minimum rate for cracking, it is to be expected that cracks will sometimes stop propagating, particularly below the threshold stress. Such non-propagating cracks are indeed observed below the threshold60s6'.Moreover, in constant-load or constant-strain tests, the strain rate diminishes with time after loading, by creep exhaustion if the stress remains sensibly constant, and it is found that the stress-corrosion results are sensitive to the relative times at which the stress and electrochemical
MECHANISMS OF STRESS-CORROSION CRACKING
8:25
65 10-
10-5 Strain ratel s
10'
Fig. 8.10 Effect of strain rate upon the cracking propensity of a Mg-A1 alloy immersed in a chromate-chloride solution
conditions for cracking are established, i.e. creep at constant load, prior to the establishment of the electrochemical conditions for cracking, delays or prevents cracking@*6'. However, the most convincing demonstration of the importance of strain rate is obtained from tests in which the strain rate is superimposed, rather than allowed to vary in the inevitable manner of constant load tests. Figure 8.10 shows the effects of various strain rates applied to a Mg-7Al alloy whilst immersed in chromate-chloride solutions, the tests being conducted to total failure and the maximum load achieved being a sensitive measure of whether or not stress-corrosion cracks were produced@. If stress-corrosion cracks are not produced then failure is by ductile fracture at the normal UTS for the material, but in the presence of stress-corrosion cracks the maximum nominal stress achieved prior to failure is markedly reduced. It is apparent from Fig. 8.10 that stress-corrosion cracking only occurs within a restricted range of strain rates and that at higher or lower values ductile fracture occurs, as confirmed by fractography. Experiments on a carbon steel in a carbonate-bicarbonate solution at a cracking potential with the pre-cracked specimens loaded as cantilevers but with the beam displaced at various rates by a device that replaces the conventional load pan, produced the results shown in Fig. 8.11. The changes in net section stress in these tests at various strain rates amounted to less than a few per cent, but the results clearly indicate a lower limiting strain rate below which crack propagation is not observed, followed by a region in which the intergranular stress-corrosion crack velocity is independent of strain rate and then, at relatively high strain rates, a transition to fast transgranular tearing. The strain-rate independent region-is to be expected since once the strain-rate is sufficiently high to create bare metal at the crack tip at a faster rate than filming can render the bare metal inactive, the factor controlling the crack velocity will be the rate of metal dissolution which is governed by equation
8:26
MECHANISMS OF STRESS-CORROSION CRACKING
10
-.10' E
-E
-u
2>
10
1 U
2
V
10'
lo-
10-837
10-7
10-6
Beom deflection rate ( c m / s l
Fig. 8.1 I
Effect of beam deflection rate of cantilever beam specimens upon stress-corrosion crack velocity of carbon steel in carbonate-bicarbonate solution
8.5. The complementary functions of stress and electrochemistry in this
model, involving the creation of bare metal at the crack tip by plastic strain, imply a strong dependence of the strain rate range for cracking, or of the threshold stress in constant load tests, upon the environmental conditions. Such effects are indeed observed, the curve in Fig. 8.10 being capable of a marked shift along the strain-rate scale according to the composition of the environment and whether or not small anodic or cathodic currents are applied, whilst the limiting beam deflection rate below which cracking is not observed in the experiments to which Fig. 8.11 refers can be changed by two or three orders of magnitude by changes in applied potential. There are indications then that where an active path mechanism is operative, the function of stress in stress-corrosion cracking is to create plastic deformation and therefore that such cracking will be more likely with the lower strength ductile metals. Where the mechanism of cracking involves embrittlement of the metal in the crack-tip region a strain energy argument is involved, and this implies, in relation to equation 8.1, that plastic strain should be minimised and elastic energy maximised for failure, conditions that are most readily met with high yield strength materials. It is well established that the hydrogen embrittlement of steels becomes more marked the higher the yield strength, although changes in structure or composition that result in a change in yield strength, or fracture toughness, may also influence
MECHANISMS OF STRESS-CORROSION CRACKING
8:27
electrochemical reactions, and such parameters as hydrogen diffusivity and these may be as significant as any change in strength in influencing stresscorrosion behaviour. It is also possible that strain rate, as opposed to stress intensity, could be of significance in the stress corrosion of high-strength steels if the environment concerned is one that may lead to filming and the stifling of the reactions that involve the release of hydrogen from the environment or its ingress into the metal. Certainly, there are indications that stress-corrosion cracking in some of the high-strength steels is sensitive to the loading rate and there is a marked similarity to the crack velocitystrain rate curve of Fig. 8.11, and the crack velocity-stress intensity curves6* obtained from tests on precracked specimens in a typical high-strength steel are shown in Fig. 8.12. Curves similar to the latter have been obtained for high-strength aluminium alloys and for titanium alloys, and the question again arises as to whether it is stress intensity or the strain rate that the latter produces that is important, especially in view of results similar to those shown in Fig. 8.10 for a titanium alloy4’.
Conclusion The interdependence of the variables in stress corrosion, namely structure, electrochemistry and response to stress, supports the suggestion that these may interact in a variety of ways and if rationalisation of the situation is to be attempted this is more appropriately achieved through the concept of a continuous spectrum of mechanisms rather than a single mechanism. The critical balance between activity and passivity is altered by changes in the structure and composition of the alloy, the response of the latter to the application of stress through changes in mechanical properties and by changes in the environmental conditions. Thus, if the structure and composition of the alloy are such that almost continuous paths of segregate or precipitate exist, usually at the grain boundaries, and which are electrochemically different from the matrix, then a latent susceptibility to intergranular corrosion may be activated by the presence of stress. In the absence of pre-existing active paths, or even in their presence if other conditions hold, the stress may generate active paths by rupturing a protective surface film or by activating dissolution at emerging slip lines. The transformation from a pre-existing to a strain-generated active path mechanism may result not only from physico-metallurgical change in the state of the alloy, but also from changes in the environmental conditions or the crack-tip strain rate. This greater role of stress or strain in moving away from the pre-existing active path end of the spectrum is continued through to those alloys that undergo local embrittlement of the metal in the crack-tip region. Table 8.1 indicates some of the systems, of metal and environment, that result in stress corrosion, arranged in a series that ranges from those in which the mechanisms are thought to be dominated by dissolution processes to those in which stress, or strain, occupies the more important part of the proposition. While it is very much a matter of opinion as to where specific systems fall in this scheme, perhaps the most significant point about such an arrangement is that it should help to serve as a reminder of the interdependence of the
8:28
MECHANISMS OF STRESS-CORROSION CRACKING
I
1
I
0 Fig. 8.12 Effect of applied stress intensity upon crack velocity for high-strength (180 GN/rn2 UTS) quenched and tempered steel (AFC 77) in distilled water (after Spiede16')
variables and that avoidance of stress-corrosion failure in a specific instance is no guarantee that the preventative action will be equally successful in other circumstances. For instance, whilst nickel additions to a steel are beneficial in relation to caustic cracking, they have little effect upon nitrate cracking and are quite harmful from the viewpoint of cracking in chloride, in that they promote a subsceptibility to cracking in the latter not observed in carbon steels. Thus, the avoidance of cracking by a mechanism occupying one part of the spectrum may induce failure in another part by a different mechanism if the interdependence of the variables is ignored. R. N. PARKINS
Table 8.1
Stress corrosion spectrum
Corrosion dominoted (solution requirements highly specij7c) Intergronulor corrosion steels in
Some AI alloys in C I solns, high potentiols
Fe-Cr-Ni cu-zn o~oys steels in CIin NHFsolns
Intergranular fracture along pre-existing paths
soh
Stress dominoted (solution requirements less specific) Ti olloys in Cu-Zn olloys Mg-AI olloys merhonol. in NO, in c r o ~ o~~oys low , soh c'- solns potentiols
Transgranular fracture along straingenerated paths
+
High steels in C I -
solns
Mixed crack paths by adsorption, decohesion 01 fracture of brittle phase
Brittle fracture
8:30
MECHANISMS OF STRESS-CORROSION CRACKING
REFERENCES 1 . Parkins, R. N., Briiish Corrosion Journal, 7, IS (1972)
2. West, J. M., Metal Science Journol, 7, 169 (1973) 3. Despic, A. R., Raicheff, R. G. and Bockris, J. O M . , Journalof ChemicolPhysics, 49,926 (1968) 4. Dix, E. H., Trans. Amer. Inst. M i n . Met. Engrs., 137, 1 1 (1940) 5. Parr, S. W. and Straub, F. G., Univ. Ill. Bull., 177 (1928) 6. Swann, P. R., Corrosion, 19, 102t (1963) 1. Pickering, H . W. and Swann, P. R., Corrosion, 19, 373t (1963) 8. Swann, P. R., from The Theory of Stress Corrosion Crocking in Alloys, Edited J. C . Scully, NATO, Brussels, 113 (1971) 9. Silcock, J. M. and Swann, P. R., from Environment-Sensitive Fracture of Engineering Moterials, Edited by Z. A. Foroulis, TMS-AIME, Warrendale, Pa, 133 (1979) 10. Nielsen, N. A., Second International Congress on Metallic Corrosion 1 1 . Robertson, W. D., Grenier, E. G., Davenport, W. H. and Mole, V. F., from Physical Metallurgy of Stress-CorrosionFracture, Edited by T. N. Rhodin, Interscience, 273 (1959) 12. Pugh, E. N., Corrosion, 41, 517 (1985) 13. Edeleanu, C. and Forty, A. J., Phil. Mag., 5 , 1029 (1960) 14. Sieradzki, K. and Newman, R. C., Phil. Mag., A, 51, 95 (1985) 15. Sieradzki, K . and Newman, R. C., J. Phys. and Chem. ofsolids, 48, p 1101 (1987) 16. Uhlig, H. B., Ref. 19, p 86 17. Birnbaum, H. K., Ref. 23, p733 18. Galvele, J. R., Corros. Sci., 27, 1, (1987) 19. Staehle, R. W., Forty, A. J. and van Rooyen, D. (eds.), Proc. Conf. on Fundamental Aspects of Stress Corrosion Crocking, NACE, Houston (1%9) 20. Proc. Conf. on Stress Corrosion Crocking and Hydrogen Embriitlement of Iron Base Alloys, NACE (1975) 21. Latanision, R. M. and Pickens, J . R. (eds.), Atomistics of Fracture, Plenum Press New York (1983) 22. Sangloff, R. P. and Ives, M. B., (eds.) Environment-Induced Cracking of Metals, NACE, Houston (1990) 23. Bruemmer, S. M., Meletis, E. I., Jones, R. H., Gerberich, W. W., Ford, F. P. and Stachle, R. W. (eds.) Porkins Symposium on Fundamental Aspects of Stress Corrosion Cracking, TMS-AIME, Warrendale, Pa, (1992) 24. Doig, P. and Edington, J. W., Brit. Corr. J., 9, 88 (1974) 25. Syrett, B. C. and Parkins, R. N., Corros. Sci., 10, 197 (1970) 26. Green, J. A. S., Mengelberg, H. D. and Yolken, H. T., J . Electrochem. SOC., 117, 433 ( 1970) 21. Vermilyea, D. A., Ref. 20, p 208 28. Pugh, E. N., Ref. 19, p 118 29. Yu, J., Parkins, R. N., Zu, Y., Thompson, G. and Wood, G. C. Corros. Sci.. 27, 141 (1987) 30. Sieradzki, K . , Kim, J . S., Cole, A. T. and Newman, R. C., J . Electrochem. Soc., 134, 1635 ( 1987) 31. Poulson, B. S. and Parkins, R. N., Corrosion, 29, 414 (1973) 32. Hoar, T. P., and West, J. M., Proc. R. SOC.,A m , 304 (1962) 33. Swann, P. R. and Embury, J. D., from High Strength Materiols, Edited by V . F. Zackay, Wiley, New York, p. 327 (1965) 34. Staehle, R. W., Ref. 8, p 233 3s. Yu, J., Holroyd, N. J. H., and Parkins, R. N., from Environment Sensitive Fracture: Evoluotion ond Comporison of Test Methods, ASTM STP 821, Edited by S. W. Dean, E. N. Pugh and G. M. Ugianski, p 288 (1984) 36. Coleman, E. G . , Weinstein, D. and Restoker, W., AcIo Met., 9 , 491 (1961) 37. Parkins, R. N., Ref 20, p 601 38. Craig, I. H., and Parkins, R. N., Brit Corr. J., 19, 3 (1984) 39. Oriani, R. A., Corrosion, 43, 390 (1987) 40. Beacham, C. D., Met. Trans., 3, 437 (1972) 41. Daw, M. S. and Baskes. M. I., Sandia Report SAND 86-8863, Sandia Natl. Labs, Albuquerque, N.M. 42. Scully, J . C. and Powell, D. T., Corros. Sci., 10, 719 (1970)
MECHANISMS OF STRESS-CORROSION CRACKING
8:31
Pugh, E. N., Ref. 21, p997 Beck, T. R., Ref. 8, p 6 4 Ebtehaj, K . , Hardie. D. and Parkins, R. N., Corros. Sci., 25, 415 (1985) Parkins, R. N., Corros. Sci., 20, 147 (1980) Parkins, R . N., from The Use of Synthetic Environmentsfor Corrosion Testing, ASTM STP 970, Edited by P. E. Prancis and T.S. Lee, p 132 (1988) 48. Parkins, R . N., Holroyd, N. J. H. and Fessler, R. R. Corrosion, 34, 253 (1978) 49. Congleton, J.. Shoji T. and Parkins, R. N., Corros. Sci., 25, 633 (1985) 50. Brown, B. F., from The Theory of Stress Corrosion Cracking in Alloys, Edited by J. C. Scully, NATO, Brussels, p 186 (1971) SI. Gangloff, R. P., (ed.), Embrifflementby the Localised Crack Environment, AlME (1984) 52. Turnbull, A., (ed.), Corrosion Chemistry within Pits, Crevices and Cracks, HMSO, London (1987) 53. Parkins, R . N.. Craig, I. H. and Congleton, J., Corros. Sci., 24, 709 (1984) 54. Parkins, R. N., Liu, Y. and Congleton, J., Corros. Sci., 28, 259 (1988) 55. Gerber, T. L., Garud, Y. S. and Sharma, S. R., Thermal and Environmental Eflecls in Fatigue: Research Design Interface, ASME, P V P Vol. 71, p 155 (1983) 56. Hudak, S. J . , Jr., Davidson, D. L. and Page R. A., Ref. 51, p 173 57. Ford, F. P., Corrosion/86, Paper No. 327, NACE (1986) 58. Parkins, R. N., Corrosion, 43, p 130 (1987) 59. Parkins, R. N. and Singh, P. M., Corrosion, 46, 485 (1990) 60. Wearmouth, W. R., Dean, G. P. and Parkins, R. N., Corrosion, 29, 251 (1973) 61. Parkins, R. N., Stress Corrosion Cracking- The Slow Strain Rate Technique,ASTM STP 665, Edited by G. M., Ugianski and J. H., Payer. ASTM Philadelphia, Pa, p 5 (1979) 62. Spiedel, M. 0.. Conference on Hydrogen in Metals, NACE, 1975 43. 44. 45. 46. 47.
8.2 Stress-corrosion Cracking of Ferritic Steels
The incidences of stress-corrosion failure in ferritic steels, as with most alloys, continues to increase in frequency with the passage of time, probably as the result of the avoidance of general corrosion, the more efficient use of steels, Le. by employing higher operating stresses, the more extensive use of methods of fabrication that leave relatively high internal stresses in structures, and as diagnostic efficiency has improved. Thus, the cracking of riveted boilers in strong caustic solutions’ has been experienced for over 80 years, and the failure of evaporating equipment containing ammonium nitrate’ for little less. The failure of plant used in the cleaning of coal gas3 and of equipment used in sour oil wells4 created particularly severe problems 30 to 40 years ago, as did the cracking of anhydrous ammonia storage vessels’ a decade later. Other environments that have been associated with stress-corrosion failures in ferritic steels have ranged from ferric chloride solution, through acids ranging from fuming sulphuric to hydrocyanic, to a sodium phosphate solution, a fairly comprehensive list having been provided by Logan6. More recently the cracking of low-strength ferrite steels by carbonatebicarbonate environments has become recognised in chemical process plant ’** and in high-pressure gas transmission pipelinesg. Most of these failures have been associated with cracks that followed an intergranular path, but transgranular fractures have been observed in carbon steels in environments including industrially important H,O-CO-CO, mixtures Io. The compositions and structures of the steels and the properties of the environments involved in these various instances of failure are so widely varying as to suggest that rationalisation of all of these experiences in a single explanation would be difficult if not unreal, i.e. it is probable that a number of different mechanisms are involved (Section 8.1). This is not to suggest that some systematisation is not possible, since from some of the steel environment systems that have been appropriately studied some common trends have emerged.
8:32
STRESS-CORROSION CRACKING OF FERRITIC STEELS
8:33
32(
28(
24( h
?J
E
z
2 20( m m
E
5 16C 9 0
S
m
E
S
12t
e
8C
4(
0
I
I
I
0.04
008
0 12
2
0 16
Carbon (wt%) Fig. 8.13 Effect of carbon content of annealed mild steels upon threshold stress for cracking in boiling 4 N NH4NO3
Effects of Steel Composition and Structure Most of the early work carried out in relation to these aspects of the problem used nitrates as the cracking environment where low-strength steels have been the objects of interest. Consequently most of what follows refers to cracking in boiling concentrated nitrate solutions except where otherwise stated. The medium and higher strength steels, such as involved in sour oil well equipment and other applications, are more frequently tested in chloride- or sulphide-containing environments related to service conditions, but the failure of these steels is dealt with elsewhere (see Section 8.4). For normal commercial-quality mild steels in the annealed or normalised conditions in which they are almost invariably used, various workers have shown that the carbon content of the steel is the major factor determining intergranular cracking susceptibility. Figure 8.13 shows the threshold stresses for a series of commercial mild steels of different carbon contents caused to crack in boiling 4 N NH,NO,. The trend of the result suggests
STRESS-CORROSION CRACKING OF FERRITIC STEELS
0
0
I
I
I
1
I
1
I
0 001
0.008
0.012
0.016
0.020
0.02L
0.028
Carbon (wt."Io)
Fig. 8.14 Effect ofcarbon content of very low carbon steels quenched from 920°C on cracking in a calcium nitrate-ammonium nitrate solution (after Long and Uhlig ")
that pure iron should be more susceptible to cracking than any steel, but the results ' I shown in Fig. 8.14 indicate that cracking susceptibility goes through a minimum as the carbon is reduced to very low levels, and taken with the results shown in Fig. 8.13 demonstrates the important influence of carbon content in relation to the intergranular cracking of steels in nitrates. Such information as exists indicates that similar trends are observed in relation to the cracking of steels in the carbonate solutions that constitute coalgas liquors and in strong caustic solutions. Whilst it appears to be widely accepted that the carbon content of the steel is important, there are differences of opinion as to how the carbon operates. Flis and Scully" and Long and Uhligll regard the effect of carbon on the mechanical properties of the steel as the means whereby this element is important in stress corrosion cracking, whereas others13 regard its electrochemical effects as being more important. Where the cracking is intergranular it appears most probable that segregation of some element or elements to the grain boundaries is likely to be involved and it is well established that carbon can segregate to ferrite grain boundaries. However, other elements can segregate to grain boundariesI4 and it is possible that other species than carbon can promote intergranular cracking of ferritic steels. Thus, it is likely that nitrogen can act in a similar manner to carbon,
STRESS-CORROSION CRACKING OF FERRITIC STEELS
a:35
the most convincing demonstration being that due to Uhlig and Sava Is who, starting with decarbonised electrolytic iron, which was resistant to cracking, found that the introduction of 0.043% N produced marked susceptibility. Increasing amounts of nitrogen in steel appear to decrease resistance to cracking in nitrates, in the same manner as carbon, while there are results available I 6 that suggest essentially similar trends in relation to cracking in boiling hydroxide solutions. Lea and Hondros” prefer phosphorus as the cause of intergranular stress-corrosion cracking of ferritic steels, having defined susceptibility in terms of a ‘fragility index’ (a product of the propensity of an element to segregate to grain boundaries and its relative harmfulness, atom for atom, once at the grain boundary). From tests upon ingots of mild steel to which different elements were added the data are presented as: Fragility index = 20%P
+ 1.9VoCu + 1%Sn + 0.9VoSb + 0.4VoAs
+ 0.3%Zn + 0.2%Ni ( + 700%S + 27%Ca + l%Al) (8.10)
It is claimed that since S, Ca and AI will be present as precipitates they would not in general be detected as grain boundary segregants and their ineffectiveness is indicated by the brackets in equation 8.10. Lea and Hondros do not consider the possible roles of carbon or nitrogen in the cracking of their steels, but from the data obtained phosphorous had the most deleterious effects. More recently Krautschick et ai. measured the cracking responses of iron-phosphorus alloys containing 0.003 to 2 wt% P. They conclude that phosphorus segregation is not necessarily the origin of intergranular stresscorrosion of mild steels in nitrate solutions and that low phosphoruscontaining carbon steels could show susceptibility to cracking. On the other hand, studies by Bandyopadhyay and Briant l 9 involving the exposure of various low-alloy steels to concentrated sodium hydroxide solution show that phosphorus segregation to the grain boundaries in steels containing up to 0.06 wt% P has a markedly deleterious effect on cracking resistance. However, the same authors indicate that carbon and molybdenum also have deleterious effects in these steels, although within the concentration limits they studied, phosphorus segregation had more deleterious effects. The results of Lea and Hondros” showing a deleterious effect from sulphur in relation to cracking by a nitrate solution is interesting in view of the effects of that element in the transgranular cracking of nuclear reactor pressure vessel steels in high-temperature water2’. It is thought likely that the sulphur, which exists mostly as manganese sulphide inclusions, creates a localised environment in the region of inclusions, from which cracks are most often initiated. The thought derives support from the adverse effects of the addition of sulphur anions to the bulk environment. Against such effects of sulphur is the report by Bandyopadhyay and Briant l9 of no effect on the intergranular cracking of their low-alloy steels exposed to hydroxide solution. Where cracks follow intergranular paths due to electrochemical effects related to grain boundary heterogeneity it is likely that selective attack should be observed in such locations by exposure to potent environments even in the absence of applied stress. This has been observed with, low-
8:36
STRESS-CORROSION CRACKING OF FERRITIC STEELS
strength ferritic steels exposed to nitrate, hydroxide or carbonate-bicarbonate solutions. Such attack does not penetrate far along grain boundaries in the absence of stress, but anodic polarisation can cause virtual intergranular disintegration of unstressed steel exposed to a nitrate solution”. Bandyopadhyay et al.22have observed boundary etching in unstressed lowalloy steel, which correlates with phosphorus segregation and cracking propensity as the result of exposure to a sodium hydroxide solution. Such correlations between intergranular stress corrosion cracking propensity and selective attack in grain boundary regions would appear unlikely if the role of grain boundary segregants was solely related to mechanical effects at boundaries, although that is not to imply that such effects are of no consequence in crack growth. The sometimes contradictory results from different workers in relation to the elements mentioned above extends to other elementsz3. Some of these differences probably arise from variations in test methods, differences in the amounts of alloying additions made, variations in the amounts of other elements in the steel and the differing structural conditions of the latter. Moreover, the tests were mostly conducted at the free corrosion potential, and that can introduce further variability between apparently similar experiments. In an attempt to overcome some of these difficulties, slow strain-rate tests were conducted on some 45 annealed steels at various controlled potentials in three very different cracking environmentsz3since, if macroscopic
- 0.5
-0 7
-0.6 Potential
Fig. 8.15
V
-0 8
SCE
Effects of potential upon the stress-corrosion cracking of various steels in C03-HCO, solution in slow strain rate tests (after Parkins ef 0 1 . ~ ~ )
STRESS-CORROSION CRACKING OF FERRITIC STEELS
8:37
0 +
0.2
0
-0.2
-0.4
-0.8
-0.6
Potential
V
-1.0
-1.2
SCE
Fig. 8.16 Effects of potential upon the stress-corrosion cracking of various steels in boiling 8.75 N NaOH in slow strain rate tests (after Parkins et a/.23)
electrochemical properties play an important role in determining cracking response, there is no reason to expect that the effects of different alloying elements will be the same irrespective of the environment. The results are expressed in terms of the effect of applied potential upon the time-to-failure ratio, the latter being derived from the time-to-failure in the test solution divided by the time-to-failure in an inert environment (oil) at the same temperature. A ratio of 1 indicates no susceptibility to cracking and increasing departure of the ratio from 1 indicates increasing susceptibility. Figure 8.15 indicates the beneficial effects that may be derived from additions of specific amounts of chromium, nickel or molybdenum in relation to cracking by a carbonate-bicarbonate solution. Increasing amounts of those elements had the effect of increasing resistance to cracking in that environment, a point returned to below. However, not all alloying additions to ferritic steels are invariably beneficial. Figure 8.16 shows that, while chromium and nickel additions were beneficial in relation to cracking by a sodium hydroxide solution, silicon and molybdenum had quite the reverse effect, particularly in extending the potential range over which cracking was observed. The areas bounded by curves such as those in Figs. 8.15 and 8.16 provide a convenient means of conducting a regression analysis on all of the data. This provides a stress-corrosion index (SCI) that reflects the effects of potential and the severity of cracking, indicating beneficial or deleterious effects according to the direction of change. The regression analysis for the tests in sodium hydroxide gave SCI,,
= 105 - 45VoC - 40VoMn - 13.7VoNi - 12.3VoCr - 11VoTi
+ 2.5VoAI + 87VoSi + 413%Mo
(8.1 1)
8:38
STRESS-CORROSION CRACKING OF FERRITIC STEELS
reflecting the beneficial effects (negative coefficients) shown in Fig. 8.16 for specific additions of chromium or nickel and the deleterious effects (positive coefficients) from specific additions of silicon and molybdenum. The corresponding equations from tests in the nitrate and the carbonatebicarbonate solutions were SCI,,,
= 1777 - 996VoC - 390VoTi - 343VoAl( - 132VoMn)
- 111VoCr - 90VoMo - 62V0Ni + 292VoSi
(8.12)
and SCIco, = 41 - 17.3VoTi - 7.8VoMo - 5.6VoCr - 4.6VoNi( - 2.9VoMn) ( 1.7VoSi) ( + 5.6VoA1) ( + 15VoC) (8.13)
+
When the t ratio (coefficient/standard error of the coefficient) was less than 2 for any element, the latter is bracketed in the regression equation, implying that only the remaining elements should be regarded as having significant effects upon the cracking propensity. The SCI values in the different environments reflect the decreasing potential range for cracking and the decreasing severity of cracking in the order nitrate, hydroxide, carbonatebicarbonate. It is probable that several factors are involved in the effects reflected in equations (8.11), (8.12) and (8.13). Thus the coefficients indicate that chromium, manganese and titanium, additions are consistently useful for all three environments, followed by nickel and aluminium in terms of effectiveness, with silicon appearing to be consistently objectionable. This approximate order of merit bears some relationship to the carbide-forming tendencies of the alloying elements. However, it is clear that this is not the only factor that determines the effectiveness of these alloying elements in relation to cracking propensities, since molybdenum is more effective than chromium or manganese as a carbide former, but this is only reflected in the cracking resistance that molybdenum confers in relation to the carbonatebicarbonate environment, its effect in relation to cracking by hydroxide being deleterious. The electrochemical influences of the alloying elements also appear to be reflected in their effects upon cracking response, with both the dissolution and filming tendencies operative. Thus, there is a general correlation between the effects of the alloying additions upon cracking behaviour and the corrosion of these elements in solutions of similar pH. Aluminium, molybdenum and titanium are well known to show good corrosion resistance in more neutral solutions, with poorer resistance in strongly alkaline or acid environments, except for oxidising acids in the cases of aluminium and titanium. The stress-corrosion results broadly reflect such behaviour. Similarly, nickel is well known to show increased corrosion resistance with increasing pH, while silicon is a very reactive element over a wide range of pH values especially in hot solutions. Notwithstanding possible explanations of the effects of these various alloying elements, perhaps the most important message arising from the effects reflected in equations (8.11). (8.12) and (8.13) is that the effects of alloying elements vary with the environment. It follows that a steel resistant to cracking in one environment may not be resistant in others, with the effect
8:39
STRESS-CORROSION CRACKING OF FERRITIC STEELS
8I
,
0.8
,
0 Ia a w
0.6.
U =?
w
T I-
0.2 0 0 A-
8 A-
Ni steel - C-Mn steel
I
- 0.2
-0.4
-0.6
POTENTIAL
-0.8
-1.0
V(SCE)
Fig. 8.17 Effects of applied potential upon the time to failure ratio in slow strain rate tests of C-Mn steel, with and without a 6% nickel addition, in boiling 8 M NaOH, I M NaHCO, + 0.5 M Na2C03 at 75"C, and boiling 4.4 M MgCI, (after Parkins et 0 1 . ~and ~ Poulson and park in^^^)
of molybdenum in markedly improving the cracking resistance in carbonatebicarbonate but being markedly deleterious in relation to hydroxide solutions making that point. Similar effects may be observed in relation to nickel additions. Thus, when almost 6% nickel is present, the resistance to cracking at various potentials by both hydroxide and carbonate-bicarbonate solutions is good, and considerably better than without the presence of nickel, as is apparent from Fig. 8.17. However, in boiling 42% magnesium chloride solutions the nickel-containing steel cracks vary readily24,yet an unalloyed steel shows no propensity for failure in such a solution. Clearly an alloyed steel developed to have low susceptibility to stress-corrosion cracking in a particular environment will not necessarily show such behaviour in a different environment, a rather obvious point widely recognised in relation to other forms of corrosion but not always recognised in the context of stress corrosion.
The Effects of Heat Treatments For steels that are most frequently used in the annealed or normalised condition the most important structural parameter that can be influenced by heat treatment is the grain size, although the extensive use of welding as a means of fabricating mild steels means that martensitic and tempered martensitic
8:40
STRESS-CORROSION CRACKING OF FERRITIC STEELS
400
320 -
1 160E
si
v I y.s. s.c f s 5 % f.s
x
0
I
I
I
I
I
I
1
I
1
2
3
4
5
6
7
6
1
I-bfmm’h) Fig. 8.18 Effects of grain size on lower yield stress, 5 % flow stress and stress-corrosion fracture stress for 0.08VoC steel in 8 N Ca(NO,), (after Henthorne and Parkins*’)
structures may also be encountered. That ferrite grain size has an effect upon stress-corrosion propensity is apparent in the results shown in Fig. 8.18, from which it is clear that coarse-grained steels fracture at appreciably lower stresses than those of smaller grain size. Such results may be interpreted in a number of ways from the more likely saturation of grain boundaries by segregate of limited quantity if the grain size is large, to the effect of grain size on the mechanical properties of steel, which as has already been mentioned are matters of importance in stress-corrosion cracking. The effect of relatively fast rates of cooling from the austenitising temperature is more marked than the grain size effect achieved by differing austenitising temperatures. Thus water quenching from 920°C appears to render steel more susceptible t o cracking than oil quenching, and further decreases in the cooling rate through air cooling to furnace cooling further increase cracking resistance. However, it needs to be stressed that these trends are relative and that even with very slow cooling, especially from high austenitising temperatures, many ferritic steels are very susceptible to stress corrosion in certain environments. The effects upon cracking tendencies of tempering following quenching is, in general, for the marked susceptibility of the water-quenched condition to be mitigated if the tempering temperature is high enough. However, there are some other differences between the results published by various workers. Houndrement, et a/.25agree with most other workers in showing that tempering above about 300°C increases cracking resistance and that the benefits are maximised when the tempering temperature is 600°C or above. On the other hand, the results of Uhlig and Sava’j show the full benefits of tempering at temperatures from 250°C upwards with a return to marked
8:41
STRESS-CORROSION CRACKING OF FERRITIC STEELS
susceptibility at 700°C. The latter temperature is dependent upon the time of tempering, marked susceptibility returning after tempering at only 500°C if this is carried out for about 10 h or more. The effects of tempering quoted by Long and LockingtonZ6for a 1% Mn alloy with very little carbon are in complete contrast to the results just mentioned. They show that tempering above 200°C increases susceptibility initially and that increasing resistance to cracking begins to be observed when the tempering temperature exceeds about 500°C. The explanation for the variability in these results may lie in the differing carbon contents of the steels used or in the test methods employed, but none of the papers quotes any results from structural studies upon the steels following the various heat treatments. That the different carbon contents of the steels used by these various workers is a factor in these apparently contradictory results on the effects of tempering quenched carbon-steels is apparent from a studyz7of a range of such steels tested in a nitrate solution. Figure 8.19 shows the threshold stress values as a function of carbon content for the range of steels in the annealed and water-quenched conditions. The deleterious effects of water quenching upon the cracking resistance of the higher carbon steels is readily apparent, whilst with carbon contents below about 0.1070 increased resistance to cracking is observed. Those data refer to constant strain tests, but slow strain-rate tests showed the same trends. The implications of the results shown in Fig. 8.19 are that subsequent tempering may be expected to increase susceptibility of the lower carbon steels but decrease that of the higher carbon materials. Figure 8.20 shows that these trends were observed
I
I
I
I
0.16
0.08
I
J 0.24
% CARBON Fig. 8.19 Threshold stresses in a boiling nitrate solution for annealed and quenched steels of different C contents (after Parkins et 0 1 . ' ~ )
8:42
STRESS-CORROSION CRACKING OF FERRITIC STEELS
I
1
I
1
I
Fig. 8.20 Effects of different tempering times upon the time to failure ratio of two steels tempered at various temperatures (after Parkins ef a/.”)
in tests upon tempered specimens of two of the steels, with similar trends shown with the other steels in the series, depending upon their carbon content. Obviously tempering the 0.05VoC steel at temperatures in the region of 700°C causes a marked deterioration in cracking resistance, that temperature being lowered with longer tempering times, confirming the trends observed by Uhlig and Sava” using a 0.06VoC steel. On the other hand, with the 0.15VoC steel, tempering at 700°C for 1 hour gave the highest cracking resistance in agreement with the results of Houdrement etal.” using a 0.26VoC steel. These various effects of quenching and tempering treatment upon cracking tendency appear to correlate with microstructural changes”. High susceptibility is associated with simple, unbranched, crack paths and relatively high crack velocities and occurs at the prior austenite grain boundaries of the higher carbon quenched steels and at the recrystallised ferrite grain boundaries of steels tempered above 500°C for the low carbon contents and above 700°C for the higher carbon materials, for 1 h treatments. High resistance to cracking is associated with lower crack velocities and a marked tendency for multiple branching to develop for short lengths along lath boundaries. These effects are observed in quenched low-carbon steels, where the main cracks follow the prior austenite boundaries and branches develop along the lath boundaries containing auto-tempered carbides, and in the higher carbon steels when tempered to precipitate carbides in the lath boundaries but without recrystallisation occurring. Repeated branching may be expected to reduce the rate of crack propagation, to extend the failure time or increase the chances of a crack ceasing to propagate by the accumulation of corrosion products in the enclave. When the tempering temperature is
STRESS-CORROSION CRACKING OF FERRITIC STEELS
a:43
sufficiently high, and the time sufficiently prolonged, susceptibility increases, in agreement with the observation that prolonged subcritical annealing of pearlite structures to promote carbide spheroidisation at the ferrite grain boundaries increases susceptibility to cracking2". The effects upon cracking tendency from tempering higher carbon mild steels mentioned above in relation to cracking by nitrate solutions have also been observed for cracking by carbonate-bicarbonate solutions. In addition, Bandyopadhyay et af.22 have commented upon the role of preferential attack on large chromium-rich carbides in blunting cracks and reducing crack velocities in low-alloy steels exposed to hydroxide solutions.
Effects of Environment Composition It has frequently been stated that the environmental requirements for stress corrosion cracking are highly specific, but the relatively extensive list of environments that have been reported as promoting cracking raises queries as to the validity of such statements. It is clear that cracking environments are specific in the sense that not all possible environments promote cracking, but to state that the solution requirements are highly specific may lead to a false sense of security in certain practical situations. It is clear that the propagation of a stress-corrosion crack requires the reactions that occur at the crack tip to proceed at a considerably faster rate than any dissolution processes that take place at the exposed surfaces of the metal, including the crack sides, since otherwise general corrosion or pitting only will be observed. For an inherently reactive metal like mild steel most of the exposed surface will only remain inactive if the surface is passivated, and so environments in which stress corrosion occurs are likely to have considerable oxidising potential. Nitrates and hydroxides, which are of course anodic inhibitors of the corrosion of carbon steels in appropriate circumstances, have such characteristics and are those anions associated with the earliest identified instances of cracking of mild steels. Other anodic inhibitors of the corrosion of such steels are also capable of promoting stress-corrosion cracking in appropriate circumstances. Thus, carbonate-bi~arbonate~~ and phosphate solutions30 promote dissolutionrelated cracking in certain potential ranges which, as with nitrates and hydroxides, can be predicted by appropriate electrochemical measurements (Section 8.1). Figure 8.21 shows the current density differences between fast and slow sweep-rate polarisation curves (Fig. 8.5) at various potentials for mild steel immersed in hydroxide, carbonate-bicarbonate and nitrate solutions. Also shown in Fig. 8.21 are the results from controlled potential slow strain-rate tests involving the same solutions, and it is clear that the potential ranges in which cracking occurred are those predicted for each of the three solutions from the electrochemical measurements. Moreover, the severity of cracking, reflected in the time-to-failure ratio, reflects the magnitudes of the current density differences for the three solutions at different potentials, reflecting the trend shown in Fig. 8.7. The potentiodynamic polarisation curves measure the tendencies for the occurrence of dissolution and filming processes, i.e. the combination that is required to promote and retain crack geometry for dissolution-related cracking (Section
8144
STRESS-CORROSION CRACKING OF FERRITIC STEELS
1000
-
cu 100-
6
\
a
E
-
io-
I 1
-i.2
L
- 06
I
I
0
Potential
+ 0.6
I
I
i1.2
V (SCE)
Fig. 8.21 Current density differences between fast and slow sweep rate polarisation curves and stress corrosion cracking suspectiblity as a function of potential for a C-Mn steel in nitrate, hydroxide and carbonate-bicarbonate solutions
8. l), and for low-alloy ferritic steels in a variety of environments they appear to give reasonable indications of cracking propensities, hence they may be useful in assessing the potencies of environments for which cracking data do not exist.
STRESS-CORROSION CRACKING OF FERRITIC STEELS
8:45
The above mostly refers to intergranular cracking, but there are some media that promote transgranular cracking in ferritic steels. It is likely that the mechanisms of failure in these cases are different from those for intergranular cracking, with strain-generated active paths or localised embrittlement of material in the crack-tip region playing important roles. Perhaps the most topical of these is the cracking of ferritic steels by high temperature water 20. Although not universally accepted, the evidence tends to support a dissolution-related mechanism of crack growth, rather than one based upon hydrogen ingress. Transgranular cracking of low-carbon steels has been reported as occurring in HCN containing 2.6-3.5 g/l of HCN, and in FeC1, solutions and chloride-bearing slurries containing ferric oxides and hydrated oxides at 3 16°C 30. C02-CO-H20 environments promote transgranular failure", increasing quantities of CO, some of which must be present for cracking to be initiated, decreasing the time-to-failure and lowering the stress required". The results may be interpreted in terms of the CO acting as an inhibitor of the attack upon iron by the acidic C02-H20, a passive film of increasing effectiveness being formed as the proportion of CO is increased, which in turn suggests a film rupture mechanism. Moist H,S has been reported34as causing the failure of cold-drawn high-carbon steel wire when loaded to only 40% of its breaking load in air, although this failure may have resulted from hydrogen-embrittlement in the light of the work that has been carried out in relation to failure in sour oil well equipment 35. In some environments it appears likely that more than one mechanism of failure may exist, depending upon the potential. Thus ammonium carbonate solutions, which promote intergranular cracking at higher potentials can give rise to transgranular cracking due to a dissolution-related mechanism at somewhat lower potentials, and at much lower potentials to hydrogenrelated cracking in slow strain-rate tests. Sodium hydroxide solutions also can promote transgranular hydrogen-induced cracking at potentials appreciably below those that sustain intergranular cracking. Such hydrogeninduced cracking in mild steels in various environments at sufficiently low potentials, although readily produced in slow strain-rate tests, is not normally regarded as a significant problem in industrial situations and so the results mentioned may simply reflect the severity of the slow strain-rate test method. However, it is possible that with cyclic loading, as opposed to the static loading conditions assumed to operate in plant that displays stress corrosion, the environment-sensitive cracking due to hydrogen reflected in slow strain-rate tests may be of relevance. Of course, with high-strength ferritic steels hydrogen-embrittlement is often regarded as the only mechanism of environment-sensitive fracture, for static or dynamic loading. But the two ranges of potentials for cracking, often separated by a regime in which there is insensitivity to cracking, mentioned above in relation to lowstrength steels, are sometimes observed with high-strength varieties and suggest to some that more than one cracking mechanism may exist for such materials.
8:46
STRESS-CORROSION CRACKING OF FERRITIC STEELS
Effects of Additions to Cracking Environments From the indications already mentioned, that cracking environments render most of the exposed surface inactive whilst allowing dissolution to soluble species at the crack tip when the potential is within the appropriate range, it is possible to categorise the effects of additions to cracking environments as follows. Where cracking occurs at the free corrosion potential, additions to the environment may cause the potential to move outside the cracking range and so prevent failure. Conversely, the free corrosion potential may normally not coincide with the cracking range and certain additions to the environment may result in cracking simply because they cause the free corrosion potential to move into the critical potential range. Alternatively, where the potential remains within the cracking range, additions may influence cracking by their effect upon the passivation or dissolution reactions or both36.Such an approach permits rationalisation of the confusing situation that appears to exist from a reading of the literature. Thus, in relation to the cracking of carbon steels in caustic alkalis, laboratory studies have frequently shown that cracking could not be reproduced unless the solution was contaminated with oxidising salts” or with oxygen itself3’. The picture is further complicated by the fact that, while substances such as KMnO,, NaNO, and Na,CrO, added to NaOH solutions promote cracking at the boiling point, at 250°C they act as inhibitors. Similarly, whilst a little dissolved oxygen apparently promotes cracking, a high concentration prevents failure 38. The effects of additions in causing the free corrosion potential to move in relation to the potential range for cracking would appear to have considerable practical significance. Thus, the addition of relatively small amounts of nitrates to concentrated NaOH solution, an approach that is sometimes used in treating boiler feed waters in an attempt to avoid caustic cracking, causes the free corrosion potential to become significantly more positive than the cracking potentials in alkali, and failure is no longer observed. However, the same addition will not prevent cracking if the potential is maintained, by potentiostatic control, within the cracking range, so that nitrates should be regarded as unsafe inhibitors of caustic cracking. Na,SO, should also be placed in this ~ a t e g o r y -even ’ ~ though the maintenance of a Na,SO,/NaOH ratio in boiler feed water in excess of 2.5 is still widely practised as a means of preventing caustic cracking in boilers. In some circumstances additions to environments may promote cracking by causing the free corrosion potential to move into the cracking range. Thus, the free corrosion potential of some mild steels is likely to lie at the boundary of the cracking range in hydroxides so that stress corrosion does not occur. The addition of a small quantity of lead oxide causes the free corrosion potential to move into the cracking range and failure then occurs readily at the free corrosion potential, i.e. without external potentiostatic control. Similarly, the well established effect of oxygen in high temperature pure water in promoting cracking in nuclear reactor pressure vessel steel is probably the result of the oxygen raising the potential into the cracking range3’. Whilst some additions operate in the manner already indicated it is clear that other substances influence cracking propensity not so much by any effect they may have upon potential, but by modifying the reactions involved in
STRESS-CORROSION CRACKING OF FERRITIC STEELS
8:41
cracking. Tannins and phosphates, and to a lesser extent silicates, prevent cracking in strong NaOH solutions even when the potential is maintained within the cracking range36. The reasons for this type of behaviour have not been studied but the relatively high anodic current peak shown by mild steel in 30% NaOH is reduced, by about two orders of magnitude, in the presence of NaH,PO,, and to a lesser extent by Na,SiO,, which may indicate that these substances operate by hindering the dissolution of ferrite. Tannins have little effect upon the anodic polarisation curve when added to NaOH and their influence in stopping cracking may be related to interference with the cathodic reaction. Results from various laboratories have shown that additions to nitrate environments appear to influence cracking according to whether they affect the pH of the solution, are oxidising substances or form insoluble products with iron3’. The effect of the pH upon the cracking potency of nitrates is apparent from studies in which various cations have been incorporated in the solution. Nitrates of acidic cations are the most potent at equivalent strengths and temperatures and this suggests that the cracking capacity of any nitrate solution may be controlled by adjustment of its pH. This is found to be so, small additions of HNO, to lower the pH of a given nitrate increasing the cracking tendency, whilst raising the pH with a sufficient quantity of OH- ions will prevent cracking at the free corrosion potential. This effect of OH- ions is related to their influence on potential however, since with potentiostatic control in the cracking range, nitrates at a pH of 10 or so will promote failure. Oxidising additions such as KMnO, and K,Cr,O, all accelerate cracking in nitrates, whilst substances such as H3P0,, Na,HPO, and CO(NH,),, which may be expected to form insoluble products with iron, retard or prevent cracking. One of the difficulties associated with much of the data on the effects of additives to potent stress-corrosion cracking environments is that they are the result of ad hoc experiments, often without recognition of the importance of potential upon cracking. Free corrosion potentials in laboratory tests can be very different from those that exist in service situations for the same metal-environment combination. Moreover, in view of the balance required between dissolution and passivity for cracking the concentration of an inhibitive substance may be critical. The point is illustrated by the results shown in Fig. 8.22, which indicates the average crack velocities for mild steel specimens tested in a carbonate-bicarbonate solution with various additions of sodium chromatew. Cracking is inhibited for all practical purposes by the addition of 0.16 wt% Na,CrO,, but Fig. 8.22 shows that at about 0.017% Na,CrO, there is enhanced cracking beyond that observed, with lesser, including zero, additions of chromate. The explanation for such a result, which has been observed with other substances, is that, while sufficient inhibitor will ensure that only passive behaviour is observed, at some intermediate concentrations the grain surface may be effectively passivated but the more active sites, the grain boundaries, are not so protected, thereby increasing the current density in such regions. Clearly it is important to ensure an adequate supply of inhibitor, and Fig. 8.22 also offers a possible explanation for the contradictory statements in the literature of the effects of some additions to cracking environments.
8:48
STRESS-CORROSION CRACKING OF FERRITIC STEELS
L
f 30-
E, l-
5
20-
9w
>
I.: 1-i--... 04 0
I
0.01
002
003
CONCENTRATION OF Na+r04
/--
0.08 C 6
wt%
Fig. 8.22 Average crack velocities observed in mild steel specimens tested in 0.5 M Na,C03 + 1 M NaHCO, at 75°C with various additions of Na,CrO,. Results refer to potential of most severe cracking at each chromate concentration; variability in crack velocity in replicate tests shown by lengths of scatter bars (after Terns and Parkins@)
Methods of Prevention The incidence of stress-corrosion cracking requires a susceptible alloy to be exposed to a specific environment at stresses above some limiting value, from which it follows that control of the problem may be through manipulation of any or these three parameters4'. In more detail the choices are: Metallurgical control
-change alloy composition -change alloy structure -use metallic or conversion coating Environmental control -apply anodic or cathodic protection -remove offending species -add inhibitor -use organic coating -modify temperature Mechanical control -reduce operating stresses -relieve fabrication stresses -avoid stress concentrators -introduce surface compressive stresses In a particular service situation, other considerations may preclude some approaches to prevention; for example, weight, strength or economic requirements may dictate the use of an alloy susceptible to cracking. More-
STRESS-CORROSJON CRACKING OF FERRITJC STEELS
8:49
over, the philosophy associated with prevention can differ according to the use of the structure or component involved. Thus, the attitude to slow crack growth in a pressure vessel is likely to be appreciably different from that where the consequences of cracking are no more than the seepage of an innocuous liquid from cracks. Ideally, approaches to prevention should begin at the design stage, but it is not infrequently the case that cracking occurs in an existing plant when such failure had not been anticipated. In such circumstances the approaches to prevention are likely to be restricted and in some situations it may be necessary to accept the continuance of cracking but still attempt to control it by minimising crack growth rates. Some of the implications for prevention or control can be deduced from the factors discussed earlier, but some of the approaches listed above need further brief mention, particularly some of those mentioned under mechanical control. Many practical instances of stress-corrosion cracking in the lower strength ferritic steels result from the presence in the structure of residual stresses, which are usually in the region of the yield stress, rather than the normally appreciably lower design stresses. Thermal stress relief is likely to be beneficial in such cases, but applying full stress relieving heat treatments to structures can present problems due to inadequate furnace capacity for large fabrications or distortion of the latter at the temperature involved (usually about 650°C for ferritic steels). However, partial stress relief by heating to lower temperatures than that required for full stress relief can be adequate in those cases where the residual and operating stresses can be reduced below the threshold stress4’. Moreover, these lower temperatures can be achieved with less distortion for furnace annealing or by locally applied heating with large structures. Since stress corrosion requires the presence of tensile stresses of appropriate magnitude it follows that if compressive stresses are introduced into a surface, cracking should not occur. Shot peening and grit blasting have been shown to be effective in preventing or reducing the incidence of environment-sensitive cracking in ferritic steels and hammer peening, which also leaves surfaces in compression when properly applied, can have similar effects4*.It is also sometimes possible to leave surfaces at which cracking would otherwise occur in a state of compression by localised heating techn i q u e ~ These ~ ~ . have been developed in relation to the stainless steel pipe cracking problem in boiling water reactors, but the principle should be equally applicable to ferritic steel structures of appropriate geometry. While plant operating conditions are largely dictated by considerations other than that of control of environment-sensitive cracking, it is worth noting that pressure or temperature fluctuations are likely to induce stress cycles that lower the threshold stress for cracking below that obtained with static loading. Obviously, unnecessary pressure or temperature excursions should be avoided or minimised. While temperature variations leading to thermal stresses may aggravate the problem, especially during plant start-up or shut-down, temperature is an important parameter in another sense in stress corrosion. Thus, the conditions for dissolution-related cracking and the crack velocity are typical thermally activated processes in most instances of cracking in ferritic steels, so that lower temperatures are less likely to result in cracking or be associated with lower crack velocities.
8:50
STRESS-CORROSION CRACKING OF FERRITIC STEELS
While ideally structures should be designed and fabricated so that environment-sensitive cracking is avoided, in practice it is sometimes necessary to live with the problem. This implies an ability to detect and measure the size of cracks before they reach the critical size that may result in catastrophic failure. Such inspection has important implications for plant design, which should be such as to allow inspection at relevant locations. The latter are regions of high residual stress (welded, bolted or riveted joints) and regions of geometrical discontinuity (notches, crevices, etc.) where stress or environment concentration may occur. R. N. PARKINS REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
17. 18. 19. 20.
21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.. 36. 37. 38. 39.
Stromeyer, C. E., J. Iron St. Inst., 79, 404 (1909) Jones, J. A., Trans. Faraday Soc., 17, 102 (1921) 1st Rept. Inst. Gas Engrs., Comm.No. 398, Brit. Welding Res. Assoc. F.M.9 Ctee (1951) Fraser, J. P., Eldredge, G. C. and Treseder, R. S . , Corrosion, 14, 517t (1958) Phelps, E. H. and Loginow, A. W., Corrosion, 18. 299t (1962) Logan, H. L., The Stress Corrosion of Metals, Wiley, New York, pp. 5-7 (1966) Parkins, R. N., Alexandridou, A. and Majumdar, P., Mats. Perf., 25, 20, (1986) Parkins, R. N. and Foroulis, Z. A., Mats. Perf., 27, 19 (1988) Parkins, R. N. and Fessler, R. R., Materials in Engineering Applications, I, 80 (1978) Kowaka, M. and Nagata, S., Corrosion, 24, 427, (1968) Long, L. M. and Uhlig, H. H., J. Electrochem. Soc., 112,964, (1%5) Flis, J. and Scully, J. C., Corrosion, 24, 326 (1968) Parkins, R. N. and Green, J. A .S., Corrosion, 24, 66,(1%8) Hondros, E. D. and Seah, M. P., Int. Met. Rev., 22, 262, (1977) Uhlig, H. H. and Sava, J., Trans. A.S.M., 56, 361 (1963) Bohmenkamp, K. Proc. Conf. on Fundamental Aspecrs of Stress Corrosion Cracking, Edited by R. W. Staehle, A. J. Forty and D. van Rooyen, NACE, Houston, p 374, (1969) Lea, C. and Hondros, E. D., Proc. Roy. Soc., 377A. 477 (1951) Krautschick, H. J., Grabke, J. H. and Diekmann, W., Corros. Sci., 28, 251 (1988) Bandyopadhyay, N. and Briant, C. L., Corrosion, 41, 274 (1985) Scott, P. M., Proc. 3rd Symp. on Environmental Degradation of Materials in Nuclear Power Systems - Water Reactors. Edited by G. J. Theus and J. R. Weeks, AIME - The Metallurgical Society Inc., Warrendale, Pa., p 15, (1987) Henthorne, M. and Parkins, R. N., Brit. Corrosion J., 5, 186 (1967) Bandyopadhyay, N., Briant, C. L. and Hall, E. L., Met. Trans. A . , 16, 1333, (1985) Parkins, R. N., Slattery, P. W. and Poulson, B. S., Corrosion, 37, 650 (1981) Poulson, B. S. and Parkins, R. N. Corrosion 29, 414 (1973) Houdrement, E., Bennek, H. and Wentrup, H., Stahl und Eisen, 60,575 and 791 (1940) Long, L. M. and Lockington, N. A., Corros. Sci., 7 , 447 (1967) Parkins, R. N., Slattery, P. W., hliddleton, W. R. and Humphries, M. J., Brit. Corrosion J . , 8, 117 (1973) Parkins, R. N., J. Iron Steel Inst., 172, 149 (1952) Sutcliffe, J. M., Fessler, R. R., Boyd, W. K. and Parkins, R. N., Corrosion 28,313 (1972) Parkins, R. N., Holroyd, N. J. H. and Fessler, R. R., Corrosion, 34, 253 (1978) Congleton, J., Shoji, T. and Parkins. R. N., Corros. Sci., 25, 633 (1985) Huckholtz, H. and Pusch, R., Stahl und Eisen, 62, 21 (1942) Brown, A., Harrison, J. T. and Wilkins, R., Corros. Sci., IO, 547 (1970) Rees, W. P., Symposium on Internal Stresses in Metals and Alloys, Inst. of Metals, London, p 333, (1948) Schultz, A. E. and Robertson, W. D., Corrosion, 13, 33 (1957) Humphries, M. J. and Parkins, R. N., Corros. Sci., 5, 747,(1%7) Schroeder, W. E., Berk, A. A. and O'Brien, R. A., Metals and Alloys, 8 , 320 (1937) Radeker, W. and Grafen, H., Stahl und Eisen, 76, 1616 (1956) Parkins, R. N., from Stress Corrosion Cracking and Hydrogen Embrittlement of Iron Base Alloys, Edited by R. W.Staehle, J . Hochmann, R. D. McCright and J. E. Slater, NACE, Houston, p a l , (1977)
STRESS-CORROSION CRACKING OF FERRITIC STEELS
8:51
40. Terns, R. D. and Parkins, R. N., Proc. 5th European Symp. on Corrosion Inhibitors, Ann. Univ. Ferrara, p 857, (1975) 41. Parkins, R . N . , Mars. Perj. 24, (8),9-20 (1985) 42. Pearson, C. E. and Parkins, R . N., Welding Research, 3, 95t (1949) 43. Danko, J. C., Paper N o 162, CORROSION/84, NACE, Houston, (1984)
8.3 Stress-corrosion Cracking of Stainless Steels
lntoduction For the purpose of this section, stainless steels will be assumed to cover the group of alloys which rely mainly on the addition of chromium to iron to impart corrosion resistance. Even within this restricted group there are many alloy types with very different microstructures and mechanical properties and a wide range of susceptibility to stress-corrosion cracking. It is convenient to subdivide these stainless steels into groups under the general headings: martensitic, ferritic, duplex and austenitic, although there are some high-strength grades produced by precipitation hardening heat treatments that could be considered as an additional group. Other groupings are possible, but the above is convenient for discussing resistance to stresscorrosion cracking. The phenomenon of stress-corrosion cracking can be defined as the occurrence of macroscopic brittle fracture of a normally ductile metal due to the combined action of stress and some specific environment. The environment need not be chemically aggressive in that high general dissolution rates are not required and the phenomenon is complicated by the fact that many different mechanisms can give rise to such cracking. It is often difficult to differentiate between the roles of anodic dissolution and hydrogen absorption on cracking. Also, microstructural changes such as the creation of martensite in the region of the crack tip, and the effect on crack propagation of quite small alternating stresses superimposed on a mean stress, further complicate attempts at complete understanding of stress-corrosion cracking. These aspects are covered in Section 8.1. In this section, the effects of gaseous hydrogen and of deliberate cathodic charging with hydrogen will not be discussed. Even excluding such instances of hydrogen-assisted cracking, the literature on stress-corrosion cracking of stainless steels is very extensive. A computer search using the key words stress-corrosion cracking, (SCC) and stainless steels generated more than two thousand references in the English language from a single database for the period 1985 to 1990. Fortunately, there have been some excellent reviews published in recent years and the present chapter draws heavily on those as well as on data presented in several recently published books on corrosion of stainless steels. 8:52
STRESS-CORROSION CRACKING OF STAINLESS STEELS
8:53
The good general corrosion resistance of stainless steels is derived from the nature of the oxide generated on iron-chromium alloys containing more than about 12% chromium. Stress-corrosion cracking can occur in these alloys by intergranular cracking, transgranular cracking or as a mixture of both. Intergranular cracking is often promoted by the precipitation of carbides in the grain boundaries of the steel. Carbon diffuses more rapidly through the iron lattice than chromium because of the differences in size of the atoms. Thus, the precipitation of chromium carbide at a grain boundary site can be associated with local denudation of chromium in the adjacent matrix so that preferential corrosion can occur along the low-chromium content grain boundaries (Section 3.3). In contrast, transgranular cracking occurs quite readily in austenitic stainless steels because they tend to have low stacking fault energies so that planar slip is common (Section 20.4). The large slip steps formed at the metal surface can break the protective oxide layer and expose bare metal to the environment. Dissolution occurs rapidly in this region because a small anode is created there that is surrounded by a large cathodic region and dissolution is encouraged to penetrate along the slip planes. Passivation of the flanks of the propagating crack can generate overall crack-like geometry and stress intensification at the crack tip can maintain the creation of bare metal there to sustain the rapid dissolution rate needed for fast crack growth. The above simple concepts need to be modified and expanded to allow discussion of stress-corrosion cracking of stainless steels in general, especially as the role of hydrogen in assisting crack growth must be accounted for.
Martensitic Stainless Steels As an approximate guide, the martensitic grades of stainless steels can be defined as those alloys of iron and chromium in which %Cr - 17 x %C c 12.5, but which still contain more than 11.5% Cr to give adequate corrosion resistance. On quenching such alloys from high temperatures they will traverse the y loop in the iron-carbon equilibrium diagram and martensite will be formed to an extent that depends upon the carbon content of the steel. Carbon contents for such steels range from 0.15 to 1.2070, depending upon the strength requirement for the steel. As their name implies, they are used in the quenched and tempered condition for components such as turbine blades, bolts, springs, valve components, cutlery etc. and in the steam generating and chemical industries for many components. Tempering in the range 400-650°C can be detrimental to the mechanical properties and corrosion resistance (Fig. 8.23)'. The latter is considered to occur because of depletion of chromium from the matrix adjacent to the precipitated carbides2. In a review of instances of stress-corrosion cracking reported to a supplier of stainless steels for various clients, Truman cites 15 instances of cracking that were all for high hardness material, Le. 350-650 Hv for quenched and tempered, and 380-430 Hv for precipitationhardened martensitic types of steel3. Typical of this group of stainless steels are the 13Cr turbine blade steels that combine high hardenability, good damping properties, good thermal shock resistance, fatigue resistance and resistance against hot pressurised molecular hydrogen4.
8:54
STRESS-CORROSION CRACKING OF STAINLESS STEELS
2
120
I
I
I
1
c
r
5
-
c
I
1
3 I I-
z
0
100-
P
TENSILE STRENGTH
0 0
N
I '
I
80
-
2: YIELD STRENGTH
D
?
60-
z 0
CORROSION RATE IN 3 % SODIUM CHLORIDJ,' 4 0 -SOLUTION AT 20'C
I-
5 ln
. v)
u
E /
-
0
100
200
300
400
500
600
700
800
TEMPERING TEMPERATURE ( I HOUR).'C
Fig. 8.23
Effect of tempering on the mechanical properties and corrosion resistance of type 420 stainless steel (after Sedriks')
The martensitic steels are commonly quoted to the AIS1 400 series specifications although that range of numbers also includes some ferritic grade^"^. In the as-quenched or as-welded condition the steels are very susceptible to hydrogen cracking and immediate tempering is advisable, especially for the higher carbon content steels. In the tempered state to strength levels typical of many design applications, the steels are susceptible to both stress-corrosion cracking and to hydrogen-assisted cracking. Despite the complication that hydrogen can be generated within cracks and enclaves in acidic solutions and that hydrogen may enter the metal and contribute to the crack propagation mechanism, it is convenient to separate the response of the steel into two categories: (1) cracking under anodic dissolution control and (2) cracking under cathodic charging conditions. The reason for this is that if the anodic dissolution reaction is removed for situations where it is the rate-controlling process then no more hydrogen can be generated within the crack. Also, if anodic dissolution at the crack tip is occurring then some part, if not all, of the crack growth must arise from removal of atoms from the crack tip by corrosion, even if an additional mechanism is needed to account for the observed crack growth rates or to explain the morphology of the fracture surfaces produced. Type 410 (UNS S41000) stainless steel is very susceptible to cracking in 70% NaOH solution and in aqueous chloride solutions4. At 1410 MPa yield strength level a USS 12Cr-Mo-V steel cracked in marine and in semi-industrial environments when loaded to 75% of yield'. Cathodic polarization decelerates stresscorrosion cracking in NaCl solutions but accelerates it in NH4CI solutions, whereas anodic polarisation always accelerates stress-corrosion cracking in all chloride solutions4. It would seem easy in principle to separate cracking that proceeds by anodic dissolution from hydrogen-assisted cracking by investigating the effects of polarisation on the crack growth rate, time to failure or some
STRESS-CORROSION CRACKING OF STAINLESS STEELS
8:55
other parameter convenient for assessing susceptibility to stress-corrosion cracking. Making the potential more anodic might be expected to increase susceptibility if cracking were under anodic dissolution control whereas cathodic polarisation should decrease susceptibility. The opposite should be the case for systems under hydrogen-assisted cracking control. The importance of these comments to practical problems is that it would be unwise to apply cathodic protection for corrosion control to a system that is known to be liab!e to exhibit hydrogen-assisted cracking. Some systems exhibit both types of cracking within different ranges of potential whereas for others there is a virtual overlap, e.g. martensitic steels in boiling NH4C1solutions at pH 5.1 where it is considered that increased anodic dissolution generates hydrogen that is absorbed into the steel to cause enhanced cracking. It is presumably for this reason that Spaehn adopts the terminologies anodic stress-corrosion cracking and cathodic stress-corrosion cracking so that the mechanism of cracking is not necessarily implied by the description of the cracking. Tempering of martensitic stainless steels is performed to improve their toughness, but the tempering causes precipitation of carbides at the grain boundaries which has two main effects on the material. First, it can denude the grain boundary regions of chromium making them less resistant to corrosion. This effect can be exacerbated by local galvanic cell conditions set up at the grain boundaries. Second, precipitation at the grain boundaries will alter the mechanical properties of the grain boundaries relative to that of the matrix. Annealing at 500°C gives continuous grain boundary precipitates rather than discrete precipitates and the former are very deleterious and cause intergranular cracking*. There are numerous quoted examples of intergranular stress-corrosion cracking in tempered martensitic stainless steels9-13 and abundant work relating this to the generation of non-equilibrium solute content profiles at the grain boundaries due to intergranular precipitation 14-’*. Thus, for a Super 12Cr-Mo-V stainless steel tested in a boiling 0.01 M NaOH plus 0.1 M NaCl solution, susceptibility was worst when a continuous chromiumdepleted concentration profile was produced at the grain boundaries by temperingIg. Further heat treatment that caused coarsening of the M,,C, precipitates generated overlapping diffusion fields that removed the continuous chromium-depletion zones and removed susceptibility to cracking. Temperature, electrode potential and solution pH are also important. For a Type 403 stainless steel tested in 0.01 M Na,SO,, Bavarian etal. have shown that at a potential chosen to lie within the potential range for cracking at l0OoC, cracking was also obtained at 75°C but not at 25°C or 5OoC2O. A large amount of stress-corrosion testing has been performed on precracked specimens (Section 8.9). Crack growth rates of almost m/s have been recorded for a martensitic stainless steel tempered at 475OC” and tested in distilled water (K = 5OMPaG) but the crack growth rate decreased for material tempered at higher temperatures. For a Type 431 steel it was found that K,,,, increased and the crack growth rates decreased by several orders of magnitude, depending upon the applied stress level, when the as-quenched steel was tempered at 650°C (Fig. 8.24)22. Spaehn4 suggests that as the favoured industrial tempering temperature range for such steels
8:56
STRESS-CORROSION CRACKING OF STAINLESS STEELS
SCC Growth Rate of Martensitic SS
3
a
10'1
Stress Intensity K in MN m-''a Fig. 8.24 Influence of heat-treatment conditions on the sub-critical stress corrosion growth rate of a nickel-bearing SS as a function of stress intensity. In the asquenched condition, the steel shows much faster crack grown rates (after Spaehn4)
is 700-75OoC, much higher K,, values than the value of about 2 0 M P a 6 quoted by Speidel and much lower crack growth rates than 10-'m/s should be expected even at high K values for properly tempered material. The occurrence of stress-corrosion cracking in the martensitic steels is very sensitive to the magnitude of the applied stress4.For instance, a 13% chromium martensitic steel tested in boiling 35% magnesium chloride solution (1253°C) indicated times to failure that decreased abruptly from more than 2500h to less than 0.1 h as the applied stress was increased from 620MPa to about 650MPa (Fig. 8.25). However, the effects of stress on time to failure are not always so dramatic. For instance, in the same set of experiments times to failure for a 17Cr-2Ni martensitic steel gradually decreased from more than 800h to about 8 h as the applied stress was increased from 500MPa to 800MPa. Thus, the dominant parameter controlling the anodic stress-corrosion cracking resistance of martensitic stainless steels is the tempering temperature subsequent to quenching. However, because tempering causes precipitation of carbides and concentration profiles at the grain boundaries which induce intergranular cracking, the susceptibilityto stress-corrosion cracking is not simply related to the hardness attained by the steel. The allowable hardness to give freedom from cracking depends upon the type of martensitic steel and the environment to which it is exposed. For instance, a hardness of 350Hv or less is considered safe for operation in boiler feed water, condensing steam or boiler water, but the maximum allowable hardness may be different in other environments4. Indeed, work by Doig et al. has shown
STRESS-CORROSION CRACKING OF STAINLESS STEELS
8:57
SCC Resistance of Martensitic SS
Fig. 8.25 Long-time constant-load tests demonstrating a distinct stress-corrosion cracking threshold stress in the case of a straight 13Cr martensitic SS as opposed to a nickel-bearing SS (after Spaehn4)
that the allowable hardness of a 12Cr-Mo-V martensitic steel is a function of the tempering temperature for the four-point bend tests they performed in a 0.01 M NaOH/O.l M NaCl solution at a stress level equal to 90% of yield j 9 . Some variation in allowable hardness existed at each tempering temperature, being greater for heat treatments that generated narrow concentration profiles than those that generated wide concentration profiles, although the differences decreased with increasing tempering temperature. It is not surprising that hardness is important because the mechanical toughness can be expected to decrease with increasing hardness, and the level of residual stress present will also depend on the hardness of the steel, especially for welded components. Thus, the important role of the microstructure in influencing susceptibility to stress-corrosion cracking is consistent with the observation that hardness levels are a good guide to stress-corrosion resistance, but they should not be used universally without due consideration of the specific alloy and the environment in which it is to be used.
Cathodic (Sulphide) Stress-corrosion Cracking The terminology cathodic (sulphide) stress-corrosion cracking is borrowed from Spaehn’s review4 for the reasons previously mentioned. Generally,
8:58
STRESS-CORROSION CRACKING OF STAINLESS STEELS
cracking of martensitic stainless steels under cathodic conditions will be hydrogen-induced and sulphur or sulphur species, and various other hydrogen recombination poisons, enhance the take-up of hydrogen into the metal, thereby increasing the susceptibility to cracking. Typical of such cracking is that of a pump shaft used in a soot water circuit where the environment was saturated with H,S and contained solid soot particles4. The material was Type 410 stainless steel and had a hardness of >350 H , . Similarly, a Super 12Cr (X-20Cr-Mo-V 12.1) German steel, which is similar to Type 422 stainless steel but without tungsten present, used in the manufacture of a start-up heater in an ammonia plant failed by hydrogen cracking that was caused by condensing water on the outer tube surface. The crack was in the heat affected zone (HAZ) of a weld where the hardness was >310Hv. There is a need for care in welding procedures to prevent hydrogen cracking. The martensitic steels must be allowed to cool to about 80°C after welding to ensure complete transformation to martensite, a mandatory requirement for gaining the necessary toughness on subsequent tempering at 750°C. In the as-quenched state the steel is very susceptible to cathodic stress-corrosion cracking if condensation is allowed to occur on the surface of the steel. Indeed, microsections taken from a similar steel have been known to crack during metallographic observation. Thus, it is extremely important to perform the tempering immediately after the weld has cooled to a sufficiently low temperature to ensure adequate toughness after tempering. The welding and tempering schedule must be well enough defined to avoid both inadequate cooling, and/or excessive delay, prior to tempering. To mitigate the generation of excessively high hardness in welded components, modifications in welding procedures have been developed, such as: 1. austenitic welding by applying preheat to maintain temperatures of about 400°C during welding; 2. martensitic welding with a preheat in the range 100-200°C; 3. partial martensitic welding with a preheat to maintain the temperature in the range 250-400°C during welding.
An alternative to the above is to use more weldable varieties of martensitic steels, such as the low-carbon nickel martensites. These have 12-17% Cr, 3-6% Ni and about 0.05% C. The low-carbon gives better weldability and the high-nickel prevents the formation of delta-ferrite. The development of these steels has been described by Irvine et ai. 23,24. These steels are also susceptible to hydrogen cracking, for example in the petrochemical and gas industries where H, S in chloride-containing environments may be very damaging. A NACE standard procedure exists that can be used to assess such steels when proposed for operation in very aggressive conditionsZ5. Data for a range of steels, including a low-carbon nickel martensitic steel, tested by the NACE procedure indicated no correlation between performance and hardness level4. Perhaps of more importance and concern, however, is the observation that cracking occurred at very low stress levels in some cases, e.g. at about 70 MPa sustained stress level in the NACE test. However, if no relevant test data for a specific application are available, it is probable that a choice of the lowest hardness that can be allowed for a particular design situation is likely to provide the greatest resistance to cracking 25-27.
STRESS-CORROSION CRACKING OF STAINLESS STEELS
8:59
Another aspect of the presence of HISin working solutions is that it can reduce the pitting resistance of the steel”. The acid conditions within the pits can generate hydrogen. Double tempering treatments are used for some steels but the effects produced are complicated. Whereas resistance to sulphide stress cracking can be increased, the fracture mode changing from intergranular to transgranular, the double tempering impairs pitting resistance and cracks can be initiated at the pits via chloride stress-corrosion cracking.
Ferritic Stainless Steels The ferritic stainless steels have Cr% - 17 x C% > 12.5, so that in cooling from high temperatures they remain completely ferritic, although the formation of austenite is possible in some grades. The main disadvantage of a totally ferritic structure stems from the fact that bcc metals exhibit a ductile to brittle transition with decreasing temperature. The value of the ductilebrittle transition temperature, T,, is very dependent on the ferrite grain In welding ferritic stainless steels it is difficult to prevent excessive grain growth in the HAZ adjacent to the welds, and unacceptable low toughness regions can be generated. The ferritic 400 series of alloys tend to have high ductile-brittle transition temperatures”, i.e. well above room temperature, even before welding. T,increases with increasing chromium content so that improving the corrosion resistance by increasing the chromium content brings a penalty of increased brittleness3’. The ferrite grain size alters T,because the length of a dislocation pile-up possible at a grain boundary increases with increasing grain size. T,is also dependent upon the flow stress of the material. Interstitial alloying elements such as carbon and nitrogen will increase the flow stress by locking and by the generation of precipitates. Thus, low carbon and nitrogen levels are advantageous for improving toughness. However, as well as the effect of carbon and nitrogen on toughness, the precipitation of carbides and nitrides in the steels can give sensitization to stress-corrosion cracking because of local depletion of chromium in the matrix adjacent to the precipitates. As precipitation is easier if heterogeneous nucleation is possible, for instance at grain boundaries, the latter can become more prone to chemical attack if they are decorated with precipitate^^'-'^. The degree of sensitisation can be reduced by the addition of stabilising elements such as Ti and Nb, but heat treatment at about 800°C is required for complete removal36(see Section 1.3). The advent of vacuum melting and argon-oxygen decarburisation techniques have allowed the production of low-interstitial ferritic stainless steels in recent years. To achieve low ductile-brittle transition temperatures, C + N is kept to less than 100ppm or to less than 400 ppm if Ti or Nb are present as stabilising elements. The improved mechanical properties are accompanied by better resistance to intergranular corrosion”. The lowinterstitial Cr-Mo and Cr-Ni-Mo-ferritic stainless steels are very resistant to chloride cracking and are used for the manufacture of heat exchangers in the chemical industry4. Data exists indicating that the high-strength 18Cr-2Mo steels are resistant to high chloride content oxygenated river water at
a:60
STRESS-CORROSION CRACKING OF STAINLESS STEELS
+
temperatures up to 13OoC4.However, cracking in 42% LiCl thiourea has been reported for high temperature annealed 26Cr-1Mo stainless steel at an applied stress equal to 90% of yield3*. For material so heat treated, the open circuit potential moved in the active direction relative to mill-annealed material during the test. A prestrain of 5% prior to loading enhanced cracking. The work indicated that susceptibility to stress-corrosion cracking depended upon the inherent corrosion and repassivation rates of the alloy, which could be altered by thermal and by mechanical treatments. Tests in boiling magnesium chloride solutions have shown that 17Cr ferritic stainless steels exhibit cracking if copper, nickel and/or cobalt are present at greater than certain amounts, and Schmidt and Jarleborg3' suggest that the total nickel copper content of 17Cr steels be kept below 0.5%. Small amounts of Ni and Cu can be tolerated by low interstitial ferritic grades in both MgCI, and CaCI, solutions, which are both very severe test environments. The data indicate that up to 0.17% Cu and up to 0.6% Ni d o not cause susceptibility to cracking in these environments. Between 0.6% Ni and 0.8% Ni the allowable Cu decreases to almost zero for tests in MgCI, at 140°C but a little Cu can be tolerated and up to 1.2% Ni without cracking occurring in CaCI, solutions at 13OoC4O.However, as has been pointed out by Staehle4', misleading indications can sometimes occur from tests in such aggressive environments. The example quoted by Staehle relates to cracking of alloy 600 in high-temperature water, but similar caution is wise with other materials and other environments. Bond and Dundas' data on the effects of nickel, copper and molybdenum on the stress-corrosion cracking of a large series of alloys indicated that cracking occurs when (Ni% 3 x Cu%) > 1.1%. Molybdenum accentuates the effect of nickel in promoting stress-corrosion cracking. Alloys with up to 5 % Mo, but free of nickel, copper or cobalt were highly resistant to stresscorrosion cracking 40*42. Although cracking can be induced in some ferritic alloys in aggressive solutions, the ferritic grades of stainless steel tend to be more resistant to cracking than the martensitic grades tested in the same solutions. Despite the fact that, as has been already mentioned above, the cracking in Bond and Dundas' work was transgranular in nature, too much emphasis should not be placed on the mode of cracking. For instance, they also refer to work by Streicher indicating that commercial type 430 and 466 stainless steels heat treated at 1095°C then water quenched cracked mainly in a transgranular mode in a MgCI, solution, but in an intergranular mode in a NaCl solution. The heat treatment had sensitised the steels to intergranular corrosion. Dundas also performed tests on similarly annealed and on welded ferritic steels which indicated susceptibility to intergranular corrosion and cracking at high stress levels in boiling artificial sea water. Annealing at 815°C made the steels immune to cracking. In any alloy system the four important factors controlling stress-corrosion cracking are: (1) alloy susceptibility; (2) stress level; (3) environment; and (4) electrode potential. The importance of electrode potential should not be overlooked. For instance, Newburg and Uhlig43 have shown that the susceptibility to cracking of 18Cr alloys of various nickel contents in MgCI, at 130°C can be altered by polarising the specimen. Cracking was induced by anodic polarisation, the degree of polarisation required depending on
+
+
STRESS-CORROSION CRACKING OF STAINLESS STEELS
8:61
the nickel content of the alloy, but the minimum potential for cracking did not vary systematically with increasing nickel content. The experimentally determined minimum potentials for stress-corrosion cracking were more negative than the free corrosion potential for those alloys that cracked under open circuit conditions. An 18Cr-1.5Ni alloy was made to crack in 42% MgCl, solution at both anodic and cathodic applied potentials, but was immune to cracking in the potential range -650 to 250mV SHE". Shimoda et al. have demonstrated the adverse effects of increasing the nickel content of a ferritic alloy in the range up to 4% Ni when tested in 42% MgCI, solution. Thus it was demonstrated that the range of alloys investigated underwent hydrogen cracking after cathodic charging at 25°C and that the cracking of a 1.75Ni alloy could be prevented in slow-strain-rate tests in deaerated 42% MgCl, at 140°C by cathodic polarisation. Uhlig and coworkers have also demonstrated the susceptibility of ferritic stainless steels to cracking under hydrogen charging condition^^'*^*^', with transgranular cracking being the dominant fracture mode, As has been mentioned previously, deliberate hydrogen charging will not be discussed in this section. The difficulties associated with the interpretation of data when hydrogen cracking versus anodic dissolution are discussed have already been mentioned and are probably of little interest to the practising engineer, although in the fulness of time a detailed understanding of the mechanisms of fracture should assist in alloy development programmes. The ferritic stainless steels tend to have quite good resistance to pitting. Pits can give stress concentration effects and acidic environments that generate hydrogen as well as the localised corrosion that might in itself be a precursor to cracking by the linking together of pits to form a trench, tunnel etc. If pitting does not occur, some other crack initiation mechanism is required. One possibility is slip step-induced cracking, and for this to occur, the slip step must be large enough to fracture the oxide layer on the steel surface and thereby expose bare metal to the environment. If this is achieved, cracking is likely to be controlled by the magnitude of the local dissolution current density at the exposed bare metal relative to the repassivation rate for the alloy at that location. Locci e t d 3 ' have obtained slip-step height and stress-corrosion data for a low-interstitial (E-Brite) and a highinterstitial 26Cr-1Mo steel. Slip step heights were similar for both steels but varied significantly with heat treatment and by introducing cold work. The heat-treated states that generated the largest slip-step heights induced greater susceptibility to stress-corrosion cracking. The difference in susceptibility of the two alloys was therefore due to the difference in corrosion and repassivation rates that arose from the different chemical compositions, rather than due to the difference in slip-step heights. The local dissolution rate, passivation rate, film thickness and mechanical properties of the oxide are obviously important factors when crack initiation is generated by localised plastic deformation. Film-induced cleavage may or may not be an important contributor to the growth of the crack4' but the nature of the passive film is certain to be of some importance. The increased corrosion resistance of the passive films formed on ferritic stainless steels caused by increasing the chromium content in the alloy arises because there is an increased enhancement of chromium in the film and the
8:62
STRESS-CORROSION CRACKING OF STAINLESS STEELS
films are thinner and more protective. In Ceislak and Duquette’s passivation treatments caused Cr enrichment in the films. At 260°C’ thicker films were formed than at lower temperatures. It was observed that chloride ions were not incorporated into the films and it was postulated that the thicker films formed at 260°C might be more defective than those formed at lower temperatures and that chloride ions might attack such defective regions in the film. The thickness of the film might be important if cracking is initiated by localised plastic flow, evidence for which exists for austenitic stainless steels in high temperature Another aspect concerning the use of ferritic stainless steels that should be remembered relates to the form of the equilibrium diagram for Fe-Cr alloys and the effect of low temperature annealing treatments (Fig. 8.26). Sigmaphase embrittlement can occur, especially with the higher chromium content alloys. Prolonged use at temperatures greater than 280°C may cause the often referred to 475°C embrittlement, which is due to the decomposition of ferrite into the low Cr a and high Cr a’ forms. Providing that such conditioning factors are accounted for, Spaehn4claims that the immunity of the low-interstitial Cr-Mo steel to hot chloride and caustic solutions has been proven, as has the immunity of nickel-bearing ‘superferritic’ steels in industrially important concentrated chloride solutions.
Binary Fe-Cr Alloysa -,a’-and c-Phase Domains
v magnetic
-
-
-
n‘
\
\
-
I
Chromium in Weight-percent Fig. 8.26 Binary constitution diagram for Fe-Cr alloys (after Spaehn4)
STRESS-CORROSION CRACKING OF STAINLESS STEELS
8:63
Duplex Stainless Steels The development of duplex stainless steels arose from the desire in the chemical industry for alloys with both higher strengths than available with austenitic grades and good corrosion resistance. The alloys developed do have these properties as well as good pitting resistance and high threshold stress values for stress-corrosion cracking. The alloys usually contain between 30 and 70% ferrite. Low ferrite alloys contain 17-21070 Cr and 5-12'70 Ni, whereas the higher ferrite content alloys contain 18.5-26% Cr and 4.5-7'70 Ni. The proportions of ferrite and austenite present can be altered by heat treatment as well as by changing the chemical composition of the alloy. Moreover, diffusion rates for alloying elements are much higher in ferrite than in austenite. At 700"C, chromium diffuses about 100 times faster in ferrite than in austenite and interstitial alloying elements also diffuse faster in the less close packed bcc lattice than in the close packed austenite. Thus, during heat treatments and during welding, significant partitioning of alloying elements can occurs3. Below 1 OOO°C carbides form in the ferriteaustenite grain boundaries and there is a difference in chromium in solid solution for the two phases, the higher chromium content being in the ferrite phase. M23(26 carbides nucleate in the high-chromium ferrite phase and in growing they denude the adjacent area of chromium causing it to transform to austenite. With carbon contents less than 0.03%, there are insufficient carbides precipitated to decorate all of the austenite-ferrite grain boundaries. Typical alloy compositions are reported by Hochmann etal. (Table 8.2)j2. Molybdenum or copper is added to some of the alloys. Edeleanu (54) was the first to record the good stress corrosion resistance of 17Cr ferritic stainless steel compared with that of austenitic stainless steels. He also noted that a duplex stainless steel performed better than an austenitic steel when tested in 42% MgCl, solution and that cracking of the austenitic stainless steel could be prevented by coupling it to a 17Cr or 20Cr ferritic steel. This was due to the ferritic steel cathodically protecting the austenitic steel. The resistance of duplex steels to stress-corrosion cracking in 200°C water containing 87.5 ppm NaCl increases with increasing ferrite content up to about 40% ferritessss6. The relative merits of duplex steels vis-&vis austenitic steels depend somewhat upon the test solution. In some cases, little difference exists for tests performed in very concentrated solutions of chlorides, but the duplex stainless steels are generally better than the austenitic grades in more dilute solutions. Data due to Suzuki et ai." confirm the maximum resistance to cracking in 42% MgCl, solution for alloys with about 40% ferrite as does the data of Shimodaira et aL4'. The data due to the latter authors clearly indicate the relative roles of stress and of the corrosion resistance of the ferrite and austenite phases in stress-corrosion cracking. Austenite grains deform at lower stress levels than ferrite grains. However, as austenite is cathodically protected by the adjacent ferrite, cracks cannot initiate at the slip steps formed in the austenite grains. At higher stress levels, strain incompatibility at the austenite-ferrite grain boundaries can encourage cracks to propagate along those interfaces. At high stress levels, transgranular cracks can propagate through both the ferrite and austenite grains. The plastic
Table 8.2 Typical chemical compositions for duplex stainless steels (after Hochmann et C Alloys which are predominantly austenitic: < 0.05 URANUS 50 (Creusot-Loire) (AFNOR 2 5 CNDU 21-8) 0.03 CF 3A (A C I) 0.08 CF 8A (A C I) 7 M)’.
Methanolic environments In methanolic environments stress-corrosion fracture of a similar kind is seen in titanium alloys. In a-alloys transgranular cleavage is observed2’ and a wide range of velocities’. With additions of HCl, however, an aditional type of fracture is seen in which intergranular separation is seen resulting from a dissolution mechanism accompanied by hydrogen pick-up by the lattice2*. This aggressive behaviour is therefore different from that observed in neutral aqueous or methanolic environments. It is inhibited by water additions, the amount depending upon the concentration of HC12’. Unstressed specimens undergo intergranular attack under open circuit conditions and this is accelerated by the application of anodic p ~ l a r i s a t i o n ~Solution ~. additions that stimulate the cathodic reaction increase the rate of intergranular attack and thereby shorten times to failure, e.g. Hg2+,Cuz+,Pd2+.The addition of H,SO, and HCOOH also increases attack, as do Br, and 1230. Raising the viscosity of the solution by additions of glycerol increases the times to failure3’. In aggressive methanolic environments no K,,,, is observed. Instead very slow intergranular fracture proceeds at an increasing velocity with increasing values of K until it is superseded by the more rapid cleavage. Where the alloy is not susceptible to cleavage the fracture will be intergranular up to overload failure. The transition in a-alloys depends upon the aluminium and oxygen content, and the degree of cold work2’. No pre-cracking is required in aggressive environments and in dynamic straining tests fracture occurs at all crosshead speeds below a maximum, since repassivation is not possible2’. Additions of water will eventually remove this first stage, but not the second unless the alloy is not susceptible to transgranular cleavage in distilled water. Impressed cathodic currents tend to prevent cracking, the current density required increasing as the water content is lowered. Anodic currents stimulate cracking and there is a linear relationship between velocity and potential up to the pitting potential’. Exposure of stressed specimens to aggressive methanolic environments followed by fracture in air results in transgranular fractures similar to stress-corrosion fractures, indicating that some species is absorbed from the environment2’. This has been postulated to be hydrogen as in aqueous environments2’. This mechanism has not been firmly established for either environment but there is an increasing amount of evidence to support it. Thus the embrittlement of specimens exposed in the unstressed condition to aggressive methanolic environments can be removed by ageing so that subsequent fracture in air reveals no cleavage32. In addition, notched specimens of Ti-A1 alloys charged with hydrogen and
8 : 122
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
broken in laboratory air have been shown to cleave along the same plane as the stress-corrosion fracture33. Higher alcohols have not been investigated extensively but failures do occur3'. Other organic liquids cause transgranular fracture, e.g. CCl,, C,H,I, and a range of commercial Freons which are fully substituted halide compounds'. Where these are not aggressive a K,,,, is observed and it is likely that a pre-crack or a dynamic test is required in order to produce stresscorrosion fracture. In those compounds causing fracture which do not appear to contain hydrogen, failure must be the result of residual moisture, providing hydrogen is the species responsible for cracking, but this point has not been established. In both aqueous and organic environments the crack velocity is related to the instantaneous stress intensity factor, as shown in Fig. 8.53. Three regions may be observed: I, I1 and 111. Regions I and I11 are not always observed and the specific relationship observed depends upon the alloy composition and heat treatment, the environmental composition and the experimental conditions'. Other environments Other environments have been shown to cause stresscorrosion failure although the amount of work done on such failures has not been large. High-purity red N,O, caused failure of a Ti-6A1-4V pressure vessel during testing3,. Cracking could be prevented by additions of NO or H,O, but not by additions of NOCl. K,,,, decreases with increasing temperature. Cracking is both intergranular and transgranular by cleavage in Ti-A1 alloys and occurs at a slower rate and at lower K values than in neutral NaC135.Cracking occurs at noble potentials and it seems unlikely that hydrogen plays any part in the fracture. Very high crack densities are sometimes observed (25/mm2). It is thought that fracture is associated with film breakdown and/or the formation of a non-protective film by chemical means: Ti
+ N,O,
-+
TiO(N03), + NO'
+ NO, + e
(unprotective film)
2Ti0(N03), --* 2Ti0,
+ 2N,O, + 0,
Commercial titanium and all alloys crack in red fuming HNO, containing 20% NO,. Eliminating NO, causes cracking in only some alloys while the addition of 2% H,O removes susceptibility completely'. Molten salts containing halides also cause stress corrosion36. Mixed chlorides and bromides at 350°C promote both intergranular and transgranular fracture with maximum velocities as high as 7mm/s. Cracking is very dependent upon both the temperature and the amount of halide present. Some liquid metals have been observed to cause embrittlement in many titanium alloys. In mercury, for example, Ti-8Al-1Mo-1V exhibits both intergranular and transgranular fracture36 with velocities as high as 10cm/s. Heat treatment affects this behaviour in a manner similar to that observed in aqueous and methanolic solutions. Some alloys are embrittled by liquid cadmium and zinc. More surprising, perhaps, is the observed solid metal embrittlement which has been found on titanium alloy components coated with cadmium, silver or zinc3'*38. Service failures of cadmium-plated Ti-6A1-4V fasteners have been reported35,and cracking of this alloy and
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
S t r e s s intensity factor
8 :123
K
Fig. 8.53 Relationship between instantaneous stress intensity factor and crack velocity for a susceptible titanium alloy in I O N HCl=. Activation energy for Stage I = 113 kJ/mol and for Stage I1 = 20.9 kJ/mol
of Ti-8Al-lMo-lV has been produced in laboratory tests on coated specimens in the temperature range 38-316°C38. While the mechanism of such failures has not been established, cadmium has been detected on the fracture surface and the fracture process appears similar to that occurring in liquid metal embrittlement. Hydrogen is not thought to be an important factor since such failures are observed in components coated both electrolytically and by vapour deposition. In addition to the failures in C1, and HC1 already referred to, cracking also occurs in hydrogen gas. Stressed Ti-A1 alloys bombarded with lowenergy protons have failed in a manner similar to that observed in hot salt cracking’. Other studies have shown that hydrogen gas can cause slow crack growth in many titanium alloys resulting in fracture surfaces very similar to those resulting from aqueous stress-corrosion failures3’.
8 : 124
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
Recent Developments Nearly all the later work on titanium alloys has focused upon the role of absorbed hydrogen in causing stress-corrosion crack propagation. Studies of similarities between stress-corrosion fracture surfaces and fracture surfaces produced in hydrogen, residual hydrogen levels in fracture surfaces and consideration of metallurgical events occurring ahead of the crack tip have all contributed to a clearer understanding of the fracture process. In one study of hot salt cracking, analysis of hydrogen in the fracture surface was madea and the conclusion was drawn that it was the species responsible for causing the fracture. There is not universal agreement on this point, however. A study of Ti-8AI-1Mo-1V alloy in a molten salt mixture of highly purified LiCVKCI at 375°C led to the conclusion that hydrogen was not responsible for the cracking process4' since it was claimed that there was no source of hydrogen in the mixture. Crack velocities in this mixture were very much higher than those normally reported in hot salt cracking. The observed fractures were similar to those observed in aqueous solutions at lower temperatures and it was suggested that some other undetermined factor was responsible for cracking under the two different exposure conditions4'. Several studies have contributed support to the hypothesis that absorbed hydrogen causes stress-corrosion cracking at ambient temperatures. Reversible embrittlement experiments on a Ti-0 alloy exposed to CH, OH/HCI solution have shown clearly that an absorbed species is responsible for transgranular cleavage4'. In dynamic strain-rate stress-corrosion tests additions of Hgz+ increased the amount of cleavage occurring in fractures, whereas additions of Pt *+ produced less cleavage. These effects were attributed to the increased and decreased amount of hydrogen absorbed, respectively, as a result of the additions. The fracture surfaces of specimens of a Ti-8AI1Mo-1V alloy after stress-corrosion cracking and those of the same alloy broken by slow strain-rate hydrogen-embrittlement (SSRHE) have been shown to be virtually indisting~ishable~~. Specimens broken in dry ultra-pure argon at various high crosshead speeds have resulted in failures that were completely dimpled but such fractures became increasingly brittle as the crosshead speed was lowered, an effect seen previously" and attributed to the presence of a small concentration of internal residual hydrogen. In aqueous solutions and under SSRHE conditions the resulting cleavage-like fractures were much more pronounced and indistinguishable from each other. The discontinuous transgranular fractures were accompanied by periodic acoustic emission signals45. Glancing angle electron diffraction revealed the presence of an fcc hydride phase on the fracture surface. The propagation process was considered to consist of the repeated formation and fracture of the hydride phase. Cracking occurred on the [ lOf7) plane, as had been previously reported by several workers but, in addition, two specimens with a texture which had this plane tending to lie parallel to the stress axis exhibited ( 100) fractures. Both these planes are hydride habit plane^^*^'. An analysis of loading mode effects has also provided evidence of the critical role of hydrogen. A stress-intensity factor ( K ) can be achieved in either a tensile loading mode (mode I) or a shearing mode (mode 111) (Section 8.9). Under mode I conditions the volume of metal immediately in
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
3.5% NaCl
Cantilever beam 00 0
0
0
8 : 125
/
0
O-D
I 1
3.5% NaCl and 10 ppm AS I 10 Time to failure min
I 100
Fig. 8.54 Susceptibility of Ti-8AI-IMo-1V to stress-corrosion cracking in 3.5% NaCl under both tensile and torsional loading, corresponding to mode I and mode 111, respectively. The ordinate consists of the ratio of failure value in solution to failure value in air 50
front of the crack is subjected to a high triaxial stress. This is not the case for mode 111conditions. Dissolved hydrogen atoms accumulate in regions of high triaxial stress@. It has been argued4’ that an alloy failing by stresscorrosion cracking mahly as the result of absorbed hydrogen would exhibit a markedly different susceptibility according to the loading mode employed. Such a difference was observed on a high-strength steel, with no failure occurring under mode 111 condition^^^. Such an approach was used by Green etal.” who examined a Ti-8Al-lMo-lV alloy in 3% NaCl solution. The results are shown in Fig. 8.54. Under potentiostatic conditions the alloy was not susceptible to stress-corrosion cracking when tested under mode 111 conditions. Under mode I conditions, however, the value of K in solution was lower than the value in air, with the normalised ratio of the two falling to around 0.7.As a further indication of the important role of hydrogen, the addition of a cathodic poison, As, to the solution lowered the ratio of 0.6. These two effects of loading mode and cathodic poison addition are readily interpretable with respect to a hydrogen-embrittlement model of stresscorrosion crack propagation. Two additional points concerning the results shown in Fig. 8.54 can be made. First, the authors argued5’ that any cracking process in which the main cause of propagation was the result of some form of anodic dissolution would occur equally under either loading mode. Any difference in loading mode behaviour would only be seen when hydrogen was the embrittling species. In support of this proposed distinction the authors presented results on the intergranular cracking of a-brass specimens in a concentrated ammoniacal solution which is considered to occur as the result of a dissolution
8 : 126
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
mechanism. Specimens were found to be equally susceptible under both loading modes of testing. Secondly, under open circuit conditions the corrosion potential of Ti-8Al-1Mo-1V alloy was about -800 mV (SCE). If As was added to the NaCl solution under those conditions no increased embrittlement was observed since at that potential AsO; ions will react with water to form ASH,, arsine, which is a gas. No As atoms will adsorb on the alloy and no effect due to the addition of the poison will be observed. At -500 mV (SCE) adsorption of As atoms on to the alloy surface will occur and a poisoning effect is therefore observed. It may be added that As additions are inhibitive to the corrosion process, mainly because of it impeding the hydrogen evolution reaction and, possibly, as a result of increasing the solution pH, if added at sufficiently high concentrations. Strain-rate effects have been examined by several authors. The stresscorrosion crack velocity has been shown to be dependent upon the strainrate at the crack tip”. The K,,,, value has been observed to be dependent upon the loading rate for a Ti-6A1-6V-2Sn alloy in a 3.5% NaCl solutions*.The value reached a minimum at an intermediate loading rate which was considered to be indicative of hydrogen embrittlement through hydride formation. The calculated strain rate was considered to correspond closely to the theoretical minimum strain rate for hydrogen transport by dislocat i o n ~ Intergranular ~~. fracture observed in region I of the velocity/K curve has been attributed to there being a low density of hydrogen-carrying dislocations which arose preferentially in grain boundary regions and gave rise to grain boundary hydrides 52. Intergranular fracture in commercial purity Ti in a CH,OH/HCl mixture was attributed to absorbed hydrogen53. The role of hydrogen in the stress-corrosion cracking of P alloys has not been widely investigated. In Ti-13V-llCr-3Al alloy cracking in aqueous solutions at ambient temperatures occurs as a cleavage process on or close to [ 100) planes54as a discontinuous process”. In CH,OH/HCl solutions reversible embrittlement experiments were interpreted as showing that cleavage was caused by absorbed hydrogen. Consistent with such an analysis, AsO; additions to the solution increased the amount of cleavage observed after stress-corrosion tests, while Pt 2 + additions resulted in much less cleavage. Hg2+ additions caused more cleavage and quinoline additions less. These effects showed that the amount of cleavage observed was not related to the corrosion rate. How hydrogen embrittlement occurs in the bcc P lattice has not been determined. Reversible embrittlement has also been observed in the P-I11 alloys without the formation of hydrides56which was also the conclusion drawn from work on a Ti-28Mo alloy. The evidence for the role of hydrogen in many cases of stress-corrosion cracking of Ti alloys is strong but it must be remembered that identical fractures can be caused by liquid metal embrittlement in which hydrogen appears to play no role and in other environments which are claimed to have no hydrogen source. Since there is more than one species capable of promoting the transgranular cleavage characteristically observed in CY alloys, where such corrosion fractures are observed any analysis must seek to establish which is the most rapidly acting embrittling species.
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
8: 127
Magnesium Stress-corrosion cracking occurs in magnesium alloys in aqueous environments at ambient temperatures. It is found mainly in wrought alloys, although isolated examples of cracking in castings have been reported. The major alloying element responsible for producing susceptibility is aluminium. In practice failure commonly appears to arise from the action of residual stresses. Mechanism
The mechanism of stress-corrosion cracking in magnesium alloys has not been fully elucidated, but many features have been observed. The r61e of the electrolyte appears to lie in concentrating the corrosion attack in a relatively small number of points resulting in pitting which often precedes cracking. Where little or no pitting occurs cracking is only initiated if the value of the stress is sufficiently high, probably causing rupture of the surface film by plastic deformation of surface metal grains. In polycrystalline specimens the threshold stress below which failure does not occur will be below the stress values below this level, but in single crystals the yield point clearly must be exceeded to produce plastic deformation and this has been observed to be necessary for stress-corrosion cracking to occur”. In such environments the capacity to repair ruptured films must not be too great otherwise this process will occur in place of crack initiation. If environments are too aggressive then widespread corrosion may occur in place of pitting and cracking. In laboratory tests mixtures of NaCl plus Na,Cr,O, are commonly used to cause cracking which is preceded by deep pitting; a common mixture consists of 3.5% NaCl plus 2% K,CrO,. The aggressive anion is important and cracking is not confined to the C1- ion. Thus in 1 N solutions it has been observed5’that the rate of stress-corrosion cracking decreases according to: Na,SO, > NaNO, > Na,CO, > NaCl CH,COONa, whereas the rate of corrosion of unstressed samples decreases according to: NaCl >Na,SO, > NaNO, > CH,COONa > Na,CO,. In NaCl plus K,CrO, mixtures the rate of cracking depends upon both the absolute concentration of each species and their ratio. It has been established that KHF, solutions also cause stresscorrosion cracking. Little investigation of the action of this electrolyte has been carried out, but since the F- ion is an inhibitor for the corrosion of magnesium at least part of the electrochemical explanation may lie in inhibited film breakdown and repair kinetics. Thus cracking does not occur in fluoride solutions above a certain limit of concentration. In non-fluoride solutions stress-corrosion crack initiation is inhibited at pH values greater than 10.259and this is also probably related to the greater ease of film formation that occurs in highly alkaline solutions on magnesium surfaces. Under open-circuit conditions or the application of anodic polarisation, only pitting is observed in unstressed specimens under all conditions of heat treatment in environments that cause cracking in stressed specimens. Cracking is generally observed to be transgranular, but this may change depending upon the heat treatment, the grain size and possibly the environmental pH. Thus in the Mg-6.5Al-1Zn alloy transgranular cracking occurs
+
8 : 128
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
if specimens are water-quenchedmI 61, a process that produces a very fine transgranular dispersion of FeAl which occurs as platelets62 that are cathodic to the solid solution matrix. Cracking may include the dissolution of material adjacent to the precipitates which may also be solute depleted. In furnace-cooled material the same alloy exhibits intergranular cracking and this may be associated with the grain boundary precipitate of MgI7All2 which occurs during furnace cooling. This precipitate is also cathodic to the matrix but the potential difference developed is lower than that between FeAl and the matrix. This may be a partial electrochemical explanation for the observation that as the iron content is increased the proportion of transgranular fracture also increases, even in furnace-cooled alloys63. Above a grain size of about 0.03 mm cracking is transgranular irrespective of the heat treatment employed. In two-phase alloys intergranular cracking occurs in specimens aged at 150°C. The path of transgranular cracking has not been clearly established and it may vary from one alloy to another. It has been reported as following crystallographic planesa, possibly the basal palne@ ,' while others report that no well-defined fracture plane can be discerned in large-grain-size specimens65,that the crack occurs at a high angle to the basal planes7, and that fracture occurs as cleavage along (OOOl), (1070) or (1071) depending upon which plane is most nearly perpendicular to the operative tensile stressM. Cracking is considered to occur by a combination of dissolution and mechanical fracture. It is not necessary therefore to account for the relatively high propagation rates by anodic dissolution since this would require high current densities, e.g. 0.6 mm/min requires 14A/cm2. An anodic reaction appears to be occurring during cracking, and cathodic protection prevents crack initiation and arrests crack propagation. During cracking the tip is active and hydrogen is evolved. Potential fluctuations have been detected67 in Mg-SA1 alloys in KHF, solutions although not in Mg-7A1 alloys in Cl-/Cr,O:mixtures6'. One explanation6'" has been given for the mechanical part of the fracture which is also indicated by fractographic 69, O' and by irregular specimen extension". This is that cleavage occurs on (3140) planes, as a result of hydrogen absorption. The amount of current required to effect cathodic protection increases with increasing stress on specimens. A threshold stress below which cracking does not occur is observed in some environment^^^^ 71. It is dependent upon the alloy composition and heat treatment, and upon the testing environment. In 3% NaCl solutions it is not well defined and this may arise from the pitting that occurs which can be expected to lower the cross-sectional area locally and gradually raise the effective stress acting across that region. Below the threshold stress, cracks occur which d o not propagate to sufficient length to cause complete fracture of specimens69. The maximum length of crack observed I , is related to the proof stress of the material u by the relationship u21 = constant. Such cracks have been observed in fatigue and corrosion fatigue7' and their density and iength are very dependent upon the mechanical properties of the alloy73. From this analysis emerges a description of alloys undergoing pitting at low stress values, pitting and cracking without complete failure at
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
8: 129
higher stress values up to the threshold level above which time to failure decreases with increasing value of the initial All0 y Susceptibility
Cracking is observed in commercial Mg-Al-Zn alloys in the range 3-10% A1 in which the ratio AVZn 2 2, and susceptibility increases with increasing A1 content. For a given alloy increasing amounts of Fe increase susceptib i l i t ~ Copper ~~. additions also appear to raise su~ceptibility’~.It must be emphasised, however, that Mg-A1 alloys exhibit stress-corrosion susceptibility even if manufactured from elements of the highest purity available. Most aluminium-free alloys appear to be non-susceptible in most aqueous environments, but KHF, solutions appear to be able to cause failure in most alloys including nominally pure magnesium metal in which failure is intergranular. Thus Mg-Mn alloys, for example, are immune in NaCl plus K,CrO, solutions (unless additions of 0.5 Ce are made to the alloys) while they exhibit cracking in KHF, solution67. Mg-14Li alloys, which have a b.c.c. structure, exhibit intergranular fracture in humid air although this can be prevented by a stabilising treatment consisting of heating for 24 h at 149°C after quenching”. The Mg-3Zn-0.7Zr alloy has been reported67to fail in distilled water and KHF, solution. Preventative Methods
Stress relieving at low temperatures is commonly used to lower stresscorrosion susceptibility in Mg alloys, e.g. 8 h at 125°C for Mg-6.5Al-1Zn0.3Mn7,, since higher temperatures tend to lower the yield point. Similar treatments are advisable for welds which can be a source of high residual stresses. Treatments designed to put the surface in a state of compressive stress also tend to prolong stress-corrosion life. Shot peening, surface rolling7’ and abrasion all produce beneficial effects. Surface oxidation followed by anodising is also to increase stress-corrosion life. Susceptible alloys can be clad with non-susceptible Mg alloys, but where the edge is exposed, wetting of both the alloy and clad layer is important in order to achieve cathodic protection of the former. The replacement of susceptible alloys with non-susceptible alloys or with alloys exhibiting less susceptibility, can often be undertaken with little loss of mechanical properties. Refining the grain size lowers susceptibility. Heat treatment produces changes in the threshold stress76and alterations in crack morphology as already described.
Aluminium Stress-corrosion cracking occurs in certain aluminium alloys which have been developed for medium and high strength by employing variations in composition, cold work and heat treatment” 77* 78. The main alloys are based upon A1-Mg, AI-Mg and A1-Cu, but stress corrosion also occurs in AI-Ag, Al-Cu-Mg, AI-Mg-Si, Al-Zn and Al-Cu-Mg-Zn alloys. It has
8 : 130
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
not been observed in pure aluminium. In alloys susceptibility appears to increase as the amount of alloying addition that can be taken into a supersaturated solid solution is raised. In ternary and higher element alloys susceptibility is also influenced by the ratio of solute elements7'. Small additions of Cr, Mn, Zr, Ti, V, Ni and Li can reduce the susceptibility of wrought products in high-purity binary, ternary and quaternary alloys79s Stresscorrosion cracking in castings is not common but it is found occasionally". The large majority of failures occur in aqueous environments and therefore most attention has been focused upon them, but failure can occur' in N,O,, mineral oils, alcohols, hexane and mercury. It has not been established whether failure in these environments is the result of residual moisture. Failures in service arise often from the action of residual stresses which can occur in components as a result of quenching followed by machining. The stress level required to initiate cracking is often very much below the yield stress. Alloys brought to a high-strength condition are particularly susceptible. Mechanism
The mechanism of cracking in aluminium alloys has not been elucidated, but many factors have been examined. Cracking is nearly always intergranular. Stress-corrosion life is very dependent upon the grain shape and orientation in relation to the acting stress. Stress-corrosion resistance is lowest in the short transverse direction of wrought components since many grain boundaries are then lying orthogenally to the applied stress. Notice of such effects is commonly taken in the design of components. In plane-strain tests a relationship between crack velocity and stress intensity factor is found', similar to that shown in Fig. 8.53 for titanium alloys. A large number of alloys exhibit only Stages I and 11. Others exhibit Stage I11 and others two 'plateaus' or Stage I1 regions'. Results similar to those shown schematically in Fig. 8.53 are also obtained. The crack velocity may vary over nine orders of magnitude and determining KIscccan be difficult since too high a value may be obtained if the velocity-detecting apparatus is not sufficiently sensitive or the length of time of the experiment too short. It has been suggested' that KlsCcmight be defined as corresponding to a crack velocity of lo-'cm/s. In equiaxed specimens crack branching is observed in the region of Stage I1 at a value of K about 1.4 times the K value at the lower end of the stage. Since cracking occurs at low stress values it is not altogether surprising that the precipitation of corrosion products within existing cracks can sometimes exert relatively appreciable stresses which result in crack propagation. The effect of environmental variables upon the logarithm of velocity vs. K relationship has been examined' for a few alloys in some conditions of heat treatment. While it cannot be certain that similar results would be obtained with all alloys, the results reported'. s2 do show interesting features that may have points in common with all alloys. For an Al-Zn-Mg-Cu alloy (7075-T65 1) the stress-corrosion plateau velocity was a maximum in 5 M KI solution under potentiostatic conditions at -520 mV (vs. S.C.E.), reaching about 2 x to 5 x 10-4cm/s, whereas in 3% NaCl under open-circuit
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
8: 131
conditions the plateau velocity was about 10-6cm/s. Stage I in both cases was the same. The plateau velocity was very sensitive to moisture and depended linearly upon the water vapour of the testing environment'. 82. Crack propagation does not occur in argon or hydrogen unless moisture is present83. Many alloys exhibit plateau velocities when tested in distilled water that are similar to those observed in moist atmospheres. Additions of C1-, Br- and I- increase the plateau velocities observed in distilled water by as much as l b times under open-circuit conditions, but many other anions have no effect upon the logarithm of velocity vs. K relationship under open-circuit conditions or over a wide range of potential values. In C1-, Br- and I - solutions the plateau velocity depends upon the halide concentration for some alloys, e.g. the velocity increases with increasing iodide concentration above a certain minimum for 7079-T651, but this is not universally true. In neutral solutions cathodic polarisation lowers the plateau velocity while anodic polarisation increases it until pitting occurs. In strongly acidic solutions the velocity is less sensitive to potential changes and no cathodic protection is observed'. The pH value of the environment does not appear to cause changes in the plateau velocity under open-circuit conditions, but acidic conditions move Stage I to the left in Fig. 8.53 thus giving rise to higher velocities for a given K value. Such effects have been described as indicating that the size of the cathodic zone within the crack where hydrogen is released controls the rate of crack propagationa4. Generally, lowering the pH shortens the time to failure of specimens, an effect that will include the influence of pH upon the initiation process. Under potentiostatic conditions lowering the pH can cause appreciable increases in the plateau velocity on the cathodic side. Temperature effects indicate an activation energy of 113 kJ/mol for Stage I and 16 kJ/mol for Stage I1 in 7079-T651 alloy. Crack velocity in Stage I1 is lowered as the solution viscosity is increased. No mechanism for cracking in N,O, has been establisheda5.In organic media crack velocities are similar to those obtained in distilled water. Lowering the water content results in lower velocities. Not all authors attribute failures in organic liquids to the residual moisturea6.Furthermore, part of the fracture may be transgranulars6. Water additions to methanol increase crack velocities as do halide additions. In oils velocities are similar to those in organic liquids and distilled water. Much of the extensive work on crack velocity described here has been carried out over a long period by Spiedel'. Detailed studies of velocitydependent and velocity-independent parameters reveal how complex the phenomenon is. The three major alloy systems will now be discussed.
AI-Mg (5000 Series) and AI-Mg-Si (6000 Series) In the binary alloy system strength is obtained mainly by strain hardening. Stress corrosion is thought to be associated with a continuous grain boundary film of Mg,Al, which is anodic to the matrix". Air cooling prevents the immediate formation of such precipitates, but they form slowly at ambient temperatures. Thus only low Mg alloys are non-susceptible (A1-3% Mg). Widespread precipitation arising from plastic deformation" with carefully controlled heat-treatment conditions can lower susceptibility. AI-5Mg alloys of relatively low susceptibility are subjected to such treatments. Mn and Cr
8 : 132
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
additions improve the stress-corrosion resistance of A1-6Mg and AI-7Mg alloysa9.Mn increases the precipitation rate and both elements promote the formation of elongated grains. AI-7Mg alloys are usually very highly susceptible, but a recent development’ indicates that an alloy of this kind can be produced of low susceptibility. Al-Mg-Si alloys are strengthened by precipitation hardening in which Mg,Si is formed. They are not very susceptible to stress-corrosion cracking77. which only occurs in specimens subjected to a high solutiontreatment temperature followed by a slow quench77.Ageing such material eliminates sus~eptibility~~.
AI-Cu and AI-Cu-Mg (2000 Series) These alloys are strengthened by precipitation hardening. Under conditions of natural ageing these alloys are highly susceptible to stress-corrosion cracking. Susceptibility is associated with slow quench rates which also result in grain-boundary corrosion in unstressed specimens, which is thought to arise from electrochemical effects between CuAl, and solute-depleted zones formed during quenching”. Since thick-section material cannot be quenched rapidly quench-rate effects determine the type of component that any particular alloy can be used to make. During artificial ageing susceptibility passes through a maximum just before peak hardness is achieved. Similar changes occur in the potential difference developed between grains and grain boundaries9’. After further ageing, precipitation of the equilibrium CuAl, occurs within the grains and the potential difference between grains and boundaries then disappears. A recent testg3provides a rapid means of indicating susceptibility to intergranular attack and stress-corrosion cracking. The specimen’s potential is measured in a mixture of absolute methyl alcohol and carbon tetrachloride. Corrosion of the grain boundary provides sites for deposition of dissolved copper, whereas an absence of corrosion results in deposits of copper which are non-adherent. The former develops a potential of about -300 mV (vs. S.C.E.) whereas the latter develops a potential of about -1100mV. Ai-Zn-Mg and Al-Zn-Mg-Cu (7000 Series) These alloys are strengthened by precipitation hardening. Cr, Mn and Zr additions produce elongated grain shapes and inhibit grain growth. High-purity ternary alloys exhibit the highest plateau velocities and although much research has been done on them they are not used in practice. Commercial low-copper alloys are particularly susceptible, and although overageing is generally beneficial the effects of such a treatment are less pronounced with these alloys. Artificial ageing is beneficial but susceptibility in the short transverse remains troubleSilver additions to the alloy are reported94 to improve stresscorrosion resistance. The effect appears to arise from the stimulation of precipitation processes which minimise the width of the precipitate-free zone which arises either from vacancy or solute depletion during quenching. Other workers find that silver gives no improvement either in strength or stress-corrosion r e ~ i s t a n c eAn ~ ~ .explanation for this difference appears to lie in experimental procedures’. A general conclusion is that ageing temperature and not chemical composition is the most important factor governing short-transverse stress-corrosion resistance in these alloys9’.
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
8: 133
Recent developments, particularly in producing alloys suitable for thick sections, is reviewed by Spiedel', together with the particular problems of welding. General
Much discussion of stress-corrosion cracking mechanisms in aluminium alloys has been concerned with the development of anodic areas at grain boundaries. The origin of such areas can be caused by the action of the stress, and susceptible alloys do not necessarily suffer from intergranular corrosion in the absence of stress. Thus in some conditions Al-Mg-Si suffers from intergranular corrosion, but not stress corrosion%, 7039-T64 suffers 1 suffers from stress corrosion but not intergranular ~ o r r o s i o n 7075-T65 ~~, from both, 7075-0 from neither. The electrochemical effects may arise from solute-depleted zones, precipitates anodic or cathodic to the adjacent matrix, or from the rupture of films at the crack tip by plastic deformation. The effect of relative humidity upon the plateau velocity suggests that there may not be a volume of water at the crack tip', a possibility which if established would demand a careful re-examination of possible electrochemical reactions. From a metallurgical viewpoint the effect of grain shape has been described. On a microstructural level the precipitate-matrix interface properties appear to be important. In alloys of A1-6Zn-3Mg aged to peak hardness, slip occurs in a relatively small number of bands which develop a high density of dislocations. Overageing, which lowers susceptibility, results in plastic deformation occurring in much more diffuse bands of dislocation^^^. Grain-boundary precipitates are important both for electrochemical and mechanical reasons and the precipitate-free zone width (as well as the solutedepleted zone width) may also be important. The precise relative significance of these three micro-structural features has not been fully ascertained and it is a subject of a considerable discussion99-'02.Much of this centres around the r61e of preferential deformation in the precipitate-free zone resulting in selective dissolution, a process that has not been demonstrated experimentally. Selective corrosion of solute-depleted regions, hydrogen adsorption, tensile-ligament dissolution and general adsorption have also been invoked as major components of mechanistic processes'. There is evidence that acidity develops within the region of the crack tip, a pH of 3.5 being observedlo3and the mass-transport-kinetics modelIMappears to explain the plateau velocity as being limited by the kinetics of halide-ion transport to the crack tip. A number of worker^'^^-'^^ have provided some evidence that absorbed hydrogen may be at least partly responsible for cracking. Preventative Methods
The importance of grain shape and the orientation of the applied stress to the short transverse direction has already been pointed out. Overageing also generally lowers strength and stress-corrosion susceptibility. Both the design and manufacture of components are important. Quenched components often have high internal tensile stresses and subsequent machining of such
8: 134
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
pieces may result in surfaces that readily nucleate cracks. This possibility can be removed by manufacturing components close to the final required size before heat treatment. Shot peening is a beneficial surface treatment since it puts the surface into a state of compression and generally obscures the grain structure. Subsequent painting of the peened surface is often useful. If pitting occurs then cracking can be expected in susceptible material when the attack penetrates the depth of the compressed surface layer. Paint coatings can be effective in preventing stress corrosion but it is not always a simple matter to produce and maintain a perfect complete coverage. Galvanic coatings based upon electroplated layers or metal pigmented paints, are commonly used. Such layers do not need to be perfect, but the protection afforded to breaks (or 'holidays') will depend very much upon the localised electrochemical conditions. Metal spraying is also employed and for highly susceptible alloys a thin cladding sheet of aluminium is employed, either on one or both sides. These clad composites are employed for general corrosion resistance and not merely to combat stress corrosion. Anodising is generally not recommended. Cathodic protection is effective but is often not practicable.
Recent Developments Later work on aluminium alloys has also focused more closely upon the role of hydrogen which had not previously been widely considered as an embrittling species in the stress-corrosion cracking process for these alloys. The idea was not new, however. Reports of intergranular failure under cathodic charging conditions had been made at a much earlier time 108.1w.A reduction in stress-corrosion life and alloy ductility in a high purity Al-5Zn3Mg alloy had been found in specimens pre-exposed to a 2% NaCl solution1Io,an effect that was accentuated if specimens were stressed'". In more recent work embrittlement in water vapour-saturated air and in various aqueous solutions has been systematically examined together with the influence of strain rate, alloy composition and loading mode, all in conjunction with various metallographic techniques. The general conclusion is that stress-corrosion crack propagation in aluminium alloys under open circuit conditions is mainly caused by hydrogen embrittlement, but that there is a component of the fracture process that is caused by dissolution. The relative importance of these two processes may well vary between alloys of different composition or even between specimens of an alloy that have been heat treated differently. The role of humid air has been examined in the embrittlement both of high-purity Al-Zn-Mg alloys and also for a few commercial compositions 112-114 . Loss of ductility in unstressed specimens is a reversible process11o.Such an effect, when observed, is readily attributable to absorbed hydrogen. In unstressed specimens hydrogen must enter through the unbroken surface film either in an atomic or a protonated form. The thickness, composition and morphology of surface films are all likely to be important factors controlling the rate of hydrogen or hydrogen ion entry. This point was emphasised at an early stage when it was observed that
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
8: 135
solution heat treatment of a high-purity A1-6Zn-3Mg alloy in the temperature range 450-500°C resulted in an increased sensitisation to pre-exposure embrittlement 'I2. The change in sensitisation could be correlated with the formation of crystalline MgO which occurred preferentially at alloy grain boundaries. It appeared that MgO crystallites facilitate the entry of hydrogen into the alloy grain boundariesIl2. Oxide thickening resulted in a reduced rate of embrittlement at room temperature. In much subsequent work a lot of attention has been focused upon the presence of Mg in precipitate-free grain boundaries and its possible role in pre-exposure embrittlement and the stress-corrosion process The retention of this segregation after precipitation has been much discussed 122*123. High temperature solution heat treatment, slow quenching and overageing may reduce the level of segregated Mg and thereby reduce the hydrogen entry rate. This would account for the beneficial effects that these procedures have upon the stress-corrosion resistance of Mg-containing alloys. Alloying effects were also examined in this study'I2. Additions of 1.7Cu or 0.14Cr to the high-purity alloy reduced the rate of embrittlement. The chromium-containing alloy and a commercial 7075 alloy both recovered their ductility after exposure to water vapour-saturated air at 20°C unlike the high-purity alloy. The effect of chromium is shown in Fig. 8.55. The highpurity alloy did not recover its ductility in dry air or after storage for 12 h at 68°C in a vacuum of lo-' torr. The 7075 recovered ductility a little more rapidly than the chromium-containing alloy. The presence of hydrogen in pre-exposed specimens was revealed by straining specimens in vacuo. Hydrogen evolution occurred in the elastic region of the stresdstrain curve, an effect that had been shown to be very much reduced by electropolishing pre-exposed specimens prior to testing 134,
1 .o
0 .-c
I >.
c
E 0.5 c. 0
3
n
0
*!-
at 70°C-
-wvsa
I 2
1
Lab. air at 20°C
6 Pre-exposure time, d 4
8
I 10
Fig. 8.55 The effect of a 0.14Cr addition on the recovery of ductility of A1-6Zn-3Mg alloys during storage in laboratory air at 2OoC after pre-exposure to water vapour-saturated air for 5 days at 70°C. The ductility ratio is the ratio of elongation-to-fracture of specimens broken under the cited conditions and under vacuum conditions 'I2
8: 136
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
but which had no effect upon the measured ductily. During the plastic straining and at the point of fracture hydrogen evolution from pre-exposed specimens was detected'24. The embrittlement caused to aluminium alloys by pre-exposure to moist atmospheres or stress-corrosion environments is thought to be due to hydrogen in the atomic form. Intergranular bubbles of hydrogen, formed in association with certain precipitates, have been observed by HV TEM 112,125 and are associated with a lowered degree of embrittlement. Increased resistance to stress-corrosion cracking in AI-Zn-Mg alloys resulted in an increased propensity for hydrogen trapping and a decrease in the permeation rate of hydrogen through unstressed alloy membranes L26. The ability to trap hydrogen as innocuous bubbles improves the embrittlement resistance of AI alloys 12'. Thus it appears that fracture occurs by hydrogen-induced grain boundary decohesion once all available sites for hydrogen trapping are saturated. High purity AI-Zn-Mg alloys have relatively few sites and therefore embrittle readily. Alloying additions to these alloys and commercial alloys in general result in microstructures that have a much higher density of potential trapping sites for hydrogen. The mechanism of hydrogen embrittlement of aluminium alloys has not been established. Experiments on a high purity A1-5.6Zn-2.6Mg alloy hydrogenated by exposure to water vapour saturated in air at 70°C indicated that internal hydrogen embrittlement occurs by the formation and rupture of a hydride phase at grain boundaries 12'. Electron diffraction revealed a very thin layer of AlH, (- 1pm thick), formed probably as the result of a stress-induced mechanism. Two stages of embrittlement were noted: stage I, in which the diffusion of hydrogen into the region ahead of the advancing crack tip was necessary to provide a sufficient concentration of hydrogen to produce AlH,; and stage 11, in which sufficient hydrogen was already present at the grain boundaries to form AIH, and hydrogen diffusion during stressing was not therefore required. Such a distinction explains both the strong dependence of stage I upon strain rate, stage I extending as the strain rate is lowered, and the absence of any strain-rate dependency in stage 11. In one interrupted stress-corrosion test in a NaCl solution a thin layer on the fracture surface at the intergranular/dimple transition region was observed, although no diffraction pattern was obtained. The authors noted that the stress-corrosion fracture surfaces frequently do not show such a layer. They recognised that stress-corrosion cracks may propagate by a competing and basically different mechanismL2'. Fractographically, failure has been seen to occur discontinuously Iz9-I3', an observation interpreted as being the result of repeated pre-exposure embrittlement. Matched arrest markings have been seen in specimens broken by stress-corrosion in chloride solutions, in water and in some service failuresI3'. For the two alloys examined, 7071 and 7179, the average striation spacing was not a strong function of the applied stress intensity factor. Acoustic emission also indicated that cracking was discontinu~us'~~. The striation results are in agreement with observations that the effect of stress intensity on stress-corrosion crack propagation in Al-Zn-Mg alloys by the hydrogen-embrittlement mechanism appears to increase the rate of crack jumping rather than to alter the magnitude of the crack advance"*. If the number of available trap sites for the embrittling hydrogen atoms is fixed by
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS 10 ppm As added
-
1.o
8: 137
Air
\ X
g-
0.9 Torsion
2In(0
* I
a,
3
0.8
CJl
7075 T6 AI alloy exposed to 3.5% NaCl 3.0% K2Cr307pH 3.2 alternate immersion
3
0
c
2 2
*
0.7
0
2
U
10 ppm, As added\
0
0.6
0.5
10
I
100 Time to failure, hrs
I 1000
Fig. 8.56 Susceptibility of 7075-T6 A1 alloy to stress-corrosion cracking in 3.5% NaCl K, Cr, 0, under both tensile and torsional loading 50
+ 3%
the alloy composition and its thermal history then the stress operating across the grain boundary must affect the grain boundary diffusion rate such that all the available trap sites are saturated more rapidly and brittle grain boundary failure can be induced by any further accumulation of hydrogen atoms. In this way the kinetics of pre-exposure embrittlement are accelerated by applied stress"'. The role of loading mode on the stress-corrosion cracking of an A1 alloy has been examined with a 7075 alloy in the T6 condition5' and for 5083133, with similar results. Figure 8.56 shows results obtained with the 7075 alloys0. In the tension test the alloy was more susceptible to cracking than in the torsion test. Unlike Fig. 8.54 for a titanium alloy, however, some cracking did occur under the torsion mode of testing which indicated that cracking occurs both by hydrogen embrittlement and by dissolution with the first factor being more important. In the tension mode the addition of the cathodic poison, As, resulted in more rapid failure, a result entirely consistent with a hydrogen-embrittlement mechanism. The beneficial effect of the As addition in the torsion mode was probably the result of an inhibitive effect upon the dissolution reaction. The role of the stress in embrittlement and stress-corrosion processes has been examined in some detail by employing the slow strain-rate technique 134,135 . Specimens of alloy 7179-T651 tested in air or in vacuum after pre-exposure to water at 70°C or in water at various potentials at ambient temperature exhibited a reversible embrittlement in excess of that arising from testing in moist air 134. The embrittlement was attributed to hydrogen absorption, and recovery was thought to be due to loss of hydrogen (particularly under vacuum) or to diffusion to traps. Potentiostatic tests revealed
8: 138
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
two potential regions of embrittlement corresponding to one cathodic and one anodic to the open circuit potential. Specimens of alloy 7049-T651 also exhibited a reversible pre-exposure at low strain-rates, with the critical strain-rate decreasing in less aggressive environments 135. Recovery from pre-exposure embrittlement was only observed when specimens were subsequently strained in an inert environment. In laboratory air or seawater pre-exposure and subsequent strain-rate effects were additive. Potentiostatic tests revealed, as with alloy 7179, that there were two potential regions of embrittlement . Fractography and overageing effects both indicated that the major embrittling species at the free corrosion potential was hydrogen embrittlement. It appeared that hydrogen absorption led first to transgranular fracture and then to intergranular fracture, with the transition occurring at lower local hydrogen concentrations as the strain rate was decreased. Similar results and conclusions were drawn from experiments on A1-6Zn-3Mg, A1-6Zn-3Mg-1.7Cu and A1-6Zn-3Mg-0.14Cr alloys preexposed in the solution-treated condition to moist vapour at 1 15°C'36. In addition to examining pre-exposure effects, the slow strain-rate testing technique has been used increasingly to examine and compare the stresscorrosion susceptibility of aluminium alloys of various compositions, heat treatments and forms. A recent extensive review 137 draws attention to differences in response to the various groups of commonly employed alloys which are summarised in Fig. 8.57. The most effective test environment was found to be 3% NaCl 0.3% H,Oz. The most useful strain rate depends upon the alloy classification. The susceptibility of AI-Li alloys to stress-corrosion cracking has been
+
100dav
10dav I
.-0
5
Approx test duration 2.5 h 1 dav I
1
15 min
100 s
I
l-
0.6
. .
I 10-8
10-7
I 10-6 10-5 Nominal strain rate (s-') I
I 10-4
I
10-3
Fig. 8.57 Strain rate regimes for studying stress corrosion cracking of 2 OOO, 5 OOO and 7 OOO series alloys '37. The ductility ratio is the ratio of elongation-to-fractureor reduction in area measured in solution to that measured in a control environment
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
8 : 139
examined to a limited extent 13'. In alternate immersion crack initiation testing the alloys are less susceptible than the extensively used aerospace alloys. Restricted geometry conditions and thin-film de-aerated electrolyte conditions promote cracking, however, probably as a result of making the necessary development of alkaline crack-tip conditions occur more readily. The pH at the crack tip in these alloys is about 9 and is controlled by the Li+ c~ncentration'~'.A reversible embrittlement effect has been detected in these alloys thereby suggesting a possible role for absorbed hydrogen in the cracking process. Susceptibility to stress-corrosion cracking is highly dependent upon the Cu content of AI-Li-Mg alloys containing 2-2.5Li and 0-0.6Mg. Crack initiation in plane specimens did not occur in the absence of Cu. Restricted specimen geometry and thin film electrolyte conditions promoted cracking even in Cu-free alloys. In AI-Li and Al-Li-Zr alloys crack initiation did not occur in plane specimensL3'.In DCB specimens, however, crack propagation occurred from notches in a 95% environment at 40°C. Cracking in an AI-2.8Li under alternate immersion conditions has also been reported 139. J. C. SCULLY
REFERENCES 1. Stress-Corrosion Cracking in High Strength Steels and in Aluminium and Titanium Alloys (ed. B. F. Brown), NRL, Washington D.C. (1972) 2 . Jackson, J. D. and Boyd, W. K., DMIC Technical Note, Battelle Memorial Institute, Columbus, Ohio (1966) 3. Stress Corrosion Cracking of Titanium, ASTM STP 397, ASTM, Philadelphia (1966) 4. Logan, H. L., Fundamental Aspects of Stress Corrosion Cracking (ed. R. W. Staehle, A. J. Forty and D. van Rooyen), NACE, Houston, 662 (1969) 5. TML Report No. 88, Battelle Memorial Institute, Columbus, Ohio (1957) 6. Peterson, V. C. and Bomberger, H. B., Reference 3, 80 (1966) 7. Kirchner, R. L. and Ripling, E. J., First Interim Report, Materials Research Laboratory, Richton Park, Illinois (1964) 8. Rideout, S. P., Louthan, M. R. Jr., and Selby, C. L., Reference 3, 137 (1966) 9. Ondrejcin, R. S., Met. Trans., 1 , 3031 (1970) 10. Gray, H. R., Corrosion, 25, 337 (1969) 11. Gray, H. R., Aerospace Structural Materials Conference, No. 2 (1969) 12. Gray, H. R. and Johnston, J. R., Met. Trans., 1, 3101 (1970) 13. Boyd, W. K., Reference 4, 593 (1969) 14. Adams, R. E. and Von Tiesenhausen, E., Reference 4, 691 (1969) 15. Weber, K.E. and Davis, A. D., Lockheed California Co., NASA CR 981 Dec. (1967) 16. Brown, B. F., Lennox, T. J., Jr., Newbegin, R. L., Peterson, M. H., Smith, J. A. and Waldron, L. J., NRL Memorandum Report 15'74, November (1964) 17. Beck, T. R. and Blackburn, M. J., J.A.I.A.A., 6, 326 (1968) 18. Blackburn, M. J. and Williams, J. C., Reference 4, 620 (1969) 19. Beck, T. R., Reference 4, 605 (1969) 20. Sanderson, G. and Scully, J. C., Corros. Sci., 8 , 541 (1968) 21. Gerberich, W. W., 2nd International Conference on Fracture, 919 (1969) 22. Sanderson, G., Powell, D. T. and Scully, J. C., Reference 4, 638 (1969) 23. Powell, D. T. and Scully, J. C., Corrosion, 24, 151 (1968) 24. Brown, B. F., Fujii, C. T. and Dahlberg, E. P., J. Electrochem. SOC.,116, 201 (1969) 25. Scully, J. C. and Powell, D. T., Corros. Sci., 10,719 (1970) 26. Feeney, J. and Blackburn, M. J., The Theory of Stress Corrosion Cracking in Alloys (ed. J. C. Scully), N.A.T.O., Brussels, 355 (1971) 27. Fager, D. N. and Spurr, W. F., Trans. Am. Soc. Metals, 61, 283 (1968) 28. Menzies, I. A. and Averill, A. F., Electrochim. Acta, 13, 807 (1968)
8 : 140
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
29. Mori, K., Takamura, A. and Shimose, T., Corrosion, 22, 29 (1966) 30. Sedriks, A. J., Corrosion, 25, 207 (1969) 31. Sedriks, A. J. and Green, J. A. S., Corrosion, 25, 324 (1969) 32. Spurrier, J . and Scully, J . C . , Corrosion, 28, 453 (1972) 33. Mauney, D. A., Starke Jr., E. A. and Hochman, R. F., Reference 1 1 34. King, E. J., Kappelt, G. K. and Fields, C., Bell Aerospace Systems Report (1966) 35. Battelle NASA Report, NASr 100(09) (1969) 36. Beck, T. R., Blackburn, M. J., Smyrl, W. H. and Spiedel, M. O., The Boeing Co. Report, Contract NAS 7-489. No. 14, December (1969) 37. Duttweiler, R. E., Wagner, R. R. and Antony, K. C., Reference 3, 152 (1966) 38. Fager, D. N. and Spurr, W. F., The Boeing Co. Report D6-22691 39. Nelson, H. G., Williams, D. P. and Stein, J . E., Met. Trans, 3, 469 (1972) 40. Binxi, Y., J. Chinese Society of Corrosion and Protection, 3 , 41 (1983) 41. Smyrl, W. H. and Blackburn, M. J., Metal/, Trans. A . , 31, 370 (1975) 42. Scully, J . C. and Adepoju, T. A., Corros. Sci., 17, 789 (1977) 43. Koch, G. H . , Bursle, A. J. and Pugh, E. N., MetaN. Trans. A , 9, 129 (1978) 44. Williams, D. N., J. Iron Steel Inst., 91, 147 (1962-3) 45. Koch, 0. H . , Bursle, A. J., Liu, R. and Pugh, E. N., Metall. Trans. A , 12, 1833 (1981) 46. Paton, N. E. and Spurling, R. A., Metall. Trans. A , 7, 1769 (1976) 47. Boyd, J. D., Trans. A.S.M., 62, 1977 (1969) 48. Wreidt, H. A. and Oriani, R. A., Acta Met., 18, 753 (1970) 49. St. John, C. and Gerberich, W. W., Metall. Trans. A , 4, 589 (1973) 50. Green, J . A. S., Hayden, H. W. and Montague, W. G., Eflect of Hydrogen on Behavior of Materials, ed. Thompson, A. W. and Bernstein, 1. M., AIME, Warrendale, Pennsylvania, p. 200 (1976) 5 1 . Adepoju, T . A. and Scully, J. C., Corros. Sci., 15, 415 (1975) 52. Muskowitz, J. A. and Pelloux, R. M., Metall. Trans. A , 10, 509 (1979) 53. Ebtejah, K., Hardie, D. and Parkins, R. N., Corros. Sci., 25, 415 (1985) 54. Lycett, R. W. and Scully, J. C., Corros. Sci., 19, 799 (1979) 5 5 . Katz, Y. and Gerberich, W. W., Int. J. Fract. Mech., 6, 219 (1970) 56. DeLuccia, J. J., Final Report, NADC076207-30 (June 1976) 57. Meller, F. and Metzger, M., U.S.N.A.C.A. Tech. Note No. 4019 (1957) 58. Romanov, V. V., Stress Corrosion Cracking of Metals (translated from the Russian), 61 (1961) 59. Loose, W. S., Magnesium, ASM, Cleveland, Ohio, 173 (1946) 60. Priest, D. K., Beck, F. H. and Fontana, M. G., Trans. ASM, 47, 473 (1955) 61. Priest, D. K., Stress-Corrosion Cracking and Embrittlement (ed. W. D. Robertson), Wiley, New York, 81 (1956) 62. Heidenreich, R. D., Gerould, G. H. and McNulty, R. E., Truns. AIME, 166, 15 (1946) 63. Pardue, W. M., Beck, F. H. and Fontana, M. G., Trans. ASM, 54, 539 (1961) 64. George, P. F. and Diehl, H. A., Mer. Prog., 62, 121 (1952) 65. Logan, H. L., J. Res. Mat. Bur. Stand, 65C, 165 (1961) 66. Fairman, L. and West, J. M., Corros. Sci., 5 , 711 (1965) 67. Perryman, E. C. W., J . Inst. Met., 78, 621 (1950-51) 68. van Rooyen, D., Corrosion, 16, 421t (1960) 68a. Chakrapani, D. G. and Pugh, E. N., Met. Trans., 6A, 1155 (1975) 69. Wearmouth, W. R., Ph.D. Thesis, University of Newcastle-upon-Tyne (1967) 70. Logan, H. L., Stress Corrosion Cracking, Wiley, New York, 217 (1966) 71. Timonova, M. A,, Intercrystalline Corrosion and Corrosion of Metals Under Stress, Consultants Bureau, New York, 263 (1962) 72. Forrest, P . G., Fatigue of Metals, Pergamon, 146 (1962) 13. Brown, B. F. and Beachem, C. D., Corrosion Sci., 5, 749 (1965) 74. Hunter, M. A., Metals Handbook, ASM, Cleveland, Ohio, 234 (1948) 75. Loose, W. S., The Corrosion Handbook (ed. H. H. Uhlig), Wiley, New York, 173 (1948) 76. Loose, W. S. and Barbian, H. A., Stress Corrosion Cracking of Metals, ASTM/AIME, 273 (1944) 77. Sprowls. D. 0. and Brown, R. H., Reference 4, 466 (1969) 78. Graf, L. and Neth, W., Z. Metallunde, 60,789 and 860 (1969) 79. Petri, H . G., Siebel, G. and Vosskuhler, H., Aluminium, 26, 2 (1944) 80. Engell, H . J., Neth, W. and Suchma, A., 2. Metallkunde, 61, 261 (1970) 81. Rogers, T. H., Corrosion 1961, Butterworths, London, 605 (1962)
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
8: 141
82. Spiedel, M. O., The Theory of Stress Corrosion Cracking (ed. J. C. Scully), N.A.T.O., Brussels, 289 (1971) 83. Watkinson, F. E. and Scully, J . C., Corros. Sci., 11, 179 (1971) 84. Sedriks, A. J., Green, J. A. S. and Novak, D. L., Met. Trans., 1, 1815 (1970) 85. Lorenz, P. M., Technical Report AFML-TR-69-99 (1969) 86. Procter, R. P. M. and Paxton, H. W., ASTM J. of Materials, 4, 729 (1969) 87. Binger, W. W., Hollingsworth, E. H. and Sprowls, D. O., Aluminium (ed. K. R. van Horn), VoI. I, ASM, 209 (1967) 88. Anderson, W. A., US Patent 3 232 796, February 1 (1966) 89. Niederberger, R. B., Basil, J. L. and Bedford, G. T., Corrosion, 22, 68 (1966) 90. Chadwick, R., Muir, N. B. and Granger, H. B., J. Inst. Metals, 82, 75 (1953-54) 91. Hunter, M. S., Frank, G. R., (Jr.) and Robinson, D. L., Second International Congress on Metallic Corrosion, N.A.C.E., Houston, 604 (1966) 92. Mears, R. B., Brown, R. H. and Dix, E. H. (Jr.), Symposium on Stress Corrosion Cracking of Metals, ASTM/AIME, 329 (1944) 93. Horst, R. L. (Jr.), Hollingsworth, E. H. and King, W., Corrosion, 25, 199 (1969) 94. Rosenkranz, W., Aluminium, 39, 741 (1963) 95. Staley, J . T., Final Report, Naval Air Systems Command Contract N00019-C8-C-0146 ( 1969) 96. Gruhl, W., Metall., 19, 206 (1965) 97. Helfrich, W. J., Corrosion, 24, 423 (1968) 98. Spiedel, M. O., Reference 4, 561 (1969) 99. Sedriks, A. J., Slattery, P. W. and Pugh, E. N., Trans. ASM, 62, 238 (1969) 100. Polmear, I. J., J. A u t . Inst. Metals, 89, 193 (1960) 101. Deardo, A. J. and Townsend, R. D., Me?. Trans., 1, 2573 (1970) 102. Watkinson, F. E. and Scully, J. C., Corros. Sci., 12, 905 (1972) 103. Brown, B. F., Fujii, C. T. and Dahlberg, E. P., J. Electrochem. SOC., 116, 218 (1969) 104. Beck, T. R., Blackburn, M. J. and Spiedel, M. 0..Quarterly Progress Report, No. 11, Contract NAS 7-489 (1969) 105. Gruhl, W., Leichtmetall-Forschunginstitutof Vereinigte Aluminium-Werke AG, Bonn, Report 1970 106. Gest, R. J. and Troiano, A. R., Corrosion, 30,274 (1974) 107. Montgrain, L. and Swann, P. R., Hydrogen in Metals (ed. I. M. Bernstein and A. W. Thompson), A.S.M., Ohio, 575 (1974) 108. Troiano, A. R., Trans. A.S.M., 52, 54 (1960) 109. Tromans, D. and Pathania, R. S., The Electrochemical Society: Extended Abstracts N , 62 (1969) 110. Gruhl, W., Z. Metallkunde, 54, 86 (1963) 111. Gruhl, W. and Brungs, D., Metall., 23, 1020 (1969) 112. Scamans, G. M., Alani, R. and Swann, P. R., Corros. Sci., 16, 443 (1976) 113. Alani, R., Scamans, G. M. and Swann, P. R., Brit. Corros. J., 12, 80 (1977) 114. Scamans, G . M., J. M a f . Sci., 13, 27 (1978) 115. Vismanadham, R. M., Sun, T. S. and Green, J. A. S., Corrosion, 36, 275 (1980) 116. Vismanadham, R. M., Sun, T. S. and Green, J. A. S., Metall, Trans. A , 11, 85 (1980) 117. Sun, T. S., Chen, J. M., Vismanadham, R. M. and Green, J. A. S., A p p . Phys. Letts., 31, 580 (1977) 118. Chen, J . M., Sun, T. S., Vismanadham, R. M. and Green, J. A. S., Metall. Trans. A , 8, 1935 (1977) 119. Scamans, G. M., Environmental Degradation of Engineering Materials. ed. Louthan, M. R., Jr., McNitt, R. P. and Sissons, R. D., Jr., Virgina Polytechnic Institute, p. 153 (1981) 120. Scamans, G. M. and Rehal, A., J . Mat. Sci., 14, 2459 (1979) 121. Malis, T. and Charturvedi, M., J. Mat. Sci., 17, 1479 (1982) 122. Pickens, J. R., Precht, W. and Westwood, A. R. C., J . Mat. Sci., 18, 1872 (1983) 123. Holroyd, N. J . H. and Scamans, G. M., Scripfa Met., 19, 915 (1985) 124. Montgrain, L. and Swann, P . R., Hydrogen in Metals, A.S.M., Metals Park, p. 575 (1974) 125. Takano, M. and Nagata, T., Corr. Eng. Japan, 32, 456 (1983) 126. Scamans, 0. M. and Tuck, C. D. S., Environment Fracture of Engineering Materials, ed. Foroulis, 2. A., AIME, Warrendale, Pennsylvania, p. 464 (1974) 127. Christodoulou, L. and Flower, H. M., Acta Met., 18, 481 (1980)
8: 142
S.C.C. OF TITANIUM, MAGNESIUM AND ALUMINIUM ALLOYS
128. Carialdi, S. W., Nelson, J. L., Yeske, R. A. and Pugh, E. N., Hydrogen Eflects in Metals, ed. Bernstein, I. M. and Thompson, A. W., AIME, Warrendale, Pennsylvania, p. 437 (1981) 129. Scamans, G. M., Scripta Met., 13, 245 (1979) 130. Scamans, G. M., Metall. Trans. A , 11, 846 (1980) 131. Scamans, G. M., Reference 36, p. 467. 132. Wood, W. E. and Gerberich, W. W., Metall. Trans. A , 5 , 1285 (1974) 133. Pickens, J. R., Gordon, J. R. and Green, J. A. S., Metall. Trans. A , 14, 925 (1983) 134. Hardie, D., Holroyd, N. J. H. and Parkins, R. N., J . Mat. Sci., 14, 6 0 3 (1979) 135. Holroyd, N. J. H. and Hardie, D., Corros. Sci., 21, 129 (1981) 136. Yuen, L. and Flower, H. M., Annual Rep., Imp. Coll. of Sci. and Tech, London (Sept 1980) 137. Holroyd, N. J. H. and Scamam, G. M., Slow Strain-Rate Stress Environment-Sensitive
Fracture: Evaluation and Comparison of Test Methods, ed. Dean, S . W., Pugh, E. N. and Ugiansky, G. M., A.S.T.M., Philadelphia, p. 202 (1984) 138. Holroyd, N. J. H., Gray, A., Scamans, G. M. and Hermann, R . , Aluminium-Lithium III, (eds Baker, C., Gregson, P. J., Harris, S. J. and Peel, C. J.) Institute of Metals, London p 310 (1986) 139. Christodoulou, L., Struble, L. and Pickens, J. R., Aluminium-Lithium II, ed. Starke, E. A., Jr. and Sanders, T. H., A.I.M.E., Warrendale, Pennsylvania, p. 561 (1984)
8.6 Corrosion Fatigue*
Introduction Corrosion fatigue can be defined as a materials failure mechanism which depends on the combined action of repeated cyclic stresses and a chemically reactive environment. The total damage due to corrosion fatigue is usually greater than the sum of the mechanical and chemical components if each were acting in isolation from the other. This simple definition describes a subject of great complexity, combining as it does, many facets of metallurgy, chemistry and mechanical engineering. Numerous laboratory investigations have been carried out emphasising one or more of these aspects, stimulated either by practical requirements for engineering design data, failure analysis, or academic motivations to learn something of the mechanisms of interactions between cyclically deformed materials and their environments. In many respects, there are close parallels with stress-corrosion cracking. However, crack nucleation and self-sustaininggrowth under the combined action of a constant, not cyclic, tensile stress and a chemically reactive environment is confined to a relatively small number of material-environment combinations. On the other hand, environmental enhancement of a fatigue process can occur with a much wider range of materials and environments because of the ability of the mechanical fatigue process to maintain sharp crack tips in circumstances where non-cyclic stresses could not. Nevertheless, it will also be shown that the dividing line between stress corrosion and corrosion fatigue is not always clear from either a mechanisms or failure analysis viewpoint, in part because it is difficult in practice to be sure that a component, or indeed a test specimen, is subject to a literally constant stress. Although corrosion fatigue has been recognised and studied for many decades, certainly since World War I, it was not until 1971 that an international conference was held to review the subject'. Several reviews are also available from the same period including those of Waterhouse' in previous editions of this book and Gilbert3. From the time corrosion fatigue was first recognised and described until the 1960s, virtually all experimental investigations used smooth cylindrical specimens which were cyclically stressed until they failed or survived some pre-determined target number of *The work described in this section was undertaken as part of the Underlying Research Programme of the UKAEA.
8: 143
8: 144
CORROSION FATIGUE
stress cycles. Data from this type of experiment are typically presented in the form of an S-N curve (Fig. 8.58) which shows the number of cycles to failure, N , as a function of the cyclic stress range, S.The same technique and method of results presentation is still used today, particularly in the context of engineering qualification tests on components and welded connections.
A I R TEST DATA
A
LOG
I NUMBER
OF CYCLES1
Fig. 8.58 S-N curves for air and corrosion fatigue tests (schematic). A , Corrosion fatigue showing retarded initiation at high stress; E , Corrosion fatigue giving a general lowering of fatigue strength (after Congleton and Craig)
With the development of linear elastic fracture mechanics (Section 8.9) in the 1960s and the recognition that fatigue crack growth rates per cycle, da/cW, could be expressed as a simple function of the cyclic crack-tip stressintensity value, AK, (Fig. 8.59), increasing attention has been focused on measuring the rates of corrosion fatigue crack growth processes. This approach has an important conceptual advantage since it is clear that if a time-dependent process, such as corrosion, is combined with a non-timesensitive, but stress cycle dependent, fatigue crack nucleation and growth process to give corrosion fatigue, then the cyclic frequency becomes an extremely important variable. It is normally difficult, if not impossible, to investigate fully cyclic frequency effects in an integrated lifetime S-N test. This is because high test frequencies, often considerably greater than 10 Hz, are necessary if the complete S-Ncurve, including low stress ranges and high cyclic lives greater than lo6 cycles say, is to be defined in an acceptably short period of time. On the other hand, a test which measures the rate of failure development is not nearly so severely constrained in studying cyclic frequency effects and therefore the time dependent aspects of corrosion fatigue. It should also be noted that the relationship between corrosion fatigue crack growth and S-N data is not necessarily straightforward and will be discussed later on.
8: 145
CORROSION FATIGUE
'
lo2
0
-
'
L
CORROSION FATIGUE CRACK GROWTH
THRESHOLD, AKO I 3 1071 10 RANGE O F STRESS INTENSITY FACTOR,AK(MhTm)
Fig. 8.59 Features of a corrosion fatigue crack growth curve
The diversity of practical corrosion fatigue problems investigated during this century illustrates the range of material-environment combinations which must be considered. For example, during World War I carbon steel towing ropes used in mine sweeping exhibited very short lifetimes which were not improved by increases in steel wire strength. Galvanising did prove to be effective, however. More recent examples associated with the marine environment have concerned the integrity of submarine hulls and of offshore structures for oil and gas production. Aircraft components must also be proved against corrosion fatigue from environments as diverse as water spray affecting undercarriage components to very hot gas environments typical of jet engines. Heat exchangers of all types which can be subjected to water hammer or cyclic thermal stresses associated with their operation also widen the range of materials from steels to nickel, aluminium, copper- or titanium-base alloys and environments from sea or river water to the carefully controlled water chemistries typical of modern boilers and nuclear power reactors. One method of ordering or categorising this great diversity of materials and environment combinations which will be followed here, is to divide the
8: 146
CORROSION FATIGUE
subject matter up into general groups as follows: (1) gaseous oxidation and adsorption, (2) other non-aqueous environments such as liquid metals, (3) aqueous systems subject to general corrosion and/or pitting, (4) aqueous systems in which the materials are immune to general corrosion, usually by virtue of a barrier coating or cathodic protection, and ( 5 ) aqueous systems in which the materials form protective oxide or passive films. Examples will be used below to illustrate corrosion fatigue behaviour of metal-environment combinations falling into each of these categories in order to deduce underlying principles and common themes. It is important to note that, apart from possibly the first and third categories above, an environment need not necessarily be ‘corrosive’ in the normal sense of the word for it to exert a substantial effect in corrosion fatigue. This arises because local strains associated with the formation and propagation of fatigue cracks can fracture or greatly thin protective films and/or expose highly, chemically reactive, fresh metal surface which can behave chemically in a radically different way to other unstrained surfaces. In the succeeding sections of this chapter a brief description of the mechanisms of fatigue crack initiation and growth as presently understood is giyen together with an indication of the various ways in which corrosion may influence these mechanical processes. After that, illustrative examples of corrosion fatigue crack growth and corrosion fatigue endurance in various alloy-environment combinations using the categories given in the previous paragraph are described. The chosen order of presentation of endurance data following crack growth data is done deliberately so that the influence of corrosion on the relative contributions of the crack nucleation and growth phases of failure development in endurance tests can be assessed and thereby linked to the final section on practical applications.
Mechanisms An important development over the last few decades has been an improved understanding of the mechanisms of how fatigue cracks initiate and grow in metallic materials. An essential first step is the localisation of cyclic, plastic deformation onto favourably orientated slip planes. Any oxidation or adsorption process may prevent slip step reversal and continuing slip on adjacent planes leads to closely spaced groups of slip planes known as persistent slip bands (PSB) (Fig. 8.60). If the surfaces are initially smooth, these slip processes can be shown to be accompanied by intrusions and extrusions of material at the slip band with incipient cracks forming at the intrusion. The irreversible movements of dislocations associated with slip band formation are very complex and vary significantly with metallurgical In polyphase materials, the sites for strain localisation may be inclusions, grain boundaries, metallic precipitates or precipitate-free zones (PFZ) or simply a mechanical stress concentration such as a notch or corrosion pit. Following the initial localisation of strain, visible cracks then form on shear planes (stage I) and may propagate in this mode across one or several grains until one dominant crack takes over and propagates perpendicular to the imposed principal tensile stress (stage 11). Failure occurs when the remaining ligament breaks by plastic collapse or brittle fracture. Some materials such
8: 147
CORROSION FATIGUE EXTRUSION
COARSE SLIP
5-15um COARSE 5LIP
N
\
I/
INTRUSION
I /
I I
\ \
GRAIN\ BOUNDARY
(a)
P F Z OR I / P S B ‘I
lb)
(C)
Fig. 8.60 Schematic diagrams showing common surface profiles produced during fatigue: (a) coarse slip and crack initiation adjacent to grain boundaries; (b)extrusions and intrusions; (c) coarse slip within a persistent slip band (after Lynch’)
as steels exhibit definite fatigue or endurance limits in air or vacuum and at cyclic stress ranges below the limit, fatigue failure does not occur. In such cases the endurance limit stress range coincides with the cyclic stredstrain yield point which is often about half the tensile strength. In other cases a definite fatigue limit is not observed and endurance limits for a suitably large number of cycles, say lo7 or lo8, is quoted where the slope of the falling S-N curve is shallow (Fig. 8.58). It is plain from the description above that the boundary between crack initiation and crack growth is not clear cut. Indeed many would regard the distinction as semantic, or state that most of the fatigue life is spent in crack propagation, however small those cracks might be. Nevertheless, the proportion of cyclic life occupied by the various stages can vary greatly with metallurgical structure, magnitude of the applied cyclic and mean stress, geometry and environment. Only stage I1 of the growth process (Fig. 8.61) can be properly characterised in terms of the linear elastic parameter, the cyclic range of the crack-tip stress-intensity factor, AK. Even this is subject to the conditions that the crack-tip plasticity be contained within an elastic continuum and that the crack is large compared to microstructural dimensions. When these conditions are satisfied and the environment is benign, the familiar Paris equation can characterise the crack growth rate per cycle, da/dN, over a wide range of AK (Fig. 8.59). Stage I crack growth, or stage I1 growth under high strain conditions, requires more specialised methods of analysis to represent the driving cracktip stress field. This problem is now known as the ‘short crack’ problem, defined roughly by cracks 0.01 to l.Omm deep dependent on alloy strength and raises unique issues with regard to the influence of chemically reactive environments. Nevertheless, successful quantitative representations of high strain, cyclic endurance by the Coffin-Manson equation6 predate attempts to characterise fatigue crack growth explicitly and are widely used in low cyclic fatigue design. It is important to note, however, that the shear decohesion processes (Fig. 8.60) associated with fatigue failure in ductile metallic materials are essentially the same throughout all the stages of crack initiation
8 : 148
Fig. 8.61
CORROSION FATIGUE
Fatigue crack propagation across specimen section (after Tomkins and Wareing 16)
and growth whether the net section stresses or strains are above or below the elastic limit. Difficulties in characterising fatigue crack growth quantitatively arise only from difficulties in providing adequate descriptions of the cracktip driving force over the full range of stresses and strains and crack sizes. However, the characteristic crack growthlarrest markings known as striations which are commonly visible on ductile metal fatigue fracture surfaces in benign or mildly oxidising environments are normally associated with stage I1 growth. Their formation is illustrated in Fig. 8.62. Before discussing the influence of corrosion on the mechanical deformation processes of fatigue crack initiation and growth in some specific systems, it is useful to have a general mechanistic framework to which specific examples can be related. In the early stages of fatigue crack nucleation, the main effect of corrosion is to accelerate the plastic deformation and slip processes which precede the formation of stage I cracks. These may be broadly classified into four groups: (1) where oxide films interfere with slip reversibility, (2) where adsorbed species influence slip by facilitating the
8: 149
CORROSION FATIGUE TENSION
LOCAL
b
9
SHEAR
STAGE IN STRESS CYCLE
Fig. 8.62 Crack-blunting model of stage I1 crack progagation (after Laird)16
nucleation of dislocations, (3) where corrosion processes result in the injection of an embrittling species such as hydrogen, and (4) where corrosion removes plastically deformed material. The same processes can also influence the later stages of crack growth although two additional considerations come into play. One is the role of oxides or other corrosion products in impeding crack closure and consequent effects on the effective range of the crack-tip stress-intensity. The second is the effect of a long, narrow diffusion path for reactants and products along a macroscopic crack and the possibility of chemical modifications to crack-tip environments relative t o the external bulk environment. Owing to the nature of the fatigue process in ductile metals, it is clear that a constant supply of atomically clean, new surface is presented to interact with any environment. Oxygen and water will normally adsorb strongly and very quickly on these new clean metal surfaces. Any oxidising agent, whether gaseous or aqueous, will react rapidly with fresh metal surface exposed by the fatigue process and the extent of reaction per cycle will then clearly
8 : 150
CORROSION FATIGUE
depend on the duration of the cycle period. With gaseous oxidants there is usually little ambiguity about the nature of the processes leading to environmental acceleration or in some cases retardation. Oxygen and air for example at normal ambient temperatures usually generate sufficient oxide to impede any rehealing during the compressive part of the cycle of the new metal surfaces created during crack opening. These processes normally also do not seriously impede crack closure of longer stage I1 cracks. At higher temperatures, however, the formation of thicker oxide layers can impede crack closure thereby raising apparent crack growth thresholds and reducing crack growth rates. Alternatively, hydrogen and hydrogen sulphide gases are generally thought to accelerate fatigue by supplying hydrogen to cause embrittlement of the plastically deformed metal created by the fatigue process. Thus any processes impeding hydrogen entry can reduce the embrittling effect. Some liquid metal environments can transport carbon to or from the crack-tip zone and thereby alter the rate of crack growth. One author’ has proposed that liquid metals facilitate the nucleation of dislocations and, on the basis of fractographic evidence, has drawn parallels with many other environments to suggest that the same mechanism operates in other cases. Water vapour and aqueous solutions are much more difficult to interpret unambiguously. In aqueous systems at the free corrosion potential, anodic processes such as dissolution of persistent slip bands or the crack tip cannot proceed without a corresponding cathodic process. Since dissolved oxygen cannot penetrate far down cracks or crevice geometries, hydrogen evolution is the most likely supporting cathodic reaction in long cracks. Thus, the dominance of dissolution or hydrogen-embrittlement processes in accelerating fatigue crack growth is difficult if not impossible to prove. Even potentiostatically controlled tests can be difficult to interpret when the problems of establishing the crack-tip potential and the chemistry of crack-tip environments are considered. In passing, one may note that the common assertion that acid environments form in stress-corrosion and corrosion fatigue cracks often does not stand up to examination. This is because precisely known preconditions must be satisfied for this to occur; i.e. a potential difference must exist between the crack tip and exterior surfaces which is normally provided by an oxygen concentration cell; the dissolved cation must be hydrolysable; and an acid forming anion must also be present. Persistent slip band or crack-tip dissolution can also act to slow down corrosion fatigue cracking by blunting the crack tip. Clearly, whether dissolution processes lead to cracktip blunting depends on the kinetics of the dissolution process and the counter effect of stress cycle dependent mechanical sharpening by fatigue. Precipitation processes within cracks can also lead to significant perturbations to crack closure with consequent effects on rates and thresholds as well as effects on diffusion of chemical species into and out of cracks. The complexity of these chemical and mechanical interactions is such that each metal-environment system must be examined on an individual basis to determine the important processes influencing corrosion fatigue crack nucleation and growth rates. Thus, in the ensuing sections, examples are quoted to illustrate commonly occurring phenomena or establish more general principles with reasonably wide applicability for particular classes of metal/environment combinations. It should be noted, however, that when
CORROSION FATIGUE
8 : 151
it becomes necessary to evaluate new metal-environment combinations, there is no unified theory of corrosion fatigue which can avoid the need for experimental data. Studies of corrosion fatigue in metallic materials have, for the most part, aimed at measuring S-Ncurves on plain cylindrical specimens or the growth rates of macroscopically large cracks as a function of the cyclic stressintensity factor, AK (Fig. 8.58 and 8.59). One point that requires attention in S-N testing is the diameter of the test section relative to the expected loss of section size by general corrosion. Excessive corrosion could lead to premature failure by simply increasing the effective net section stress which would not then genuinely reflect the behaviour of larger components. Most of the common types of pre-cracked fracture mechanics specimen have been used in investigations of corrosion fatigue crack growth although the compact tension (CT) specimen has been the most popular. The CT specimen has a number of important experimental advantages for this type of work, among them a high mechanical advantage allowing the lowest applied loads of any type of specimen to achieve a given AK value and relatively simple indirect monitoring (i.e. non-visual) of crack size by compliance or electrical resistance methods. General experimental techniques for corrosion fatigue tests including guidelines on environment containment and monitoring of environmental chemistry, including most importantly, the specimen corrosion potential in aqueous solutions, have been extensively described'. Two aspects of experimental design requiring constant vigilance are unwanted electrochemical effects from containment materials and other fixtures and prevention of interactions between electrical equipment such as potentiostats and electrical resistance crack monitoring devices. However, concerns over possible electrochemical effects arising from the use of d.c. electrical resistance crack measurement techniques appear on present evidence to be unfounded. An important underlying assumption in much fracture mechanics-based corrosion fatigue testing is the similitude between different geometries given the same environmental conditions in which cracks are assumed to be characterised by one simple parameter, AK. This is an assumption which has to be carefully examined, particularly if the results are required for some practical application. For example, will the environmental conditions prevailing in a crack in a CT specimen imitate correctly those in a more realistic semi-elliptical crack shape? In one case for structural steel in seawater, this has been shown to be a reasonable assumption'. There is more doubt in the case of pressure vessel steels in simulated light water reactor environments where exchange of dissolved impurities between the crack environment and the bulk has a critical influence on the environmental contribution to cracking". In the case of short cracks, however, there are both mechanical reasons invalidating the representation of crack-tip strains and deformations by AK as discussed earlier" and theoretical and experimental evidence of the importance of crack size at the millimetre to submillimetre level on crack chemistry '*.
8: 152
CORROSION FATIGUE
Crack Propagation-Non-aqueous Environments Many fatigue crack propagation experiments are carried out in laboratory air at normal ambient temperatures without any concern that air oxidation or reactions with water vapour might be contributing to the crack extension process. In many cases, but not all, the use of air data as a baseline against which to compare other environmental influences is a reasonable working assumption. Nevertheless, it is well known that in steels, for example, crack growth rates obtained in vacuum are two to three times less than those obtained in.airI3. It may also be noted in passing that fatigue crack growth in many different materials in air can be plotted in a relatively narrow scatter band if expressed as a function of AK/E, where E is the elastic modulus, rather than AK alone. Crack growth thresholds are also often adversely influenced by air oxidation at ambient temperatures relative to vacuum 1 3 . At higher temperatures, for example those relevant to gas turbines, air oxidation and other corrosion processes become progressively more important contributors to fatigue crack growth in iron- and nickel-base alloys of commercial interest. An order of magnitude increase in crack growth rates in iron- or nickel-based superalloys for aero engines is not uncommons~'3. However, oxide wedging at low AK values which reduces crack-tip opening displacements can actually raise the apparent crack growth threshold under constant amplitude fatigue loading. Such a mechanism could be less effective under complex spectrum loading sequences where compressive forces can grind up an accumulating oxide scale. At even higher temperatures above about one third of the melting temperature of an alloy, creep effects also begin to contribute to the crack extension process as well as air oxidation and the resulting crack growth behaviour can vary in a very complex way with loading and environmental variables'4.'5. Once material failure processes such as creep make a significant contribution to crack growth, non-linear deformation processes occurring in front of the advancing crack invalidate the stress-intensity factor K,or its cyclic range AK, as a sensible parameter characterising the near crack-tip stress field. The difficulties inherent in finding an acceptable characterising stress field parameter for crack growth under creep-fatigue conditions have been discussed extensively by Tomkins and co-workers I 4 * l 6 . For temperatures below the creep range it has been suggested that oxidation can only accelerate fatigue crack growth to an upper limit defined by half the maximum crack-tip opening per cycle as shown by the examples in Fig. 8.63. This is an important principle to understand because of its potential use in design problems, as are the circumstances under which the principle breaks down. From considerations of the feasible geometry of crack tips, fatigue cracks growing in ductile materials by a shear decohesion process cannot be greater than half the crack-tip opening displacement per cycle. The reason that fatigue cracks usually grow at less than this rate is because in real hardening materials, crack-tip strains are not accommodated on one shear plane emanating from the crack tip but on many planes which spread plastic flow to the crack flanks immediately behind the crack tip. Thus any corrosion process which is indiscriminate in removing material from the crack tip or sides will cause blunting if the resulting combination of environ-
8: 153
CORROSION FATIGUE
1
I
I
I
I
5
6
7
8
I I 910
I 15
STRESS-INTENSITY
I
I
20
25
I
I
1
30 35 LO
AMPLITUDE, MN,-%
Fig. 8.63 Effect of environment on fatigue crack growth rate in 1% Cr-Mo-V steel at 550°C (after Tomkins and WareingI6)
mental and mechanical fatigue damage exceeds the theoretical maximum crack-tip opening. Equally, any alternative, potentially self-sustaining failure process such as creep or, as will be seen later, stress-corrosion cracking, that contributes to crack growth will not be contained within this theoretical maximum fatigue crack growth rate defined by the maximum crack-tip opening displacement. At high temperatures, an obvious alternative corrosion process which would cause this generalisation to break down would be hot salt corrosion, particularly sulphidation which occurs in marine gas turbines, and leads to rapid intergranular crackingL3.It has also been observed that rapid diffusion of oxygen down grain boundaries in some superalloys at high temperatures causes large increases in fatigue crack growth which are unrelated to creep effects6. Where creep cavitation occurs, extremely large accelerations in crack growth rate are possible when the cyclic crack-tip opening is of the same order as the cavitation spacing. Such rapid failure processes can be likened to opening a zip fastener through the material. Some of the most interesting work.on the mechanisms of corrosion fatigue crack growth has been done on steels and high strength aluminium alloys in carefully controlled water vapour or hydrogen gas environments. Great care is needed in this type of work to ensure the removal of adsorbed species under vacuum prior to admitting the gaseous environment of interest which
8: 154
CORROSION FATIGUE
must also be very pure*. The fracture surfaces of AISI 4340 steel resulting from stress corrosion or corrosion fatigue and the reaction kinetics between water vapour and iron crystals of known orientation have been studied by Auger electron spectroscopy and low energy electron diffraction 1 7 , .~ These results showed that the rate limiting step in the environmental contribution to corrosion fatigue is the reaction between water vapour and iron or possibly iron carbide. Observations of transients in crack growth rates following changes of cyclic frequency strongly suggested that hydrogen produced from the reaction between iron and water vapour is primarily responsible for the environmental enhancement of fatigue crack growth in high strength AISI 4340 steel. The zone of hydrogen damage was also deduced to be somewhat greater (approximately 0.1 to 1.0mm) than the calculated reversed plastic zone size at the tip of the crack or relevant microstructural dimensions. Similar measurements and deductions have been made for high-strength aluminium alloys 19. Interest in the role of hydrogen embrittlement in corrosion fatigue, particularly in steels, but also high-strength aluminium alloys and the hydride forming metals such as titanium and zirconium, has prompted much research using hydrogen or hydrogen sulphide gases. In addition there have also been industrial uses and failures of these combinations of materials and environments which have given added impetus to the work. In steels, the influence of hydrogen on fatigue crack propagation shows close parallels with behaviour in low temperature aqueous environments”. For example, research work following a catastrophic failure in 1974 of a 3+ Ni-Cr-Mo-V steel end ring component of a 500 MW generator operating in 5 bar pressure hydrogen established that the high yield stress of the material of 1 250 MPa rendered it susceptible to hydrogen-induced crack growth at constant crack-tip stress intensity. Parallel corrosion fatigue experiments showed the classical above and below K,,,, behaviour seen in high-strength steels in low temperature aqueous chloride solutions (see Fig. 8.59 and next section). Thus any fraction of the cycle period spent with the stress intensity above the static threshold for hydrogen cracking resulted in large increases in corrosion fatigue crack growth rates and the coincident presence of intergranular or brittle facets on the resulting fracture surfaces. At yield strengths below 1 100 MPa these end ring steels were not susceptible to hydrogen cracking under constant loads but there was nevertheless a significant residual frequency-dependent hydrogen-environment effect on fatigue crack growth rates, (Fig. 8.64). Such effects are enhanced by increasing hydrogen pressure (Fig. 8.64) and by the presence of hydrogen sulphide gas, and substantially decreased by air contamination of the hydrogen atmosphere*~20*21. This and other work has pointed to the importance of adsorption on fresh metal surfaces created by the fatigue process at the crack tip. The mechanisms by which hydrogen enhances fatigue crack growth once absorbed into the metal remains as much a mystery in this as in other hydrogen-embrittlement research. The reduction in the adverse effect of low-pressure hydrogen gas atmospheres at low cyclic frequencies (Fig. 8.64) is particularly difficult to explain. It may be due to the mismatch between hydrogen and dislocation mobility within the plastic zone, since hydrogen is rather weakly bound to dislocations, or due to minor impurities
30
s I
E
z I 0
-
-
ti 2
e--*
PRESSURE
WAVEFORM
L2 B A R
SINE
2 BAR
SINE
2 BAR
SQUARE
0
II Y
Fig. 8.64 Influence of hydrogen pressure, frequency and waveform on the enhancement of fatigue crack growth in 708M40 steel AK = 30 MNm-3’2 (after McIntyreZL)
in the gas atmosphere slowly poisoning active metal surface sites for hydrogen adsorption at the crack tip”. Corrosion fatigue crack growth in high-strength aluminium alloys is strongly influenced by the presence of water vapour typically between 100 and 10 OOO ppm in air or other inert or oxidising gases at normal ambient temperaturesL9’22. The adverse effect of water vapour tends to saturate at the higher partial pressure. The influence of water vapour has been attributed to hydrogen embrittlement. However, although no systematic studies of hydrogen gas atmospheres on fatigue crack growth are available, gaseous hydrogen has not been found to influence appreciably total fatigue life in these alloys. The apparent discrepancy may be due to the extreme reactivity of new aluminium surfaces created at fatigue crack-tips with any oxidising impurity in the gaseous environment and the impervious nature of aluminium oxide films to hydrogen diffusion. Fatigue crack propagation has been studied extensively in stainless steels over a wide range of temperatures and oxidising environments because of important actual or potential applications in nuclear reactors and steam raising plant Environments such as nitrogen, argon and liquid sodium at temperatures up to 500°C have little influence on fatigue crack growth in ductile stainless steels such as types 304 and 316 compared with vacuum over the same temperature range or room temperature air. Air at 500°C
8: 156
CORROSION FATIGUE
produces more than an order of magnitude increase in propagation rates over a wide frequency range whereas steam at the same temperature causes nearly a further two orders of magnitude increase in crack growth rates. Clearly these are very large and significant effects of environment apparent under conditions where creep interactions are insignificant. Pressurised water at 300°C is not nearly so aggressive, however, at least on solutionannealed stainless steel in the absence of dissolved oxygen (see next section), indicating that strongly thermally activated oxidation processes must operate at higher temperatures. A considerable technical literature exists on liquid metal embrittlement, but relatively little work has been done on corrosion fatigue crack growth in liquid metals. Most work relates to the influence of sodium at relatively high temperatures around 600°C on stainless steels in the context of core components for fast reactors15. In low oxygen (5-10 ppm) sodium, fatigue crack growth rates in type 316 stainless steel are equivalent to those measured in vacuum or inert gases. Carburising or decarburising sodium can enhance these rates by up to a factor of five, however. Other possible contaminants such as lead, tin or zinc may also have very adverse effects. An additional environmental aspect of nuclear reactor components, particularly those in close proximity to or part of the core, is neutron irradiation damage. In general, there does not seem to be a serious adverse effect of irradiation on fatigue crack growth in ferritic or stainless steels until very high doses, say greater than one displacement per atom, are encountered. In these circumstances, helium bubble formation, in particular from n, a reactions with boron within the metal, accompanied by physical swelling occurs and considerable frequency-dependent degradation of fatigue and creep fatigue properties is possible. As in the case of creep cavitation, the most severe effects are observed in stainless steels when the crack-tip opening is of the same order as the helium bubble spacing.
Crack Propagation-Aqueous Environments Steels
A great deal of experimental work has been carried out using carbon and low-alloy steels in either 3.5% sodium chloride solution or seawater. At the medium-to-low-strength levels, say less than 1 O00 MPa yield strength, such materials are not normally susceptible to environmentally-induced cracking (by hydrogen embrittlement) under constant applied loads in aqueous environments unless there are additional sources of hydrogen such as from hydrogen sulphide contamination or excessive cathodic polarisation. By contrast, fatigue crack propagation rates are markedly increased both at the free corrosion potential and at more cathodic potentials consistent with reasonable levels of cathodic protection. The increase in fatigue crack growth rates due to corrosion can be represented by a simple multiplying factor on the corresponding in-air rates like that given earlier in Fig. 8.64 for hydrogen gas environments. Similar observations have been made for quite a wide variety of low-alloy steels freely corroding in 3.5% sodium chloride
CORROSION FATIGUE
8: 157
solution or seawater 8*z4.z5. It is seen that aqueous environmental influences on fatigue crack growth are negligible at frequencies of 10Hz and above and tend to reach a limiting factor at 10-’Hz or lower frequencies. At even lower frequencies, there is evidence that the environmental effect declines due to crack-tip blunting and in combination with low values of AK, cracks can actually be arrested because the crack-tip pitting rate is faster than the crack growth rate (on a time base). This has been demonstrated particularly well for intermittent wetting and drying conditions for structural steel in seawater representing splash zone environments on offshore structures’. These crack-tip corrosion processes are also thermally activated and an activation energy of about 40 kJ/mole can be deduced from temperature effects on corrosion fatigue crack growth rates around normal ambient temperaturesz4. In addition to the cyclic frequency, the shape of the cyclic waveform also has a marked effect on the environmental contribution to crack growth as originally shown by Barsomz6. It has been demonstrated in several steel/ aqueous environment combinations that the primary environmental contribution to crack extension occurs during the increasing load part of the cycle. Thus, for cycle waveforms of the same period, sine, triangle and positive sawtooth shapes show similar environmental effects, whereas square and negative sawtooth waveforms (Le. those with a very fast leading edge or rising load) show negligible contributions from the aqueous environment to crack growth when compared with normal laboratory air test results. Clearly, the mechanistic significance is that the environmental influence depends on the length of time in the cycle that new metal surface is being exposed to the chemically reactive solution in the crack enclave. There is now a considerable body of evidence that points to hydrogenembrittlement as being primarily responsible for the accelerations in fatigue crack growth seen in steels freely corroding in ambient temperature aqueous environmentsz8.For example, as pointed out above, the frequency response of corrosion fatigue in low-alloy steels in hydrogen gas closely resembles that in aqueous environments ”. In addition, numerous transient effects during changes of experimental conditions seem inexplicable except on the basis of hydrogen embrittlement of a zone of metal just in front of the crack tip. One might anticipate on this basis that cathodic polarisation might increase corrosion fatigue rates with decreasing potential. In fact, a slightly more complex situation arises in which a minimum in the environmental effect is seen at about 100 to 200 mV below the free corrosion potential which then rises as the potential is moved increasingly in the negative dire~tion~.’~. A possible explanation for the effect of cathodic polarisation has been provided by work on hydrogen permeation rates through low-alloy steel crevices subjected to cathodic polarisation at the crevice mouthz8. Hydrogen permeation rates at the base of a crevice as a function of externally applied potential exactly match the trend of the environmental contribution to corrosion fatigue rates. This can be readily understood in terms of crack-tip acidification enhancing hydrogen production at the crevice tip at the free corrosion potential, but being reduced at slightly more negative potentials by the accumulation of alkaline cathodic reaction products until finally the rate of hydrogen production increases again as the overpotential for hydrogen evolution becomes greater.
8: 158
CORROSION FATIGUE
The influence of crack-tip chemistry and electrochemistry on corrosion fatigue crack growth in steels in salt water environments has been extensively reviewed by Turnbull ’*. Corrosion fatigue crack-tip chemistry in large cracks (> 10 mm deep) is surprisingly little disturbed compared with static crevices by mechanical pumping effects, at least at low frequencies such as 0.1 Hz or below. Thus much of the understanding which has developed in recent years concerning crack-tip chemistry in relation to stress-corrosion cracking is also relevant to corrosion fatigue. In ionically conductive solutions such as 3.5% sodium chloride solution or seawater, ohmic drops down cracks or crevices are not large, at least at externally imposed potentials within, say, 500mV of the free corrosion potential. Thus in seawater, cathodic protection to say - 850 mV (versus Ag/AgCl) will give a crack-tip potential of the order of -800mV (versus Ag/AgCl). This gives rise to a complication for seawater whereby calcareous scale can precipitate both within and outside a crack as a consequence of alkali-forming cathodic reactions. This hard calcareous scale which forms on the crack flanks can have a large effect in reducing the degree of crack opening for a given applied load range with the result that cracks which would otherwise grow at low rates, slow down and even arrest when cathodically polarised in seawaterg. This phenomenon is most in evidence at crack growth rates approaching the in-air threshold hK, Le. at AK values of less than 15 MPa The final major parameter influencing corrosion fatigue crack growth rates in low-alloy ferritic steels in ambient temperature aqueous environments in addition to cyclic frequency, waveform and electrochemical potential is the mean stress level about which the cyclic stress oscillates. It is normal in work on fatigue crack growth to define the mean stress conditions in terms of the stress ratio, R , equal to the ratio of the minimum to the maximum stress or stress intensity in the cycle (Fig. 8.59). The stress ratio has also been found to increase crack propagation rates above the threshold AK for crack growth in low-alloy steels in aqueous environments whereas there is little influence of stress ratio in air except on crack growth thresholds themselves. Figure 8.65 shows the combined effects of potential and stress ratio on fatigue crack growth in a structural steel exposed to seawater either at the free corrosion potential or at - 1.1 V (versus Ag/AgC1)9. Increasing stress ratio increases crack growth rates but the effect apparently saturates between R = 0.5 and 0.7. No good mechanistic model has been proposed to explain this effect of R ratio except that in general terms it is clear that a greater proportion of the cyclic crack-tip opening is converted into crack extension at the higher stress ratios in the presence of the aqueous environment. It will be appreciated from the discussion so far concerning the effect of chemical precipitates in cracks and dissolution rates at crack tips, that when these processes are combined with the influence of R ratio on crack growth On the thresholds, a rather complex set of interactions is fea~ible’~’’~. whole, higher stress ratios which result in the crack faces being held wider apart than with lower R ratios tend to reduce the influence of crack-tip precipitates and their effect on crack closure. Even in the absence of the complicarion of precipitates in cracks, a good deal of variability is found in crack growth thresholds in salt water environments relative to those found in air25*29.30. It is perhaps not surprising that at low crack growth rates the effect of crack-tip dissolution and any consequential hydrogen-embrittlement
a.
8: 159
CORROSION FATIGUE
3 x 10-6
1 o-6
I-
--
M E A N AIR DATA LINE X6 8 5 4 3 6 0 : 500 STEEL
+'d
10-9
1
oO"
I
/
I
I
I I I I
CYCLIC STRESS INTENSITY FACTOR, A K M ~ Pam
Fig. 8.65
Corrosion fatigue crack growth data for structural steel in seawater at 0.1 Hz, R = - 1 to 0.85 and -1.lOV (Ag/AgCl) (after Scottz4)
8: 160
CORROSION FATIGUE
can reduce or increase thresholds dependent on the precise competing kinetics of the electrochemical, mechanical and metallurgical damage processes. An interesting feature of Fig. 8.65(b), which shows corrosion fatigue crack growth results for a medium strength steel somewhat over-cathodically protected in seawater, is the appearance of a stress-corrosion or plateaulike feature, particularly at high R ratios (cf. Fig. 8.59). In fact there is no evidence that this steel in its as-received condition is at all susceptible to stress-corrosion cracking under these environmental conditions. Nevertheless, a period of crack growth independent of AK as in Fig. 8.65(b) is a clear indication of the intervention of a rate-limiting process unrelated to AK; in this case most probably the rate of evolution of hydrogen near the crack tip or the rate of diffusion of hydrogen to the crack-tip process zone. Such features are commonly observed in corrosion fatigue tests in alloys which may or may not be susceptible to stress-corrosion cracking3' and provide a clear indication of how the dividing line between corrosion fatigue and stress corrosion is far from being well defined. It will be seen later when corrosion fatigue systems are discussed in which the breaking and re-healing of passive, protective oxide films are critical to the'crack advance mechanism, that the concept of environmental cracking processes dependent on the application of a continuing dynamic strain is not novel. Indeed the so-called slow strainrate test or constant extension rate test for stress-corrosion susceptibility has been specifically designed to cope with such circumstances and 'windows' of strain rates are commonly found in which environmentally-induced cracking is possible and outside which it is not. Thus we may have environmentally controlled cracking processes in corrosion fatigue dependent on dynamic straining of the crack tip by fatigue forces over a specific range of frequencies but whose rate is not a function of AK or any other cyclic crack-tip plasticity parameter because chemical reaction rates or diffusion processes are rate controlling. In the case of high-strength steels (yield strengths around or greater than about 1 OOO MPa), simple models can be employed which superimpose stress-corrosion cracking (by hydrogen embrittlement) on the fatigue process In this case, hydrogen-embrittlement cracking of high-strength steels under constant stress can be well represented as a time-dependent rate, d d d t , as a function of K , the stress-intensity factor, with a well-defined threshold, K,,,, . If a fatigue force is applied, then any fraction of the cyclic AK which exceeds KIscccauses a marked increase in observed corrosion fatigue crack growth rates as illustrated in Fig. 8.59. The fact that this relatively simple model works so well indicates that there is comparatively little strain-rate sensitivity in the constant stress hydrogen cracking process itself, either on KIsccor on the plateau growth rate. As indicated earlier, many other metal-environment systems in which mixed fatigue and stress-corrosion-like crack growth processes are possible are not so amenable to such a simple superposition model because the rate of environmental attack is itself strain-rate sensitive. An example in which this has been extensively examined is the case of pressure vessel steels exposed to simulated light water reactor coolants at c. 300°C'0. It is known that the rate of crack growth in corrosion fatigue tests on medium-strength reactor pressure vessel steels (A533-B and A508) is very sensitive to the dissolved oxygen concentration between 25 and 100 ppb (which has a strong ''s3*.
CORROSION FATIGUE
8 : 161
influence on corrosion potential), the sulphur impurity content of the steel, the sulphur anion concentration in the water and the linear water flow rate. The influence of all these factors has been rationalised on the basis that a sulphur anion rich environment in the crack enclave greatly enhances electrochemical dissolution reaction rates on emerging slip planes at the crack-tip (and consequently, also, enhances nearby cathodic hydrogen evolution reactions). However, these reactions can only take place as the protective oxide film of magnetite (or magnetite plus haematite depending on oxygen concentration) formed rapidly at these high temperatures is broke4 at the crack tip. This in turn depends on the crack-tip loading rate or strain rate. By representing the environmentally controlled rate of crack growth as a function of crack-tip strain-rate, it has been possible to construct a predictive model which is still basically a superposition model, but one in which the environmental contribution depends not only on the cracktip stress-intensity exceeding a critical minimum value but also on the effective crack-tip loading rate. Predictions of the influence of frequency and R ratio from the model fit known experimental data very well indeed. The most obvious consequences of this modification of the superpositioy principle are a dependence of the plateau corrosion fatigue rates on f-7 rather than f -' of non-strain-rate-sensitive models and the existence of a specific 'window' of cyclic frequencies only within which it is possible to observe any environmental influence on crack growth rates at all. The study of the growth by fatigue of physically short cracks usually less than 0.1 to 1 .0 mm deep is a topic of much current research interest. The study of environmental effects appears to have been confined so far to the influence of high temperature air oxidation of superalloys for aero-engines (see next section for more details) and to steels in salt-water environments. Even in the absence of reactive environments, short cracks grow considerably faster than long ones when expressed as a function of the linear elastic fracture mechanics parameter, AK. This can be due to uncontained plasticity at the tip of the crack or microstructurally important features of similar dimensions to the crack size, both of which invalidate the representation of the crack-tip driving force by AK. One commonly applied technique to take account of reduced mechanical constraint at a short crack-tip is to plot the crack growth results from both short and long cracks (i.e. conventional fracture mechanics specimens in the second case) as a function of AKeR where a correction is made to the nominal AK to allow for the minimum stress intensity at which the crack closes. This point is often detected experimentally by electrical potential drop methods. When such corrections are made, short and long crack data are normally self-consistent as a function of Me,. Similar successes of the Me, approach have been achieved in the context of oxide blocking of cracks and pressure effects in viscous liquids. Another older method, though no less successful for low cycle fatigue, has been to express crack growth rates as a power law function of the applied plastic strain range6. Crack size effects in corrosion fatigue crack growth have, however, been observed to persist to larger crack sizes than those associated with plasticity and microstructural effects. Notably, increases by up to a factor of 5 0 0 in small surface crack growth at depths up to 3 mm compared to longer cracks have been observed in high-strength A4130 low-alloy steel immersed in 3 % NaCl solution", At the high steel strength levels used in these tests, short
8 : 162
CORROSION FATIGUE
crack effects due to mechanical or metallurgical reasons were only detectable below 0.1 mm. Later experiments on a lower strength HY 130 low-alloy steel under the same test conditions showed that short cracks grew two to five times faster than long cracks while a low-strength carbon-manganese steel showed little influence of crack size on growth rates34.The effect of environment on crack growth in all these examples was attributed to hydrogen embrittlement (in common with many other similar metal/environment combinations described earlier). It was further argued, supported by some difficult calculations based on necessarily simplified models of corrosion fatigue cracks, that the enhanced environmental effect seen in short cracks was due to the increased availability of hydrogen ions for reduction to embrittling hydrogen atoms. It was suggested that as short crack lengths increased, the rate of hydrogen ion reduction increased to a characteristic maximum whereas oxygen reduction would dominate at or very close to the surface. The decrease after the maximum at even longer crack lengths was thought to be due to transport limitations of the kinetics of the hydrogen ion reduction reaction while the different responses of the three steels was attributed to their inherently differing sensitivities to hydrogen embrittlement. Turnbull has also pointed out that high-strength, low-alloy steels contain significant amounts of chromium which on dissolution and hydrolysis can lower the crack pH much more than is possible from the hydrolysis of ferrous ions3. Nevertheless, irrespective of the detailed mechanistic interpretation, the observations reported are an important reminder that the principle of similitude of corrosion fatigue crack growth rates as a function of AK cannot always be taken for granted and should always be checked when data are required for practical applications. Another example where this similitude principle may break down was described earlier for pressure vessel steels in high temperature aqueous environments. Iron- Chromium- Nickel Alloys
Compared with ferritic carbon and low-alloy steels, relatively little information is available in the literature concerning stainless steels or nickelbase alloys. From the preceding section concerning low-alloy steels in high temperature aqueous environments, where environmental effects depend critically on water chemistry and dissolution and repassivation kinetics when protective oxide films are ruptured, it can be anticipated that this factor would be of even more importance for more highly alloyed corrosionresistant materials. One steel which has received more attention than most is Type 403 (12% Cr) stainless steel in a medium yield strength condition of 650MPa3' because of its importance for turbine blades. For this type of application, cyclic frequencies are relatively high and most of the data relate to frequencies around 30 Hz. At this frequency, distilled water up to the boiling point, steam, seawater and even sulphurous acid environments increase fatigue crack growth rates by up to a factor of five compared to air, with sulphurous acid the most aggressive. As might be anticipated, crack propagation rates observed at lower frequencies and high stress ratios lead to more severe environmental effects. Crack propagation data for distilled water and salt NaCl) at 100°C show roughly order of water solutions (0.01 and 1 . 0 ~
CORROSION FATIGUE
8: 163
magnitude increases in crack growth rates, particularly for AK values above 20 MPa f i and frequencies between 40 Hz and 0.1 Hz. Chloride concentration and pH values between 2 and 10 appear to have little influence. However, lower frequencies of lo-* and Hz can yield many orders of magnitude increase in growth rates even in distilled water. In a hardened condition, Type 403 stainless steel with yield strengths between 1200 and 1 600 MPa will also suffer from stress corrosion in distilled water at ambient temperature 34. A comparison between austenitic, austeno-ferritic and ferritic stainless steels in 3% sodium chloride has shown distinct differences in the environmental component of crack growth by up to an order of magnitude, even in corrosion fatigue tests at 200 Hz. These environmental effects were shown to be more severe at 0.5 Hz, with the ferritic stainless steel the best of the group and the austenitic stainless steel the worst. Since stainless steels usually depend on oxygen in solution to form protective, passive oxide films, the availability of oxygen down the crack becomes crucial to the interpretation of results such as these. At low cyclic frequencies, oxygen access to the crack tip is unlikely on theoretical grounds” and there is some experimental evidence to support this ~ o n t e n t i o n At ~ ~high . frequencies such as 20 or 30 Hz, the importance of pumping becomes more important. Nevertheless, the ranking of these different stainless steels on corrosion.fatigue crack growth seems to be more related to their crevice corrosion resistance rather than general corrosion resistance. It appears clear, therefore, that an improved understanding of corrosion fatigue crack growth in these alloys will come about if attention is focused on those factors which affect repassivation rates at crack tips; for example oxygen access and the electrical resistivity of crack enclave solutions and their impact on crack-tip polarisation and dissolution kinetics. High-frequency experiments do not normally allow enough time for processes more akin to stress-corrosion cracking to appear in corrosion fatigue tests, as made clear in the previous sections on carbon and low-alloy steels. Some evidence that stress corrosion can occur during corrosion fatigue crack growth in stainless steels has been observed in tests at 3 Hz on austenitic stainless steel (type 304) in various halide solutions at ambient temperature where ‘plateaux’or periods of constant crack growth rate over specific ranges of AK were observed34. The influence of pure water environments at temperatures up to 300°C is not large, however, in solution-annealed stainless steels23. One particular technological problem worthy of special mention concerns environmentally-induced intergranular cracking in type 304 sensitised stainless steels in Boiling Water Reactor environments, typically at 260 to 290°C. Sensitisation of type 304 steel causes chromium depletion at the grain boundaries in the heat-affected zones of type 304 stainless steel pipe welds. A great deal of work has been done to characterise the mechanism of environmental attack. There is little doubt that intergranular cracking in this material is due to selective dissolution of the chromium-depleted zones at the relatively high corrosion potentials achieved in normal oxygenated (200 ppb) BWR coolants36. Further, extensive slow strain-rate stresscorrosion tests have shown that the rate of cracking depends on the imposed strain rate. Similarly, in corrosion fatigue tests, intergranular cracking can also be detected provided both the frequency and the applied AK values are low enough. By contrast, no evidence of intergranular cracking is found when
8: 164
CORROSION FATIGUE
either the value of AK is too high (> 20 MPa &) or the frequency is too high (> 0.01 Hz). Relatively little work on corrosion fatigue crack growth in nickel-base alloys has been published34.Such alloys are normally selected for their inherently high resistance to corrosion, crevice corrosion and stress-corrosion cracking so that it is not surprising that aqueous environmental effects where measured have not been large. One notable application of a nickelbase alloy, Alloy 600, is for steam generator tubes in pressurised water reactors. Stress-corrosion cracking in Alloy 600 exposed to water environments between 290 and 350°C is exceedingly slow and sensitive to many metallurgical and environmental variables. It can be seen that with stress-corrosion rates typically of the order of 3 x 10-8mm/s at 325°C and an activation energy of 180 J/mole3’, exceedingly low frequency cycles would be needed to pick up an effect in normal water environments associated with the PWR. However, in concentrated caustic environments (which can accumulate by hide-out mechanisms on the boiler water side of steam generators), stresscorrosion cracking rates are more rapid. As an extreme example, the rate of stress-corrosion crack growth in Alloy 600 in molten caustic soda at 335°C is about mm/s and significant increases in corrosion fatigue crack growth rates due to this cause are apparent at frequencies of less than 1 Hz as illustrated in Fig. 8.6634.This diagram illustrates clearly how careful an investigator must be to conclude that environmental effects on fatigue crack growth are absent or minimal in a particular metal-environment combination. If stress-corrosion rates are very low, as is the case with Alloy 600 in pure water environments even at high temperature, then cyclic frequencies must also be very low to observe an environmental effect in corrosion fatigue. From Fig. 8.66, we can predict that cyclic frequencies less than Hz would be necessary to observe superposition of stress-corrosion
w 3
0
2
+
lo-9
-?a
10
-
SOLUTION A N N E A L E D SENSITIZED
\
ENVIRONMENT N o O H , 3 3 5 OC A K = L1 M N . m -% I
I
I
I
I
\ I
I
I
I
I
I
I
Fig. 8.66 Effect of frequency on the growth rate of corrosion fatigue cracks in alloy IN600 (after Speide134)
CORROSION FATIGUE
8 : 165
mm/s in water at 325°C. Thus, cracking in Alloy 600 at a rate of 3 x relatively small accelerations of the order of a factor of 2 for crack growth rates reported for sensitised or solution-annealed Alloy 600 in pure deoxygenated water at 288°C are not s ~ r p r i s i n g ~ Air-saturated ~. water with 7ppm dissolved oxygen at 288°C was found to be only slightly more aggressive. On the other hand, the high-strength, precipitation hardened version of Alloy 600, Inconel X750, under the same conditions gave very large accelerations in fatigue crack growth which were found to be highly sensitive to heat treatment. Aluminium Alloys
Three broad classes of aluminium alloys will be considered here; the heattreatable high-strength aluminium-copper 2000 series and aluminium-zincmagnesium 7000 series alloys and the non-heat-treatable lower strength aluminium-magnesium 5000 series alloys which are used extensively in marine applications. In a previous section it has already been observed that high-strength 2000 and 7000 series alloys are sensitive to the presence of water vapour in corrosion fatigue tests. Stress-corrosion susceptibilities of these alloys in low temperature aqueous solutions and the effect of composition and heat treatment have been widely in~estigated~~. It is not surprising therefore that when subjected to corrosion fatigue in similar environments, substantial environmental effects can be observed particularly at low frequencies of less than 1 Hz and AK values above KIscC31339. These environmental effects tend to be accompanied by increasing proportions of brittle striations or intergranular cracking when the stress-intensity exceeds the threshold for stress-corrosion cracking, K,,,, . Cyclic waveform at low frequencies does not appear to have a major influence on corrosion fatigue crack growth rate in these cases, probably because the predominant mode cracking is related to stress-corrosion susceptibility which itself is not in this case strongly strain-rate sensitive. Differences between 3.5% sodium chloride solution, natural seawater and simulated seawater and the effect of flow rate for a 7000 series alloy have all been observed to be small or negligible. Relatively little information on corrosion fatigue crack propagation is available for 5000 series alloys which is surprising in view of their marine applications3'. At high frequency, 30Hz, only a slight influence of a seawater environment has been found. For frequencies around 0.1 Hz, a distinct but small effect of seawater on fatigue crack growth has been measured at AKvalues greater than 10 MPa This is of a similar order to that found on low- and medium-strength structural steels. Cathodic polarisation and deoxygenation of the environment are also beneficial. The mechanism of environmental degradation by stress-corrosion cracking or corrosion fatigue has generally been attributed to hydrogen embrittlement '9*22. However, the reactivity of freshly created aluminium surfaces with any oxidising agent rapidly leads to repassivation. Since the oxide on aluminium is relatively impervious to hydrogen diffusion, and hydrogen diffusion rates are in any case very slow in aluminium, dislocation transport and pumping of the fracture process zone ahead of the crack tip is
A.
8: 166
CORROSION FATIGUE
generally invoked as the mechanistic explanation. Precipitate-matrix interfaces are particularly important sites where separation and crack formation can occur. An alternative explanation has been provided which is based on anodic dissolution of the crack tip and which leads to good quantitative predictions of the influence of aqueous environments on aluminium alloys, even the high-strength ones.36Critics point to the adverse effects of water vapour which are similar to those of aqueous environments and where electrochemical explanations are inappropriate. In addition, the adverse effects of aqueous corrosion prior to fatigue tests which can be partially reversed by heat treatment to remove hydrogen are also noted.
Titanium and Zirconium Alloys
These two groups of alloys are discussed together because of their ability to absorb hydrogen and internally precipitate hydrides. Titanium alloys are quite complex from a metallurgical viewpoint and corrosion fatigue crack growth in them is strongly dependent on m i c r o s t r u ~ t u r e ~ lMost ’ ~ ~ . work appears to have been carried out using a Ti-6A1-4V alloy in various heat treatment conditions leading to varying proportions of a (hexagonal) and 0 (cubic) phases, although many of the other available titanium alloys have also been investigated from time to time. Nearly all the work on corrosion fatigue crack growth has concentrated on the influence of salt water environments (3.5% sodium chloride or seawater or simulated saline solutions) at normal ambient temperatures. In common with many of the alloy-environment systems described so far, if the alloy is not susceptible to stress-corrosion cracking under constant stress or stress intensity, then little or no effect of environment on fatigue crack growth is observed. In these cases, frequency, R ratio and potential within the passive or cathodically protected ranges for titanium have no effect on growth rates. Many high-strength titanium alloys are susceptible to stress corrosion, however, in environments as diverse as aqueous chloride solutions, chloride contaminated methanol and molten salts. The mechanism is generally accepted to be hydrogen embrittlement with the formation of internal hydrides on slip planes, which impede slip and promote cleavage4’.When tested under corrosion fatigue conditions, those alloys which exhibit stresscorrosion cracking show large environmental effects on fatigue crack propagation when the static stress modes participate. A feature of especial interest in such titanium alloys is the manner in which the apparent threshold for the onset of high crack growth rates (at constant R ratio) varies with cyclic frequency as shown in Fig. 8.67. This behaviour should be contrasted with high-strength steels and aluminium alloys where a single frequencyinsensitive threshold parameter, K,,,, , and a constant plateau rate of stresscorrosion crack growth are sufficient to account for the observed cracking rates in corrosion fatigue when superimposed on the inert environment fatigue crack growth rate. Strain rate sensitivity of the stress-corrosion threshold and plateau rate parameters have already been highlighted in connection with lower strength stainless and non-stainless steels under passive conditions. There is evidence too of a frequency dependent threshold to the
CORROSION FATIGUE
8: 167
Fig. 8.67 Effect of frequency on corrosion fatigue crack growth behaviour of Ti-6A1-4V in aqueous 0.6 M NaCl (after PellouxZ9)
onset of high plateau corrosion fatigue crack growth rates in mediumstrength steels in high temperature aqueous environments. Thus, although the frequency sensitivity of corrosion fatigue crack growth in titanium alloys shown in Fig. 8.67 was regarded as unique when first observed, there is a growing body of evidence for similar effects in other alloy-environment systems. Zirconium alloys have been much less thoroughly studied than titanium alloys. The main application of interest has been for nuclear reactor components where good corrosion resistance combined with a low neutron capture cross-section has been required. Corrosion fatigue crack growth in these alloys in high temperature (260-290°C) aqueous environments typical of
8: 168
CORROSION FATIGUE
BWR and PWR coolants have been reviewed4'. Provided hydride precipitation and thereby stress-corrosion susceptibility is avoided, especially at the normal operating temperatures of water reactors, environmental effects on fatigue crack growth are small. A possible exception arises when zirconium alloys are subjected simultaneously to irradiation, high temperature aqueous corrosion by oxygen-containing BWR coolants and cyclic stresses. Under these circumstances, rather high environmental contributions ( x 10) to corrosion fatigue crack growth have been observed. There is clearly a need for further work in this area to sort out the relative importance of dissolved oxygen and irradiation effects both in terms of neutron damage to the material and their effects on oxidising potential. Copper Aiio ys
Remarkably little has been published on corrosion fatigue crack propagation in copper and its alloys. In general little or no influence of marine environments has been observed in crack propagation experiments on manganese and nickel-aluminium bronzes although the frequencies employed were quite high ( > 2.5 H z ) ~ ' . ~ ~ .
Corrosion Fatigue Endurance It has to be stated from the outset in this section that there is rarely a one-toone correspondence between the effects of environment observed in endurance tests on plain specimens and crack propagation tests on pre-cracked specimens (assuming the same materials, environments and fatigue test variables). In certain circumstances such as welded connections or other components with built-in pre-existing defects, such a correlation is possible. On the other hand, more than 90% of the cyclic life of smooth cylindrical specimens can be spent in propagating a stage I crack across one or two grains in inert environments and little or no relationship exists with standard crack growth test results. The lack of a general correlation shows us that the effects of corrosion on the early stages of crack nucleation and growth are usually different to those observed on macroscopic crack growth. This is despite a general recognition that most of the fatigue life of any artefact, including plain specimens, is taken up in developing a crack, however small, nucleated early in life. Thus in many circumstances, it must be the case that the effects of corrosion on stage I crack nucleation and growth are quite different to those on stage I1 growth. In addition, it has been noted already that environments themselves can be modified by chemical and diffusion processes set up in long cracks. In view of the above, it is therefore necessary to summarise separately the contents of several detailed reviews of the observations of corrosion fatigue endurance properties of many metalalloy-environment combinations2.3.13,IS. 19,22.25,31,39,44-46
8 : 169
CORROSION FATIGUE
Gaseous Environments
Early work on the fatigue strength of various metallic alloys including steels, aluminium alloys, copper alloys and nickel-base superalloys in vacuum and in air clearly demonstrated that fatigue performance improved in vacuum2. 3. 15.44.45 . At room temperature, these effects are not generally large but in high-strength steels, aluminium, titanium and magnesium alloys, significantly improved fatigue strength or cyclic lives have been observed in dry air compared with moist air15'31*44. Other environments such as inert gases and liquid sodium with low partial pressures of oxygen also enhance the fatigue lives of steels compared with air environments as do those which increase bulk material strength such as carburising liquid sodium or neutron irradiation damage Is. At elevated temperatures, the adverse influence of air oxidation on stainless steels and nickel-base superalloys increases 15,44. An example is shown in Fig. 8.68 for a nickel-base super alloy where a marked temperature effect on fatigue life was observed in air but which disappeared in vacuum6. Such obviously large effects of air oxidation on fatigue life at high temperatures has led to some difficulties in determining the relative importance of oxidation and creep damage in environment-creep-fatigue interactions where the environmental contribution has not been separately investigated. Detailed studies of the frequency dependence of corrosion fatigue lives of superalloys and stainless steels in air at high temperatures has revealed the existence of critical frequencies, typically about 1.O Hz, above which no effect of air oxidation is found. At lower frequencies, the Coffin-Manson equation can be modified by a frequency-dependent term which successfully correlates all the corrosion fatigue data (Fig. 8.68). This equation in turn can be simply derived by integrating a crack growth power law expressed
a w
A286-
v-10
593'C
a
=
I"
I"
10'
CONSTANT
10
106
CYCLES TO FAILURE, N f
Fig. 8.68 Plastic strain versus fatigue life for A286 in air and vacuum at 593°C. Numbers adjacent to test points indicate frequency, Y , in c.p.m. K and p are material and environment constants (after Coffin')
8 : 170
CORROSION FATIGUE
as a function of the plastic strain range with the same frequency-dependent term6. Duquette has discussed various hypotheses and supporting observations of mechanisms by which adsorption or oxidation by oxygen and water vapour can influence early slip behaviour, slip band formation and the growth of stage I cracks". At elevated temperature, there is good evidence that intergranular oxidation during preheating of stainless steels and superalloys to the test temperature creates an effective notch for premature crack initiation. At normal ambient temperatures there is much more controversy about how adsorbed species or oxides promote or inhibit slip or rewelding during the compressive part of the cycle and how environments can alter the tensile properties of oxide films. Nevertheless, there is little doubt that water vapour can be a potent cause of hydrogen-embrittlement effects at least in high-strength ferrous, aluminium, magnesium and titanium alloys. Aqueous Environments
Prior to the modern day preoccupation with the application of fracture mechanics to fatigue and corrosion fatigue crack growth, a very large technical literature of S-N corrosion fatigue results on metal alloys in aqueous environments was published. Gilbert summarised a great number of S-N test results on various alloys in environments such as distilled water, tapwater and seawater. The main effect of corrosion was to decrease by very considerable margins the effective fatigue strengths at any given cyclic life. Sometimes, however, strength was improved at short cyclic lives by very aggressive environments which presumably blunted out incipient fatigue cracks. Examples of typical corrosion fatigue S-N results for carbon steels
Fig. 8.69 Effect of air and aerated or deaerated distilled water and 3% NaCl solution on fatigue behaviour of steel at 25°C (after Duquette and Uhlig)
8 : 171
CORROSION FATIGUE
are given in Fig. 8.69. It is noteworthy that these large effects of corrosion were observed despite the frequent use of high test frequencies (> 10 Hz) when environmental effects on stage I1 crack growth rates would be negligible. In some cases no fatigue limit was observed at long cyclic lives. There must be some doubt about the often inferred wide applicability of this observation, however, since specimens were often quite small and general or localised corrosion could reduce the cross-sectional area very significantly in many cases. Endurance limits or fatigue strengths at specific cyclic lives were found to be insensitive to metallurgical condition showing no correlation with tensile strength (in contrast to that observed in air). Corrosion resistance, often specifically pitting resistance, was much more important in determining the endurance limit. Various compilations of fatigue endurance limits as a function of alloy strength have been published but the most recent and most comprehensive due to SpeidelI3 are reproduced here in Fig. 8.70 and 8.71. From these figures, it can be concluded that the increased strength of an alloy can only be exploited in corrosion fatigue if first it is resistant to corrosion by the environment. However, it must not be of such a high strength as to be susceptible to hydrogen embrittlement. Speidel has shown how this philosophy has been used to practical advantage in steam turbine blade specifications3'. Ferritic 12% Cr steels are widely and effectively used in good quality steam but where aggressive condensate is encountered a Ti-6A1-4V alloy is necessary since 12% Cr steels suffer severe losses of fatigue strength under such conditions. The importance of the prevailing corrosion conditions in determining corrosion fatigue strength is further emphasised by the response of the S-N curve to electrochemical potential and in some instances corrosion ULTIMATE TENSILE STRENGTH, UTS, [ksg
40 600
80
60
120
100
!
160
1 LO
I
$a1
FATIGUE AND CORROSION FATIGUE STRENGTH,
0.
y
50
;so
0
CARBON STEEL
- 70
O D
A
NORMALIZED
0.
A
QUENCHED AND TEMPERED
60
1
I
0 c
50 AMBIENT TEMP.
0
FATIGUE STRENGTH-0 5xUT
z
E 40 $
9 30
d
20 10
200
1
I
I
I
I
I
I
I
I
300
LOO
500
600
700
800
900
1000
1100
1200
ULTIMATE TENSILE STRENGTH, U T 5 [MN/mZ]
Fig. 8.70 The fatigue strength of carbon steels of varying tensile strengths in air aerated water and seawater (after Speidel 1 3 )
8: 172
CORROSION FATIGUE ~~
I
TYPICAL CORROSION FATIQUE STRENGTH, N=10,0 R=-1, IN AERATED SALT SOLUTIONS AND SEA WATER.
5c
COBALT ALLOYS TITANIUM ALLOYS
0
N
E 40
\
NICKEL- BASE
Z
I
SUPER ALLOYS
Y
I I(3
Z W (L
30 LLI
i
DUPLEX STAINLESS STEEL!
FERRITIC STAINLESS STEEL:
3
a -
WITH > 25"/0 CHROMIUM
I-
\A
Sentinel h o b
Design corrosion allowanca
Fig. 9.1 I
Sentinel hole method of monitoring corrosion of a pipe wail
Hydrogen probes are mainly used in refineries to detect the onset of conditions when H,S cracking of carbon-steel equipment could become a real risk. As a qualitative monitoring technique, it has a long and proven service of worth. Weight-loss coupons are the most used and most abused of corrosionmonitoring methods. The technique is abused by the often repeated mistake of coupons being placed in such a position that the fluid flow around them is totally unrepresentative of that experienced by the equipment they are intended to simulate. The flow around a specimen projecting into a flowing piped stream may result in totally different corrosion conditions from that experienced by the pipe walls. A less precise result from a spool piece inserted into a pipeline may be far more typical of true corrosion rates in the pipe than a highly precise result from a corrosion coupon. Sometimes, however, it is possible to get close to actual flow conditions. Thus, in agitated vessels, specimens bolted to the outer edge of the agitator blade, in the same orientation as the blade, will give very useful information on agitator corrosion rates. Corrosion coupons are probably most usefully used to rank materials of construction and to detect the permanent onset of a significant change in
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
9:31
corrosivity. Coupons integrate corrosion damage over a period and are of only marginal use in a situation where rapid and large increases in corrosion rate can occur.
Electrical Resistance Monitors
Electrical resistance monitors use the fact that the resistance of a conductor varies inversely as its cross-sectional area. In principle, then, a wire or strip of the metal of interest is exposed to the corrodent and its resistance is measured at regular intervals. In practice, since the resistance also varies with temperature, the resistance of the exposed element is compared in a Wheatstone bridge circuit to that of a similar element which is protected from the corrodent but which experiences the same temperature. In process streams where there are large changes in process temperature over a short time, the fact that the temperature of the protected element will lag behind that of the exposed element can give rise to considerable errors. The most recent development is the use of test and reference elements that are both exposed to the corrodent. The comparator element has a much larger area than the measuring element so that its resistance varies much less than that of the measuring element during their corrosion. Several drawbacks to this type of monitor, deduced from service experience, may be quoted: 1. If corrosion occurs with the formation of a conducting scale, e.g. FeS or Fe30,, then a value of the measured element resistance may be obtained which bears little relation to the loss in metal thickness. 2. Pitting or local thinning of the measured element effectively puts a high resistance in series with the rest of the element and thus gives a highly inflated corrosion rate. 3. Wire form measured elements tends to suffer corrosion fatigue close to the points where it enters the support. This is particularly true in turbulent-flow conditions, and strip-type elements are preferred in such cases. 4. Where a solid corrosion product is formed, meaningful results are only obtained after a ‘conditioning’ period for a new measured element. Even so, the conditions under which the scale is laid down may not be the same as that for the original equipment. This objection applies equally to coupons or spools, and points to one of the basic objections of using anything other than the plant itself to monitor corrosion rates. 5 . The more massive the measured element, the longer its useful life, but the less sensitive the monitor is to small changes in cross-sectional area. Thus, a compromise between long life and sensitivity has to be decided upon, depending on the application. The advantages of this type of monitor are that it can be automated to produce print-outs of corrosion rate at regular intervals and that it can be used to monitor corrosion in any type of corrodent, e.g. gaseous, non-ionic liquid or ionic electrolyte. Such monitors are in wide use, especially in refinery applications.
9:32
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
Linear Po/en’ation Measurement
Linear polarisation measurement is based on the Stern-Geary equation:
[z]
b, b,
E,,,,,
= -2. 3iC0,,( b ,
icon.
=K
+ b,)
(2)
There has been considerable talk recently in the literature about errors in this equation, but the modifications to it proposed are minor compared with the practical errors introduced by its use (see also Section 19.1):
of b, and b,, i.e. The Tafel constants of the anodic and cathodic polarisation curves, first have to be measured directly in the laboratory or deduced by correlating values of AE/Ai measured on the plant with .,i values deduced from corrosion coupons. The criticism is that the K value is likely to be inaccurate and/or to change markedly as conditions in the process stream change, Le. the introduction of an impurity into a process stream could not only alter i, but also the K factor which is used to calculate it. 2. The equation assumes that for a given A E (usually 10 mV) shift, the corresponding change Ai is solely attributable to an increase in metal dissolution current. However, in solutions containing high redox systems, this may be very far from the case. 1. The values
Practical experience with the technique has been that in some simple electrolyte solutions, ‘reasonably good’ correlation is achieved between corrosion rates deduced by linear polarisation and from corrosion coupons. ‘Reasonably good’ here seems to be considered anything better than a factor of two or three. However, the a.c. linear-polarisation technique has been used with considerable success to control inhibitor additions to overcome corrosion in ships’ condensers while operating in estuarine waters l8 and the d.c. technique has been used in controlling the corrosivity of cooling waters. Although it can only be used in ionic electrolyte solutions, results have indicated that the necessary conductivity is not as high as was once thought to be the case. To summarise: the technique is very much in its infancy as a monitoring method and must be used with caution until proven in specific applications.
Corrosion Potential Measurement
The application of this method of corrosion monitoring demands some knowledge of the electrochemistry of the material of construction in the corrodent. Further, it is only applicable in electrolyte solutions. The nature of the reference electrode used depends largely on the accuracy required of the potential measurement. In the case of breakdown of passivity of stainless steels the absolute value of potential is of little interest. The requirement is to detect a change of at least 200 mV as the steel changes from
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
9:33
the passive to the active state. In this case a wire reference electrode, e.g. silver if there are chloride ions in solution to give a crude reversible silver/ silver chloride electrode, may well be sufficient. Alternatively, the redox potential of the solution may be steady enough to be used as a reference potential by inserting a platinum wire in the solution as the wire electrode. However, in the case of stress-corrosion cracking of mild steel in some solutions, the potential band within which cracking occurs can be very narrow and an accurately known reference potential is required. A reference half cell of the calomel or mercury/mercurous sulphate type is therefore used with a 1iquidAiquid junction to separate the half-cell support electrolyte from the process fluid. The connections from the plant equipment and reference electrode are made to an impedance converter which ensures that only tiny currents flow in the circuit, thus causing the minimum polarisation of the reference electrode. The signal is then amplified and displayed on a digital voltmeter or recorder. Corrosion potential measurement is increasing as a plant monitoring device. It has the very big advantage that the plant itself is monitored rather than any introduced material. Some examples of its uses are: 1. To protect stainless-steel equipment from chloride stress-corrosion cracking by triggering an anodic protection system when the measured potential falls to a value close to that known to correspond to stresscorroding conditions. 2. To trigger off an anodic protection system for stainless-steel coolers cooling hot concentrated sulphuric acid when the potential moves towards that of active corrosion. 3. To prompt inhibitor addition to a gas scrubbing system solution prone to cause stress-corrosion cracking of carbon steel when the potential moves towards a value at which stress-corrosion cracking is known to occur. 4. To prompt remedial action when stainless-steel agitators in a phosphoric-acid-plant reactor show a potential shift towards a value associated with active corrosion due to an increase in corrosive impurities in the phosphate rock. It can be seen that in each case considerable knowledge is required before the potential values associated with the equipment can be interpreted.
Monitor Retractability
Corrosion coupons require periodic weighing, resistance-probe elements require renewing and reference electrodes develop faults. Since the emphasis is on monitoring plants which remain on-line for long periods, careful consideration has to be given to how the monitor is going to be serviced. Systems are now marketed which enable such servicing to be carried out with the plant on-line and these do not rely on the monitoring being installed in a by-pass line or in line with a duplicated piece of equipment such as a pump, which may not always be in use. Figure 9.12 shows a system based on a tool used for under-pressure break-in to operating plant.
9:34
CORROSION IN CHEMICAL A N D PETROCHEMICAL PLANT
Spring-loaded plug for fitting and removing of coupon holder
,
Internally threaded
__ Coupon or probe holder
c
Fig. 9.12 System based on a tool used for under-pressure break-in to opera1ting plant
Phase Five
- Remedial Measures
Figure 9.13 summarises the tools which are at the corrosion engineer’s disposal in the solution of a corrosion problem once it has appeared. The solution adopted will frequently be a combination o f these, and economics and convenience will determine the course adopted if there is an option.
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
lnstitute
Install cathodic-
material
MEASURES
Alter process variable (s)
Institute planned maintenance
9:35
Install anodic-
equipment design
Improve feedstock purity
Fig. 9.13 Options for remedying corrosion problems in process plant
Summery
Corrosion control in chemical plant is a continuous effort from the inception of the design to the closure of the plant. Economics dictate the risks which are taken at the design stage with respect to corrosion and the extent of the precautions taken to prevent it. Errors in design and changes in operation will occur which increase the risk of corrosion. Corrosion-monitoring systems give advance warning and enable remedial measures to be worked out and adopted.
Recent Developments Introduction
The format of the original section has been adopted in this update. Because of the ‘timeless’ nature of the original material relating to phases 2 and 3, updating has been confined to phases 1, 4 and 5. Phase One
- Plant and Process Design
Information/Knowledge Systems Computers have revolutionised the basic process/mechanical design processes, and are beginning to impact significantly on the corrosion engineer’s role of predicting material performance’’. Apart from the increasing availability of computerised databases, significant effort is being expended on the development of computer-aided management and expert systems. There has been much debate around desirable and practicable objectives for such systems2’, but most are directed at one or more of the basic elements of education, failure diagnosis and materials selection. The development of such systems is expensive and time consuming, and the major industrial initiatives have been undertaken on a collaborative
9: 36
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
basis. Of particular note are the UK ACHILLES”, the European PRIMEz2 and the North American Materials Technology Institute (MTI), National Association of Corrosion Engineers (NACE) and National Institute for Standards and Technology (NIST) projects which when complete will provide expert systems relating to a wide range of process industry environments and corrosion control technologies.
Major Industry Corrosion Problems A number of specific problems have achieved ‘industry’ status over the past ten years, owing to their cost and/or threat to plant integrity. All have significant design implications. 1. ‘External’ corrosion of surfaces beneath thermal insulation and fire-
proofing systems, resulting in general corrosion of carbon and lowalloy steel, and chloride-induced stress-corrosion cracking of austenitic stainless steel. Key preventative measures are to keep water out of such systems, to allow it to be removed should it get in, the specification of appropriate insulating/fireproofing materials, and the use of protective coatings23. 2. The effects of hydrogen on carbon and low-alloy steel equipment: (a) It has become recognised that 0.5 C-Mo grades of steel can suffer more hydrogen ‘damage’ at elevated temperatures than indicated by the API ‘Nelson’ curves, the most recent edition of which draws attention to the problemx. (b) The various forms of ‘wet HIS’ cracking and blistering, familiar in the oil and gas production industry, have been experienced in storage and pressure vessels in the refining industry2’, and have contributed to at least one major failure26. Preventive measures similar to those utilised in oil and gas production, including hardness control and stress relief”, are necessary to avoid cracking. 3. Environmentally-induced cracking has emerged as a significant problem in the following fluids: (a) Anhydrous ammonia. This potential problem is now widely recognised in ammonia storage/processing equipment28. Oxygen and water promote and inhibit cracking, respectively. The problem has been heavily researched on an ‘industry’ basis in EuropeB, and key factors for controlling the problem recognised. (b) Amine-based acid gas removal systems. Cracking can occur in both C 0 2 and HIS removal units utilising MEA, DEA, MDEA and DIPA, and has been reported in all types of equipment, including absorbers/contactors, exchangers and piping. An industry survey has been undertaken 30. (c) ‘Deaerated‘ water. Following some problems in the pulp/paper industry, it is now clear that process and utility industry deaerated water storage vessels, and possibly other steam/water circuit equipment, can suffer environmentally induced cracking31. The origins of the problem remain rather obscure, but there are probable parallels to the well understood nuclear pressure vessel cracking problems, where critical levels of oxygen promote ~racking’~. The evidence to date suggests that thermal stress relief prevents cracking in all three environments.
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
Ph8se Four
9:37
- Corrosion Monfiotfng
A number of techniques have been developed since the original material was written”. Some instrumentation/transducer developments also merit comment, Electrochemical Techniques Although the linear polarisation resistance technique has moved beyond the ‘infancy’ status attributed to it in the original material, its inherent limitations remain, i.e. it is a perturbation technique, sensitive to environmental conductivity and insensitive to localised corrosion. Two developments have occurred: 1. A.c. impedance. Measurements of the frequency variation of impe-
d a n ~ e ’ allow ~ ~ ’ ~separation of the ‘change transfer resistance’ from the contributions to the total impedance of the environment resistance, surface films, adsorbed layers, etc. Robust instruments utilising a twofrequency technique have been 2. Electrochemical noise. Fluctuations in potential or current from baseline values during electrochemical measurements are particularly prominent during active/passive transitions. This so-called ‘electrochemical’ noise is of particular value in monitoring localised corrosion, i.e. pitting, crevice and deposit corrosion and stress-corrosion 39. cra~king’~, Instruments providing simultaneous measurement of a number of parameters on multi-element probes have been developed, including potential ‘noise’, galvanic coupling, potential monitoring, and a.c. impedance3’. Reported plant applications of a.c. impedance and electrochemical noise are rare, but include stainless steels in terephthalic acid (TA) plant oxidation and fluegas desulphurisation (FGD) liquors 35, nuclear fuel reproce~sing~~, scrubber systems3’. Radioactivation Techniques Neutron and thin layer (TLA) activation are non-intrusive techniques offering the prospect of continuous, direct component monitoring, in addition to coupon or probe, monitoring. In principle, localised corrosion can be monitored using a double-layer technique. Process plant applications of the technique have been limited to dateao. AcousticEmission (AE) Conventional, periodic internal inspection of process equipment is highly expensive, particularly where an in-service deterioration mechanism, e-g. stress-corrosion cracking or corrosion fatigue is suspected. The potential for AE as a basis for plant integrity monitoring has been recognised over the past 10 years4’. Monsanto have been particularly active in extending technology developed initially for fibre reinforced plastic (FRP) equipment to the assessment of metallic equipment42. The technique utilises arrays of transducers attached to the external surfaces of the equipment, which detect small-amplitude elastic stress waves emitted when defects ‘propagate’. Using sophisticated computational techniques, ‘events’ can be characterised in terms of their severity and location. Conventionally, the technique has been used off-line, to provide information on the structural integrity of equipment, typically during a pressure test. However, the technique can be used on line by periodically raising the pressure some 5-10070 above the maximum operating pressure and one system
9:38
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
for the continuous monitoring of stress corrosion cracking in blast furnace plants has been described43. Complementary technologies such as conventional non-destructive examination (NDE) and fracture mechanics are needed to size and determine the significance of defects revealed by AE. Probe/Instrumentation Developments The principles of good practice in the design, construction and location of corrosion probes have been reviewed”. Specific probe designs which acknowledge hydrodynamic influences45and the combined effects of mass and heat transfer3’ have been developed. Computers have impacted significantly on corrosion monitoring instrumentation and data management&. Cableless corrosion monitoring utilising radio techniques has recently become available4’. Phase Five
- Remedial Measures
Significant developments have occurred in many of the basic corrosion prevention technologies over the past 10 years. Metallic Materials Stainless steel technology has been revolutionised by the combined effects of argon oxygen decarburisation (AOD) and nitrogen alloying (0.1-0- 25%) producing a range of alloys with improved localised corrosion (including chloride stress corrosion) resistance, and in specific cases oxidising or reducing acid resistance, compared with the basic 18Cr-8Ni grades48. The principal groups are: 1. Ferritic Fe-Cr-Mo compositions with 18-30% Cr, 1 4 % Mo and in some cases up to 4% Ni.
2. Duplex ferritic-austenitic alloys with 18-26% Cr, 5-7VoNi and up to 4% Mo. 3. High nickel austenitics, with 25-35% Ni, 20-22070 Cr and up to 6% Mo, with good resistance to reducing acids. 4. High chromium austenitics with 24-25% Cr, 20-22% Ni and up to 2% Mo, with good nitric acid resistance. 5. High silicon austenitics, containing 4-6% Si, with good resistance to highly oxidising nitric and sulphuric acids.
Nickel alloy technology has also been influenced by AOD melt processing, allowing the production of more weldable variants of the basic ‘B’, ‘C’ and ‘G‘ families of alloys. Additional improvements have come from alloying around the basic Ni-Mo and Ni-Cr-Mo compositions49. Non-Metallic Materials Numerous engineering thermoplastics have been commercialised50 including materials such as polyetherether ketone (PEEK) and polyether sulphate (PES) with much improved thermal/chemical resistance. The usage of FRP equipment has increased, and fluoropolymer lining technology/applications have come of age. Of particular interest is the development of stoved, fluoropolymer coating systems for process industry equipment.
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
9 : 39
Electrochemical Protection Potential control technology has developed considerably in recent years beyond the more traditional applications in sulphuric acid storage, cooling etc. Numerous applications have been identified in the pulp and paper industry, including the control of stresscorrosion cracking, pitting and crevice corrosion5'. Systems have also been developed for plate heat exchangers, FGD scrubbers, and phosphoric acid storage vessels. J. A. RICHARDSON D. FYFE REFERENCES American Petroleum Institute Publication No. 941, 1st edn, July (1970) Collins, J. A. and Monack, M. L., Mats. and Perf., 12, 11, June (1973) Bates, J. F., Ind. and Eng. Chem., 55, 18, Feb. (1966) Loginow, A. W. and Phelps, E. H., Corrosion, 18, 299 (1962) 5. Atkins, K., Fyfe. D. and Rankin, J. D., Safety in Air and Ammonia, Conf. Proc., Chem. Eng. Prog. Pub., Vancouver (1973) 6. Loginow, A. W., Bates, J. F. and Mathay, W. L., Mats. Perf., 11, 35, May (1972) 7. Corrosion Resistance of Hastelloy Alloys. Stellite Div., Cabot Corp. 8. Design of Chemical Plant in WigginNickelAlloys, and Wiggin CorrosionResistingAlloys, Henry Wiggin & Co. Ltd. 9. Corrosion Resistance of Titanium, New Metals Div., Imperial Metal Industries 10. Uranus Stainless Steels for Severe Corrosion Conditions, CAFL, France 11. Rabald, E., Corrosion Guide, Elsevier, Amsterdam, 2nd edn (1968) 12. Polar, J. P., A Guide to Corrosion Resistance. Climax Molybdenum Co. 13. Corrosion Data Survey, NACE Pub., Houston, USA, 2nd edn (1971) 14. Ross, T.K., British Corrosion J., 2, 13 1 (1967) 15. Hix, H. B., Mats. Prof. and Perf., 11, 28, Dec. (1972) 16. Clark, W. D. and Sutton, L. J., Weld. and Met. Fab., 21, Jan. (1974) 17. Charlton. J. S., Heslop. J. A. and Johnson, P., Phys in Tech., to be published 18. Rowlands, J. C. and Bentley, M. N., British Corrosion J., 7 No. 1,42 (1972) 19. Strutt, J. E. and Nicholls, J. R.. (Eds.), Plant Corrosion Prediction of Materials Performance, Ellis Horwood Limited, Chichester (1987) 20. Hines, J. G., Corrosion information and computers. Br. Corr. J . , 21, 81-86 (1986) 21. Westcott, C., Williams, D. E., Croall. I. F., Patel, S. and Bernie, J. A. The development and application of integrated expert systems and databases for corrosion consultancy. In Plant Corrosion Prediction of Materials Performance, Ibid. 22. Bogaexts. W. F.. Ryckaert, M. R. and Yancoille, M. J. S., PRIME-the European ESPRIT project on expert systems for materials selection. In Proceedings of Corrosion 88, St Louis, 1988, paper 121, NACE, Houston (1988) 23. Pollock, W.I. and Barnhatt, J. M., (Eds.), Corrosion of Metals Under Thermal Insulation. ASTM, Philadelphia (1985) 24. Steels for Hydrogen Service at Elevated Temperatures and Pressures in Petroleum Refineries and Petrochemical Plants, API Publication 941 API, Washington (1990) 25. Merrick, R. D., Refinery experiences with cracking in wet H2S environments. Materials Performance, 27, 30-36 (1988) 26. McHenry, H. I., Read, D. T. and Shives, T. R., Failure analysis of an amine-absorber pressure vessel. Materials Performance, 26. 18-24 (1987) 27. Cantwell, J. E., LPG storage vessel cracking experience. In Proceedings of Corrosion 88, St Louis, 1988, paper 157, NACE, Houston (1988) 28. Cracknell, A., Stress corrosion cracking of steel in ammonia: an update of operating experience. In Proceedings of AIChE Symposium on Safety in Ammonia Plants and Related Facilities, Los Angeles, 1982 (1982) 29. Lunde. L., and Nyborg. R., Stress corrosion cracking of different steels in liquid and vaporous ammonia. In Proceedings of Corrosion 87, San Francisco, 1987, paper 174, NACE, Houston (1987) 30. Richert, J. P., Bagdasarian, A. J. and Shargay, C. A., Stress corrosion cracking of carbon steel in amine systems. Materials Performance, 27, 9-18 (1988) 1. 2. 3. 4.
9:40
CORROSION IN CHEMICAL AND PETROCHEMICAL PLANT
31. Kelly, J. A., Cuzi, C. E. and Laronge, T. M., Deaerator cracking-industry update. In Proceedings of Corrosion 88, St Louis, 1988, paper 350 NACE, Houston (1987) 32. Jones, R. L.. Overview of international studies on corrosion fatigue of pressure vessel steels. In Proceedings of Corrosion 84, New Orleans, 1984, paper 170, NACE, Houston (1984)
33. Richardson, J. A., Innovations in techniques for corrosion monitoring. In Proceedings of the Conferenceon Advances in Materials TechnologVfor Process Industry Needs.Atlanta, 1984, NACE, Houston, pp. 200-218 (1985) 34. Hladky, K., Callow, L. M. and Dawson, J. L., Corrosion rates from impedance measurements: an introduction. Br. Cow. J., 15, 20-25 (1980) 35. Robinson, M. J. and Strutt, J. E.. (1988) The assessment and prediction of plant corrosion using an on-line monitoring system. In On-LineMonitoring of ContinuousProcess Plants, Ed. Butcher, D. W., SCI/Ellis Horwood, Chichester, pp. 79-94 (1988) 36. Hladky. K.,Lomas, J. P.,John, D. 0.. Eden, D. A. and Dawson, J. L., Corrosion monitoring using electrochemicalnoise: theory and practice. In Proceedings of the Conference
on Corrosion Monitoring and Inspection in the Oil, Petroleum and Process Industries, London, 1984 (1984) 37. Shaw, R. D., Operating experience and research as a basis for design advice. In Plant Corrosion Prediction of Materials Performance, Ed. Strutt, J. E. and Nicholls, J. R., Ellis Horwood Chichester, pp. 53-64 (1987) 38. Cox, W. M., Phull, B. S., Wrobel. B. A. and Syrett, B. C.. On-line monitoring of FCD scrubber corrosion with electrochemical techniques. Materials Performance, 25, 9-17 (1986) 39. Stewart, J.. Scott, P.M.,Williams, D. E.and Cook. N.
M.,Probabilitiesof initiation and propagation of scc in sensitised stainless steel. In Proceedings of Corrosion 88, St Louis, 1988, paper 285, NACE, Houston (1988) 40. Asher, J., Conlon, T. W. and Tolfield, B. C., Thin layer activation-a new plant corrosion monitoring technique. In On-LineMonitoring of ContinuousProcess Plants, Ed. Butcher, D. W., SCI/Ellis Horwood, Chichester pp. 95-106 (1983) 41. Guidance Notes on the Use of Acoustic Emission Testing in Process Plants, ISGHO, Institution of Chemical Engineers, Rugby (1985) 42. Fowler, T. J., AE of process equipment. In Proceedings of Third International Symposium on Loss Prevention and Safety Promotion in the Process Industries, Basle, 1980, (1980) 43. Stevens, P. 0.and Webborn. T. J. C., On-line monitoring for possible stress-corrosion cracking by acoustic emission analysis. In Proceedings of UK Corrosion '83, Birmingham, 1983, ICST, Birmingham, pp. 109-115. (1983) 44. Turner, M. E. D., Probe design and application. In Corrosion Monitoring in the Oil, Petrochemical and Processlndustries, Ed. Wanklyn, J., Oyez Scientific and Technical Services Limited (1 982) 45. Robinson, M. J., Strutt, J. E., Richardson, J. A. and Quayle, J. C., A new probe for online corrosion monitoring. Materials Performance, 23 No. 5,46-49 (1984) 46. Thompson, J. L., Automated corrosion monitoring and data management permits new philosophy in plant operation. In Proceedings of UMIST Conference on Corrosion Monitoring and Inspection in the Oil, Petroleum and Process Industries, London, 1980 (1980) 47. Webb, G., Pacslink: cableless corrosion monitoring. Br. Corr. J.. 23, 74-75 (1988) 48. Sedriks, A. J., Corrosion of Stainless SteeLs. John Wiley, New York (1979) 49. Muzyka, D. R. and Klarstrom, D. L., Innovations in high performance alloys and their applications to the process industries. In profeedings of the Conference on Advances in Materials Technology for Process Industry Needs, Atlanta, 1984, NACE, Houston, pp. 89-103 (1985) 50. Nowak, R. M.,The expanding world of engineering thermoplastics. Ibid. 51. Thompson, C. B. and Garner, A.. Electrochemicalcorrosion protection of process plant equipment. Br. Corr. J., 21. 235-238 (1986) BIBLIOGRAPHY
Henthorne, M., Corrosion and theProcessPlant, collectionof papers from Chem. Eng. (19711 1972).
9.3 Design for Prevention of Corrosion in Buildings and Structures The prevention of corrosion must begin at the design stage, and full advantage should be taken of the range of protective coatings and corrosionresistant materials available. Furthermore, at this stage particular attention should be paid to the avoidance of geometrical details that may promote or interfere with the application of protective coatings and their subsequent maintenance. Consideration should also be given to the materials to be used, the methods of protection, fabrication and assembly, and the conditions of service. The corrosion of metal components in buildings may have important and far-reaching effects, since: 1. The structural soundness of the component may be affected. 2. Where the component is wholly or partly embedded in other building
materials, the growth of corrosion products on the face of the metal may cause distortion or cracking of these materials; trouble may also arise when the metal is in contact with, although not embedded in, other building materials. 3. Failure of the component may lead to entry of water into the building. 4. Unsightly surfaces may be produced. 5 . Stresses produced in the metal during manufacture or application may lead to stress-corrosion cracking.
The Corrosive Environment The conditions to which a metal may be exposed can vary widely', but broadly the following types of exposure may arise. Exposure to external atmospheres The rate of corrosion will depend mainly on the type of metal or alloy, rainfall, temperature, degree of atmospheric pollution, and the angle and extent of exposure to the prevailing wind and rain. Exposure to internal atmospheres Internal atmospheres in buildings can vary; exposure in the occasionally hot, steamy atmosphere of a kitchen or bathroom is more severe than in other rooms. Condensation may occur in 9:41
9:42
DESIGN IN BUILDINGS AND STRUCTURES
roof spaces or cavity walls. One particularly corrosive atmosphere created within a building and having its effect on flue terminals is that of the flue gases and smoke from the combustion of various types of fuel. Embedment in, or contact with, various building materials Metal components may be embedded in various building mortars, plasters, concrete or floor compositions, or else may be in contact with these. Similarly, they may be in contact with materials such as other metals, wood, etc. Contact with water or with water containing dissolved acids, alkalis or salts Many details in building construction may permit rain water to enter and this may be retained in crevices in metal surfaces, or between a metallic and some other surface. Water may drip on to metal surfaces. These conditions, which can involve a greater risk of corrosion than exists where a metal is exposed to the normal action of the weather, are more severe when the water contains dissolved acids, alkalis or salts derived from the atmosphere or from materials with which the water comes into contact. Normal supply waters can also cause corrosion. Contact between dissimilar metals Galvanic action can occur between two different bare metals in contact if moisture is present, causing preferential corrosion of one of them (see Section 1.7). It is thus important to consider all types of exposure. If a building is to be durable and of good appearance, special attention must be paid to the design of details, especially those involving metals, and precautions must be taken against corrosion, since failure which is not due to general exposure to the external atmosphere often occurs in components within or structurally part of the building.
Ferrous Metals Faulty geometrical design is a major factor in the corrosion of ferrous metals. A design may be sound from the structural and aesthetic points of view, but if it incorporates features that tend to promote corrosion, then unnecessary maintenance costs will have to be met throughout the life of the article, or early failure may occur. Some of the more important points that should be observed are noted below2. Where these cannot be implemented, extra protection should be provided. Air 1. Features should be arranged so that moisture and dirt are not trapped. Where this is not practicable consideration should be given to the provision of drainage holes of sufficient diameter, located so that all moisture is drained away (Fig. 9.14). 2. Crevices should be avoided. They allow moisture and dirt to collect with a resultant increase in corrosion. If crevices either cannot be avoided, or are present on an existing structure, they can often be filled by welding or by using a filler or mastic.
DESIGN IN BUILDINGS AND STRUCTURES
,Water
9:43
and dirt
/
Poor
Better Fig. 9.14 Channels and angles
3. Joints and fastenings should be arranged to give clean uninterrupted lines. Welds are generally preferable to bolted joints, and butt welds to lap welds. If lap joints have to be used, then appropriate welding or filling may be necessary to avoid the entrapment of moisture and dirt (Fig. 9.15). 4. Condensation should be reduced by allowing free circulation of air, or by air-conditioning. Storage tanks should be raised from the ground to allow air circulation and access for maintenance and provision should be made for complete drainage (Fig. 9.16). 5. All members should either be placed so that access is provided for maintenance, or so thoroughly protected that no maintenance will be required for the life of the equipment or structure.
Crevices colleci liquids a n d dust
Solder, weld or f i l l with caulking compound
Dl Fig. 9.15
Welded and riveted joints
9:44
DESIGN IN BUILDINGS AND STRUCTURES
Incomplete drainage
Access for pa in t ing
Circulation of air
Fig. 9.16 Storage tanks
6. Where practicable, rounded contours and corners are preferable to angles, which are subject to mechanical damage at edges and are difficult to coat evenly3. Tubular or rolled hollow sections could often advantageously replace ‘I’ or ‘H-sections (Fig. 9.17). 7. Corrosion is often particularly pronounced on sheltered surfaces where the evaporation of moisture is retarded. Design features of this kind should either be avoided or additional protection provided. 8. Steel should not be exposed to contact with water-absorbent materials and care must be exercised when using steel in contact with wood. Not only is wood absorbent, but the vapours from it may be corrosive in enclosed spaces. 9. Large box-section girders can be enclosed by welding-in bulkheads near the ends; the welds must not have gaps or condensation may occur within the box section. 10. Features that allow moisture to drip on to other parts of a structure should be avoided, and in this connection particular attention should be paid to the siting of drainage holes. 11. Where steel members protrude from concrete, or in similar situations, attention should be given to the position of abutment. This should be arranged so that water drains away from the steel. 12. When surfaces are being bolted, the holes should coincide and bolts
Poor
Better
Fig. 9.17 Contour used in construction
DESIGN IN BUILDINGS AND STRUCTURES
9:45
must not be forced into undersize holes, since the resulting stresses may result in stress-corrosion failure. 13. The corrosion of painted mild-steel window frames is often troublesome, especially on horizontal members where moisture tends to collect. This effect can be reduced by bevelling the edges with putty before painting, to help drainage. It is preferable, however, to use more resistant materials such as galvanised steel, stainless steel or aluminium for window frames.
Materials of Construction There are three broad categories of steel4: 1. Mild steels to which only small amounts of alloying elements are deliberately added, e.g. manganese (Section 3.1). 2. Low-alloy steels to which 1-2% of alloying elements are added (Section 3.2). 3. Highly alloyed steels, such as stainless steels, which contain 12-20'70 Cr and sometimes up to 10% Ni and 3% Mo (Section 3.3).
Mild Steels
Most steels fall into this category, ranging from large structural sections to thin sheet, and minor variations in composition do not markedly affect their corrosion resistance. This is not generally important since such steels are usually protected by some form of coating which is specific for the condition of service. Account should be taken of this fact when planning to use this type of steel, so that the coating can be applied at the stage at which the maximum benefits will result. At the same time, thought should also be given to the type of coating and its method of application, since one geometrical design may be more suitable for one particular type of coating application than another. The more automatic the method of coating application, the more economical and efficient it is, since automation lends itself more readily to more even coatings than do manual methods, e.g. large surface areas lend themselves more readily to spraying techniques, whereas open work structures are more suitable for dipping methods. The coating should also be applied to a specified minimum thickness which is adequate for the serviceconditions and life envisaged. Surface preparation is of prime importance, and optimum performance of modern protection coatings can be achieved only if the surface of the steel has been adequately treated. The method of surface preparation depends on the shape and size of the structure or component. Thus it is preferable to blast-clean an openwork steel structure by manual methods, since with this type of structure automatic blast cleaning would lead to excessive impingement of the abrasive on the machine itself. Steel, whether in structural form or as a sheet, can be protected by many different coating systems, such as paint, plastic materials, concrete and other metals, either singly or in combination (such as a metal coating followed by a paint system, or a plastic coating). Examples of this compound type of
9:46
DESIGN IN BUILDINGS AND STRUCTURES
protection are the new Forth road bridge and the Severn bridge, both of which are protected by sprayed metal plus a paint system. (See also Section 12.4.) Lo w-all0y Steels
Low-alloy steels usually contain small percentages of alloying elements, such as copper, chromium and nickel, up to a total of 1-2%. Under favourable conditions they tend to corrode less rapidly than the ordinary carbon steels when exposed freely in air. Under sheltered conditions, or in crevices, they may well corrode at the same rate as mild steels’. When such steels are exposed bare, the initial appearance of the rust is similar to that on mild steel, although in time it tends to become darker, more compact, and of a more even texture than ordinary rusts. When low-alloy steels are considered for use in the bare condition, the remarks made earlier with regard to design must be given even greater attention, particularly in relation to crevices and sheltered areas. Also, the appearance and performance must be acceptable for the specific application, and care should be taken to ensure that adjacent concrete and stonework is not stained brown in the early stages by moisture dripping from the rusted steel. This can be accomplished by the following: attention to design, the careful siting of the rainwater drainage system, the use of loose gravel that can be raked over, painting the concrete or in several other ways. Low-alloy steels can be obtained as structural sections or in sheet form, and must be blast-cleaned to remove the millscale before exposure. Such material has been widely used in North America for highway bridges and for architectural purposes, and also to some extent in the UK and Europe. After suitable surface preparation, e.g. blast cleaning, low-alloy steels can be coated by paints, sprayed metal coatings, etc. and there is some evidence that such coatings last longer than on mild steel under similar conditions of exposure6. Stainless Steels
There are many grades of stainless steel, and some are virtually noncorrodible under ordinary atmospheric conditions. Their resistance results from the protective and normally self-repairing oxide film formed on the surface. However, under reducing conditions, or under conditions that prevent the access of oxygen, this film is not repaired, with consequent corrosion. Since stainless steels are generally unprotected, the design points discussed earlier are particularly applicable to them, and features such as crevices should be avoided. It is recommended that advice be sought when choosing the type of stainless steel to be used. Under severe conditions it may be necessary to use an Fe-18Cr-lONi-3Mo type, but under milder conditions a much lower grade such as an Fe-13Cr steel may be satisfactory. Frequently the deciding factor will be cost, since in general, the greater the content of alloying metals, particularly Ni and Mo, the higher the cost. The following points should be considered:
DESIGN IN BUlLDlNGS AND STRUCTURES
9:47
1. The environment to which the steel will be exposed. 2. Types and concentration of solutions that may be in contact with the steel. This is particularly important where the failure may be due to local concentrations of dilute solutions. For example, the small chloride content of tap waters is unlikely to cause any trouble, but if it concentrates at the water level due to heating and evaporation of the water, then attack may occur. 3. Operating temperatures and pressures. 4. Mechanical properties required. 5. Work to be performed on the steel. 6. Fabrication and welding techniques to be used. In connection with welding it should be emphasised that the correct grade of steel and electrode or filler rod must be used.
Stainless steel is not generally made in the large sizes offered in the cheaper steels, but a range of sections, tubes, flats, rods and sheet is obtainable. Some savings in thickness and weight are possible, however, because of its superior corrosion resistance. If the strength requirements go beyond the point where the use of stainless steel becomes economic, it is possible to use clad material. Stainless steel is often used as cladding and for window frames, doors, etc. for prestige buildings.
Coated Steel Sheet
Probably the most familiar coated steel sheets are the ubiquitous galvanised corrugated roofing and cladding sheets which have been used for many years, particularly for farm buildings, either painted or unpainted. In addition to zinc other metallic coatings are available, e.g. hot dip aluminium and hot dip aluminium-zinc alloys. Nowadays, however, zinc-coated steel sheets, either continuously galvanised or electroplated, are often used as a basis material for overcoating with plastic materials or paints. The coatings are usually applied continuously and have a range of uses both externally and internally. Many surface finishes are obtainable, e.g. plain or embossed, and in an extensive range of colours, to suit almost any requirement '. Some of the uses of such precoated materials are roofing, cladding, decking, partitions, domestic and industrial appliances, and furniture. The formability of these materials is excellent and joining presents no problems. The thickness of the coating varies according to the material used and the service conditions which the end product has to withstand. Vast amounts of continuously galvanised steel sheets are produced, and unless they are painted or otherwise coated, their life depends on the thickness of the galvanising and the service environment in which they are used. Similarly in the case of steel sheets coated with aluminium or aluminiumzinc alloys, their performance is dictated by their coating thickness (see Section 13.4). A problem often associated with such material is corrosion at the cut edges. From work carried out by BISRA and others' it has been shown that providing the bare steel edge is less than 3mm in width, the amount of corrosion is minimal and the life of the sheet is not adversely
9:48
DESIGN IN BUILDINGS AND STRUCTURES
affected, although rust staining will occur. Aluminium and aluminium-zinc alloy coatings are not as effective as zinc coatings in sacrificially protecting cut edges. If staining is important because of the appeareance, the cut edge should be orientated so that the stain does not run over the sheet. The edge could, alternatively, be beaded over or painted with a suitable painting scheme. If appearance is important it may be advantageous to paint overall.
Protective Coatings Zinc, aluminium, aluminium-zinc alloys and other materials such as paints and plastic coatings are often used as protective coatings for steel. These metals act not only as a barrier, but where breaks occur in the coating, corrode preferentially under most conditions and thus sacrificially protect the underlying steel. Aluminium is normally less negative than zinc, but provides adequate sacrificial protection in industrial and marine environments. The corrosion protection afforded by aluminium-zinc alloy coatings lies between that of aluminium and zinc. Two alloys are currently used: 5% aluminium whose properties are more akin to zinc and 55% aluminium which is closer to aluminium. With metal coatings the life expectancy depends on the coating weight, which is generally synonymous with thickness. The thickest coatings are produced by dipping or by spraying, thinner coatings by diffusion and in the case of zinc by electrodeposition (see Chapter 12). The metal spraying operation using zinc or aluminium as a protective coating is usually followed by a painting scheme. The choice of sprayed metal and paint scheme depends on the service conditions’, but normally this type of system is used on prestige buildings or structures, where longevity is of prime importance and maintenance requirements need to be kept to a minimum. Paint is the most widely used protective coating for steelwork and normally acts as a barrier between the metal and environment. The choice of type of paint and the final thickness required depends on the conditions of service, and the more severe the conditions the thicker and more resistant the paint film needs to be. Also the more sophisticated the paint system the more demanding is the surface preparation required. Often steelwork will initially be painted before final fabrication, and problems that may arise when maintenance painting becomes necessary may not
Inadequate accass for maintenan ce pa inti ng Fig. 9.18
Access for maintenance
DESIGN I N BUILDINGS AND STRUCTURES
9:49
be fully appreciated" (see Fig. 9.18). The '1'-beam can be painted in the shop, but access for maintenance may be inadequate and either the distance t should be increased or the gap closed so that maintenance is not required. An actual case of failure occurred where the rolled steel joists carrying the floor of a refrigeration chamber were placed so close together that they could not be reached for painting''. Heavy condensation led to dangerous rusting on the inner surfaces of the joints and in consequence the steelwork had to be replaced prematurely (Fig. 9.19). 203mmx127mm FILLER BEAMS
TRANSVERSE SECTION
LONGITUDINAL SECTION
Fig. 9.19 Design of reinforced concrete floor. For the old joists A was 7 in (178 mrn) leaving a I in (25 mm) gap between the toes; for the new joists A was increased to 18 in (457 mm)
Designers should always bear in mind the necessity to inspect and maintain all parts of a structure that may be corroded and should provide adequate access for these purposes. The choice of protective system will be determined by many factors such as the importance of the structure, the environment and its proposed life. Having chosen a suitable system or systemsI2, it is essential that requirements including adequate inspection are specified exactly, and that there is the fullest possible collaboration between the paint suppliers, the contractors, the architects and all other parties concerned. It is not always appreciated that the life of a coating depends not only on the material but also on other factors such as surface preparation and the application of coating to give the required dry film thickness. A duplex coating of a cathodic coating and an organic coating may well have a life greater than the sum of the expected life of both coatings.
Bimetallic Corrosion When other metals are used in conjunction with steel, careful consideration must be given to the possibilities of galvanic attack (Section 1.7). The rate of corrosion and damage caused to the more negative metal will depend upon the relative sizes of the anodic (corroding metal) and cathodic areas. A small anode and a large cathode will result in intensive corrosion of the anodic area. On the other hand, if the anode is large compared with the cathode, the corrosion of the anodic area will be more general and less likely to result in rapid failure. For example, a steel rivet in a copper plate will be rapidly attacked in sea-water, whereas a copper rivet in a steel plate may lead only to slightly accelerated corrosion of the steel in the area adjacent to the rivet. Prediction of the rate of corrosion of the less noble metal
9:50
DESIGN IN BUILDINGS AND STRUCTURES
in a galvanic cell is difficult, but there is always the possibility of serious trouble if two dissimilar metals are in contact, particularly under immersed conditions. The safest way of avoiding this is to ensure that dissimilar metals are not in contact. If this is impracticable, the following will help to reduce or stop attack on steel: 1. Use more noble metals for fastenings. 2. Insulate the metals from each other by suitable gaskets, washers, etc. 3. Paint the surfaces of both metals. Avoid painting only the less noble metal because if the coating is damaged severe attack may result at the damaged area. 4. Prevent moisture dripping from the more noble metal on to the less noble metal.
From the reversible potential of zinc, accelerated corrosion would be expected t o occur when zinc is coupled with many other metals commonly used in buildings. Aluminium, contrary to its reversible potential, is generally found to be slightly cathodic to zinc and is protected when the two metals are coupled together, as when aluminium sheet is fixed with galvanised nails. In practice, although some small acceleration in the corrosion rate of zinc will be expected in the immediate area of contact with another metal, the effect is usually severe only when it is in contact with copper. For example, where zinc and aluminium gutters or zinc and cast-iron gutters are fitted together, very little accelerated corrosion of the zinc is normally found. BrassI3, with its 30-40% zinc, is very much less active than copper, and brass screws and washers can be used for fixing zinc with little or no accelerated corrosion troubles; but with copper, rapid failure occurs. Drainage water from copper affects zinc in a similar way. Zinc sheets must never be fixed with copper nails, nor should copper roofs drain into zinc or galvanised gutters. Copper lightning arrestors provide further potential hazards to zinc work; when a copper lightning strip has to pass over or near a zinc roof, it should be either well insulated or heavily tinned.
Non-ferrous Metals and Plastics For some purposes where the strength and ductility of steel are not prerequisites, other metals or materials may be used to advantage, particularly when the component or article is not a load-bearing one. Some of the nonferrous metals and plastics materials are extremely useful in this respect, especially the latter with their excellent corrosion-resistant properties and ease of formability. Non-ferrous metals in sheet form are often used as roof covering. In such situations they could well become subject to condensation. Condensation could be the result of thermal pumping or internal conditions. Under conditions in which condensation can occur, copper is not normally attacked, but lead, zinc and aluminium may be attacked and corrode from the inside of the building outwards. Copper l4 Although copper is weather-resistant under normal conditions of exposure, certain precautions are necessary to avoid the risk of premature failure. For instance, copper that is exposed to high concentrations of flue
DESIGN IN BUILDINGS AND STRUCTURES
9:51
gases, as may happen within a metre or so of chimney exits, may become corroded within a relatively short time. To avoid this the chimney should be built to a reasonable height above the roof. For similar reasons ventilators should not be made of copper where highly sulphurous fumes may be encountered. The use of potentially corrosive materials as underlays for copper roofing may also result in failure. Bare copper exposed indoors will slowly tarnish. Transparent lacquers may be used, however, to retain a bright surface without the need for frequent cleaning. Neither copper nor any copper alloy will remain bright and polished without maintenance or coating. Lead ' Corrosion of lead gutters and weatherings is usually associated with slate roofs on which vegetable growth such as algae, moss or lichen is present. These produce organic acids and carbon dioxide which significantly increase the acidity of rain water running over the roof. New cedar-wood shingles also contain acids which are slowly washed out by rain, thus intensifying the attack that would in any case slowly occur owing to vegetable growth on the roof. Probably the simplest way of avoiding this type of failure is to protect the lead with a thick coating of bituminous preparation extending well underneath the edge of the roof. Lead is relatively easily corroded where acetic acid fumes are present and under such conditions it either should not be used or should be efficiently protected. Generally, any contact between lead and organic material containing or developing acids will cause corrosion; for instance, unseasoned wood may be detrimental. Trouble from this cause may be prevented by using well-seasoned timber, by maintaining dry conditions, or by separating the lead from the timber by bitumen felt or paint. Lead is also subject to attack by lime and particularly by Portland cement, mortar and concrete, but can be protected by a heavy coat of bitumen. A lead damp-proof course laid without protection in the mortar joint of a brick wall may become severely corroded, especially where the brickwork is in an exposed condition and is excessively damp. Aluminium The resistance of aluminium and certain of its alloys to atmospheric corrosion is fairly high. Nevertheless, corrosion does occur, especially on under-surfaces, e.g. of bus shelters. Normally, in simple exposure the corrosion reaction stifles itself and the rate falls to a low value. With a few alloys, however, atmospheric corrosion may lead to severe attack, and layer corrosion may occur. It is important therefore to pay attention to materials, design and protection. Intermetallic contacts, crevice conditions, horizontal surfaces, etc. should be avoided. Materials for sections, plate and sheet, where strength is important, should be restricted to primary alloys. All heat-treated alloys should be painted, using first a chromate priming paint containing not less than 20% zinc chromate pigment, or an equivalent chromate paint. Crevices should be packed with a suitable composition such as chromate jointing compound or impregnated tape. By paying attention to these points, aluminium should behave satisfactorily.
Zinc Zinc surfaces corrode more slowly in the country than in either marine atmospheres or in industrial areas where sulphur pollution constitutes the main danger both to them and to many other building materials.
9:52
DESIGN IN BUILDINGS AND STRUCTURES
Sulphur and its compounds in the air can become oxidised to sulphuric acid; this forms soluble zinc sulphate, which is washed away by rain”. The purity of the zinc is unimportant, within wide limits, in determining its life, which is roughly proportional to thickness under any given set of exposure conditions. In the more heavily polluted industrial areas the best results are obtained if zinc is protected by painting, and nowadays there are many suitable primers and painting schemes which can be used to give an extremely useful and long service life under atmospheric corrosion conditions. Primers in common use are calcium plumbate, metallic lead, zinc phosphate and etch primers based on polyvinyl butyral. The latter have proved particularly useful in marine environments, especially under zinc chromate primers Is. Zinc has been used extensively as a roofing material, but its life, especially in industrial areas, is somewhat dependent on the pitch or slope of the roof; those of steep pitch drain and dry more rapidly and therefore last longer. Irrespective of the locality, exterior zinc-work may fail prematurely if the design is unsuitable or the installation faulty. Many failures arise from a combination of purely mechanical reasons and secondary corrosion effects, white-rusting being the most important of these. This tendency can, however, be reduced by chromating. When used inside factoriesI6zinc coatings have been found to be satisfactory in withstanding attack by many industrial gases and fumes. The protection of fabricated structural steelwork by hotdip galvanising as is used in current constructions, has allowed lightweight concrete-clad steel sections to be used with complete safety. Zinc in contact with wood Zinc is not generally affected by contact with seasoned wood, but oak and, more particularly, western red cedar can prove corrosive, and waters from these timbers should not drain onto zinc surfaces. Exudations from knots in unseasoned soft woods can also affect zinc while the timber is drying out. Care should be exercised when using zinc or galvanised steel in contact with preservative or fire-retardant-treated timber la- Solvent-based preservatives are normally not corrosive to zinc but water-based preservatives, such as salt formulated copper-chrome-arsenic (CCA), can accelerate the rate of corrosion of zinc under moist conditions. Such preservatives are formulated from copper sulphate and sodium dichromate and when the copper chromium and arsenic are absorbed into the timber sodium sulphate remains free and under moist conditions provides an electrolyte for corrosion of the zinc. Flame retardants are frequently based on halogens which are hygroscopic and can be aggressive to zinc (see also Section 18.10).
Zinc-alloy diecastings used indoors Zinc-alloy diecast fittings have good corrosion resistance. Generally, such castings may be used in buildings without further protection by painting, but it is of advantage, especially where conditions of permanent dampness may occur, that they should be chromate-treated or phosphated, and then enamelled, or coated with an etch primer and painted after installation. Where a chromium-plated finish is used, it is important that an adequate basis of electroplated copper and nickel plating is provided. Soluble sulphates and chlorides in brickwork, plaster and other walling materials provide a more serious source of corrosion under damp condi-
DESIGN IN BUILDINGS AND STRUCTURES
9:53
tions. Under such circumstances, or where zinc or zinc-alloy fittings are to be placed in contact with breeze, concrete or black ash mortar (made from ground ashes) the metal may be protected with two coats of hard-drying bitumen paint.
Zinc as a protective coating to building components Perhaps the most important use of zinc in building is as a protective coating to steel. In spite of the initial cost, a substantial coating of zinc (or of aluminium) is of great value, and often saves the cost of remedying troubles caused by corrosion. For general purposes it can be accepted that the effectiveness of the coating depends on the weight of zinc coat applied and not on the method of application.
Metals in Contact with Concrete Little information is available about the corrosion of metals in concrete, although it seems likely that all Portland cements, slag cement and highalumina cement behave similarly”. Concrete provides an alkaline environment and, under damp conditions, the metals behave generally as would be expected; e.g. zinc, aluminium and lead will react, copper is unaffected, while iron is passivated by concrete. Aluminium reacts vigorously with a wet, freshly prepared concrete mix and the reaction, in which hydrogen is evolved, has been used for preparing lightweight cellular concrete. When the concrete has set, however, its reactivity is reduced. The degree of corrosion experienced by aluminium depends upon its alloy type”. Whilst the extent of corrosion may not reduce the structural strength of the aluminium, the more voluminous corrosion product formed can lead to cracking and spalling of the concrete. Zinc will initially react with cement-based materials with the evolution of hydrogen. This reaction can be controlled by the presence of soluble chromate either in the cement (over 70 ppm) or as a chromate passivation treatment to the zinc surface. Zinc can therefore be used to provide additional protection to steel in concrete. It is more effective in carbonated concrete than in chloride-contaminated concrete. The reaction of lead with concrete differs from that of aluminium and of zinc in that it is not normally rapid during the early wet stage. It is, however, progressive in damp conditions, and this is said to be due to the fact that the concrete prevents the formation of a protective basic lead carbonate film on the surface of the lead. The packing of lead cables in plaster of Paris is reported to be of doubtful value in preventing corrosion from surrounding concrete. Little information is available on the performance of copper and of copper alloys in contact with concrete, but concrete sometimes contains ammonia, even traces of which will induce stress-corrosion cracking of copper pipe. The ammonia may be derived from nitrogenous foaming agents used for producing lightweight insulating concrete. The corrosion behaviour of iron and steel in contact with concrete is of great importance, not only because of the amount of metal involved, but also because the metal is frequently load-bearing, and the stability and
9:54
DESIGN IN BUILDINGS AND STRUCTURES
durability of a structure may depend upon the control of corrosion. The alkaline reaction of the adjacent concrete may, however, damage sensitive paints and protective finishes. The corrosion of steel reinforcements in concrete is discussed below. Effects of Composition of Concrete
Concrete" made with ordinary Portland cement is an alkaline material having a pH in the range 12.6-13.5. Steel embedded in such a material will be passive. However, like most alkaline materials concrete will react with the acid gases in the atmosphere, e.g. sulphur dioxide, carbon dioxide with a reduction in alkalinity. Carbon dioxide is the reactant which effects a chemical change in the concrete reducing the pH to a level at which steel is no longer passive. This process is known as carbonation. Carbonation spreads in from the surface of the concrete and when the carbonation front reaches the steel the steel is at risk from corrosion. The rate at which carbonation occurs depends upon the porosity, permeability, cement content, water/cement ratio and other factors but the depth is normally proportional to 4.The depth of concrete cover to the steel reinforcement therefore has a significant bearing on the corrosion protection provided by the concrete to the embedded steel. In general terms the thicker the cover the longer the concrete provides protection to the steel. Unfortunately, the protection provided by concrete can be overcome by contamination of the concrete by chloride. Chloride, when entering the concrete as a contaminant of the mix constituents, is to a large extent (about 90'70)complexed within the cement matrix and only a small percentage is free in the pore solutions. The present codes of practice2' ban the use of chloride-bearing additives and restrict the amount of chloride present in concrete. For normally reinforced concrete made with ordinary Portland cement it should be not more than 0.4% chloride ion with respect to the cement content weight/weight. When mature concrete is contaminated by chloride, e.g. by contact with deicing salts, the cement chemistry is more complex, and less chloride is taken up by the cement hydrate minerals and a larger proportion is free in the pore solutions and can therefore pose a greater hazard. When embedded steel corrodes, the production of a more voluminous corrosion product pushes the concrete from the steel with resultant cracking and spalling of the concrete. There is no objection to the use of slag aggregates for reinforced concrete provided the slag meets the various sulphur-content specifications, and similar considerations apply to lightweight aggregates, although it has been claimedI9 that the sulphur content of blast-furnace slag is not dangerous. Clinker aggregates, on the other hand, are not permitted in the UK because they cause corrosion of reinforcement. The corrosiveness of clinker and boiler slag is due probably to the high sulphur content.
DESIGN IN BUILDINGS A N D STRUCTURES
9:s
The Corrosion of Steel Reinforcements in Concrete Normal Reinforcement
In the middle of the last century, the tensile properties of concrete were improved by the introduction of steel to reinforce the concrete. This practice has developed since then to such an extent that reinforced concrete is now one of the major structural materials used in construction. In general it has proved to be a good durable material with some of the structures erected at the turn of the century still providing satisfactory service in the late 1970s. Normally concrete is reinforced with plain carbon steel, but under conditions where rapid carbonation can occur or there is a risk of chloride contamination, corrosion-protected or more corrosion-resistant reinforcing steels may be necessary. Currently there are three reinforcing bars which have enhanced corrosion resistance: 1. Galvanised steelz2provides increased corrosion resistance in carbonated concrete. In concrete with more than 0.4'70 chloride ion with respect to the cement content, there is an increased risk of corrosion and at high chloride contents the rate of corrosion approaches that of plain carbon steel. In test conditions the rate of corrosion is greater in the presence of sodium chloride than calcium chloride. 2. Fusion-bonded epoxy-coated steelU performs well in chloride-contaminated concrete up to about 3.9% chloride ion in content. 3. Austenitic stainless steels" resist corrosion at levels of chloride contamination greater than that which can be resisted by epoxy-coated bar.
Prestmssed Reinforcement For prestressed concrete, either high-tensile steel wires or occasionally bars of steel alloy containing manganese and silicon, can be used. Galvanised wires may also be used for prestressed concrete, but it is recommended that they be chromated before use. In a normal reinforced concrete structure, the tensile stress in the steel is comparatively low, but in prestressed concrete the steel is held permanently in tension with a stress equivalent to about 65% of its breaking load. It is necessary, therefore, in prestressed concrete, to consider the possibility of the occurrence of stress corrosion. Surface rusting or corrosion of prestressed wires can affect the working cross-sectional area of the reinforcement, and pitting, which might be unimportant on a 12 mm bar, could cause failure on a 2.5 mm diameter stressed wire. The number of reported failures of prestressed concrete due to fracture of reinforcement is very low and in general the behaviour of steel in prestressed concrete is no different from that of steel in ordinary reinforced concrete. Prestressed concrete is made from materials of slender section using higher working stresses than are customary for ordinary reinforced concrete. The concrete is also of a higher quality.
9 : 56
DESIGN IN BUILDINGS AND STRUCTURES
Work carried out at the Building Research Station*’ suggests that the most significant influences on the corrosion of prestressed steel wire in concrete are: (1) the presence of chloride, (2) the composition of the concrete, (3) the degree of carbonation of the concrete, (4) the compaction of the concrete around the ’wire ensuring that voids are absent, and ( 5 ) chloride promoting pitting attack, leading to plastic fracture and not stress-corrosion cracking of prestressing steel. Gilchrist25(a) considers that prestressing steels may fail by either hydrogen cracking or active path corrosion, depending on conditions; most service failures have been due to hydrogen cracking. Prestressed steel in concrete should thus be durable if a dense, impervious and uniform concrete free of chloride surrounds the steel and adequate depth of concrete is given to the steel.
Materials in Water-supply Systems The most important non-ferrous metals for handling water are lead, copper and zinc; the last, however, is used chiefly as a protective coating on steel or alloyed with copper to form brass. The choice of materials for most applications in domestic water supply is governed by consideration of mechanical properties and resistance to corrosion, but the cost, appearance and ease of installation should also be considered when the final choice has to be made between otherwise equally suitable materials 26. Many plastic materials are also now being used in domestic water systems, in the form of pipes and fitments. Features that should be avoided for all materials (particularly ferrous metals) in liquid environments and points that should be followed are2’: 1. Crevices, because they collect deposits and may promote corrosion by
causing oxygen depletion in the crevice, thereby setting up a corrosion cell in which the areas receiving less oxygen corrode at a higher rate. 2. Sharp changes in direction, especially where liquids are moving at high velocities, and re-entrant angles, dead spaces and other details where stagnant conditions may result should be avoided. This is particularly important if inhibitors are to be used. 3. Baffles and stiffeners inside tanks should be arranged to allow free drainage to the bottom of the vessel. The bottom should slope downwards and have rounded corners. Any drain valves or plugs should fit flush with the bottom (Fig. 9.16). 4. Wherever possible different metals should not be connected in the same system. If they have to be used they should be insulated from each other, and the cathodic metals placed downstream of anodic ones. Galvanised steel pipes Threaded mild-steel tube is the cheapest material for water pipes, but it is not normally used owing to the amount of rust introduced into the water as a result of corrosion. Galvanised mild-steel tube overcomes this problem and may be used for nearly all hard waters, but it is not satisfactory for soft waters or those having a high free-carbon-dioxide content. The ability of a water to form a ‘scale’is, therefore, of prime importance when considering the suitability of galvanised steel for an installation.
DESIGN IN BUILDINGS AND STRUCTURES
9: 57
The Langelier index (Section 2.3) gives useful guidance to this, but it is only an approximation. The scale can be deposited either as nodules covering a relatively small area of metal, or as a thin scale covering a large area. Provided the deposit is not porous, the latter has the greater effect in reducing corrosion, and it has been found that waters derived from rivers tend to form more useful scales than those from wells. Galvanised steel tubing is cheaper than lead or copper tubing, but is, however, more costly to install because it cannot be bent without damage to the galvanising. A full range of preformed bends, tees, etc. is available, but cutting the tubes to length, threading the ends and screwing up the joints are slow processes. Consequently, a galvanised steel installation is cheap for long straight runs of pipes, but for complicated systems it is liable to be more expensive than copper because of the high cost of installation.
Lead pipes The corrosion resistance of lead is generally excellent, but it is attacked by certain waters. This is usually of little significance so far as deterioration of the pipe is concerned, but is important because of danger to health, since lead is a cumulative poison; even very small doses taken over long periods can produce lead poisoning'*. It is for this reason that its use for carrying potable water has been discontinued. Copper pipes For plumbing above ground, copper is supplied in both halfhard and hard conditions. It has sufficient strength to require only few supports, and can be bent cold, in the small sizes, either by hand or with a portable bending machine. Copper is also supplied in the fully soft condition in coils for laying underground, for heating-panels, etc. Light-gauge copper tube may be joined by autogenous welding or by bronze welding. These processes, which produce neat strong joints, are usually applied to the larger sizes of tube. For the tubes used for domestic water supply, capillary-soldered fittings, or compression fittings are normally employed. Two types of corrosion may be experienced. The first is analogous to plumbo-solvency, with the copper being dissolved evenly from the surface of the tube. With some watersz9 it is potentially dangerous to use galvanised hot-water tanks and copper pipes. In domestic systems, premature failure of galvanised hot-water tanks connected to copper circulating pipes, due to pitting corrosion of galvanised steel, is encouraged by more than about 0.1 p.p.m. of copper in the water. Failures of galvanised cold tanks due to copper in the water are often the result of back circulation of hot, copper-bearing water in badly designed systems where the cold-water tank is installed too close to the hot-water cylinder. Hot water carried into the cold tank via the expansion pipes when the water is allowed to boil may also sometimes be responsible. Secondly, under certain conditions copper may suffer intense localised pitting corrosion, leading sometimes to perforation of the tube, in quite a short time. This form of attack is not common and depends on a combination of unusual circumstances, one of which is the possession by the tube of a fairly, but not entirely, continuous film or scale that is cathodic to the copper pipe in the supply water; this can set up corrosion at the small anodes of bare copper exposed at faults or cracks in the film. Carbon films give rise to such corrosion, but since 1950, when the importance of carbon films was
9 :58
DESIGN IN BUILDINGS AND STRUCTURES
first discovered, manufacturers have taken precautions to avoid as far as possible producing tubes containing them. (See also Sections 1.6 and 4.2.)
Aluminium pipes Aluminium might become an important material for carrying water if its liability to pitting corrosion could be overcome. Very soft waters are difficult to accommodate when normal pipe materials are used, and it is for these that aluminium offers most promiseM. The possibility of using it for domestic water pipes, however, appears at present to depend upon finding a cheap and effective inhibitor that could be added to the water, or upon the use of internally clad tube, e.g. AI-1 -25 Mn alloy clad with a more anodic alloy, such as Al-1Zn. Such pipes are at present mainly used for irrigation purposes3’. Stainless steels Thin-walled stainless steel (Fe-18Cr-8Ni) tubes are now frequently used for domestic installations in place of copper pipe”. Care is required, however, in the design of stainless steel equipment for use in waters with a high chloride content, or where the concentration can increase, since pitting attack may occur. It may also be susceptible to failure by stresscorrosion cracking under certain conditions. Plastic pipes Pipes made from plastic materials such as unplasticised P.v.c., Polythene, ABS and GRP are now widely used for carrying domestic cold water, wastes and rain water. Joining varies according to pipe diameter and service condition, but is generally relatively simple (see Section 18.6). Buried pipes Pipes to be laid underground must resist corrosion not only internally but also externally. Light sandy soils, alluvium, or chalk are generally without appreciable action but made-up ground containing a high proportion of cinders is liable to be exceptionally corrosive as also is heavy clay containing sulphates. The latter provides an environment favourable to the growth of sulphate-reducing bacteria, which operate under anaerobic conditions, reducing sulphates in the soil to hydrogen sulphide, and causing severe corrosion especially of steel. Aluminium is believed not to be susceptible to this form of attack, but, like copper or galvanised steel, it is severely attacked by cinders. Wet salt marsh, although it has little effect on copper and only slightly more on lead, causes severe corrosion of galvanised steel or aluminium. These materials are also severely corroded in London clay, in which copper could probably be used unprotected. When ferrous metal service pipes or piles, etc. are buried in the ground it is advantageous in almost all cases to coat it in some way even if the coating is just a simple dip into a bituminous solution. If, however, the soil is aggressive, or the component is vital or irreplaceable, a more resistant coating should be used, and consideration should be given to the application of cathodic protection (Section 10.4). The coating used can be an epoxide type or something similar, or a plastic coating which is wrapped or extruded onto the pipe wall. Galvanised steelwork buried in the soil in the form of service pipes or structural steelwork withstands attack better than bare steel, except when the soil is more alkaline than pH 9.4 or more acid than pH 2-6. Poorly aerated soils are corrosive to zinc, although they d o not necessarily cause pitting. However, soils with fair to good aeration containing high concentrations of chlorides and sulphates may do so. Bare iron may be attacked five
DESIGN IN BUILDINGS AND STRUCTURES
9 : 59
times more rapidly than zinc in well-aerated soils low in soluble salts, or in poorly aerated soils; and if the soil is alkaline and contains a high proportion of soluble salts the rate may be even higher. Only in soils high in sulphide content does iron corrode less rapidly than zinc. Plaster and concrete Domestic water pipes are often used in contact with plaster, concrete or flooring materials3’. Copper is unaffected by cement, mortars and concrete, which are alkaline in reaction, but it should be protected against contact with magnesium oxychloride flooring or quick-setting materials such as Keenes cement, which are acid in character. Materials containing ammonia may cause cracking at bends or other stressed parts of brass or copper tubes-some latex cements used for fixing rubber flooring come in this category and contact with these should be avoided. Lead is not affected by lime mortar but must be protected from fresh cement mortar and concrete, either by wrapping or by packing round with old mortar or other inert materials. Galvanised coatings are not usually attacked by lime or cement mortars once they have set, but aluminium is liable to be attacked by damp concrete or plaster, even after setting.
Materials for Tanks Copper hot-water tanks These are usually made Cylindrical with domed tops and bottoms, because this form of construction produces a strong tank from light-gauge sheet. They are normally trouble-free except for occasional cases of leakage at the seams, which are usually welted or overlapped and then brazed. Brazing brasses, containing 40-50% zinc, often give good service, but are susceptible to dezincification in some waters. Dezincification, which is most likely to occur in acid waters or waters of high chloride content, can be avoided if cylinders are brazed with an alloy such as Cu-14Ag-SP. It is more expensive but is fairly ductile, and if used in conjunction with capillary-gap seams, makes an economical as well as a sound job.
In water where copper tanks might be subject to pitting corrosion it is good practice to fit an aluminium rod” inside the tank. This corrodes sacrificially within the first few months of service, and during this period a protective film is built up on the copper surface. Galvanised steel hot-water tanks These may be of cylindrical or rectangular form, the latter being popular where space is limited. In hard or moderately hard waters, galvanised steel hot-water tanks, with galvanised circulating pipes and cast-iron boilers, usually give trouble-free service, but failure by pitting occurs occasionally. Sometimes this is due to extraneous causes, such as rubbish left inside the tank when it is installed. Iron filings left in the bottom of the tank or deposits of inert material are liable to interfere with the formation of the protective scale by the water, and can lead to failure. Another cause of trouble is overheating, especially during the early life of the tank. Above 70°C a reversal of polarity may take place, the zinc becoming cathodic to the iron. Above this temperature protection of exposed iron is not to be expected. Persistent overheating is frequently the result of fitting
9:60
DESIGN IN BUILDINGS A N D STRUCTURES
a hot-water tank too small for the heating capacity of the boiler. A lowering of temperature by as little as 5-10°C can add years to the life of a tank. It has also been that large-capacity immersion heaters operated intermittently are more beneficial to tank life than small-capacity heaters operated continuously. Magnesium anodes 3' suspended inside a galvanised hot-water tank and in electrical connection with it afford cathodic protection to the zinc, the alloy layer and the steel, at high temperatures as well as in the cold. The magnesium is eventually consumed but it is probable that in the interim a good protective scale will have formed on the inside of the tank, so that the magnesium anode will then no longer be necessary. One of the difficulties of this method, however, is the maintenance of a sufficiently even current distribution over the inside of a tank to protect the whole surface, especially in waters of low conductivity. The method is therefore unlikely to be applicable to soft waters. Cold-water tanks Domestic cold-water tanks are usually made of galvanised steel. As with hot tanks, it is important to avoid leaving filings, etc. in the tank when it is installed and it should be covered to prevent rubbish falling in later. In most waters, galvanised cold-water tanks give good service, the zinc coating protecting the iron while a protective scale is formed. With very soft waters, however, or with waters of high free carbon dioxide content, which do not produce a scale, there may be trouble. Steel or galvanised steel tanks for use in such waters can be protected by coating with bituminous paint or, alternatively, reinforced plastics may be used. For larger cold-water storage tanks, sectional steel or cast-iron tanks protected by several coats of cold- or hot-applied bitumen or bituminous paints are often used. It is important, however, to ensure that all millscale, dirt, etc. is removed before applying the protective coating. Stainless steels can also be used for this purpose. Cold-water tanks made from Poiythene or GRP are generally available, especially in domestic sizes, and are now often used in domestic installations.
Water Fittings
Brass water fittings give no trouble except that dezincification may occur in acid waters or waters of high chloride content, especially when hot. This dezincification has three effects. Firstly, the replacement of brass by porous copper may extend right through the wall of the fitting and permit water to seep through. Secondly, the zinc which is dissolved out of the brass may form very voluminous hard corrosion products and eventually block the waterway-this is often the case in hot soft waters. Thirdly, and often the most important, the mechanical properties of the brass may deteriorate. For instance, a dezincified screwed union will break off when an attempt is made to unscrew it and a dezincified tap or ball-valve seat is readily eroded by the water. Brass water fittings are normally produced from two-phase brass by hot pressing. Unfortunately this material is vulnerable to dezincification in certain water areas. In areas where the hot pressed fittings are vulnerable,
DESIGN IN BUILDINGS AND STRUCTURES
9:61
fittings manufactured from single-phase brasses containing 0.3% arsenic or other non-dezincifiable alloys should be used. Plastic water fittings ranging from taps to lavatory cisterns are now available and are gradually replacing items previously made in metal in the domestic field, especially in situations where condensation is the cause of unsightly corrosion products. E.E. WHITE K. 0.WATKINS R.N. COX REFERENCES 1. Jones, F. E., Chem. and Ind. (Rev.),1050 (1957) 2. Reinhart, F. M., Prod. Engng., 22, 158 (1951) 3. Rudolf, H.T., Corrosion. 11, 347t (1955) 4. B.i.s.r.a./BSC Booklet No. 1 5. Chandler, K.A. and Kilcullen. M. B., Brit. Cor. J.. 5 No. 1, 24-32 (1970) 6. Robinson, W. L. and Watkins, K. O., Steel and Coal, July 13,6568 (1962) 7. Watkins, K.0.. B.I.S.F. Symposium, ‘Developments in Methods of Prevention and Control of Corrosion in Building’, Nov. 17th (1966) 8. Anderson, E.A. and Dunbar, S. R., American Zinc Institute Report N314, page 62 9. Watkins, K. 0.. Brit. Cor. J., 9 No. 4, 204 (1974) 10. Chandler, K.A. and Stanners, J. F., I.S.I. Publication No. 122 11. Hudson, J. C. and Wormwell, F., Chem. and Znd. (Rev.), 1078 (1957) 12. Protection of Iron and Steel Structuresfrom Corrosion, CP2008,British Standards Institution, London (1966) 13. Bailey, R. W. and Rudge, H. C., Chem. and Ind. (Rev.), 1222 (1957) 14. Baker, S. and Cam, E., Chem. and Ind. (Rev.), 1332 (1957) 15. Bonner, P. E. and Watkins, K. O., 8th Fatipec Congress, 385-394 (1%6) 16. Stanners, J. F., J. Appl. Chem., 10,461 (1960) 17. Halstead, P. E., Chem. and Ind. (Rev.),1132 (1957) 18. Laidlaw, R. A. and Pinion, L. C., Metal Plate Fasteners in Trussed Rafters Treated with Preservatives of Flame Retardants-Corrosion Risks, IS 11/77, Building Research Establishment (1977) 19. Jones, F. E. and Tarleton, R. D.. E#ect of Embedding Aluminium and Aluminium Alloys in Building Materials, National Building Studies Research Paper 36, London (1%3) 20. The Durability of Steel in Concrete. Part I: Mechanism of Protection and Corrosion, Digest 263, Building Research Establishment 21. BS 8110:1985.Structural use of concrete. Part 1: Code of practice for design and construction, H.M.S.O., London 22. Treadaway, K. W. J.. Brown, B. L.and Cox, R. N., Durability of GalvankedSteel in Concrete, Special Technical Publication, ASTM 23. Treadaway, K. W. J., Davies, H. and Brown B. L., Performance of fusion bonded epoxy coated steel reinforcement, Proceedings of the Institute of Structural Engineers (in press) 24. Treadaway, K. W. J., Cox, R. N. and Brown, B. L.,Durability of corrosion resisting steels for reinforced concrete, Proceedings of the Institute of Civil Engineers (in press) 25. Treadaway, K. W. J., Brit. Cor. J., 6 No. 2, 66-72,March (1971) 25(a). Gilchrist, J. D., C.I.R.I.A. Technical Note ISSN:0305-1781, May (1975) 26. Campbell, H.S . , Chem. and Ind. (Rev.), 692 (1957) 27. Chandler, K. A. and Watkins, K. 0.. Mach, Desgn. Engng., Aug. (1965) 28. Holden, W.S.,Water Treatmentundfiaminution, J. and A. Churchill, London, 55 (1970) 29. Kenworthy, L., J. Inst. Met., 69, 67 (1943) 30. Porter, F. C. and Hadden, S . E., J. Appl. Chem., 3, 385 (1853) 31. Campbell, H.S., B.N.F. Publication No. 544. Oct. (1%8) 32. Non-Ferrous Metals Post War Building Studies, No. 13, 1944, H.M.S.O., London, 1 1 (1944) 33. Sereda, P. J., Corrosion, 17, 30 (l%l)
9.4 Design in Marine and Offshore Engineering
The field of marine and offshore engineering has massively expanded in recent times, due principally to the remarkable growth of the offshore oil and gas industry. Since the world's first steel offshore oil and gas installation was commissioned in the Gulf of Mexico in 1947, the continental shelf areas of the oceans now provide approximately 25% of the world total oil and gas production. Looking ahead, there will be a continuing development of the continental shelf areas together with exploitation of significant oil and gas reserves in certain deeper ocean basin areas of the world. These factors allied to a general decline in productivity of established onshore production provinces, will result in the proportion of oil and gas produced offshore continuing to rise steadily. There are, of course, many strong engineering links between the longestablished marine engineering industries and the newer offshore engineering industries. Today, there are many hundreds of fixed offshore oil and gas production platforms, drilling rigs and other forms of support installations around the world which require extensive and often costly programmes of maintenance and repair. Carbon-manganese steels dominate marine and offshore structural and process applications largely by virtue of their excellent range of mechanical properties, good availability and cost considerations. However, they are not particularly corrosion resistant in aqueous saline media, and corrosion protection of these steels has to be provided by effective coatings (including cladding and sheathing), cathodic protection or corrosion inhibitor treatments, depending upon circumstances. However, such protection schemes may cost many millions of pounds, both in terms of primary design and installation cost, and in terms of downstream maintenance and other ongoing commitments and costs arising at some later stages of the installation life. The design priority is thus to ensure that a functional and secure design will be suitably productive and be maintained at reasonable cost for the duration of the installation's life. The vast majority of corrosion design issues faced in marine and offshore engineering involve water in one form or another. In that regard, the two principal media involved are seawater and formation waters (oilfield brines). Seawater, of course, surrounds offshore installations though may also be used as the medium in reservoir injection and other critical offshore process 9:62
DESIGN IN MARINE A N D OFFSHORE ENGINEERING
9:63
applications. Hence it is desirable that both structural and internal process corrosion issues be faced at the design stage. There are four main structural corrosion design zones in the marine environment: atmosphere, splash/ tidal, immersed and mud. Of these, the splashltidal zone is by far the most aggressive environment. For example, unprotected low-alloy steels may show mean annual corrosion losses in the atmosphere and immersed regions of around 100 pm/y, whereas in the splashltidal regions, it may be as high as 625-875 pm/y depending upon design detail, location, the presence of floating debris or ice, and temperature of the metal’s surface. Consequently, corrosion protection measures in this zone on ships, semi-submersibles, drilling rigs and fixed production platforms require to be of the very highest and durable specification.The mud zone may or may not be a zone of serious corrosion hazard, depending upon whether or not anaerobic bacterial action is taking place. Bacterial levels and activity should be checked before installing buried pipework, piling, etc. Formation water occurs naturally with virtually all oil and gas reservoirs, and is constitutionally similar to seawater in many respects. From a corrosion point of view, however, it differs notably in the following respects: 1. It is generally more saline than seawater. Most North Sea formation
waters have salinities two to three times that of seawater. 2. It is anoxic. 3. It has a very low sulphate ion concentration. Additionally, there may be COz, H,S or bacteria present, all of which substantially increase the corrosivity of formation waters. Furthermore, whilst in a ‘young’ oil and gas well the levels of produced formation waters (termed ‘watercut’) may well be very low, at later stages of maturity, the watercut may reach values in excess of 90%. Consequently, oil and gas production systems may often be subject to increasing corrosion risk with time. The principal features of seawater and formation waters affecting the marine and offshore corrosion engineering design progress are discussed in the following sections. Chloride ion concentration Chloride (and indeed bromide and iodide) ions in sea or formation waters are particularly aggressive and troublesome species. They participate in depassivation corrosion processes on alloys such as chromium and chromium-nickel steels, aluminium and titanium alloys, particularly in the absence of oxygen. In addition, many chloride corrosion products which may be formed are highly water soluble, hence little protection is afforded to the metal surface being corroded. Conjoint corrosion phenomena such as stress corrosion cracking and corrosion fatigue may also be exacerbated in the presence of chloride - particularly at elevated temperatures as in oil and gas production situations-though this, of course, depends upon the operating circumstances of the exposed material. Chlorides are often found as the salt aerosols of the atmosphere, and consequently may strongly influence the corrosion performance of structures and plant, particularly in marine or coastal situations. This influence on corrosivity reduces proportionately with distance from the seawater surface though local environmental factors such as prevailing wind direction, level
’,
9:64
DESIGN IN MARINE AND OFFSHORE ENGINEERING
of other atmospheric pollutants such as carbon, nitrogen and sulphur dioxides, patterns of precipitation and relative humidity are also influential factors which must be considered when determining the overall corrosivity of a particular location, and hence the materials and/or protection scheme(s), if any, which require to be used2. It is also remarkable how much chloride and other salt aerosol components are ingested into air-consuming systems in offshore installations and aircraft flying over or operating near seawater on a regular basis3. In particular, the massive air-consumption requirements of gas turbine engines used for pumping or power generation applications on offshore installations, and for powering helicopters and fixed-wing aircraft, renders these units highly vulnerable to corrosion problems associated with salt aerosol ingestion, such as fluxing of passive films on nickel-based turbine blades and pitting of aluminium alloy compressor blades. The pervasive nature of salt aerosol components, also appears to have played some part in the world’s worst helicopter disaster off Sumburgh in the Shetland Isles in November 1986, when a helicopter servicing offshore oil and gas platforms in the North Sea crashed killing 45 persons on board. A critical gearwheel in the forward transmission of this aircraft displayed what appeared to be fretting corrosion in part due to the primary ingress (and entrapment) of salt4 (Fig. 9.20). Oxygen concentration Oxygen concentration is important in a number of respects. When it is high, it generally ensures that the cathodic reaction for
Fig. 9.20 Spiral bevel ring gear assembly from the forward transmission gear box of Boeing Vertol 234-LR (Chinook) aircraft registration G-BWFC which crashed into the sea off the Shetland Isles in 1986. Note the peripheral and radial fractures in the gear, which appeared to be responsible for the crash. There was evidence of fretting and galvanic corrosion which may have been responsible for initiation of the fracture sequence
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:65
most situations (and certainly in seawater within the normal pH range of 7.6 to 8.2) is one of oxygen reduction viz:
+ 2H,O + 4e S 4(OH)-
O2
(9.1)
This reaction proceeds rather sluggishly under most circumstances, and the accompanying production of hydroxide ions (which may have the effect of raising the pH and partially-passivating adjacent surfaces) results in this reaction being the rate-limiting corrosion reaction in seawater. Under certain circumstances, oxygen reduction is replaced or is accompanied by the hydrogen-evolution cathodic reaction which brings much more serious consequences such as accelerated corrosion rates, hydrogen embrittlement, disbonding of coating systems and, of course, fire or explosion hazard. The hydrogen-evolution cathodic reaction is promoted in seawater media which: 1. are anaerobic or anoxic; 2. have a lower pH than normal; 3. host a metal surface held at excessively negative cathodic potential
during cathodic protection operations. However, dissolved oxygen ensures that passive films are maintained on passivating metals and alloys. Conversely, in anoxic waters where there is no alternative supply of oxygen, corrosion rates of passivating metals and alloys may rise dramatically and are often manifested by severe pitting attack I . However, in the case of non-passivating alloys such as carbon-manganese steels, corrosion rates will be reduced. It is important to stress that in anaerobic or anoxic waters, there will be a greater risk of sulphate-reducing bacteria becoming active, producing hydrogen sulphide, and increasing corrosion rates on affected surfaces'. Consequently, the use of biocides or biostats should be carefully considered in such situations. Electrical resistivity The low electrical resistivity of seawater (and even lower values for formation waters) results in two important corrosion and corrosion-protection consequences: 1. It enables greater cathodic/anodic surface area ratios to become active in corrosion processes, thereby promoting pitting mechanisms in vulnerable materials. 2. It enables cathodic protection to be applied with relative ease.
Scaling properties The feature of precipitating scales in seawaters and formation waters may bring corrosion advantages or disadvantages depending upon circumstances. For example, the scaling of tube walls in heat exchangers or process coolers may reduce heat transfer rates and thermal efficiencies in such systems. Oil- and gas-well tubing may also be subject to scaling as a consequence of injection water breakthrough in a complex reaction with formation waters from the reservoir rock itself. This scaling may be so extensive as to plug the voids in the reservoir rock, and reduce the bore of the well tubing -both of which can seriously reduce well production rates. However, where such tube scales are coherent and intact, they can provide effective corrosion protection to affected surfaces.
9:66
DESIGN IN MARlNE AND OFFSHORE ENGlNEERlNG
Cathodic protection in seawater also results in the precipitation of a calcareous scale on the metal surface (due largely to the increase in pH (see equation 9.1)) and the scale has a largely beneficial effect in three respects. 1. It helps 'spread' the cathodic current over a greater area of surface. 2. It reduces current requirements for maintenance of a particular cathodic potential, thereby reducing costs. 3. It provides some temporary and partial corrosion protection should the cathodic protection system become ineffective for any reason.
Marine fouling Marine fouling is a design issue oniy in seuwuter and is largely determined by two factors, the first of which is the composition and nature of the exposed surface. Certain alloys such as the cupronickels, have antifouling properties in normal exposure circumstances allowing slow copper dissolution, whereas steels foul rapidly and heavily when corrosion is not proceeding rapidly. In addition, if the design of the surface is such as to produce quiescent havens of low water velocity, or are 'rough' in surface finish, then marine fouling will amass quickly and heavily. The second factor determining fouling is the zone of exposure. Marine fouling only amasses in continuous and possibly very thick films in the surface layers of seawater, hence seawater intakes (or other engineering artefacts which are required to remain substantially free of fouling) should be placed at depths of around 70 m or greater, where, at worst, only discontinuous and thin foulant cover may occur. Most of the published evidence suggests that marine fouling cover particularly where it is continuous and well established-reduces corrosion rates of steels"'. Indeed, 35% seawater is by no means the most corrosive of saline environments towards steel. Brackish water, as found in estuarine or certain other coastal areas, is considerably more aggressive towards steel6, and careful design measures should be taken to ensure that effective corrosion control is achieved in such circumstances.
Design Principles The broad principles of design that should be followed in order to effectively and economically control corrosion in marine and offshore engineering should also be subject to the overriding necessity to regard designing against corrosion as an integral part of the total planning and costing procedure which should be continuously followed at all stages from the initial plans to the finished construction. Failure to do so is likely to result in breakdown of plant (with consequential losses), costly maintenance or modifications in design (if these are practicable) and a possible reduction of safety factors. An attempt to design against corrosion as an afterthought is generally unsatisfactory, costly and often impractical. Whilst careful design and informed forethought can often minimise or even prevent corrosion at little extra cost where the environmental conditions or the conditions of service are severe (as in most forms of marine and offshore engineering) reliable, secure and cost-effective corrosion control cannot be effectively achieved without considerable expense, although even in these circumstances good design can help to significantly reduce this. In
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:67
general, the extra initial outlay involved in building structures and equipment with a level of corrosion resistance, protection or control appropriate to the service conditions concerned and the length of life required, more than compensates for the downtime, trouble and cost that stem from employing cheaper materials of inadequate resistance or with inadequate protection. Many corrosion problems may also arise through poor or inadequate QA/ QC procedures. The main principles to be observed can be summarised as follows: 1 . Features that apply, entrap or retain corrosive agents such as water, water vapour and aggressive ions should be strenuously avoided. This can be done by (a) attention to the geometry of designs and methods of construction, (b) by the provision of adequate drainage, (c) by protection against contact with hygroscopic, absorbent and/or corrosive materials, and (d) by methods of preventing or reducing such entrapment such as changing operating conditions or dehumidifying (Figs. 9.21 and 9.22).
Fig. 9.21 Water trap in steel girder assembly of a bridge. Salt aerosol, bird excrement, etc., may also find their way into this stagnant water, producing an extremely corrosive fluid
2. Where appropriate, designs should facilitate the application of adequate corrosion-protection systems that can be readily maintained. This can be achieved by attention to the geometry of the initial and any retrofitted design and methods of construction, and by making provision for good inspectability and accessibility. 3. All methods of corrosion control such as careful materials selection, including coating and cladding, inhibition and cathodic protection, should be regarded as an integral part of the design process.
9:68
DESIGN IN MARINE AND OFFSHORE ENGINEERING
I
Fig. 9.22 Waterhteam vent from a chemical plant cooling system. Note that venting water and condensed steam drips onto the lower surface, producing corrosion problems that would not exist if the vent were placed in a location not likely to create such problems
4. Care should be exercised in the use of dissimilar metals in contact or
in close proximity. If dissimilar metals must be used, they should be insulated from one another so far as is practicable. Alternatively, if they cannot be insulated, the use of a ‘middle piece’ with a suitable potential may be effective’. In any event, where a galvanic couple exists, the more active metal should have the greater exposed area. 5 . Seawater systems should be designed to avoid excessive water velocities, turbulence, aeration, particulates in suspension, rapid changes in piping section and direction. Likewise, extended periods of shutdown should also be avoided since stagnation of contained seawater, will result in bacterial activity and H,S production with consequential and perhaps serious corrosion and health and safety problems. 6 . Undue static or cyclic stressing and other features which give rise to stress concentrations should be avoided as these may lead to premature failure by stress-corrosion cracking or corrosion fatigue.
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:69
7. So far as possible, components that operate in highly turbulent-flow conditions should be designed with a view to eliminating cavitation and/or impingement erosion attack. 8. Designs should have regard to the material being employed, e.g. designs and methods of fabrication or construction suitable for steel will not be directly transferable to, or appropriate for, aluminium alloys or glassreinforced plastics. 9. Welded joints should be ground flush to enable good coating adhesion and performance. In addition, weld metal should be selected such that it is cathodic to the parent material(s) being welded (Fig. 9.23). 10. Use of material with good pitting corrosion resistance is desirable in seawater and oilfield brine (formation water) media. Some examples of how these principles should be applied, are described in the following pages.
Fig. 9.23 Welded areas of an offshore platform structural tubular displaying premature corrosion due to (a) lack of surface grinding (resulting in poor paint adhesion and performance) and (b) the fact that the weld metal is anodic relative to the parent material of the tubular
9:70
DESIGN IN MARINE AND OFFSHORE ENGINEERING
Avoidance of Entrapment of Corrosive Agents Many factors influence the corrosion of metals in the atmosphere, including the natural phenomena that make up the vagaries of climate and weather. Of these, the feature of greatest importance is moisture in its various forms, since, other factors apart, the amount of corrosion that takes place is largely a question of whether and how long a period the surface of the metal is wetted (‘time of wetness’). Although the corrosivity may not be high provided the condensed moisture remains uncontaminated, this rarely happens in practice, and in marine environments sea salts are naturally present not only from direct spray but also as wind-borne particles. Moreover, many marine environments are also contaminated by industrial pollution owing to the proximity of factories, port installations, refineries, power stations and densely populated areas, and in the case of ships’ or offshore installation superstructures by the discharge from funnels, exhausts or flares. In these circumstancesany moisture will also contain S, C and N compounds. In addition, solid pollutants such as soot and dust are likely to be deposited and these can cause increased attack either directly because of their corrosive nature, or by forming a layer on the surface of the metal which can absorb and retain moisture. The hygroscopic nature of the various dissolved salts and solid pollutants can also prolong the time that the surface remains moist. Designs should therefore avoid, as far as possible, all features that allow water (whether seawater, rainwater or moisture from any source) to be applied, entrapped or retained. These conditions are not only corrosive towards bare metals; they also adversely affect the life of protective coatings both directly and by the fact that it is often difficult at areas subject to these conditions to give sound and adequate surface preparation for good paint adhesion and subsequent performance. Good designs in these respects do not differ for metal structures exposed in marine environments from those for similar structures exposed elsewhere, several examples of which are illustrated elsewhere in this text and in Reference 9, but the frequent presence of seawater or salt-contaminated water makes observance of these designs especially necessary. The same principles also apply to ships or offshore installation superstructures and internal fittings, although there are many sites where it is impossible to ensure that water cannot collect and be retained. A few of many examples are ventilation shafts or ducting that is subject to the ingress of spray or rain, areas around wash-deck valves, junctions of mushroom ventilators with decks, junctions of horizontal stiffeners with vertical plates, and behind the linings of bathrooms, wash areas and under bulkhead and deck coverings. In these areas either extra protection should be given or designs should allow, so far as is practicable, ready accessibility for frequent inspection and/or maintenance. Bilges and ballast waters are one of the most difficult areas of this nature to deal with, especially in machinery spaces, since not only are they almost impossible to keep dry or even to dry out while the offshore installation is in operation, but effective maintenance of protective coatings at all areas is in any case quite impossible except at major overhauls and refits, because of inaccessibility or very high temperatures and humidities. Good initial
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:71
protection of these areas by a well-applied coating system during building when all surfaces are fully accessible considerably reduces subsequent maintenance problems. Designs can help by arranging for bilge water to collect in sumps or separator vessels where it can be completely drained or pumped away. Alternatively, some authorities favour the use of cathodic protection with sacrificial anodes, but this is only effective if an adequate amount of water is present and if it has a low electrical resistivity. In addition the facility to inspect and retrofit anodes should be available. The method may also become ineffective by the anodes becoming coated with oil or grease, being painted inadvertently, or removed through use or mechanical damage. Another common cause of a metal surface remaining wet is contact with absorbent materials, particularly insulants such as rock-wool, and this again can cause serious attack when seawater or chloride-contaminated water becomes entrained. In addition, the absorbent material itself can be (or may become) corrosive, as in the case of wood. Examples of trouble that can occur from this cause are wooden decks or fenders laid over steel, certain aluminium alloy frames in contact with wooden hulls, and zinc or cadmiumcoated fasteners in wooden hulls. The whole subject of the corrosion of metals by wood receives detailed treatment in Section 18.10. Lagging of process pipes can create similar trouble by absorbing moisture during shut-down periods or by becoming wet through atmospheric exposure or through condensation ”. Certain lagging materials may also contain chloride ions. Calcium silicate is the preferred lagging material and moisture absorption should be prevented by the application of a waterproof coating to the insulation and/or ensuring that any trapped moisture may be subsequently ventilated. Process pipework should, of course, receive a properly applied, high quality and compatible coating system prior to application of insulation materials and thereafter be regularly inspected in service to ensure that no breakdown in the protective coating and insulation system has occurred, otherwise exceedingly high corrosion rates may result. The other main cause of metals remaining wet under atmospheric conditions is condensation. This is particularly liable to occur in mess and pipe decks, laundries, galleys, machinery or piping areas or indeed in any enclosed space where humid conditions prevail. Such conditions also encourage the growth of bacteria and fungi which are not only a potential safety hazard but also a source of deterioration of paint coatings resulting in enhanced attack of metals due to the corrosive nature of their products of metabolism. Cadmium plating which is used extensively in electronic equipment and in the coating of fasteners is particularly vulnerable to this form of attack lo. The cost-effective solution to condensation problems in enclosed spaces is often found to be air conditioning. If this is not practicable, then the use of dehumidifiers, the provision of as much ventilation as possible or vapour-phase inhibition (VPI) treatment should be considered. Care should also be taken to design or locate equipment in such a way that free circulation of preferably dry air is not impeded. Condensation, usually contaminated with chlorides, is also prone to occur on the external surfaces of marine structures in dry dock because of the humid conditions that often prevail in such locations.
9:72
DESIGN IN MARINE AND OFFSHORE ENGINEERING
Design considerations in relation to protective coating systems A wide variety of protective coating types and systems is available for corrosion control on external and internal surfaces of structural and process plant in marine and offshore engineering. These are discussed in detail elsewhere in this text, and the purpose here is to highlight the critical importance of certain design and related operational aspects which affect both the selection and performance of protective coating systems. The following design considerations should be made:
1. Coatings should be applied to surfaces under the optimal possible environmental conditions. Shop-applied systems generally perform better than the same systems in field-applied situations. Monitoring and control of humidity and other application circumstances are critical in securing good results from high performance marine and offshore coatings. 2. Coatings should only be applied to fully and correctly prepared surfaces. Surface preparation is discussed elsewhere in this text; however, the following undesirable surface design/operational features should be eliminated wherever possible: (a) excessive numbers of holes, sharp edges and rapid changes in section or surface profile; (b) undressed welds, severe pitting or surface defects; (c) risk of subsequent mechanical, electrical, chemical or welding damage to the coating; ( d ) excessivetemperatures or temperature range, frequency of temperature change (as may be found in process plant subject to thermal cycling); (e) excessive changes in pressure (as may be found in certain intermittently-used subsea equipment, internals of hyperbaric chambers etc); (f)areas of the coated surface which when in service, cannot be inspected (or maintained when necessary).
It may be that if one or other of the foregoing deficiencies exist, then other surface design changes or changes in corrosion control procedure should be considered. Even if no deficiencies exist, it is important to realise at the design stage that if a protective coating system is to be used as part or whole means of corrosion control, there must be a ‘downstream’commitment to inspection/ maintenance/renewal of the coating system, as appropriate to the particular operating life of the coating system.
Seawater Systems General Design and Layout
The conditions under which these systems operate can be extremely severe and, although the alloys at present available and in extensive use in sea-
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:73
water systems offer good resistance to many forms of attack, even the most resistant can fail under the conditions that can arise from poor design installation or operational detail. In fact, these systems provide an outstanding example of the important part that design can play in minimising corrosion and related conjoint degradation phenomena such as corrosion erosion. The ideal design is one in which all parts can be operated satisfactorily with water flowing with the least turbulence and aeration, and at a rate of flow within the limits that the materials involved can securely withstand. These limits, with regard to flow-rate limitations, vary with the material, as described in Section 1.2, but turbulence, aeration or presence of suspended particulates can lower these limits considerably, and designs that eliminate these two factors go a long way towards preventing impingement attack, which can be the major cause of failures in sea-water systems. (See also Sections 1.6 and 2.1 .) Good designs should start at the inlets which should be shaped to produce smooth streamline flow with least turbulence and minimal ingress of deleterious substances. In the case of the ships’ inlets they should be located in the hull in positions, so far as is practicable with other requirements, where turbulence is least excessive and the amount of entrained air is as low as possible. Inlets located close under the bilge keel or immediately aft of a pump discharge are in particularly bad positions. The design should not impart a rotatory motion to the water stream, since the vortex formed (in which air bubbles will tend to be drawn and cavitation problems induced) can travel along the piping until disrupted by some change in the geometry of the system, e.g. a bend or change of section in the piping or an irregularity in the bore. The energy will then be released resulting in an excessive local speed of highly aerated water and consequent rapid impingement and wallthinning attack. Some reduction in the amount of air in the water stream can be achieved by the maximum use of air-release pipes and fittings. To ensure that the water flows through the whole of the system as smoothly as possible and with the minimum of turbulence, it is vital that the layout of pipework should be planned before fabrication starts. It should not be the result of haphazard improvisation to avoid more and more obstacles as construction proceeds. Pipe runs should be minimised or run as directly as possible with every effort made to avoid features that might act as turbulence raisers. For this reason the number of flow controllers, process probes, bends, branches, valves, flanges, intrusive fittings, or mechanical deformation or damage to the pipework, should be kept to a minimum. In some systems it may be feasible to select the sizes of pipes to give the correct speed in all branches without the need for flow-regulating devices, or a bypass may be satisfactory thereby eliminating one possible source of turbulence. Where flow control is necessary, this should be effected preferably either by valves of the glandless diaphragm type or by orifice plates, in either case set to pass the designed quantity of water under the conditions of maximum supply pressure normally encountered. Screw-down valves are not advised for throttling, nor are sluice and gate valves, as these in the partly open position cause severe turbulence with increased local speed and potentially serious erosion problems on the downstream side. For auxiliary heat exchangers and sanitary services in ships fed by water from the firemain
9:74
DESIGN IN MARINE A N D OFFSHORE ENGINEERING
(which is at a high pressure), effective pressure-reducing valves must be fitted, otherwise rapid failures by impingement attack are almost inevitable. It is very important that all flow-regulating devices should be fitted only on the outlet side of equipment. With regard to the various other.features in a piping system that can set up turbulence, careful design of these can do much to reduce, if not completely eliminate, their harmful effects. Thus, pipe bends need cause little trouble if the radius of the bend is sufficiently large. A radius of four times the pipe diameter is a practice followed in some installations as well as by some authorities, with a relaxation to a radius of times three when space is restricted. Crimping of bends should be avoided by the use of a filler during bending, but care should be taken to remove all traces of filler residues before putting the pipe into service as these can initiate corrosion. Branch pipes cause the minimum disturbance if they can be taken off the main piping by swept ‘T’ pieces rather than by right-angled junctions. Where the latter are unavoidable the diameter of the branch main ‘should be as generous as possible. If connecting pieces are not used, branch pipes should be set at a shallow angle with the main piping and should not protrude into the latter. Flanged joints are a very common cause of turbulence unless correctly made and fitted. Close tolerances should be placed on the machining of flanges to match the bore of the pipe; mating flanges should be parallel and correctly aligned, and gaskets should be fitted so that they are flush with the bore and do not protrude into the pipe. Alternatively, butt welding can be used provided pipe ends are not misaligned and weld metal roots do not protrude into the bore. Screwed union fittings also give no trouble if pipes are correctly aligned. A further precaution to lessen the risk of impingement attack is to fit straight lengths of pipe down-stream of possible turbulence raisers. Piping Materials
The characteristics of the various metals commonly used for seawater systems, chiefly, nickel and titanium alloys, galvanised steel and to a lesser extent aluminium alloys and stainless steels, are fully described in their respective sections. Reference here will be confined to mentioning some of the advantages and limitations of clad and nonmetallic piping. As regards the former, a recent development has been the cladding of steel piping with a welded overlay of cupro-nickel which shows promising application for components such as ships’ inlet trunking. Non-metallic-clad steel piping, if correctly manufactured and fitted, has the advantage of being resistant to deterioration by seawater at the speeds normally encountered. However, if the coatings possess pin-holes or other discontinuities or are too brittle to withstand a reasonable amount of shock or are damaged by subsequent welding or cutting operations, water may gain access to the steel causing rapid perforation or lifting of the coating and corrosion of the pipe walls with the possibility of a blockage ensuing. Nonmetallic materials can provide cost-effective and secure solutions
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:75
to many corrosion problems in the marine and offshore industries. They can be used as corrosion-resistant lining materials or as piping materials in their own right. In particular, glass fibre reinforced plastic (GFRP) and glass fibre reinforced epoxy (GFRE) materials have been the subjects of sustained development work. GFRP has been successfully developed for both structural and process applications, such as the hulls of minesweepers and low-pressure secondary pipework applications. GFRE, on the other hand, is largely a product of the space age, being used originally for rocketmotor cases and even today is used on Polaris and Minuteman class rockets and solid-booster motors for the space shuttle. GFRE has been mainly developed for process applications and indeed was used originally for oilfield pipework in the 1950s. Later applications include flow lines and gathering lines, secondary-recovery waterflood systems and, in the 1970s, downhole tubing and casing at depths to 3 OOO m. In recent years, the American Petroleum Institute (API) has introduced specifications for GFRP and GFRE line pipe materials, namely API 15LR for low pressure line pipe ( < 1 OOO psi) and API 15HR for high pressure line pipe (>1 OOOpsi). An interesting specification has also been produced for downhole tubing and casing (API 15AR) which is due for further reviewI8. GFRP and GFRE materials have very good general corrosion resistance in a wide variety of marine and offshore process media. In addition, they are lightweight (being only 10-2590 as dense as steel), strong, non-magnetic, and are now widely available at reasonable cost. Fabrication and assembly of pipework systems can be easily achieved by bell-and-spigot glued joints or by other commercially-available mechanical jointing systems. The principal limitations of these materials remain in terms of their relatively poor fire and mechanical damage resistance, restricted size range availability and certain mechanical and chemical property shortcomings. In addition, assembly and installational detail of pipework has to be carefully undertaken to ensure that no installational or in-service damage (such as distortion or sagging) is sustained. Nevertheless, it is clear that these materials should form an increasingly important range of corrosion-resisting materials for use in marine and offshore pipework and vessels in the years ahead. Design of Components and Fittings
Even with the velocity, turbulence and aeration of the water supplied to the equipment being within acceptable limits, if corrosion is to be avoided it is still necessary for the units themselves to be well designed. Some of the more important aspects involved are outlined below. Condensers, heat exchangers and process coolers The shape of the inlet water box and the shape and positioning of the entry should be designed to produce as smooth a flow of water as possible, evenly distributed over the tubeplate. Poor design can result in high-speed turbulent water regimes developing at certain points, which may give rise to cavitation and impingement conditions at some areas of the tubeplate and particularly at some
9: 76
DESIGN IN MARINE AND OFFSHORE ENGINEERING
tube ends; in other relatively stagnant pockets some tubes may receive an inadequate supply of water, leading to overheating or to the settlement of debris, with consequent deposit attack. Non-condensing gases can also accumulate in a stagnant area accentuating the tendency to local overheating, but this can be counteracted to some extent by the provision of adequate air-escape fittings. Local overheating may also be caused by impingement of steam due to poor baffling, and this can cause penetration of the tubes by wet steam erosion. Local overheating of Cu-30Ni cupronickel tubes may lead to ‘hot spot’ corrosion”’12.More general overheating, which can occur with some auxiliary condensers, particularly drain coolers, may cause a rapid build-up of scale on the inside of the tubes necessitating acid descaling. Design faults in two-pass condensers and heat exchangers that can cause corrosion include poor division plate seals allowing the escape of water at high velocity between the passes, and flow patterns that produce stagnant zones. Partial blockages of tubes by debris can act as turbulence raisers and should be guarded against by fitting either weed grids, filters or plastic inserts in the tube ends so that any object passing through the insert is unlikely to become jammed in the tube. These methods will not, however, prevent trouble from marine organisms that enter the system in their early stages of development. This can usually be effectively dealt with by intermittent chlorination or by other chemical dosing treatment. An alternative scheme is to fill the system with fresh water for a period, followed by removal of the dead organisms by water jetting or by an approved mechanical method. Marine organisms dislike changes in conditions such as water salinity, temperature, flow rate, etc. An important point in the design of condensers and heat exchangers is the provision of facilities that allow ready access for cleaning and maintenance. Access doors should be well sited and should be clear of pipework and fittings. Covers of smaller heat exchangers should be easy to remove well clear of the units. Small doors should be fitted at the lowest point to enable residual fluid and debris to be flushed out, thus ensuring complete drainage when this is required. Lack of these facilities understandably tends to result in maintenance procedures not being carried out as frequently or as well as they should, and this can lead to serious corrosion problems which may be insoluble in the particular plant location. The final point to be mentioned on the subject of marine condenser, heat exchanger or process cooler design is the danger of purchasing auxiliary units through ‘package deals’, which may be adequate for fresh-water service but can have an extremely limited life when operated on seawater. In addition, equipment designed for use on land even for seawater service often proves unsatisfactory when installed in ships or offshore installations. It is generally preferable for such equipment to be specifically designed for the purposes intended. Pumps and valves In addition to designing these so as to cause least turbulence in the water stream, careful design can also minimise corrosion of the pumps and valves themselves.
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:77
Pumps can be a major source of trouble, with rapid deterioration of parts due to impingement, cavitation or galvanic corrosion problems 13. The latter two issues may be the result of either poor design or exacerbated by the conditions under which the pump is operated. The remedy may be found in changes in material, since some materials resist cavitation better than others. However, even the most resistant material may have a short life if the cavitation is severe. Designs should avoid features that produce excessive turbulence which may induce cavitation, or that allow the passage of high-velocity water between the high- and low-pressure areas. Cavitation in pumps can develop if the water supply is not continuous, e.g. when pumping-out bilges, especially with choked strainers or lines, or if the water supply is controlled by a badly designed throttling device. Pumps with high suction heads are particularly prone to suffer cavitation damage. Conditions in pumps can still be severe, however, in the absence of cavitation, and high duty materials may be necessary for a reasonable service life. The use of silastomer or neoprene coatings and linings can often give good results even in reasonably severe service conditions. As regards valves, diaphragm types are the most satisfactory but most valves have to withstand extremely turbulent conditions, and as with pumps, even the best designs need to be constructed of resistant materials, or be coated with a suitably durable material such as silastomer or neoprene.
Designs to Prevent Galvanic Corrosion In general it is wise to avoid as far as possible, the use of incompatible metallic joints in marine and offshore practice, since these are often in contact with seawater or water that contains chlorides which are effective corrosive electrolytes. It is prudent to take very considerable precautions to prevent corrosion at the design and installational stages. However, the widely diverse properties required of the materials used in such installations make it impracticable to avoid all such joints. In seawater systems contact between metals and alloys with differing corrosion potentials:* is very common. Recurrent difficulties may be faced with anodic welded joint areas, for e ~ a m p l e ' ~ Internal . lining, material or procedural changes are often the only methods of dealing effectively with the problem. In condensers, iron protector slabs can be employed, although ferrous water boxes sometimes serve the same purpose; impressed-current cathodic protection systems are also finding extended use. Different alloys are also to be found in contact in valves, pumps and other equipment, and between these components and copper-alloy piping; these seldom have closely similar corrosion potentials, but many can have only slight differences that are of little consequence in practice. Even where components have a significant difference in potential, this can often be acceptable if the less noble component is large in area compared with the more noble metal, *The reversible or equilibrium potentials given in the EMF series of metals may have little significance in assessing which metal in a couple will have an enhanced corrosion rate and which will be protected.
9:78
DESIGN IN MARINE AND OFFSHORE ENGINEERING
or is of thicker section and can be allowed to suffer a certain amount of corrosion loss without effect on overall plant integrity. In fact, the galvanic sacrificial protection afforded by this means to a more noble and vital component can be most valuable, as for example if pump impellers are given some protection by the pump casing which is often of a ferrous material such as cast iron. The tendency for some otherwise highly corrosion-resistant alloys to suffer crevice attack can also often be overcome in this way. Other combinations, however, do create difficulties. For example, nonferrous components and fittings such as copper-alloy valves and sea-tubes can cause serious corrosion of adjacent steel, the galvanic attack in many of these instances being accentuated by crevicing, high water speeds and sometimes by the inadequate design of inlets which promote turbulence and contribute to the difficulty of maintaining paint films intact for any length of time. It should be noted that the extent and rate of damage is also dependent on the area ratio of the cathode to anode areas of the two metals, and this is a maximum when a large cathode area is acting on a small anode area. Corrosion in these areas is sometimes effectively controlled by cathodic protection with zinc- or aluminium-alloy sacrificial anodes in the form of a ring fixed in good electrical contact with the steel adjacent to the nonferrous component. This often proves only partially successful, however, and it also presents a possible danger since the corrosion of the anode may allow pieces to become detached which can damage the main circulatingpump impeller. Cladding by corrosion-resistant overlays such as cupronickel or nickel-base alloys may be an effective solution in difficult installational circumstances. Non-ferrous propellers in ships can cause similar trouble, especially during the ship’s fitting-out period. This may be controlled by not fitting the propellers until the final dry docking prior to carrying out sea trials; by coating the propellers prior to launching and removing the coating at the final dry docking; and by cathodic protection of the whole outer bottom. The protection of the outboard propeller shaft is most successfully achieved by a coating of epoxy resin reinforced with glass cloth. The earthing of shafts to the hull is advocated by some ship owners and by some navies, but the practical difficulties involved in maintaining good contact are substantial. One of the most important and extensive applications of two metals in proximity in ship construction is the use of aluminium-alloy superstructures with steel hulls. This practice is now widely accepted in both naval and merchant services. The increased initial cost is more than offset by certain fabrication advantages, weight saving and fuel consumption, with the concomitant advantages of increased accommodation, a lowered centre of gravity giving greater stability (which is especially valuable for ships carrying deck cargo) and a lighter draft. These advantages are so substantial that the use of these two metals in juxtaposition even in a marine environment must be accepted. Serious trouble can, however, usually be avoided if the following steps are taken: 1. The metal interface should be designed to be as high a quality fit as
possible.
DESIGN IN MARINE A N D OFFSHORE ENGINEERING
9:79
2. The faying surface of the steel should be given a metal coating of either zinc (by spraying or galvanising) or aluminium spray. 3. Jointing compounds such as neoprene tape or fabric strip impregnated with chromate inhibitor or other inhibited caulking compounds should be applied carefully to ensure complete exclusion of water. 4. Rivets or other forms of fastenings should be of a similar material to the plate with which it is in contact. 5 . The joint should be painted on both sides with the appropriate coating systems (following a suitable surface preparation programme). Regardless of the method selected, great attention must nevertheless be given to ensuring that the final detail is sound and able to be inspected and repaired if necessary, when the structure is in service. Galvanic corrosion arising from different metals not in ‘direct’ electrical contact must also be guarded against. In particular, water containing small amounts of copper resulting from condensation, leakage or actual discharge from copper and copper-alloy pipes and fittings, can cause accelerated attack of steel plating with which it comes into contact. Likewise copper pipes carrying iron or steel particles may also suffer pitting attack as a consequence of microgalvanic cells created by such particles forming wall contacts. Good design and ‘clean’operating practices can help prevent this. In addition, both the pipework and the plating should be accessible for inspection and for applying and maintaining protective coatings.
Stress-corrosion Cracking, Corrosion Fatigue and Cavitation Damage Some examples of how design can assist in counteracting these forms of attack in marine and offshore practice are given below. The three phenomena are dealt with comprehensively in Chapter 8. Stress corrosion cracking
Low-carbon and chromium-nickel steels, certain copper, nickel and aluminium alloys (which are all widely used in marine and offshore engineering) are liable to exhibit stress-corrosion cracking whilst in service in specific environments, where combinations of perhaps relatively modest stress levels in material exposed to environments which are wet, damp or humid, and in the presence of certain gases or ions such as oxygen, chlorides, nitrates, hydroxides, chromates, nitrates, sulphides, sulphates, etc. A broad indication of the principal environmental features which may promote stress-corrosion cracking is given in Table 9.4”. The list is by no means exhaustive but it is useful in preliminary stress-corrosion cracking risk assessment within the design process. In addition, the temperature of the environment may also be a contributory factor in stress-corrosion cracking processes. Consequently, the high-risk areas will include boilers, heat exchangers, process coolers, drilling equipment, downhole tubulars, wireline
9:80 Table 9.4
DESIGN IN MARINE AND OFFSHORE ENGINEERING
Environments known to promote stress-corrosion cracking in certain engineering alloys Is
Material
Environments ~~
AI alloys Mg alloys Cu alloys C steels Austenitic steels High strength steels Ni alloys Ti alloys
Chlorides, moist air Chloride-chromate mixtures, moist air, nitric acid, fluorides, sodium hydroxide Ammonia, moist air, moist sulphur dioxide Nitrates, hydroxides, carbonates, anhydrous ammonia Chlorides, sulphuric acid Moist air, water, chlorides, sulphates, sulphides Hydroxides Halides, methanol
and logging equipment, packers, wellhead and downstream oil and gas processing equipment. Clearly, the lowering of stress-corrosion cracking risk at the design stage must be an important priority when dealing with the combinations of operational and environmental circumstances outlined above. It has become something of an increasing priority in recent years owing to the fact that certain engineering design trends or priorities - such as increasing strength/ weight ratios-often lead to selection of higher alloy steels being used in highly stressed corrosive situations. In addition, there is a trend for certain industrial processes to be undertaken at progressively higher process temperatures, for example deeper (hence hotter) sour oil and gas wells being drilled and produced gives rise to increasing instances of chloride stress cracking (CSC) problems in oilfield equipmentI6. An outline of the general design options for reduction of stress-corrosion cracking risk in marine and offshore installations can be summarised as follows: 1. Attempt to ensure that residual and operational tensile or oscillatory tensile (fatigue) stresses in components are kept moderate so far as reasonably practicable. 2. Provide effective means of corrosion protection of components consistent with operational circumstances. This may involve cathodic protection, inhibition, coatings or combinations of these. 3. Remove significant environmental influences as soon as reasonably practicable, e.g. dewatering/dehumidification of produced oils and gases as close to wellhead as possible. 4. Choose a material known to have reasonable and reliable intrinsic stress-corrosion cracking resistance in the operational media. For example, a very high stress-corrosion cracking risk would be attached to the installation of a highly stressed titanium-nickel cryogenic memoryalloy pipe coupling, used in the transmission of commercial-grade methanol (which may contain ca. 2 OOO ppm water) for oil well dewaxing operations. 5 . Ensure stress-raising features such as holes, welds, edges, rapid changes in section, etc, are minimised in their exposure to the stress-corrosion cracking risk environment. Heat treatment of components may give beneficial results.
DESIGN IN MARINE AND OFFSHORE ENGINEERING
9:81
Corrosion fatigue
Marine and offshore engineering, by its very nature involving many components being subjected to oscillating stress levels in a wide variety of aggressive media, has to consider corrosion fatigue as a serious risk in many instances. Again, the influences of water, dissolved oxygen, chlorides, etc, are all known to have generally deleterious effects on the corrosion fatigue performance of many engineering alloys Is. Successful methods of counteraction may involve modification to the corrosive environment, reduction of fatigue stress levels by one means or another, corrosion protection or materials uprating. Recurrent high-risk corrosion fatigue areas, however, should be closely monitored. These areas include: 1. high-speed ship hull plates; 2. boiler and pipework components subject to pulsating stresses of thermal or mechanical origin; 3. condenser and heat exchanger tubing, where inadequate support/ clamping arrangements are made; 4. propeller, impeller and pump shafts, connecting rods of diesel engines, gas turbine and gearbox components where intake air or oil is contaminated; 5 . sucker rods in corrosive wells.
Many of the design rules applying to stress-corrosion cracking may also be considered as being potentially useful in minimising corrosion fatigue risk. Perhaps one area where cathodic protection should be applied with great caution is in circumstances where over-protection of a structure may give hydrogen evolution, which in turn may lead to hydrogen absorption of a component -particularly bolt, stud or shaft materials- which, in turn, may raise the risk of corrosion fatigue in certain marine and offshore installational circumstances. Cavitation Damage
This type of damage is dealt with comprehensively in Section 8.8. It can be particularly severe in seawater giving rise to cavitation corrosion or cavitation erosion mechanisms, and hence can be a considerable problem in marine and offshore engineering. Components that may suffer in this way include the suction faces of propellers, the suction areas of pump impellers and casings, diffusers, shaft brackets, rudders and diesel-engine cylinder liners. There is also evidence that cavitation conditions can develop in seawater, drilling mud and produced oil/gas waterlines with turbulent high rates of flow. Improvements in design that can assist in preventing cavitation damage may be concerned with the shape of the component itself, or with its surroundings. The design of propellers and impellers depends largely on the expertise of the manufacturers of these components. Where cavitation damage develops, this may occasionally be due to unsuitable design in relation to the conditions of service, but in most cases occurring in seawater systems, the trouble arises from other causes such as poor layout, air leaks
9:82
DESIGN IN MARINE AND OFFSHORE ENGINEERING
in suction piping or faulty operation. Underwater fittings should be designed to offer as little resistance as possible to movement through the water and to leave the minimum turbulence in their wake which might result in cavitation damage to other parts of the ship. Cathodic protection often proves of benefit in reducing damage t o ships’ propellers and underwater fittings. In addition, certain energy-absorbing coatings may effectively limit cavitation damage in pump casings, etc. The geometry of piping systems, accuracy of fit, smoothness of internal surface, presence of turbulence creators etc., may be important contributory factors in instances of cavitation attack. Figure 9.24 shows cavitation erosion attack in tubing pulled from an oil and gas production well. The source of the cavitation attack was a downhole valve which produced severe downstream turbulence, even in the fully-open position. Such forms of attack, in lines where flow velocities may be as high as 100 m/s, may quickly become cumulative, producing further secondary or tertiary erosion bands downstream of the first area of attack. Internal coatings such as baked phenolics may be practical in some circumstances both from the corrosion control and ‘surface smoothing’ points of view. Other design issues which may be considered when attempting to minimise cavitation damage include:
Fig. 9.24 Cavitation erosion in oil and gas production well tubing. Note the severe localised attack arising from a turbulence-creating downhole valve a few metres upstream
DESIGN IN MARINE AND OFFSHORE ENGINEERlNG
9:83
1. improve quality of fit of pipe butts, flange areas, etc., to minimise weld
2.
3. 4.
5.
root or gasket intrusion into the process flow, which may create cavitation damage. Examine geometry and size of piping/flowline systems to ensure process streams are subject to minimal pressure changes and fluctuations, changes in fluid direction and flow rates consistent with production requirements. Examine pressure heads and maximise whenever reasonably practicable to minimise vapour cavity formation. Establish corrosivity levels of process streams and choose materials and protection systems appropriately. If possible, cool process streams to minimise vapour cavity formation. L. KENWORTHY D. KIRKWOOD REFERENCES
1. Laque, F. L., Marine Corrosion, John Wiley and Sons, New York (1975) 2. Baxter K. F., NSC Conference, Trans. Inst. Mar. Eng., 91, 24 (1979) 3. Stuart Mitchell, R. W. and Kievits, F. J., Gas Turbine Corrosion in the Marine Environ-
4. 5. 6.
7.
8.
ment, Joint Inst. Mar. Eng., Inst. Corr. Sci. Tech. Conf. on Corrosion in the Marine Environment, London, (1973) Bell, S. E.and Kirkwood, D., Determination of theFatalAccident Inquiry on the Chinook Accident, Sumburgh, November 1986. Crown Office. Edinburgh, p. 40 (1987) Wilkinson, T. G., In Corrosion and Marine Growth on Offshore Structures, Ed. Lewis and Mercer, Ellis Horwood, p. 31 (1984) Southwell, C. R., Bultman, J. D. and Huimner, J. R., Influence of Marine Organisms on the Life of StructuralSteels, US Naval Research Laboratory Report No. 7672, Washington DC, USA (1974) Wadkins, L. L., Corrosion and Protection of Steel Piping in Seawater, Technical Memorandum No 27, US Army Corps of Engineers, Coastal Enquiry Centre, Washington DC, USA (1%9) Sneddon, A. D. and Kirkwood, D., Marine Fouling and Corrosion Interactions on Steels and Copper-Nickel Alloys, Proc. UK Corrosion 88 Conf., Inst. Corr. Sci. Tech. NACE
(1988) 9. Wranglen, G., An Introduction to Corrosion and Protection of Metals, Institut for Metallskydd, Stockholm, Sweden (1972) 10. Kenworthy, L., Trans. Inst. Mar. Eng., 77 (a), 154 (1965) 11. Breckan, C. and Gilbert, P. T., Proc. 1st Int. Congress on Metallic Corrosion, London 1%1, Butterworths, London, p. 624 (1962) 12. Bem, R. S. and Campbell, H. S., ibid., p. 630 (1962)
13. Robson, D. N. C., In Corrosion and Marine Growth on Offshore Structures Ed, Lewis and Mercer, Ellis Horwood, p. 69 (1984) 14. Vetters, A., Corrosion in Welds, Proc. Australian Corrosion Conference, Sydney (1979) 15. Parkins, R. N. and Chandler, K. A., Corrosion Control in Engineering Design, HMSO, London, p. 3 (1978) 16. King, J. A. and Badalek, P. S. C., Proc. UK National Corrosion Confernce, Inst. Corr. Sci. Tech., London, p. 145, (1982) 17. Pollock, W. I. and Barnhart, J. M., Corrosion of Metals under Thermallnsulation, ASTM (1985) 18. Biro. J. P., Specifications should increase the use of fiber glass downhole, Oil and Gas J . , 96. Aug. (1989)
9:84
DESIGN IN MARINE AND OFFSHORE ENGINEERING
BIBLIOGRAPHY Scheweitzer, P. A., Corrosion and Corrosion Protection Handbook, Dekker AG, Bade (1989) Ashworth, V. and Booker, C. J. L. (Eds.), Cathodic Protection; Theory and Practice, Ellis Horwood (1986) C 0 2 Corrosion in Oil and Gas Production: Selected papers and Abstracts, NACE, Houston (1984)
H2S corrosion in Oil and Gas Production, NACE, Houston (1982) Sulphide Stress Cracking Resistant Metallic Materialsf o r OilfieldEquipment, NACE Standard MR-01-75 (as revised) (1992) Svensk Standard SIS 055900-1%7, Pictorial Surface Preparation Standardsf o r Painting Steel Surfaces. Rogue, T. and Edwards, J., Sulphide Corrosion and H2S Stress Corrosion, Norwegian Maritime Research 14-23, vol. 12, no. 3 (1984) Crawford, J., Offshore Installation Practice, Butterworths (1988) Tiratsoo, E. N., Oirfields of the World, 3rd Edn. and Supplement, Scientific Press Limited (1986)
Sedriks, A. J., Corrosion of Stainless Steels, John Wiley & Sons, New York (1979) Chandler, K. A., Marine and Offshore Corrosion, Butterworths (1984) Ross, T. K., Metal Corrosion, Engineering Design Guides, No 21, Oxford University Press (1977)
Fontana, M. G., Corrosion Engineering, 3rd Edn., McGraw-Hill (1986) Schweitzer, P. A., What Every Engineer Should Know About Corrosion, Dekker AG, Bade ( 1987)
Rowlands, J., Corrosion f o r Marine and Offshore Engineers, Marine Media (1986) Ashworth, V., Corrosion, Pergamon Press (1987) Technical Note, Fixed Offshore Installations, TNA 703, Cathodic Protection Evaluation, Det . Norske Veritas (1981) Technical Note, Fixed Offshore Installations, TNA 702, Fabrication and Installation of Sacrificial Anodes, Det. Norske Veritas (1981)
9.5
Design in Relation to Welding and Joining
A jointed fabrication* is one in which two or more components are held in position (a) by means of a mechanical fastener (screw, rivet or bolt), (6)by welding, brazing or soldering or (c) by an adhesive. The components of the joint may be metals of similar or dissimilar composition and structure, metals and non-metals or they may be wholly non-metallic. Since the majority of fabrications are joined at some stage of their manufacture, the corrosion behaviour of joints is of the utmost importance, and the nature of the metals involved in the joint and the geometry of the joint may lead to a situation in which one of the metals is subjected to accelerated and localised attack. Although corrosion at bimetallic contacts involving different metals has been dealt with in Section 1.7, it is necessary to emphasise the following in relation to corrosion at joints in which the metals involved may be either identical or similar: 1. A difference in potential may result from differences in structure or
stress brought about during or subsequent to the joining process. 2. Large differences in area may exist in certain jointed structures, e.g. when fastening is used. Furthermore, many joining processes lead to a crevice, with the consequent possibility of crevice corrosion. Before considering the factors that lead to corrosion it is necessary to examine briefly the basic operations of joint manufacture.
Mechanical Fasteners These require little description and take the form of boltings, screws, rivets, etc. Mechanical failure may occur as a result of the applied stress in shear or tension exceeding the ultimate strength of the fastener, and can normally be ascribed to poor design, although the possibility of the failure of steel fittings at ambient or sub-zero temperatures by brittle fracture, or at ambient temperatures by hydrogen embrittlement, cannot be ignored. If brittle failure is a problem then it can be overcome by changing the joint design or *For definitions of terms used in this section see Section 9.5A
9:85
9: 86
DESIGN IN RELATION TO WELDING AND JOINING Anodic
Insula tors
A
Cat hod ic
situation because the large anode to cathode ratio The reverse situation may be acceptable under mild corrosive conditions
insulator between these surfaces i f component metals are dissimilar
of
(b)
which continuous (d)
(C)
lnsulat ing washer
\
mT1
lnsulat ing gasket
(e)
(f)
Fig. 9.25
Design of insulated joints'
employing a fastener having a composition with better ductility transition properties. The corrosion problems associated with mechanical fixtures are often one of two types, i.e. crevice corrosion or bimetallic corrosion'-4, which have been dealt with in some detail in Sections 1.6 and 1.7, respectively. The mechanical joining of aluminium alloys to steels using rivets and bolts, a combination which is difficult to avoid in the shipbuilding industry, represents a typical example of a situation where subsequent bimetallic corrosion could occur. Similarly, other examples of an ill-conceived choice of materials, which could normally be avoided, can be found in, for
DESIGN IN RELATION TO WELDING AND JOINING
9:87
example, brass screws to attach aluminium plates or steel pins in the hinges of aluminium windows. The relative areas of the metals being joined is of primary importance in bimetallic corrosion, and for example, stainless steel rivets can be used to joint aluminium sheet, whereas the reverse situation would lead to rapid deterioration of aluminium rivets. However, in the former case a dangerous situation could arise if a crevice was present, e.g. a loose rivet, since under these circumstances the effective anodic area of the aluminium sheet would be reduced, with consequent localised attack. In general, under severe environmental conditions it is always necessary to insulate the components from each other by use of insulating washers, sleeves, gaskets, etc. (Fig. 9.25)5, and the greater the danger of bimetallic corrosion the greater the necessity to ensure complete insulation; washers may suffice under mild conditions but a sleeve must be used additionally when the conditions are severe. The fasteners themselves may be protected from corrosion and made compatible with the metal to be fastened by the use of a suitable protective coating, e.g. metallic coating, paints, conversion coating, etc. The choice of fastener and protective coating, or the material from which it is manufactured, must be made in relation to the components of the joint and environmental conditions prevailing6. Thus high-strength steels used for fastening the fuselage of aircraft are cadmium plated to protect the steel and to provide a coating that is compatible with the aluminium. In the case of protection with paints it is dangerous to confine the paint to the more anodic component of the joint, since if the paint is scratched intense localised attack is likely to occur on the exposed metal. In general, paint coatings should be applied to both the anodic and cathodic metal, but if this is not possible the more cathodic metal rather than the more anodic metal should be painted. The use of high-strength steels for bolts for fastening mild steel does not normally present problems, but a serious situation could arise if the structure is to be cathodically protected, particularly if a power-impressed system is used, since failure could then occur by hydrogen embrittlement; in general, the higher the strength of the steel and the higher the stress the greater the susceptibility to cracking. A point that cannot be overemphasised is that, in the long term, stainless steel fasteners should be used for securing joints of stainless steel.
Soldered Joints Soldering and brazing are methods of joining components together with a lower-melting-point alloy so that the parent metal (the metal or metals to be joined) is not melted (Table 9.5). In the case of soft soldering the maximum temperature employed is usually of the order of 250°C and the filler alloys (used for joining) are generally based on the tin-lead system. The components must present a clean surface to the solder to allow efficient wetting and flow of the molten filler and to provide a joint of adequate mechanical strength. To obtain the necessary cleanliness, degreasing and mechanical abrasion may be required followed by the use of a flux to remove any
9:88
DESIGN IN RELATION TO WELDING AND JOINING Table 9.5 Soldering and brazing
Process SOLDERING hot iron oven induction ultrasonic dip resistance wave and cascade BRAZING torch dip salt bath furnace induction resistance
Temp. range ("C) 60-300
500-1 200
Typical fillers
Fluxes
70Pb-30% 40Pb-60Sn 70Pb-27Sn-3Sb 40Pb-58Sn-2Sb Sn-Zn-Pb
Chloride based Fluoride based Resin based
%AI-IOSi 50Ag-15Cu-17Zn-18Cd Ag-Cu-Ni-In 60Ag-30Cu-IOZn
Borax based Fluoride based Hydrogen gas Town's gas Vacuum
5OCu-5OZn 97cu-3P 70Ni- 17Cr-3 B-IOFe 82Ni-7Cr-SSi-3Fe 60Pd-40Ni
remaining oxide film and to ensure that no tarnish film develops on subsequent heating. In the case of carbon steels and stainless steels, and many of the nonferrous alloys, the fluxes are based on acidic inorganic salts, e.g. chlorides, which are highly corrosive to the metal unless they are removed subsequently by washing in hot water. For soldering tinplate, clean copper and brass, it is possible to formulate resin-based fluxes having non-corrosive residues and these are essential for all electrical and electronic work. Activators are added to the resin to increase the reaction rate, but these must be such that they are thermally decomposed at the soldering temperature if subsequent corrosion is to be avoided'. Corrosion is always a risk with soldered joints in aluminium owing to the difference in electrical potential between the filler alloy and the parent metal and the highly corrosive nature of the flux that is generally used for soldering. However, it is possible to employ ultrasonic soldering to eliminate use of flux. With aluminium soldering it is imperative that the joints be well cleaned both prior and subsequent to the soldering operation, and the design should avoid subsequent trapping of moisture.
Brazed Joints When stronger joints are required, brazing may be used'. The filler alloys employed generally melt at much higher temperatures (600- 1 20O0C), but the effectiveness of the joining process still depends upon surface c:ean!iness of the components to ensure adequate wetting and spreading. Metallurgical and mechanical hazards may be encountered in that the filler may show poor spreading or joint filling capacity in a certain situation
DESIGN IN RELATION TO WELDING AND JOINING
9:89
or may suffer from hot tearing, whilst during furnace brazing in hydrogencontaining atmospheres there is always the possibility that the parent metal may be susceptible to hydrogen embrittlement or steam cracking. Furthermore, brittle diffusion products may be produced at the filler base-metal interface as a result of the reaction of a component of the filler alloy with a base-metal component, e.g. phosphorus-bearing fillers used for steel in which the phosphorus diffuses into the steel. Serious damage can be caused by (u) diffusion into the parent metal of the molten brazing alloy itself when either one or both of the parent metal(s) is in a stressed condition induced by previous heat treatment or cold working, and (b) by an externally applied load which need only be the weight of the workpiece. Nickel and nickel-rich alloys are particularly prone to liquid-braze-filler attack especially when using silver-based braze fillers at temperatures well below the annealing temperature of the base metal, since under these conditions there is then no adequate stress relief of the parent metal at the brazing temperature. The problem may be avoided by annealing prior to brazing and ensuring the maintenance of stress-free conditions throughout the brazing cycle. There is a whole range of silver-, nickel- and palladium-based braze fillers of high oxidation and corrosion resistance that have been developed for joining the nickel-rich alloys; however, the presence of sulphur, lead or phosphorus in the basemetal surface or in the filler can be harmful, since quite small amounts can lead to interface embrittlement (Section 7.5). In the case of the Monels, the corrosion resistance of the joint is generally less than that of the parent metal and the design must be such that as little as possible of the joint is exposed to the corrosive media. When, in an engineering structure, the aluminium-bronzes are used for their corrosion resistance, the selection of braze filler becomes important and although the copper-zinc brazing alloys are widely used, the corrosion resistance of the joint will be that of the equivalent brass rather than that of the bronze. With the carbon and low-alloy steels, the braze fillers are invariably noble to the steel so that there is little likelihood of trouble (small cathodeAarge anode system), but for stainless steels a high-silver braze filler alloy is desirable for retaining the corrosion resistance of the joint, although stress-corrosion cracking of the filler is always a possibility if the latter contains any zinc, cadmium or tin. An interesting example of judicious choice of braze filler is to be found in the selection of silver alloys for the brazing of stainless steels to be subsequently used in a tap-water environmentg. Although the brazed joint may appear to be quite satisfactory, after a relatively short exposure period failure of the joint occurs by a mechanism which appears to be due to the break-down of the bond between the filler and the base metal. Dezincification is a prominent feature of the phenomenon" and zinc-free braze alloys based on the Ag-Cu system with the addition of nickel and tin have been found to inhibit this form of attack. A similar result is obtained by electroplating 0-007mm of nickel over the joint area prior to brazing with a more conventional Ag-Cu-Zn-Cd alloy. Brazing is generally considered unsuitable for equipment exposed to ammonia and various ammoniacal solutions because of the aggressiveness of ammonia to copper- and nickel-base alloys, but recently an alloy based
9:90
DESIGN IN RELATION TO WELDING AND JOINING Table 9.6 Typical joining processes
Joining process
Types
Mechanical fasteners Soldering and brazing Fusion welding
Nuts, bolts, rivets, screws Hot iron, torch, furnace, vacuum Oxyacetylene, manual metal are, tungsten inert gas, metal inert gas, carbon dioxide, pulsed are, fused are, submerged arc, electro slag and electron beam Spot, seam, stitch, projection, butt and flash butt Pressure, friction, ultrasonic and explosive
Resistance welding Solid-phase welding
on Fe-3 -25B-4.40Si-50.25Ni has been shown to be suitable for such applications' I . Upton'* has recently studied the marine corrosion behaviour of a number of braze alloy-parent metal combinations and has shown that compatibility was a function of the compositions of the filler and parent metals, their micro-structures and chance factors such as overheating during the brazing operation.
Welded Joints The welded joint differs from all others in that an attempt is made to produce a continuity of homogeneous material which may or may not involve the incorporation of a filler material. There are a large variety of processes by which this may be achieved, most of which depend upon the application of thermal energy to bring about a plastic or molten state of the metal surfaces to be joined. The more common processes used are classified in Table 9.6. The macrographic examination of a welded joint shows two distinct zones, namely the fusion zone with its immediate surroundings and the parent metal (Fig. 9.26). It is apparent therefore, that such processes produce differences in microstructure between the cast deposit, the heataffected zone which has undergone a variety of thermal cycles, and the parent plate. Furthermore, differences in chemical composition can be introduced accidentally (burnout of alloying elements) or deliberately (dissimilar metal joint). Other characteristics of welding include: (1) the production of a residual stress system which remains after welding is completed, and which, in the vicinity of the weld, is tensile and can attain a magnitude up to the yield point; (2) in the case of fusion welding the surface of the deposited metal is rough owing to the presence of a ripple which is both a stress raiser and a site for the condensation of moisture; (3) the joint area is covered with an oxide scale and possibly a slag deposit which may be chemically reactive, particularly if hygroscopic; and (4) protective coatings on the metals to be joined are burnt off so that the weld and the parent metal in its vicinity become unprotected compared with the bulk of the plate. Therefore, the use of welding as a method of fabrication may modify the corrosion behaviour of an engineering structure, and this may be further aggravated by removal of protective systems applied before welding, whilst at the same time the use of such anti-corrosion coatings may lead to difficulties in obtaining satisfactorily welded joints 13-'6.
DESIGN IN RELATlON TO WELDING AND JOINING
9:91
,Reinforcement .Weld bead or deposit zone
Penetration’ Root
Parent plate
, , ’
Fusion line
’
2 (a)
Heat-affected zone
-
‘ L
-
---
_ i
.
/
, ,
’/,
Cast nugget
Parent plates
N
,\‘
‘\Electrode
indentation
(b)
Fig. 9.26 Weld definitions. (a) Fusion weld and (b) resistance spot weld
Weld Defects
There is no guarantee that crack-free joints will automatically be obtained when fabricating ‘weldable’ metals. This is a result of the fact that weldability is not a specific material property but a combination of the properties of the parent metals, filler metal (if used) and various other factors (Table 9.7) 17. The consequence of the average structural material possessing imperfect weldability is to produce a situation where defects may arise in the weld deposit or heat-affected zone (Table 9.8 and Fig. 9.27). It is obvious that these physical defects are dangerous in their own right but it is also possible for them to lead to subsequent corrosion problems, e.g. pitting corrosion at superficial non-metallic inclusions and crevice corrosion at pores or cracks. Other weld irregularities which may give rise to crevices include the joint angle, the presence of backing strips and spatter (Fig. 9.29). Butt welds are to be preferred since these produce a crevicefree profile and, furthermore, allow ready removal of corrosive fluxes. carbon and Low-alloy Steels
These usually present little problem since the parent and filler metals are generally of similar composition, although there is some evidence that the
9:92
DESIGN I N RELATION TO WELDING AND JOINING
Table 9.7 Factors affecting weldability* Parent metal Composition Thickness State of heat treatment Toughness Temperature Purity Homogeneity
Filler metal
Other factors Degree of fusion (Joint formation) Degree of restraint Form factor (Transitions) Deposition technique Skill and reliability of the welder
Composition Impact strength Toughness Hydrogen content Purity Homogeneity Electrode diameter (Heat input during welding)
‘Data after Lundin ”
Table 9.8
Defect Hot cracks Underbead cracks Microfissures Toe cracks Hot tears Porosity
Weldability defects
Causes
Remedies
Large solidification range Segregation Stress Hardenable parent plate Hydrogen Stress Hardenable deposit Hydrogen Stress High stress Notches Hardenable parent plate Segregation Stress Gas absorption
More crack-proof filler Less fusion Low hydrogen process Planned bead sequence Preheating Low hydrogen process Pre- and post-heating Planned bead sequence Preheating Avoidance of notches Less fusion Cleaner parent plate Remove surface scale Remove surface moisture Cleaner gas shield
Hot (solidification) cracks
(transgranular)
Micro-f issuresl
(transgranular)
Fig. 9.27
\Root crack (transgranular)
Possible weld defects
DESIGN IN RELATION TO WELDING AND JOINING
9:93
,Surface pore
strap and root
Fig. 9.28 Possible crevice sites
precise electrode type in manual metal-arc welding for marine conditions may be important; weld metal deposited from basic-coated rods appears to corrode more rapidly than that deposited from rutile-based coatings ”. An environment containing H,S, cyanides, nitrates or alkalis may produce stress-corrosion cracking in highly stressed structures and these should be first stress relieved by heating to 650°C. An interesting development in weldable corrosion-resistant steels is the copper-bearing or weathering steels (Section 3.2) which exhibit enhanced corrosion resistance in industrial atmospheres in the unpainted condition. For optimum corrosion resistance after welding, the filler employed should be suitably alloyed to give a deposit of composition similar to that of the steel plate 19.
Stainless Irons and Steels Since stainless irons and steels (Section 3.3) are widely used for resisting corrosive environments, it is relevant t o consider the welding of these alloys in some detail. There are three groups of stainless steels, each possessing their own characteristic welding problem: 1. Ferritic type. Welding produces a brittle deposit and a brittle heataffected zone caused by the very large grain size that is produced. The problem may be reduced in severity by the use of austenitic fillers and/ or the application of pre- and post-weld heat treatments; the latter is a serious limitation when large welded structures are involved. 2. Martensitic type. Heat-affected zone cracking is likely and may be remedied by employing the normal measures required for the control of hydrogen-induced cracking. 3. Austenitic types. These are susceptible to hot cracking which may be overcome by balancing the weld metal composition to allow the formation of a small amount of 6-Fe (ferrite) in the deposit, optimum crack resistance being achieved with a 6-Fe content of 5-10%. More than
9:94
DESIGN IN RELATION TO WELDING AND JOINING
this concentration increases the possibility of a-phase formation if the weldment is used at elevated temperature with a concomitant reduction in both mechanical and corrosion properties. The corrosion of stainless steel welds has probably been studied more fully than any other form of joint corrosion and the field has been well reviewed by Pinnow and Moskowitz”. whilst extensive interest is currently being shown by workers at The Welding Institute*’. Satisfactory corrosion resistance for a well-defined application is not impossible when the austenitic and other types of stainless steels are fusion or resistance welded; in fact, tolerable properties are more regularly obtained than might be envisaged. The main problems that might be encountered are weld decay, knifeline attack and stress-corrosion cracking (Fig. 9.29). Weld decay is the result of the intergranular precipitation of chromium carbide in the temperature range of 450-800°C and material in this condition is referred to as being ‘sensitised’. Sensitisation depletes the matrix in the grain-boundary region of chromium and this region may eventually suffer intergranular corrosion (see Section 3.3). During welding some zone in the vicinity of the weld area is inevitably raised within the sensitisation temperature range and the degree of severity of sensitisation will be dependent on a number of process factors that determine the time in this temperature range, e.g. heat input, thickness of plate. For most commercial grades of stainless steel in thin section (< 10 mm) the loss in corrosion resistance is slight and seldom warrants any special measures. For a high degree of corrosion resistance, or in welded thick plate, it becomes necessary to take one of the following courses of action: 1. Thermally treat the structures to effect a re-solution of the chromium
carbide; this is often impractical in large structures unless local heat treatment is employed, but is not always satisfactory since a sensitised zone could be produced just outside the local thermally treated region. 2. Use extra-low-carbon steel. 3. Use stabilised steels, i.e. austenitic steels containing niobium, tantalum or titanium.
\
Knifeline attack (intergranular) Weld decay
Stress corrosion (tran sgranular)
$J ‘\I
/!I 0 200 400 700 1oO01500
Fig. 9.29 Corrosion sites in stainless steel welds. The typical peak temperatures attained during welding ( “ C )are given at the foot of the diagram. Note that knifeline attack has the appearance of a sharply defined line adjacent to the fusion zone
DESIGN IN RELATION TO WELDING A N D JOINING
9:95
It is important t o note that the filler metals should also be stabilised, particularly in a multi-run weld where previous deposits are obviously going t o be thermally cycled as later runs are deposited. It may also be necessary t o increase the nickel and chromium contents of the filler to offset losses incurred during welding. It should be noted that sensitisation has very little effect on mechanical properties and that intergranular attack occurs only in environments that are aggressive. France and Greene22point out that the precautions taken to avoid sensitisation are frequently unnecessary, and they have carried out a potentiostatic study of a number of electrolyte solutions to evaluate the range of potential, composition and temperature in which intergranular attack occurs. They claim that by means of these studies it is possible to predict whether the environment will be aggressive or non-aggressive to the sensitised zone, and that in the case of the latter no precautions need be taken to avoid sensitisation. This work, which has been criticised by Streicher 23, is still controversial and generally the normal precautions concerning sensitisation are taken irrespective of the nature of the environment. Titanium stabilised fillers should not be used in argon-arc welding as titanium will be vaporised and its effectiveness as a stabiliser lost. Carburising the weld seam by pick-up from surface contamination, electrode coatings or the arc atmosphere leads to increased tendency to intercrystalline corrosion. The effect of the welding process on the severity of weld decay varies according t o the process and the plate thickness so that no single recommendation is possible for every thickness of plate if resistance to attack is essential. The severity of weld decay correlates quite well with sensitisation times as calculated from recorded weld heating cycles. Under certain conditions it is possible for a weldment to suffer corrosive attack which has the form of a fusion line crack emanating from the toe of the weld; this is termed knifeline attack. It is occasionally experienced in welded stabilised steels after exposure to hot strong nitric acids. The niobium-stabilised steels are more resistant than the titanium-stabilised types by virtue of the higher solution temperature of NbC, but the risk may be minimised by limiting the carbon content of a steel to 0.06% maximum (ELC steel). Stress-corrosion cracking (Chapter 8) is particularly dangerous because of the insidious nature of the phenomenon. The residual stresses arising from welding are often sufficiently high to provide the necessary stress condition whilst a chloride-containing environment in contact with the austenitic stainless steels induces the typically transgranular and branched cracking. An increased nickel content marginally improves the resistance of the steel to this type of attack, whilst at the opposite extreme, the ferritic chromium steels are not susceptible. The only sure means of eliminating this hazard is to employ either a stress-relief anneal or a molybdenum-bearing steel, but stabilised steels must be used since the required heat treatment is in the carbide-sensitisation temperature range.
9:%
Nickel A’I
DESIGN IN RELATION TO WELDING AND JOINING
ys (Section 4.5)
In the main, welding does not seriously affect the corrosion resistance of the high nickel alloys and stress relief is not generally required since the resistance to stress corrosion is particularly high; this property increases with increase in nickel content and further improvement may be obtained by the addition of silicon. The chromium-containing alloys can be susceptible to weld decay and should be thermally stabilised with titanium or niobium, and where conditions demand exposure to corrosive media at high temperatures a further post-weld heat treatment may be desirable. For the Ni-Cr-Mo-Fe-W type alloys, SamansZ4suggests that the material should be given a two-stage heat treatment prior to single-pass welding in order to produce a dependable microstructure with a thermally stabilised precipitate. The Ni-28Mo alloy provides a special case of selective corrosion analogous to the weld-decay type of attack; it may be removed by solution treatment or using an alloy containing 2% Vzs. Of the weldability problems, nickel and nickel-based alloys are particularly prone to solidification porosity, especially if nitrogen is present in the arc atmosphere, but this may be controlled by ensuring the presence of titanium as a denitrider in the filler and maintaining, a short arc length. The other problem that may be encountered is hot cracking, particularly in alloys containing Cr, Si, Ti, Al, B, Zr, S, Pb and P. For optimum corrosion resistance it is recommended that similar composition fillers be used wherever possible, and obviously any flux residues that may be present must be removed. Aluminium A110 ys (Section 4. I l
These alloys are very susceptible to hot cracking and in order to overcome this problem most alloys have to be welded with a compensating filler of different composition from that of the parent alloy, and this difference in composition may lead to galvanic corrosion. A further problem in the welding of these materials is the high solubility of the molten weld metal for gaseous hydrogen which causes extensive, porosity in the seam on solidification; the only effective remedy is to maintain the hydrogen potential of the arc atmosphere at a minimum by using a hydrogen-free gas shield with dry, clean consumables (e.g. welding rods, wire) and parent plate. In general, the corrosion resistance of many of the alloys is not reduced by welding. Any adverse effects that may be encountered with the highstrength alloys can be largely corrected by post-weld heat treatment; this is particularly true of the copper-bearing alloys. Pure aluminium fillers impart the best corrosion resistance, although the stronger AI-Mg and Al-Mg-Si fillers are normally suitable; the copper-bearing fillers are not particularly suitable for use in a corrosive environment. Resistance welding does not usually affect the corrosion resistance of the aluminium alloys. The heat-affected zone may become susceptible to stress-corrosion cracking, particularly the high-strength alloys, and expert advice is necessary
DESIGN IN RELATION TO
WELDING AND JOINING
9:97
Table 9.9 Possible problems in less commonly welded metals Metal
Weidabiiity
Corrosion
~~~~~~~
Copper alloys
Magnesium alloys Titanium alloys
Porosity Hot cracking Hot tearing Steam explosion Porosity Hot cracking Lack of fusion Porosity Embrittlement
De-zincification De-aluminification Stress corrosion Stress corrosion Pitting Stress corrosion
concerning the suitability of a particular alloy for a certain environment after welding. In this context AI-Zn-Mg type alloys have been extensively studied26and it has been shown that maximum sensitivity appears to occur when there is a welldeveloped precipitation at the heat-affected zone grain boundaries adjacent t o the fusion line, a fine precipitate within the grain and a precipitate-free zone immediately adjacent the grain boundaries. The action of stress-corrosion cracking then appears to be a result of local deformation in the precipitate-free zone combined with the anodic character of the precipitate particles. Other Materials
Space does not permit a survey of all the various weldable metals and their associated problems, although some suggestions are made in Table 9.9. It is sufficient to state that with a knowledge of the general characteristics of the welding process and its effects on a metal (e.g. type of thermal cycle imposed, residual stress production of crevices, likely weldability problems) and of the corrosion behaviour of a materia1 in the environment under consideration, a reliable joint for a particular problem will normally be the rule and not the exception. Corrosion Fatisue (Section 8.61
A metal's resistance to fatigue is markedly reduced in a corrosive environment. Many welded structures are subjected to fluctuating stresses which, with the superimposed tensile residual stress of the joint, can be dangerous. In addition to this a welded joint is a discontinuity in an engineering structure containing many possible sites of stress concentration, e.g. toe or root of the joint, weld ripple. Protection of Welded Joints
Structural steels are frequently protected from corrosion by means of a paint primer, but these materials can have an adverse effect on the
9:98
DESIGN IN RELATION TO WELDING AND JOINING
subsequent welding behaviour and this is mainly observed as porosity 'j. Hot-dip galvanising for long-term protection can also lead to porosity and intergranular cracking after welding, in which case it may be necessary to remove the zinc coating from the faying edges prior to welding. The presence of zinc can also lead to operator problems due to the toxicity of the fume evolved unless adequate fume extraction is employed. Prior to painting, all welding residues must be removed and the surface prepared by grinding, grit blasting, wire brushing or chemical treatment. This preparation is of fundamental importance, the method of applying the paint and the smoothness of the bead apparently having little effect on the final result2'.
Recent Developments Although the problems associated with the corrosion and protection of jointed structures have been recognised since the early days of structural fabrication, they have taken on a special significance in the past 15 years. The motivation for the increased impetus is mainly one of concern over possible costly, hazardous or environmentally unfriendly failures particularly those concerned with offshore constructions, nuclear reactors, domestic water systems, food handling, waste disposal and the like. The subject of weldment corrosion in offshore engineering has been well reviewed by TurnbullZ9.Galvanic effects are possible if the steel weld metal is anodic to the surrounding parent plate and is enhanced by the high anode to cathode surface area ratio that exists. Lundin3' showed that basic ferritic weld metal has a more electronegative potential, acid ferritic weld metal is the most electropositive, whilst rutile ferritic weld metal is intermediate between the two. The nature of the surface and its prior treatment (eg. peening) seemed to have no effect. It was also noted that the heat affected zone (HAZ) was no less corrosion resistant than the unaffected plate. Millscale, an electronically conducting oxide of Fe, should be removed by mill-blasting as its presence can cause serious galvanic effects around the joints. On the other hand, Saarinen and 0nne1a3' considered that weld metal corrosion can be eliminated by using a suitably balanced electrode type, the remaining problem then being in the HAZ whose corrodibility increased with increasing Mn content. This was related to the effect of Mn on the transformation characteristics of Fe. Thus, the heat input during welding must be important since the significant factor will be the cooling rate of the HAZ after welding. These findings have been substantiated by Ousyannikov e?ai. 32 using a scanning comparative electrode probe. Increasing the heat input changed the weld metal from anodic to cathodic relative to the parent plate, although the presence of Ni reduced the magnitude of the effect. Recently, attention has been directed to a study of the problem of grooving corrosion in line-pipe steel welded by high frequency induction or electric resistance welding. In sea water, it seems to be related to high sulphur content in the weld zone, the type of environment, its temperature and v e l ~ c i t y ~The ~ * ~importance ~. of sulphur is significant since Drodten and Herbsleb have reported that localised corrosion at welded joints is more a
DESIGN IN RELATION TO WELDING
A N D JOINING
9:99
function of S, Si, microstructure, and non-metallic inclusion type and shape than of the local oxygen c ~ n c e n t r a t i o n ~ ~ . One of the major concerns in offshore construction is that of corrosion fatigue. T ~ r n b u l ldiscusses *~ this at length. Cracks usually originate at weld toes, the point of initiation being associated with crack-like defects (slag, non-metallics, cold laps, undercuts, hot tears). These can constitute sharp notches situated at a point of maximum stress concentration due to the weld geometry. It is to be noticed that although cracks initiate in the HAZ at the weld toe, the majority of crack propagation occurs in what is essentially unaffected parent plate. In air, it is possible to have cracks that grow at a decelerating rate until no further growth occurs; this is the ‘short crack’ problem widely discussed by Miller36,and the cracks are referred to as nonpropagating cracks. On the other hand, the same cracks may continue to grow at an accelerating rate in a corrosive environment even though the stress may be below the fatigue limit; this has been studied little until recently. Burns and Vosikovsky3’ have given considerable attention to corrosion fatigue of tubular joints in the BS4360:50 type steels and X65 line-pipe steels. Initiation at the toe occurs after a small fraction of life and long surface cracks can exist for over 50% of the life. On the other hand, laboratory tests on plate-to-plate welded specimens of the cruciform type show cracks which are much smaller for a larger percentage of the life but their growth rate accelerates as the depth increases. In sea water, the effects of cyclic frequency, stress ratio, electrochemical potential, oxygen content and intermittent immersion at 5 1 2 ° C have all been evaluated”. There is some evidence that at the lower temperatures, the seawater is less detrimental to fatigue life, but at all temperatures studied, crack growth rate was always faster than in air. At intermediate ranges of M ,there was a significant reduction in crack growth rate as the seawater temperature was reduced from 25°C to 0°C. Crack growth rate increased with cathodic protection as a result of absorption of H at the crack tip. Whilst the cracks are small and AK low, calcareous blocking is very effective and under these conditions cathodic protection (CP) reduces the crack growth rate. As the crack length increases, blocking becomes less effective and the increased hydrogen embrittlement can accelerate the growth rate to values greater than experienced for the unprotected joints. In the same vein, obtained data showing that CP raises the initial fatigue Nibbering et crack resistance but has little effect at a later stage of crack propagation. Even so, they considered that CP is still the most effective method for prolonging structural life under corrosion fatigue conditions. This is not unreasonable since crack initiation and early growth can represent a large proportion of the total life. Marine fouling leading to the local production of H2Sincreases crack growth rate, but what the effect is when combined with CP is uncertain. Some of the factors mentioned earlier in connection with other steel corrosion problems are important to sulphide stress-corrosion cracking, (SSCC), eg. compositions, particularly C which usefully can be reduced to below 0.05V0,S, microstructure and segregati~n~’-~. Compositional homogenisation by heat treatment can be beneficial4’, whilst the presence of Cu in the *See Sections 8.6 and 8.9.
9: 100
DESIGN IN RELATION TO WELDING AND JOINING
steel may have some merit". SSCC of weld repairs in well-head alloys was investigated by Watkins and R o ~ e n b e r gwho ~ ~ found that the repairs were susceptibleto this problem because of the hard HAZs developed by welding. Post-weld heat treatment was an essential but not complete cure compared with unrepaired castings. In the case of hydrogen-assisted cracking of welded structural steels, composition is more important than mechanical properties and the carbon equivalent should be 7%. The role of H appears to be logical from the work of Patel and Jarman who have reported the magnitude of the strain field around the tip of propagating cracks in Al-Zn-Mg”. This field is under constant review by Holroyd and Hardie76. General corrosion damage was the cause of failure of an AI alloy welded pipe assembly in an aircraft bowser which was attacked by a deicing-fluid water mixture at small weld defect^'^. Selective attack has been reported in welded cupro-nickel subjected to estuarine and seawater environments7*. It was the consequence of the combination of alloy element segregation in the weld metal and the action of sulphate reducing bacteria (SRB). Sulphidecoated Cu-enriched areas were cathodic relative to the adjacent Ni-rich areas where, in the latter, the sulphides were being continuously removed by the turbulence. Sulphite ions seemed to act as a mild inhibitor. General corrosion occurs in the weld metal and HAZ of welded Zr2.25 Nb alloys in an environment of H,SO, at temperatures greater than 343 K, the rate increasing with concentration. Above 70% H,SO, both general corrosion and IGA occur, whilst above 80% hydrogen embrittlement was found also. Sulphides were found to be deposited on the metal surface 79.80. Protection of welds, both before and after welding, is worthy of careful consideration. For example, in the electric-resistance welding of hot-dipped galvanised steel, welding had little effect on the seawater corrosion of the coated steel when compared with the uncoated steel, the latter showing considerable corrosion after 12 months exposure”. The subject of galvanising and the welding of structural steels has been given special attention by Porter8*, but by far the most common method of protection is by painting which McKelvieE3discusses in terms of fundamentals of paint as a corrosion barrier and the cleaning and coating procedures necessary to achieve protection of welded structures, In these articles he covers the type of contaminants arising from welding as well as cleaning methods, blast primers, galvanising, coating removal for repair welding, wire brushing and chemical
’’.
9 : 102
DESIGN IN RELATION TO WELDING A N D JOINING
treatments. Lloyds Register of Shipping lists the proprietary products that have no significant deleterious effects on subsequent welding work 84. Soldered joints present their own characteristic corrosion problems usually in the form of dissimilar metal attack often aided by inadequate flux removal after soldering. Such joints have always been a source of concern to the electrical Lead-containing solders must be used with caution for some types of electrical connection since Pb(OH), .PbC03 may be found as a corrosion product and can interrupt current flow. Indium has been found to be a useful addition to Sn-Pb solders to improve their corrosion resistance8'. However, in view of the toxicity of lead and its alloys, the use of lead solders, particularly in contact with potable waters and foodstuffs, is likely to decline. In the related process of brazing, crevice corrosion has been found when joining copper tubes using Cu-Ag-P fillers, the presence of scale adjacent to the joint being deemed responsible". Interface corrosion of brazed stainless steel joints has been comprehensively reviewed by Kuhn and Trimmer89whilst Lewisw has used photo-electron spectroscopy to confirm the dezincification theory. As a technical problem it has been reported as occurring in contact with the drink sake with the further complication that the eluted Fe3+ ions from the corrosion of the stainless steel gave rise to discoloration of the liquid". On the other hand, the corrosion resistance of a high temperature brazed joint in a Mo-containing low-C stainless steel exposed to drinking water gave no problemsg2. Suganuma et al. reported an unusual instance of the stress-corrosion cracking of SiN brazed with AI when subjected to an environment of water93. It was contended that the interfacial region was weakened as a result of the surface layers of the SiN being deformed by the grinding operation used prior to brazing. A pre-heat treatment of the SiN at a temperature of no less than 1 lOOK was found to remove the damage. Finally, mechanical joints, e.g. nuts, bolts, rivets etc., are still important joining methods for which attention must be given to compatibility to avoid dissimilar metal corrosion problems and crevice c o r r o ~ i o n ~ - ~ " .
Conclusions Every type of corrosion and oxidation problem can be encountered in jointed structures and it is obvious that most engineering structures must be jointed. It would appear therefore, that all structures are on the verge of disintegration. Yet, for every jointed structure that fails by corrosion, there are many hundreds of thousands which have survived the test of time. With a reasonable knowledge of the mechanics of jointing, the possible design and process factors (e.g. crevices, dissimilar materials in contact, presence of fluxes), a basic understanding of corrosion science and, above all, commonsense, few problems in the established fabrication fields should be encountered. As aptly pointed out by Scully", as with all other scientific and technological problems, experience is often the final arbiter.
R. A. JARMAN
DESIGN IN RELATION TO WELDlNG AND JOINING
9: 103
REFERENCES 1. Booth, F. F., Br. Corros. J., 2 No. 2, 55 (1967) R. D. J., Br. Corros. J., 2 No. 2, 61 (1967) 3. Layton, D. N. and White, P. E., Br. Corros. .I. 2, No. 2, 65 (1967) 4. Evans, U. R., The Corrosion and Oxidation ofMetals, 1st Suppl. Edward Arnold, London ( 1968) 5 . Layton, D. N. and White, P. E., Br. Corros. J.. 1 No. 6, 213 (1966) 6. Discussion, Br. Corros. J., 2 No. 2, 71 (1967) 7. Allen, B. M., Soldering Handbook, Iliffe, London (1969) 8. Collard Churchill, S., Brazing, The Machinery Publ. Co., London (1963) 9. Sloboda, M. H., Czechoslovak Conf. on Brazing, 18 (1969) 10. Jarman, R. A., Myles, J. W. and Booker, C. J. L., Br. Corros. J., 8 No. I , 33 (1973) and Linekar, G. A. B.. Jarman, R. A. and Booker, C. J . L., Er. Corros. J., 10 (1975) 11. Stenerson, R. N., Welding J., 48 No. 6, 480 (1969) 12. Upton, B., Br. Corros. J., 1 No. 7, 134 (1966) 13. Gooch, T. G. and Gregory, E. N., Br. Corros. J , , Suppl. issue, ‘Design of Protective Systems for Structural Steelwork’, 48 (1968) 14. Baker, R. G. and Whitman, J. G., Br. Corros. J., 2 No. 2, 34 (1967) 15. Hoar, T. P., Er. Corros. J., 2 No. 2, 46 (1967) 16. Discussion, Br. Corros. J., 2 No. 3, 49 (1967) 17. Lundin, S., Weldability Questions and Cracking Problems, ESAB, Goteborg, 2 (1963) 18. Bradley, J. N. and Rowland, J. C., Er. Weld. J., 9 No. 8, 476 (1962) 19. Slimmon, P. R., Welding J . , 47 No. 12, 954 (1968) 20. Pinnow, K. E. and Moskowitz, A., Welding J., 49 No. 6, 278 (1970) 21. Gooch, T. G. et al., W.I. Res. Bull., 12 No. 2, 33 (1971) and W.I. Res. Bull., 12 No. 5 , 135 (1971) 22. France, W. D. and Greene, N. D., Corrosion Science, 8, 9 (1968) 23. Streicher, M. A., Corrosion Science, 9, 53 (1969) 24. Samans, C. H., Meyer, A. R. and Tisinai, G. F., Corrosion, 22 No. 12, 336 (1966) 25. Lancaster, J. F., Metalhrgy of Joining, Chapman and Hall, London (1986) 26. Kent, K. G., Met. Revs., 15 No. 147, 135 (1970) 27. Keane, J . D. and Bigos, J., Corrosion, 16 No. 12, 601 (1960) 28. Scully, J . C., The Fundamentals of Corrosion, Pergamon Press, London (1968) 29. Turnbull, A., Corrosion Fatigue of Structural Steels in Sea Water, Reviews in Coatings and Corrosion, 5 (1-4). 43 (1982) 30. Lundin, S., Suetsaren (ESAB), No. 2, 2 (1967) 31. Saarinen. A. and Onnela, K.. Corr Sci., 10(11), 809 (1970) 32. Ousyannikov. V. Yu., Chernov, B. B.. Semenchenko, V. S. and Tyul ’Kin Yu, k., Suur. Proizuod., (4), 38 (1986) 33. Diiren, C., Herbsleb, G. and Treks, E., 3R Int.. 25(5), 246 (1986) 34. Duren, C., Treks, E. and Herbsleb, G . , Mater. Perform., 25(9), 41 (1986) 35. Drodten, P., Herbsleb, G. and Schwenk, W., Steel Res., 60(10), 471 (1989) 36. Miller, K. J., Fat. Eng. Materials and Slrucrures, 5(3), 223 (1982) 37. Burns, D. J. and Vosikowsky, 0..Time-Dependent Fracture, Ed Krauz, A. S., Martinus Nijhoff, Dordrecht, 53 (1985) 38. Nibbering, J. J. W., Buisman, B. C., Wildschut, H. and Rietbergen, E., Lasterchnick. (9). 187 (1986) 39. Choi, J. K., Kim, H. P. and Pyun, S. I., Korean Inst. Mer., 24(1), 14 (1986) 40. Terasaki , F., Ohtani, H., Ikeda, A. and Nakanishi, M., Proc. Insr. Mech. Eng. A , Power Process Eng., 200(A3), 141 ( 1986) 41. Kobayasghi, Y., Ume, K., Hyodo,T. andTaira, T., Corr. Science.27(10/11), 1117 (1987) 42. Goncharov, N. C., Mazel, A. G. and Golovin, S. V.. Suar. Proizuod., (4) 21 (1986) 43. Watkins, M. and Rosenberg, E. L., Mater. Perform., 22(12) 30 (1985) 44. Pircher, H and Sussek, G., Srahl und Eisen, 102(10) 503 (1982) 45. Drodten, P., Stuhl und Eisen, 102(7), 359 (1982) 46. McMinn, A., FCGR in HAZ-Simulated A533-B steel, H.M.S.O., London, Oct. (1982) 47. Ray, G. P., Jarman, R. A. and Thomas, J. G. N.. Corr. Science, 25(3), 171 (1985) 48. Trimmer, R. M. and Jarman. R. A., Metal Constr., 15(2), 97 (1983) 49. Compton, K . G. and Turley, J. A,, Galvanic and Pitting Corrosion, ASTM STP 576, ASTM, 56 (1976) 2. Everett, L. H. and Tarleton,
9: 104
DESIGN IN RELATION TO WELDING AND JOINING
Prasad Rao, K. and Prasanna Kumar, S., Corrosion, 41(4), 234 (1985) Herbsleb, G. and Stoffelo, H., Werkstovfe Korros., 29(9). 576 (1978) Tamaki, K., Yasuda, K. and Kimura, H., Corrosion, 45(9), 764 (1989) Kajimura H., Ogawa K. and Nagano, H.. Tesfsu-fo-Hagad,75(11) 2106 (1989) Miura, M., Koso, M., Kudo, T. and Tsuge, H., Quaff.J. Jpn. Weld. Soc.. 7(1), 94 (1989) Fujiwara, K. and Tomasi, H., Corros. Eng., 37(11), 657 (1988) Angelini, E. and Zucchi. F.. Br. Corns. J.. 21(4), 257 (1986) Sridhar, N., Fasche, L. H. and Kolts, J., Mater. Perform., 23(12) 52 (1984) Grekula. A. I.. Kujanpaa. V. P. and Karjalainen. V. P.. Corrosion, 40(11), 569 (1984) Chen, J. S. and Levine, T. M., Corrosion, 45(1), 62 (1989) 60. Fenn. R. C. and Newton, C. J., Mater. Sci. Technol., 2(2). 181 (1986) 61. Edling, G., Swefsen, 38(3), 61 (1979) 62. Engelhard. G., Mattern, U.. Pellkofer. D. and Seibold, A., Welding and Cuffing (Dusseldorf), 40, 19 (1988) 63. Deverell, H. E., Mater. Perform., 24(2). 47 (1985) 64. Pazebnov. P.P.. Aleksandrov. A. G. and Gorban. V. A.. Swar. Proizwod., (6). 18 (1986) 65. Watanabe, T., El, K. and Nakamura, H., J. High Temp Soc Japan, 14(4), 185 (1988) 66. Zingales, A.. Quartarone, G. and Moretti, G., Corrosion, 41(3). 136 (1985) 67. Lipodaev, V. N., Swar Proizwod., (9,4 (1989) 68. Jarman. R. A. and Cihll, V., Meful Constr., 11(3). 134 (1979) 69. Cihll, V. and Lobl, K., Mem. Efud. Sci. Rev. Mefall., 83(2), 87 (1986) 70. Baeslack, W. A.. Weld. J.. 58(3). 83s (1979) 71. Kamachi Mudali, U., Dayal, R. K. and Rnanamoorthy, P., Werksfov.Korros., 37(12), 637 ( 1986) 72. Klueh, R. L. and King, J. F., We/d. J.. 61(9). 302s (1982) 73. Kim, Y. S. and Pyum, S. I., Br. Corros. J., 18(2), 71 (1983) 74. Cardier, H.. Metal/., 34(6), 515 (1980) 75. Patel, P. S. and Jarman, R. A., Br. Corros, J., 20(1), 23 (1985) 76. Holroyd, N. J. H. and Hardie. D., Environment Sensifive Fracium. ASTM STP 821. ASTM, 202, (1984) 77. Bartsch. E., Pracf. Metallog.. 17(7) 355 (1980) 78. Little, B. and Wagner, P., Mafer Perform., 27(8), 57 (1988) 79. Polyakov, S. C., Grigorenko. G. M., Dnoprienko. L. M. and Goncharov, A. B., Zaschch. Mer., 25(3) 419 (1989) 80. Adeeva, L. I., Grabin, V. F. and Goncharov, A. B., Avfom Swarka, (I), 25 (1989) 81. Roswell. S. C . , Mefal Constr., 104(4). 163 (1978) 82. Porter, F. C., Metal Constr., 15(10), 606 (1983) 83. McKelvie, A. N.. Metal Constr., 13(11/l2), 693 and 744 (1981) 84. Approved Prefabrication Primers and Corrosion Control Coatings, Lloyds Register of Shipping, London (1983). 85. Costos, L. P., Weld. J., 61(10), 320s (1982) 86. Yamaguchi. S., WerksfofleKorros.. 33(11), 617 (1982) 87. Kartyshow, N. G., Welding Prodn., M(9) 26 (1979) 88. Stevernazel. G., Werksfofle Technik., 12(12) p 439 (1981) 89. Kuhn, A. T. and Trimmer, R. M.. Br. Corros, J., 17(1), 4 (1982) 90. Lewis, G., Corr. Science, 20(12) 1259 (1980) 91. Takizawa. K., Nakayama, Y. and Kurokawa. K., Corros. Eng.. 38(8) 417 (1989) 92. Lugscheider, E. and Minarski, P.. Schweissen Schneiden, 41(1I). 590 (1989) 93. Suganuma. K., Nihara. K.. Fujita. T. and Okamoto, T.. Conf. Proc. AdwancedMaferials, Vol. 8. Mefal Ceramic Joints, Tokyo, 1988, MRS, Pittsburg, USA (1988) 94. Szaniewaki. S., Powloki Ochronne., 11(3), 32 (1983) 95. Hahn, F. P., Ind. Corros., 2(6) 16 (1984) 96. Bauer, 1. C. 0..Aluminium, 58(5), El46 (1982) 97. Bauer. 1. C. O., Aluminium. 58(10), E201 (1982) 98. Hoffer, K., Aluminium, 57(2), E161 (1982) 50. 51. 52. 53. 54. 55. 56. 57. 58. 59.
9.5A
Appendix - Terms Commonly Used in Joining *
Automatic Welding: welding in which the welding variables and the means of making the weld are controlled by machine. Bead: a single run of weld metal on a surface. Braze Welding: the joining of metals using a technique similar to fusion welding and a filler metal with a lower melting point than the parent metal, but neither using capillary action as in brazing nor intentionally melting the parent metal. Brazing: a process of joining metals in which, during or after heating, molten filler metal is drawn by capillary action into the space between closely adjacent surfaces of the parts to be joined. In general, the melting point of the filler metal is above 500"C, but always below the melting temperature of the parent metal. Brazing Alloy: filler metal used in brazing. Butt Joint: a connection between the ends or edges of two parts making an angle to one another of 135" to 180" inclusive in the region of the joint. Carbon Dioxide Welding: metal-arc welding in which a bare wire electrode is used, the arc and molten pool being shielded with carbon dioxide gas. Covered Filler Rod: a filler rod having a covering of flux. Deposited Metal: filler metal after it becomes part of a weld or joint. Edge Preparation: squaring, grooving, chamfering or bevelling an edge in preparation for welding. Electro-slag Welding: fusion welding utilising the combined effects of current and electrical resistance in a consumable electrode and conducting bath of molten slag, through which the electrode passes into a molten pool, both the pool and the slag being retained in the joint by cooled shoes which move progressively upwards. Electron-beam Welding: fusion welding in which the joint is made by fusing the parent metal by the impact of a focused beam of electrons. *Data extracted from BS 499: Part 1 (1965). Complete copies of this standard can be obtained from The British Standards Institution, Information Department, Linford Wood, Milton Keynes, MK 14 6 LE.
9 : 105
9 : 106
TERMS COMMONLY USED IN JOINING
Filler Metal: metal added during welding, braze welding, brazing or surfacing. Filler Rod: filler metal in the form of a rod. It may also take the form of filler wire. Flux: material used during welding, brazing or braze welding to clean the surfaces of the joint, prevent atmospheric oxidation and to reduce impurities. Fusion Penetration: depth to which the parent metal has been fused. Fusion Welding: welding in which the weld is made between metals in a molten state without the application of pressure. Fusion Zone: the part of the parent metal which is melted into the weld metal. Heat-affected Zone: that part of the parent metal which is metallurgically affected by the heat of the joining process, but not melted. Hydrogen Controlled Electrode: a covered electrode which, when used correctly, produces less than a specified amount of diffusible hydrogen in the weld deposit. Manual Welding: welding in which the means of making the weld are held in the hand. Metal-arc Welding: arc welding using a consumable electrode. MIG-welding: metal-inert gas arc welding using a consumable electrode. Oxyacetylene Welding: gas welding in which the fuel gas is acetylene and which is burnt in an oxygen atmosphere. Parent Metal: metal to be joined. Pressure Welding: a welding process in which a weld is made by a sufficient pressure to cause plastic flow of the surfaces, which may or may not be heated. Resistance Welding: welding in which force is applied to surfaces in contact and in which the heat for welding is produced by the passage of electric current through the electrical resistance at, and adjacent to, these surfaces. Run: the metal melted or deposited during one passage of an electrode, torch or blow-pipe. Semi-automatic Welding: welding in which some of the variables are automatically controlled, but manual guidance is necessary. Spatter: globules of metal expelled during welding onto the surface of parent metal or of a weld. Spelter: a brazing alloy consisting nominally of 50% Cu and 50% Zn. Submerged-arc Welding: metal-arc welding in which a bare wire electrode is used; the arc is enveloped in flux, some of which fuses to forin a removable covering of slag on the weld. TIG-welding: tungsten inert-gas arc welding using a non-consumable electrode of pure or activated tungsten. Thermal Cutting: the parting or shaping of materials by the application of heat with or without a stream of cutting oxygen. Weld: a union between pieces of metal at faces rendered plastic or liquid by heat or by pressure, or by both. A filler metal whose melting temperature is of the same order as that of the parent material may or may not be used. Welding: the making of a weld. Weld Metal: all metal melted during the making of a weld and retained in the weld. Weld Zone: the zone containing the weld metal and the heat-affected zone. R.A. JARMAN
10
CATHODIC AND ANODIC PROTECTION
10.1 Principles of Cathodic Protection 10.2 Sacrificial Anodes 10.3 Impressed-current Anodes 10.4 Practical Applications of Cathodic Protection 10.5 Stray-current Corrosion 10.6 Cathodic-protection Interaction 10.7 Cathodic-protection Instruments 10.8 Anodic Protection
10: 1
10:3 10:29 10:56 10193 10.122 10.129 10:136 10:155
I O . 1 Principles of Cathodic Protection
Cathodic protection is unique amongst all the methods of corrosion control in that if required it is able to stop corrosion completely, but it remains within the choice of the operator to accept a lesser, but quantifiable, level of protection. Manifestly, it is an important and versatile technique. In principle, cathodic protection can be applied to all the so-called engineering metals. In practice, it is most commonly used to protect ferrous materials and predominantly carbon steel. It is possible to apply cathodic protection in most aqueous corrosive environments, although its use is IargeIy restricted to natural near-neutral environments (soils, sands and waters, each with air access). Thus, although the general principles outlined here apply to virtually all metals in aqueous environments, it is appropriate that the emphasis, and the illustrations, relate to steel in aerated natural environments. The text seeks to show why it is that cathodic protection is apparently so restricted in its scope of application despite its apparent versatility. Nevertheless, having recognised the restricted scope it is important to emphasise that the number and criticality of the structures to which cathodic protection is applied is very high indeed.
Historical In recent years it has been regarded as somewhat passe to refer to Sir Humphrey Davy in a text on cathodic protection. However, his role in the application of cathodic protection should not be ignored. In 1824 Davy presented a series of papers to the Royal Society in London' in which he described how zinc and iron anodes could be used to prevent the corrosion of copper sheathing on the wooden hulls of British naval vessels. His paper shows a considerable intuitive awareness of what are now accepted as the principles of cathodic protection. Several practical tests were made on vessels in harbour and on sea-going ships, inciuding the effect of various current densities on the level of protection of the copper. Davy also considered the use of an impressed current device based on a battery, but did not consider the method to be practicable. 10: 3
10:4
PRINCIPLES OF CATHODIC PROTECTION
The first ‘full-hull’ installation on a vessel in service was applied to the frigate HMS Sumarung in 1824. Four groups of cast iron anodes were fitted and virtually perfect protection of the copper was achieved. So effective was the system that the prevention of corrosion of the copper resulted in the loss of the copper ions required to act as a toxicide for marine growth leading to increased marine fouling of the hull. Since this led to some loss of performance from the vessel, interest in cathodic protection waned. The beneficial action of the copper ions in preventing fouling was judged to be more important than preventing deterioration of the sheathing. Cathodic protection was therefore neglected for 100 years after which it began to be used successfully by oil companies in the United States to protect underground pipelines’. It is interesting that the first large-scale application of cathodic protection by Davy was directed at protecting copper rather than steel. It is also a measure of Davy’s grasp of the topic that he was able to consider the use of two techniques of cathodic protection, viz. sacrificial anodes and impressed current, and two types of sacrificial anode, viz. zinc and cast iron.
Electrochemical Principles Aqueous Corrosion
The aqueous corrosion of iron under conditions of air access can be written: 2Fe
+ O2+ 2H,O
+
2Fe(OH),
(10.1)
The product, ferrous hydroxide, is commonly further oxidised to magnetite (Fe,O,) or a hydrated ferric oxide (FeOOH), i.e. rust. It is convenient to consider separately the metallic and non-metallic reactions in equation 10.1: 2Fe
0,
+
2Fe2++ 4e-
+ 2H,O + 4e- -,40H-
(10.2)
(10.3)
To balance equations 10.2 and 10.3 in terms of electrical charge it has been necessary to add four electrons to the right-hand side of equation 10.2 and to the left-hand-side of equation 10.3. However, simple addition and rationalisation of equations 10.2 and 10.3 yields equation 10.1. We conclude that corrosion is a chemical reaction (equation 10.1) occurring by an electrochemical mechanism (equations 10.2) and (10.3), i.e. by a process involving electrical and chemical species. Figure 10.1 is a schematic representation of aqueous corrrosion occurring at a metal surface. Equation 10.2, which involves consumption of the metal and release of electrons, is termed an anodic reaction. Equation 10.3, which represents consumption of electrons and dissolved species in the environment, is termed a cathodic reaction. Whenever spontaneous corrosion reactions occur, all the electrons released in the anodic reaction are consumed in the cathodic reaction; no excess or deficiency is found. Moreover, the metal normally takes up a more or less uniform electrode potential, often called the corrosion or mixed potential (Ecorr).
PRINCIPLES OF CATHODIC PROTECTION
Fez+
Environment
20H-
10:5
Fez+
r t
Fig. 10.1 Schematic illustration of the corrosion of steel in an aerated environment. Note that the electrons released in the anodic reaction are consumed quantitatively in the cathodic reaction, and that the anodic and cathodic products may react to produce Fe(0H)z
Cathodic Protection
It is possible to envisage what might happen if an electrical intervention was made in the corrosion reaction by considering the impact on the anodic and cathodic reactions. For example, if electrons were withdrawn from the metal surface it might be anticipated that reaction 10.2 would speed up (to replace the lost electrons) and reaction 10.3 would slow down, because of the existing shortfall of electrons. It follows that the rate of metal consumption would increase. By contrast, if additional electrons were introduced at the metal surface, the cathodic reaction would speed up (to consume the electrons) and the anodic reaction would be inhibited; metal dissolution would be slowed down. This is the basis of cathodic protection. Figure 10.2 shows the effect on the corrosion reaction shown in Fig. 10.1 of providing a limited supply of electrons to the surface. The rate of dissolution slows down because the external source rather than an iron atom provides two of the electrons. Figure 10.3 shows the effect of a greater electron supply; corrosion ceases since the external source provides all the requisite electrons. It should be apparent that there is no reason why further electrons could not be supplied, when even more hydroxyl (OH -) ion would be produced, but without the possibility of a concomitant reduction in the rate of iron dissolution. Clearly this would be a wasteful exercise. The corrosion reaction may also be represented on a polarisation diagram (Fig. 10.4). The diagram shows how the rates of the anodic and cathodic reactions (both expressed in terms of current flow, I)vary with electrode potential, E. Thus at E,, the net rate of the anodic reaction is zero and it increases as the potential becomes more positive. At E, the net rate of the cathodic reaction is zero and it increases as the potential becomes more negative. (To be able to represent the anodic and cathodic reaction rates on the same axis, the modulus of the current has been drawn.) The two reaction rates are electrically equivalent at E,,,,, the corrosion potential, and the
10:6
PRINCIPLES OF CATHODlC PROTECTION
02 + 2H20
I
02 + 2H2O
Environment
Metal
,1 4 e From external source Fig. 10.3 Schematic illustration of full cathodic protection of steel in an aerated environment. Note that both anodic reactions shown in Fig. 10.1 have now been annihilated and there is an accumulation of OH- at the surface
corresponding current, lCorr is an electrical representation of the rate of the anodic and cathodic reactions at that potential, Le. the spontaneous corrosion rate of the metal. That is, at E,,,, the polarisation diagram represents the situation referred to above. Namely, that when spontaneous corrosion occurs, the rate of electron release equals the rate of electron consumption, and there is no net current flow although metal is consumed, and meanwhile the metal exerts a single electrode potential. To hold the metal at any potential other than E,,,, requires that electrons be supplied to, or be withdrawn from, the metal surface. For example, at E ' the cathodic reaction rate, I;, exceeds the anodic reaction rate, I;, and the latter does not provide sufficient electrons to satisfy the former. If the
10:7
PRINCIPLES OF CATHODIC PROTECTION
--Fez+
+ 2e
E
+ 2H20 + 4e-40H-
log I I I Fig. 10.4 Polarisation diagram representing corrosion and cathodic protection. A corroding metal takes up the potential Eco,, spontaneously and corrodes at a rate given by I,,,,. If the potential is to be lowered to E' a current equal to (11i1 - 16)must be supplied from an external source; the metal will then dissolve at a rate equal to 1:
metal is to be maintained at E', the shortfall of electrons given by ( I I,'I - I:) must be supplied from an external source. This externally supplied current serves to reduce the metal dissolution rate from I,,,, to Z,l. At E, the net anodic reaction rate is zero (there is no metal dissolution) and a cathodic current equal to I," must be available from the external source to maintain the metal at this potential. It may also be apparent from Fig. 10.4 that, if the potential is maintained below E,, the metal dissolution rate remains zero (I, = 0), but a cathodic current greater than I:must be supplied; more current is supplied without achieving a benefit in terms of metal loss. There will, however, be a higher interfacial hydroxyl ion concentration. Oxygen Reduction
In illustrating the corrosion reaction in equation 10.1, the oxygen reduction reaction (equation 10.3) has been taken as the cathodic process. Moreover, in Figs 10.1 to 10.4 oxygen reduction has been assumed. Whilst there is a range of cathodic reactions that can provoke the corrosion of a metal (since to be a cathodic reactant any particular species must simply act as an oxidising agent to the metal) the most common cathodic reactant present in natural environments is oxygen. It is for this reason that the oxygen reduction reaction has been emphasised here. When corrosion occurs, if the cathodic reactant is in plentiful supply, it can be shown both theoretically and practically that the cathodic kinetics are semi-logarithmic, as shown in Fig. 10.4. The rate of the cathodic reaction is governed by the rate at which electrical charge can be transferred at the metal surface. Such a process responds t o changes in electrode potential giving rise to the semi-logarithmic behaviour.
10:8
PRINCIPLES OF CATHODIC PROTECTION
Because oxygen is not very soluble in aqueous solutions (ca. 10 ppm in cool seawater, for example) it is not freely available at the metal surface. As a result it is often easier to transfer electrical charge at the surface than for oxygen to reach the surface to take part in the charge transfer reaction. The cathodic reaction rate is then controlled by the rate of arrival of oxygen at the surface. This is often referred to as mass transfer control. Since oxygen is an uncharged species, its rate of arrival is unaffected by the prevailing electrical field and responds only to the local oxygen concentration gradient. If the cathodic reaction is driven so fast that the interfacial oxygen concentration is reduced to zero (Le. the oxygen is consumed as soon as it reaches the surface), the oxygen concentration gradient to the surface reaches a maximum and the reaction rate attains a limiting value. Only an increase in oxygen concentration or an increase in flow velocity will permit an increase in the limiting value. The cathodic kinetics under mass transfer control will be as shown in Fig. 10.5.
Fe -+Fe*+ + 2e
E
log I I I
Fig. 10.5 Polarisation diagram representing corrosion and cathodic protection when the cathodic process is under mass transfer control. The values of E,,,. and I,,,, are lower than when there is no mass transfer restriction, Le. when the cathodic kinetics follow the dotted line
Figure 10.5 demonstrates that, even when semi-logarithmic cathodic kinetics are not observed, partial or total cathodic protection is possible. Indeed, Fig. 10.5 shows that the corrosion rate approximates to the limitand the current required for protecing current for oxygen reduction (I,im) tion is substantially lower than when semi-logarithmic cathodic behaviour prevails. Hydrogen Evohtion
In principle it is possible for water to act as a cathodic reactant with the formation of molecular hydrogen:
10:9
PRINCIPLES OF CATHODIC PROTECTION
+
(10.4) 2 H 2 0 2e- H, + 2 0 H Indeed, in neutral solutions the corrosion of iron with concomitant hydrogen evolution deriving from the reduction of water is thermodynamically feasible. In practice, this cathodic reaction is barely significant because the reduction of any oxygen present is both thermodynamically favoured and kinetically easier. In the absence of oxygen, the hydrogen evolution reaction at the corrosion potential of iron is so sluggish that the corrosion rate of the iron is vanishingly small. Nevertheless, hydrogen evolution is important in considering the cathodic protection of steel. Although hydrogen evolution takes little part in the corrosion of steel in aerated neutral solutions (see Fig. l0.6), as the potential is lowered to achieve cathodic protection so it plays a larger, and eventually dominant, role in determining the total current demand. This too is demonstrated in Fig. 10.6 where, it must be remembered, the current supplied from the external source at any potential must be sufficient to sustain the total cathodic reaction, i.e. both oxygen reduction and hydrogen evolution reactions at that potential. It will be seen that to lower the potential much below E, entails a substantial increase in current and significantly more hydrogen evolution.
E
-+
A 02 + 2 H z 0 + 4e-
40H-
Fe + Fe2+ + 2e
E 2H20 + 2e
-+ 20H-
+H2
Fig. 10.6 Polarisation diagram showing the limited role hydrogen evolution plays at the corrosion potential of steel in aerated neutral solution, the larger role in determining cathodic protection currents and the dominant role in contributing to current requirements at very negative potenitals. The dotted line shows the total cathodic current due to oxygen reduction and hydrogen evolution
Methods of Applying Cathodic Protection There are two principal methods of applying cathodic protection, viz. the impressed current technique and the use of sacrificial anodes. The former includes the structure as part of a driven electrochemical cell and the latter includes the structure as part of a spontaneous galvanic cell.
,
10: 10
PRINCIPLES OF CATHODIC PROTECTION
+TI-
El~tr~nfiow
source
Corrosive environment
~
Impressed current anode in ground bed
Positive current flow (ionic)
Protected structure
Fig. 10.7 Schematic diagram of cathodic protection using the impressed-current technique
Impressed Current ~ ~ t h ~ d
Figure 10.7 illustrates the use of an external power supply to provide the cathodic polarisation of the structure. The circuit comprises the power source, an auxiliary or impressed current electrode, the corrosive solution, and the structure to be protected. The power source drives positive current from the impressed current electrode through the corrosive solution and onto the structure. The structure is thereby cathodically polarised (its potential is lowered) and the positive current returns through the circuit to the power supply. Thus to achieve cathodic protection the impressed current electrode and the structure must be in both electrolytic and electronic contact. The power supply is usually a transformerhectifier that converts a.c. power to d.c. Typically the d.c. output will be in the range 15-l0OV and 5-100 A although 200 V/200A units are not unknown. Thus fairly substantial driving voltages and currents are available. Where mains power is not available, diesel or gas engines, solar panels or thermoelectric generators have all been used to provide suitable d.c. It will be seen that the impressed current electrode discharges positive current, i.e. it acts as an anode in the cell. There are three generic types of anode used in cathodic protection, viz, consumable, non-consumable and semi-consumable. The consumable electrodes undergo an anodic reaction that involves their consumption. Thus an anode made of scrap iron produces positive current by the reaction: Fe -+ Fez+ + 2ethe ferrous iron entering the environment as a positive current carrier*. Since the dissolving anode must obey Faraday’s law it follows that the wasting of the anode will be proportional t o the total current delivered. * In practice the cathodic protection current will be carried in the corrosive environment by more mobile ions, e.g. O H - , N a + , etc.
PRINCIPLES OF CATHODIC PROTECTION
1O:ll
In practice the loss for an iron anode is approximately 9 kg/Ay. Thus consumable anodes must be replaced at intervals or be of sufficient size to remain as a current source for the design life of the protected structure. This poses some problems in design because, as the anode dissolves, the resistance it presents to the circuit increases. More important, it is difficult to ensure continuous electrical connection to the dissolving anode. Non-consumable anodes sustain an anodic reaction that decomposes the aqueous environment rather than dissolves the anode metal. In aqueous solutions the reaction may be: 2H,O
-+
0,+ 4H'
or in the presence of chloride ions: 2C1- --t CI,
+ 4e-
+ 2e-
Anodes made from platinised titanium or niobium fall in this category. Because these anodes are not consumed faradaically, they should not, in principle, require replacement during the life of the structure. However, to remain intact they must be chemically resistant to their anodic products (acid and chlorine) and, where the products are gaseous, conditions must be produced which allow the gas to escape and not interfere with anode operation. This is particularly true of the platinised electrodes because they can operate at high current density (> 100 A/m2) without detriment, but will then produce high levels of acidity (pH 300 mV when current applied. Positive shift 100 mV when current interrupted. More negative than beginning of Tafel segment of cathodic polarisation ( E - log I ) curve. 5 . A net protective current in the structure at former anodic points. 6. Polarise all cathodic areas to open circuit potential of most active anode areas. 1. 2. 3. 4.
50 mV) between these two potentials will call into question the suitability of the anode in the particular environment. The protection potential refers to the potential at which experience shows corrosion of a metal will cease. Different materials have different protection potentials (Table 10.6). Occasionally a less negative protection potential will be specified because some degree of corrosion is permissible. It should be noted that in a mixed metal system the protection potential for the most base metal is adopted. Table 10.6
Protection potentials of metals in seawater (V vs. Ag/AgCl/seawater)
Iron and steel aerobic environment anaerobic environment Lead Copper alloys
-0.8 -0.9 -0.55 -0-45 to -0.60
The driving voltage is the difference between the anode operating potential and the potential of the polarised structure to which it is connected. For design purposes, the driving voltage is taken as the difference between the anode operating potential and the required protection potential of the structure.
SACRIFICIAL ANODES
10:31
Anode Capacity and Anode Efficiency
The anode capacity is the total coulombic charge (current x time) produced by unit mass of an anode as a result of electrochemical dissolution. It is normally expressed in ampere hours per kilogram (Ah/kg) although the inverse of anode capacity, Le. the consumption rate (kg/Ay) is sometimes used. The theoretical anode capacity can be calculated according to Faraday’s law. From this it can be shown that 1 kg of aluminium should provide 2981 Ah of charge. In practice, the realisable capacity of the anode is less than the theoretical value. The significance of the actual (as opposed to the theoretical) anode capacity is that it is a measure of the amount of cathodic current an anode can give. Since anode capacity varies amongst anode materials, it is the parameter against which the anode cost per unit anode weight should be evaluated. The anode efficiency is the percentage of the theoretical anode capacity that is achieved in practice: anode efficiency =
anode capacity x 100% theoretical capacity
Anode efficiency is of little practical significance and can be misleading. For example, magnesium alloy anodes often have an efficiency ca. 50% whilst for zinc alloys the value exceeds 90%; it does not follow that zinc alloy anodes are superior to those based on magnesium. Efficiency will be encountered in many texts on sacrificial anode cathodic protection. Anode Requirements
The fundamental requirements of a sacrificial anode are to impart sufficient cathodic protection to a structure economically and predictably over a defined period, and to eliminate, or reduce to an acceptable level, corrosion that would otherwise take place. In view of the above criteria, the following properties are pre-requisites for the commercial viability of a sacrificial anode: 1. The anode material must provide a driving voltage sufficiently large
to drive adequate current to enable effective cathodic polarisation of the structure. This requirement implies that the anode must have an operating potential that is more negative than the structure material to be protected. 2. The anode material must have a more or less constant operating potential over a range of current outputs. Consequently the anode must resist polarisation when current flows; the polarisation characteristics must also be predictable. 3. An anode material must have a high, reproducible and available capacity, i.e. whilst acting as an anode, it must be capable of delivering consistently and on demand a large number of ampere hours per kilogram of material spent. The ideal anode material will not passivate in the exposure environment, will corrode uniformly thus avoiding
10:32
SACRIFICIAL ANODES
mechanical fragmentation (hence wastage), and will approach its theoretical capacity. 4. The production of large quantities of alloy material in anode form, and possessing the desired mechanical properties, must obviously be practicable and economic. Thus secondary processing such as heat treatment is undesirable.
Sacrificial Anode Materials Whilst cathodic protection can be used to protect most metals from aqueous corrosion, it is most commonly applied to carbon steel in natural environments (waters, soils and sands). In a cathodic protection system the sacrificial anode must be more electronegative than the structure. There is, therefore, a limited range of suitable materials available to protect carbon steel. The range is further restricted by the fact that the most electronegative metals (Li, Na and K) corrode extremely rapidly in aqueous environments. Thus, only magnesium, aluminium and zinc are viable possibilities. These metals form the basis of the three generic types of sacrificial anode. In practice, with one minor exception (pure zinc), the commercially pure metals are unsuitable as sacrificial anode materials. This is because they fail to meet one or more of the pre-requisitesoutlined above. In each generic type of material alloying elements are added to ensure more acceptable properties. Table 10.7 provides a list of the more important anode materials by broad category, and some indication of their operating parameters. It is at once, clear that there are major differences in performance between one generic type and another. Thus the magnesium alloys have very negative operating potentials and are therefore able to provide a large driving voltage for cathodic protection; zinc and aluminium alloys are more modest in this respect. Aluminium alloys, by contrast, provide a substantial current capacity which is more than twice that avaiIabIe from the zinc and magnesium alloys. It might appear that this implies that if the driving voltage is the most important feature in a given cathodic protection system (e.g. when there is a need for short-term high currents or a high resistivity to overcome) then magnesium alloys are to be preferred, but if a high capacity is required (e.g. steady delivery of current over a long life) aluminium alloys would be better. In practice, selection is significantly more complicated and the topic is discussed in more detail in later sections. Table 10.7 Anode potentials of various alloys used for cathodic protection Anode potential
Max current cupucity
(V vs. Ag/AgCl/Seawater)
(Ahlkg)
A-Zn-Hg A-Zn-Sn A-Zn-In
-1.0 to -1.05 -1.0 to -1.10 -1.0 to -1.15
2 830 2 600 2 700
Zn-Al-Cd
-1.05
780
Mg-Mn
-1.7 -1.5
1 230
Alloy
Mg-A-Zn
1230
SACRIFICIAL ANODES
10:33
Even within a generic type of alloy there are significant performance differences. Thus, for example, Al-Zn-In alloys provide a higher driving voltage but a lower current capacity than Al-Zn-Hg alloys. Once a decision to use a generic type of alloy has been made, these apparently small differences in performance become important in the final selection. This subject is also discussed below. Alloying additions are made to improve the performance of an anode material. Of equal importance is the control of the levels of impurity in the final anode, since impurities (notably iron and copper) can adversely affect anode performance. Thus careful quality control of the raw materials used and the manufacturing process adopted is essential to sound anode production. This too is discussed below. An intimate knowledge of the factors influencing the operation of sacrificial anodes and design parameters, is essential if a full appreciation of how best to select an anode and achievement of optimum performance is to be realised. The following considerations deal with those factors which ultimately determine anode performance.
Factors Affecting Anode Performance AIIOy Composition
The constituent elements of anode materials, other than the basis metal, are present whether as a result of being impurities in the raw materials or deliberate alloying additions. The impurity elements can be deleterious to anode performance, thus it is necessary to control the quality of the input materials in order to achieve the required anode performance. Since this will usually have an adverse impact on costs it is often desirable to tolerate a level of impurities and to overcome their action by making alloying additions. Alloying elements may also be added for other reasons which are important to anode production and performance. These matters are discussed in this section. In general, sacrificial anode alloy formulations are proprietary and covered by patents. The patent documents are often very imprecise where they relate to compositions that will produce effective anodes and quite inaccurate in ascribing the function to a given alloying element. Whilst the commercial literature is more specific where it relates to compositions, it rarely details the purpose of an alloying addition. In discussing alloy composition here, the treatment derives from the technical literature and can only be a broad-brush account. This is because the laboratory work reported in the literature has tended to be more empirical than scientific, being directed towards producing viable anodes. Impurities
Zinc is relatively low in the electrochemical series and is widely regarded as an active metal. However, when high-purity zinc is placed in hydrochloric acid it will dissolve extremely slowly, if at all. It may be encouraged to
10 :34
SACRIFICIAL ANODES
dissolve by placing it in contact with platinum metal. Hydrogen evolution will occur vigorously on the platinum and the zinc will dissolve freely. Zinc proves to be a poor cathode for hydrogen evolution and cannot, therefore, easily support the cathodic reaction that would lead to its own corrosion. The platinum provides the surface on which this cathodic reaction easily occurs. If, by contrast, the zinc is of commercial quality it will dissolve readily in the acid. This is because the impurities in the zinc provide the cathodic sites for hydrogen evolution which allows the zinc to corrode. One important feature of an anode alloy is that it should dissolve with a capacity approaching the theoretical value. That is, all the electrons released by the metal dissolving should be transferred to the structure to support the cathodic reaction there, and should not be wasted in local cathodic reactions on its own surface. In other words an anode should act like the pure zinc described above (Le. only dissolve when attached to a good cathode) rather than impure zinc. In all the generic types of sacrificial anode alloys the presence of iron is found to be deleterious. This is because an intermetallic compound formed between it and the basis metal proves to be a good cathode. Its presence will result in a substantial lowering of the capacity of an anode. Moreover, the presence of this cathodic material will often raise (make less negative) the anode operating potential and may, in the limit, promote actual passivation. Thus the driving force available from the anode is reduced or completely destroyed. For example, when the solid solubility of iron in zinc (ca. 14 ppm) is exceeded the anode operating potential becomes more positive. This has been attributed to the formation of Zn(OH),, around intermetallic precipitates of FeZn,,’. The presence of iron has a similar adverse effect in aluminium’ and magnesium alloys’. There are two ways of avoiding the iron problem: to control the iron added with the basis metal or to sequester the iron in some way to render it ineffective. In practice it is not possible to permit more than a limited iron content because sequestering is only economic and practicable within defined limits. It has been seen that iron has an adverse effect because it forms a second phase (insoluble) material in the alloy which acts as an effective local cathode. Sequestering is the technique of adding an alloying addition that will cause an alternative intermetallic compound with iron to form. This compound might form a dross to be removed mechanically. Alternatively the new intermetallic compound could be a less effective cathode in which case removal would not be necessary. Both silicon and aluminium are added to zinc to control the adverse effects of iron. The former forms a ferro-silicon dross’ (which may be removed during casting). Aluminium forms an intermetallic compound which is less active as a cathode than FeZn,;. Simflarly in aluminium and magnesium alloys, manganese is added to control the iron’”. Thus in aluminium alloys for example, the cathodic activity of, FeAl, is avoided by transformation of FeAl, to (Fe, Mn)Al:. This material is believed to have a corrosion potential close to that of the matrix and is, therefore, unable to produce significant cathodic activity”. There will be an upper limit on the level of impurity that can be overcome by alloying additions. The addition of manganese is not effective in
10:35
SACRIFICIAL ANODES
*.
Al-Zn-In alloys if the iron content exceeds 0.22% Equally there may be a limit on the level of alloying addition. This may be related to the absolute level of alloying addition present or to the permissible ratio between it and the impurity element. For example, as Fig. 10.13 shows, a progressive increase in the Mn:Fe ratio in an Al-Zn-In anode increases the capacity quite markedly, but once the 1 :1 ratio is reached an even more dramatic fall is found.
-26001 u)
5 f 2400
4
vi c
c
2 5 2000
1800 I
I
0:5
I 1 :o
I 1:5
Mn: Fe ratio Fig. 10.13 The effect of the Mn/Fe ratio on the performance of AI-Zn-In-Mn anode alloys. Alloy composition range: Zn 4.0-4.6%; In 0.020-0-029%: Mn 0.004-0.35%; Fe 0.060.30% (after Klinghoffer and Linder*)
Other heavy metal impurities (especially copper and nickel) have similar adverse effects on all generic alloy types. In their case sequestering has not proved successful and control of input quality is used to keep their concentration acceptably low9. Table 10.8 outlines the quality requirements of the basis, or primary, metal for the three generic types of anode. These are the qualities required even when sequestering is also adopted. It will be seen that two grades are listed in the case of aluminium. This is because certain patented formulations permit the lower (99.8%) grade material providing that the iron and silicon are within the limit given. Alloying Additions
We have seen that the adverse effect of impurities can, within limits, be controlled by alloying additions. Thus silicon and aluminium are added to zinc, and manganese to aluminium and magnesium, to counter the effect of iron. Additions are made for other purposes, all of which aim to improve the performance of the anode. These include lowering the anode operating
10: 36
SACRIFICIAL ANODES
Table 10.8 Suitable primary material quality requirements 99.90% Magnesium Cu 0.02 rnax Mn 0.01 rnax Sn 0.01 max Ni 0.001 max Pb 0.01 max others 0.05 max 99-90 min Mg
99.99% Zinc Pb 0.003max Cu 0.001 rnax Cd 0.003max Fe 0.002max Sn 0.001 max Zn 99.99 min
99.80% Aluminium Fe 0.12 max Si 0.08 max Cu 0.03 rnax Zn 0.03 max Mn 0.02 max Mg 0.02 max A1 99.80 min
99.99% Aluminium Pb 0.003max Cu 0.002max Cd 0.003max Fe 0-003 max Sn 0.001 max A1 99.99 min
potential to increase the driving voltage, avoiding passivation, increasing the anode capacity, improving the dissolution morphology, modifying the mechanical properties of the dissolution product to promote detachment, and improving the mechanical properties of the anode. Table 10.9 lists some common zinc anode alloys. In three cases aluminium is added to improve the uniformity of dissolution and thereby reduce the risk of mechanical detachment of undissolved anode material Cadmium is added to encourage the formation of a soft corrosion product that readily crumbles and falls away so that it cannot accumulate to hinder dissolution'. The Military Specification material was developed to avoid the alloy passivating as a result of the presence of iron9. It later became apparent that this material suffered intergranular decohesion at elevated temperatures (>50°C) with the result that the material failed by fragmentation". The material specified by Det Norske Veritas was developed to overcome the problem: the aluminium level was reduced under the mistaken impression that it produced the problem. It has since been shown that decohesion is due to a hydrogen embrittlement mechanismL4and that it can be overcome by the addition of small concentrations of t i t a n i ~ m ' ~ It. is not clear whether
'.
Table 10.9 Standard zinc alloys Alloy
ASTM
component WVO)
8418-88 TYP I
Al Cd Fe cu
Pb Si Others (total) Zn Operating potential (V vs. Ag/AgCl/seawater) Caeacitv (Ah/kn)
0~10-0~50 0*025-0*07 0.005 max 0.005 max 0.006 max
ASTM 8418-88 Type II
US Mili2
0-005 rnax
0.10 -0.50 0.025-0.07 0.005 rnax
Spec. A 18001 J
DnV Re~omrn'~ for elevated temp.
0.10 remainder
0.10-0.20 0.03-0.06 0-002max 0.005 max 0.006 max 0,125 max remainder
remainder
0.003 max 0.0014 max 0-002 rnax 0-003 max remainder
-1.05
-1.05
-1.05
-1.05
780
780
780
780
-
0.1
0.005 rnax 0.006 rnax
-
10:37
SACRIFICIAL ANODES
the titanium acts as a getter for hydrogen or simply serves to refine the grains and increase the grain boundary area thereby diluting the embrittlement effect. It is claimed that newly developed alloys with magnesium additions are also resistant to intergranular attack at elevated temperatures’*’’. Although aluminium is a base metal, it spontaneously forms a highly protective oxide film in most aqueous environments, i.e. it passivates. In consequence, it has a relatively noble corrosion potential and is then unable to act as an anode to steel. Low level mercury, indium or tin additions have been shown to be effective in lowering (i.e. making more negative) the potential of the aluminium; they act as activators (depassivators). Each element has been shown to be more effective with the simultaneous addition of zinc16. Zinc additions of up to 5% lower the anode operating potential, but above this level no benefit is gained’. Below 0.9% zinc there is little influence on the performance of aluminium anodes’. Table 10.10 lists a number of the more common commercial alloys. Table 10.10 Proprietary aluminium anode materials Alloy
component (wt%)
AI-Zn-In
Fe Si Zn Hg In Sn Mg cu Mn Ti Others (each) A1 Operating Potential (V vs Ag/AgCl/seawater) Capacity (Ahlkg)
0.12 max 0.05-0.20 2’8-6’5 0.0 1-0.02
0.006 max 0.02 max remainder
AI-Zn-In(-Mn-Mg)‘7
0.18 max 0.01-0.02 2.0-6.0
AI-Zn-Sn
AI-Zn-Hg
0.13 max 4.0-5.0
0.08 max 0.11-0.21 0.35-0.50
-
0 01-0.03
0.1-2.0 0.01 maw 0.1-0.2 0.02 max
remainder
0.1 0.01 max
0-35-0.50
0.006 max remainder remainder
-1.10
-1.10
-1.10
-1.05
2 700 max
2 700 max
variable
2 830 max
The best capacities in seawater are obtained from alloys containing zinc and mercury, but this is achieved at the expense of a somewhat more noble operating potential. Zinc and indium additions give a less noble operating potential but are associated with a lower capacity. In practice this effect on the operating potential can be quite significant. The driving voltage between Al-Zn-Hg (operating potential - 1a 0 5 V) and steel (protection potential -0.80 V) is 0-25 V. The use of Al-Zn-In provides a 20% increase in driving voltage and thereby the possibility of a higher current output. Thus, both alloys have important advantages and disadvantages. However, the toxic nature of mercury may prohibit its use in rivers or harbour waters. The Al-Zn-Sn alloys require careful heat treatment in their production. Inevitably this leads to more expense and inconvenience. The advent of the alloys containing mercury or indium rendered these alloys very much less attractive. Presently Al-Zn-Hg alloys are under some pressure because
10:38
SACRIFICIAL ANODES
of the toxicity of mercury. As a result there has been a decline in their use as compared with the Al-Zn-In alloys. Improved capacity has been reported in saline mud environments by making magnesium additions (0.1-2.OVo) to Al-Zn-In alloys These materials can age harden and hence suffer reduced ductility. Since this can subsequently lead to longitudinal cracking of the anodes they should not be cast in thin sections”. Higher levels (up to 8%), whilst improving strength and casting characteristics, incur the disadvantage of a reduced capacity’. Both titanium and boron can be added as grain refiners to ensure small grain size and hence high surface area grain boundaries’’. This reduces the risk of preferential attack at grain boundaries and promotes more uniform dissolution. Typical proprietary magnesium anode materials are given in Table 10.1 1. Magnesium anodes comprise two distinct types, the Mg-Mn and Mg-Al-Zn alloy systems. Additions of up to 1 - 5 % manganese to high-purity magnesium yields a material with an operating potential of -1 -7V vs. Ag/AgCl/ seawater. The Mg-Mn alloys therefore exhibit very high driving potential and thus find application in particularly resistive environments. Mg-Al-Zn anodes have an operating potential ( - 1 5 V vs. Ag/AgCl/seawater) 200 mV above that of the Mg-Mn alloys. This is very favourable in view of problems with overprotection. Thus they are more popular in typical environments than the Mg-Mn alloys. The alloys also contain manganese which is added to overcome the deleterious effects of iron’. Alloying additions of aluminium, zinc and manganese to magnesium serve to improve the anode capacity and reduce the operating potential, compared with that of pure magnesium’. There is however no difference between the capacity of Mg-Mn and Mg-Al-Zn anodes.
”.
-
Table 10.11 Proprietary magnesium alloys Mg-Mn No. 1
cu AI Si Fe Mn Ni Zn Others (each) Mg Operating potential (V vs. Ag/AgCl/seawater) Capacity (Ah/kg)
0.02 0.01 max
-
0.03 0.5-1.3 0.001 0.01 rnax
remainder
-I * 7 1 230
Mg-Mn No. 2
Mg- AI-Zn
0.02 max 0.05 max 0.05 rnax 0.03 max
0.08 rnax 5.3-6.7 0 . 3 max 0.005 rnax 0.25 min O a I 3 max
0.5-1’5
0.03 max 0-03max remainder -1 -7 1 230
2.5-3.5 0.03 max remainder -1 -5 1 230
Many more exotic compositions for anode materials are often encountered in the literature. It should be appreciated, however, that continual mention in texts of these materials in no way reflects their usage or acceptance commercially as viable sacrificial anodes.
SACRIFICIAL ANODES
10: 39
Metallurgical Factors
In producing anodes, the production method must not compromise the benefits of alloy formulation. A number of undesirable anomalies can occur during production which may detract from the desired anode properties. Some of these are discussed below. A detailed account of production requirements can be found elsewhere”. Although most anodes are made by gravity casting, some are made by continuous casting or extrusion. The method of casting affects the physical structure of the anode. That is, the associated cooling process will influence the segregation of alloying constituents. In some cases it is undesirable to permit segregation since this may lead to preferential attack at grain boundaries. However, it is believed that segregation of activating elements by inverse segregation benefits the performance of some alloys. This mechanism is a suggested explanation for the mercury- and indium-rich phases found on the surfaces of aluminium anodes Is. The increased surface concentrations of these elements aid activation and are therefore beneficial. Porosity within the anode is detrimental since the weight of anode material, and hence the number of ampere hours of charge unit mass available will be less for a given shape. Moreover, it is possible that hydrolysis of dissolution products will occur in the pores. This leads to local acidity and a reduction in capacity. Necking of the interconnecting pore walls during dissolution may also result in the loss of intact anode material by fragmentation, thus reducing the anode capacity further. The inclusion of extraneous matter, as a consequence of unclean foundry practices may likewise increase the tendency to fragmentation. Cracking of anodes during casting is in many cases unavoidable due to the stresses imposed by cooling. The problem is more common in Al-Zn-Hg anodes and less common when continuous casting is used. Longitudinal cracks cannot be accepted as these will lead eventually to mechanical loss of material. A greater tolerance to transverse cracks can be exercised. For example, one quality specification permits an anode completely supported by the insert to have transverse cracks of unlimited length and depth provided that there are no more than ten cracks per anode and their width does not exceed 5 mmm. This is somewhat arbitrary but emphasises the point that cracks which threaten anode integrity are of more importance than those which lead to reduced performance. The anode material must stay firmly attached to the steel insert, which is necessary to conduct the current from the anode to the structure, throughout its design life to remain effective. Consequently surface preparation (by dry blast cleaningm) of the insert prior to casting, to ensure a sound bond with the anode material, is essential. Voids at the insert/anode material interface are undesirable as these will also affect the bond integrity. Environmental Factors
The conditions of environmental exposure play a key role in determining anode performance. Indeed, specific environments often preclude, or necessitate, the use of particular anode materials.
lo:@
SACRIFICIAL ANODES
This section is not intended to deal with those environmental factors which influence cathodic current demand (e.g. oxygen availability or the presence of calcareous deposits) but those which directly affect the performance of the anodes. Temperature is of particular importance to the performance of anodes, especially when anodes are buried. Anodes may often be used to protect pipelines containing hot products. Thus temperature effects must be considered. Figure 10.14 illustrates the effect of temperature on different anodes in hot saline mud. Al-Zn-In anodes experience greatly reduced capacity in open seawater at temperatures above 7OoCz1(down to 1200Ah/kg at 100°C) and in seabed muds in excess of 50°C21,22 (900Ah/kg at 80°C). At elevated temperatures passivation of both aluminium alloys and pure zinc can occurz3. Considerable improvement in performance (capacity, and to a lesser extent operating potential) has been claimed for a range of modified Al-Zn-In anode materials 17.
3000 Lab test free running
2500 0
z = 2000 a >
+.-
1500
-
0 m
5 1000 -
w
L
3
0
500
0
10
20
30
40
50
60
Temperature
70
80
90
100
("C)
Fig. 10.14 Capacity-temperaturerelationships for anodes covered with saline mud (after Jensen and Torleif")
Zinc anodes have also experienced problems at elevated temperatures in saline mud, suffering intergranular decohesion at approximately 7OoCz4. Later work by the same authors showed the threshold for damage to be ca. 50°C. The material is not recommended above 40°C" although special zinc based materials for temperatures exceeding 50" C have been developedI3(Iw3). Pure zinc, which does not suffer intergranular decohesion, will passivate under these conditions". It is claimed that newly developed Zn-Al-Mg anodes will perform satisfactorily at elevated temperatures " . Nevertheless Al-Zn-In anodes have been specified for operation above 50°C ' I . Further-
10:41
SACRIFICIAL ANODES
more, steps are now taken to ensure that the anode design prevents anode material being exposed to elevated temperatures under buried conditions. The presence of H,S (from bacterial activity in anaerobic saline mud, for example) can result in a significant decrease (16%) in capacity and loss of operating potential for Al-Zn-In anodes2*. Environmental resistivity and chloride content will affect anode performance. Aluminium alloy anodes require the presence of chloride ions to prevent passivation. Land-based applications generally provide insufficient chloride levels for this purpose. Consequently aluminium alloy anodes only find application in saline environments. The capacity and operating potential, of aluminium alloy anodes in particular, illustrated in Figs. 10.15 and 10.16, are dependent on the degree of salinity. With reducing salinity the anode capacity will decrease and the operating potential rise. This becomes increasingly significant below 10-20% seawater strength and is important for design in estuarine conditions. Passivation of aluminium alloy anodes as a consequence of electrolyte stagnation may occur, particularly if the anode is immersed in silt or sand; zinc performs reasonably under these conditions.
0
400
1 3
I 12
I
33
I 100
Percent of seawater strength
Fig. 10.15 Anode operatingpotential in semi-salinewater exposed for 30-38 days at 15-20°C. Current densities (mA/ft*): 400 (A); 200 (0); 80 (0) (after Schreiber and Murray”)
The capacity of an anode is dependent on the anode current density2’. To some extent it will be governed by the exposure environment but, in part, is within the control of the design. Certainly wholly unsuitable current densities can usually be avoided. At lower operating current densities some anodes exhibit reduced capacity; this is shown in Fig. 10.17. Long periods of low operating current density can lead to passivation. This may result in failure to activate when the current demand increases (as can occur with anodes on coated structures when the coating deteriorates).
10:42
SACRIFICIAL ANODES
e
1200 1150 1100 -
: s
+
8'
-
,e
: 900 c
2
c 1000 0 tu (1 m
,--
, /
0
? L 3 0
al
71
U O
2
400
Mil spec Zn
200 rnNft2
0 0
300 -
I
1
J. I
I
4 mos., lab
Ai-in-Zn-Si
1.09 v
I
20
I
40
2-yr., field
I
I 1
60 80 100
I
200
I
I
1
400 600 800
Current density, mA/ft2 Fig. 10.17 Performance of commerciallycast anodes in field trials (free-running) over wide current density range in ambient Gulf of Mexico mud. Potentials: negative volts versus Ag/AgCl. Zn: 310-350 AhAb at 40 and 80 mA/ft2 (after Schreiber and Murray'')
Selecting the Appropriate Anode Material It is desirable to choose an anode material with the lowest cost per ampere hour of current supplied. However, the choice is often governed by other constraints and becomes a compromise.
10 :43
SACRIFICIAL ANODES
Zinc
Of all the anode materials, zinc is arguably the most reliable. It has, with few exceptions, reliable electrochemical performance. These exceptions lie within the area of high temperature operation. Zinc provides the lowest driving voltage of the generic alloy types. It is therefore unsuitable in highly resistive soils, as Fig. 10.18 shows, and low salinity waters. However, an operating potential of 1 -05 V vs. AgIAgCVseawater cannot lead to overprotection which is an advantage where concerns for coating disbondment and hydrogen damage of high strength steel (> 700 MPa 1 3 ) exist.
-
70001I 6000
4000
6
\
Impressed current
3000
._ 0
v,
2000 1000
1
2 Current demand (A)
3
Fig. 10.18 The effect of soil resistivity and current demand on the choice between impressedcurrent and sacrificial anode protection (after Ashworth er o / . ~ ~ )
Zinc anodes have a poor capacity (780 Ah/kg) compared with aluminium (>2500 Ah/kg). However, zinc is not susceptible to passivation in low chloride environments or as a consequence of periods of low operating current density. The reliable operational characteristics of zinc often outweigh the apparent economic attraction of aluminium which can passivate under such conditions. Zinc anodes do not find application at temperatures in excess of 50°C. Zn-Al-Cd alloys suffer intergranular decohesion, and high purity zinc will passivate. Zinc anodes are not predominant in onshore or offshore applications, but they find considerable use under both conditions. Afurninium
The great attraction of aluminium anodes is their very high capacity, over three times that of zinc. They are attractive from a cost point of view and
1Q:44
SACRIFICIAL ANODES
also offer substantial weight savings which can be of great importance (e.g. offshore structures). Aluminium anodes comprise essentially three generic types: Al-Zn-In, Al-Zn-Hg and Al-Zn-Sn. Since Al-Zn-Sn alloys have largely been superseded, they will not be discussed further. Indium and mercury are added to aluminium to act as activators, i.e. to overcome the natural passivation of aluminium. Despite this, aluminium anodes are not suitable for low chloride environments which would lead t o passivation. These anodes are therefore not used for land-based applications (although examples of use in environments such as swamps do exist). Similarly their use in low chloride aqueous environments such as estuaries must be viewed with caution. The choice between Al-Zn-In and AI-Zn-Hg may well be influenced by their respective operating potentials and capacities. Where an additional driving voltage is required (such as in seabed mud), AI-Zn-In anodes may be preferred to ensure adequate structure polarisation. Alternatively, a lower driving potential may be acceptable where the additional capacity (and hence weight saving) is the predominant factor; this favours Al-Zn-Hg anodes. Aluminium anodes are less constant in their electrochemical characteristics than zinc. This presents no major problem provided the designer is aware of their properties. They suffer from reduced capacity and increased operating potential (and hence risk of passivation) with increasing temperatures above approximately 50°C (Fig. 10.14), decreasing salinity (Figs. 10.15 and 10.16) and decreasing operating current density (Fig. 10.17). Aluminium alloys are susceptible to thermite sparking when dropped on to rusty surfaces. Consequently their use may be subject to restrictions. For example, in ships’ tanks the weight of the anode and the height that it is suspended are strictly controlled. This is because thermite sparking is dependent on the kinetic energy of the anode. Aluminium alloy anodes based on Zn-AI-In and Zn-Al-Hg have now become the work-horse materials for seawater service. Magnesium
Magnesium anodes are of two generic types, Mg-Mn and Mg-Al-Zn. Both alloy systems have a high driving voltage and therefore find application in high resistivity environments; soils and fresh, or brackish, waters for example. The Mg-Mn alloy is useful in particularly resistive environments (up to 6 OOO ohm cm) as a result of an available driving voltage 200 mV greater than Mg-Al-Zn anode. Because magnesium is non-toxic its use is permissible in potable water systems where the conductivity is low. The high driving voltage may, however, result in overprotection. Combined with relatively poor capacity (1 230Ahlkg) and high unit cost these disadvantages mean that magnesium rarely finds application in subsea environments where alternatives are available. Despite this, Mg-AI-Zn anodes have been used in seabed mud and for rapid polarisation of structures (in ribbon form). The susceptibility of magnesium to thermite sparking when dropped onto rusty surfaces can preclude its consideration for applications involving a spark hazard, e.g. tankers carrying inflammable petroleum products.
SACRIFICIAL ANODES
10 :45
Magnesium is the predominant sacrificial anode material for onshore use.
Anode Testing Tests of sacrificial anode materials are generally conducted for three reasons: for screening (or ranking), performance information and quality control. The application of sacrificial anodes for the protection of structures requires the development of suitable anode materials for the exposure environment. Screening tests enable the rapid selection of materials which show potential as candidates for the given application. These tests may typically use a single parameter (e.g. operating potential at a defined constant current density) as a pass/fail criterion and are normally of short duration (usually hours) with test specimen weights of the order of hundreds of grams. The tests are not intended to simulate field conditions precisely. Performance testing is long term (months to years). Once a potentially attractive formulation has been determined it is used to produce detailed data on its performance and behaviour as an anode material under the anticipated exposure conditions. For this reason the test should mirror as closely as possible the expected operating conditions, or where practicable be conducted in the field. Large specimens (tens or hundreds of kilograms) may be used for these tests. Quality control tests are intended to detect produced materials which deviate from manufacturing specifications, and thus may result in questionable performance. The materials are usually subjected to spectrographic analysis which is the primary quality control check. The exposure tests are necessarily of short duration (hours or days), in which the test conditions attempt to reflect the environment of operation, for example using artificial seawater for a marine application. Since a property that is reproducible and indicative of a consistent quality anode is all that is required, there is no attempt to mirror, except in the crudest fashion, current density profiles. Test methods available are the free-running test (galvanic cell), galvanostatic test (constant current) and potentiostatic test (constant potential). These are always run in conjunction with visual examinations with particular emphasis on dissolution pattern. The critical information required from testing may include one or all of the following: tendency to passivation, anode operating potential and capacity. The tests, whilst all capable of producing information on the above, tend to be particularly suited to certain applications. For example potentiostatic testing is useful for evaluating passivation tendencies but not generally appropriate to anode capacity determination.
Cathodic Protection System Design Design Parameters
Before a satisfactory cathodic protection system using sacrificial anodes can be designed, the following information has to be available or decided upon:
10 :46
SACRIFICIAL ANODES
1. the area of the steelwork to be protected; 2. the type of coating, if any, that is to be used;
3. the cathodic current density; 4. cathodic protection system life.
Area of Steel Requiring Protection and Coating Considerations The area of bare steel to be protected is usually calculated from drawings and knowledge of the actual structure and must account for all electrically continuous steelwork exposed to the electrolyte. Steelwork not specified in drawings and subsequently overlooked is a common cause of underdesign. In practice the area is usually taken assuming the steel surfaces to be flat without corrugations, indentations or surface roughness. An allowance for uncertainties in real area is normally involved. Many structures are coated. Thus the presented area far exceeds the area of steel to be protected, which is restricted to uncoated areas and holidays in the coating. It is therefore practice to assume an arbitrary level of coating breakdown for coated areas to obtain the area of metal requiring cathodic protection: Area =
presented area x
070
breakdown
100
Of course the breakdown will vary through the life of a structure with the result that the area requiring protection will change. Various estimates of coating breakdown have been made and Table 10.12 provides one such. It will be seen that Table 10.12 assumes a rate of breakdown that varies with time. The significance of the area of steelwork is that the greater the area the greater the weight and/or area of anode material required for protection. Table 10.12 Guide to coating breakdown for offshore stru~tures’~
Coating breakdo wn (%)
Lifetime Initial
Mean
Final
10
2
20 30
2 2 2
I I5
IO 30 60 90
(years)
40
25 40
Cathodic Current Densities for Protecting Steel Examples of current density requirements for the protection of steel (to achieve a steel potential of -0.8V vs. Ag/AgCl/seawater) are given in Tables 10.13 and 10.14. It should be realised that the current demand of a structure will be influenced by, inter alia, temperature, degree of aeration, flow rate, protective scales, burial status, presence of bacteria and salinity. It is important that the correct current density requirement is assigned for design purposes. If too high a value is used the structure may be wastefully overprotected, whereas a value too small will mean that the protection system will underprotect and not achieve its design life.
10 :47
SACRIFICIAL ANODES Table 10.13 Current density used in ship hull cathodic protection design
C
a
U
Deep Water t o Deep Water
0.75
1.5
2
3
5
Design Current Density
Principally Deep Water
1.25
2.5
3
5
6
mA/sq.ft.
Occasional Scour
3
4
5
6
7
Typical Coatings
Ice Damage Frequent Scour
6
6
8
10
10
Retrofits
C A
E
.a . d
0 P,
w L
2 Table 10.14
Guidance on minimum design current densities for cathodic protection of bare steel13
Current density (mA/m2)
Area Initial
Mean
Final
North Sea (northern sector, 57-62"N)
I80
90
I20
North Sea (southern sector, up to 57"N)
I50
90
100
130 130 130 130
70 70 70 70 70 60 60 40 20
90 90 90 90 90
Arabian Gulf India Australia Brazil West Africa Gulf of Mexico Indonesia Pipelines (burial specified) Saline mud (ambient temperature)
130 110
I10 50 25
80 80
40 15
System Life Cathodic protection systems may be designed with a life of between 1 and 40 years. The greater the time of protection, the greater the mass of anode material that is required. Intermittant exposure and local conditions need to be considered also. The ballast or storage tanks of ships will experience periods of complete submergence, partial coverage and may at times be empty. Similarly, the
10 :48
SACRIFICIAL ANODES
wetted areas of offshore structures may be governed by tidal and seasonal variations. Local requirements must therefore be considered in order to achieve the optimum life of the system.
Calculating the Weight and Number of individual Anodes
Firstly, the total weight of anode required to protect the structure for its projected life is calculated. This is given by: W=
i, A I8760
c
(10.14)
where W = total mass of anode material (kg) A =structure area to be protected (m’) iav = mean structure current density demand (A/m2) 1 = design life in years (1 year = 8 760 h) C = anode capacity (Ah/kg) Obviously, the total weight of the anode material must equal or be greater than the total weight, W, calculated above. Similarly each anode must be of sufficient size to supply current for the design life of the cathodic protection system. The anodes must also deliver sufficient current to meet the requirements of the structure at the beginning and end of the system life. That is, if current demand increases (as a result of coating breakdown, for example) the output from the anodes should meet the current demands of the structure.
Anode Size and Shape
In practice there is often not an extensive range of suitable anode sizes from which to select. Economics may dictate an ‘off-the-shelf choice from a manufacturer or the anode shape may have to conform with the geometric limitations of the structure. Consequently, the choice of anode size and shape is often limited. The current output from an anode will depend on its surface area. Generally, larger anodes will have a higher current output. Anodes of the same weight but differing shape, can have different outputs because the surface area to weight ratio will not be equal for all forms. Thus, for a given weight of anode the shape will offer a degree of flexibility when considering current output. Anode Output
Anode output is the current available from the anode under the design conditions. It will depend on the shape of the anode, the resistivity of the environment, the protection potential of the structure and the anode operating potential. It is defined as:
SACRIFICIAL ANODES
I=
[E2 - 4 R
1
10:49
(10.15)
where Z = anode output (A) E, = operating potential of the anode (V) E2 = protection potential (V) R = anode resistance (ohm) The protection potential of steel in aerobic environments is taken as -0-80V (vs. Ag/AgCl/seawater). Anode Resistance
Table 10.15 lists those formulae suitable for the calculation of anode resistance, R, under submerged conditions. Similar formulae exist for buried conditions 2b. Table 10.15
Resistance formulae for submerged anodes'' of various geometries
Anode type
Slender anodes mounted at least 30cm offset from platform member.
Resistance formula
= resistivity L = length of anode r = radius of anode (for other than cylindrical shapes, r = C/27r, where C = cross section periphery).
p
Slender anodes mounted at least 30 cm offset from platform member. L crn)
Fig. 10.19 Water resistivity
Anode Life
Having calculated the resistance, and hence current output the anode life, L, is checked by calculation: MU L=IE where L M U E
(10.16)
= effective life of anode (years) = mass of single anode (kg) = utilisation factor, e.g. 0-75-0.80 for bracelet anodes = consumption rate of the anode (kg/Ay) (inverse of capacity in
suitable units) Z = anode output (A)
U is purely a function of anode geometry and is the fraction of anode material consumed when the remaining anode material cannot deliver the current required 13. Excessive anode life is of no benefit. If the calculated life is unsuitable a different anode size and/or shape should be considered. However, this may not always be possible especially for short-life, coated structures, when dimensional constraints on the anodes may be imposed. Number of Anodes
The total number of anodes, N,is calculated from: (10.17)
SACRIFICIAL ANODES
10:51
This calculation should yield a practicable number of anodes, i.e. 10 or 10 OOO anodes are both clearly unacceptable for the protection of an offshore oil production platform. N x M must be equal to, or greater than, the total weight of anode material, W, required. It is difficult to achieve both the exact current output and precise weight of anode material simultaneously. Consequently a compromise is reached, but both must at least match design requirements. A check to ensure that the anodes will deliver sufficient current to protect the structure at the end of the design life should be conducted. This entails calculating the expected anode output at the end of its life and checking that it meets the demands of the structure. Generally the output is calculated using a modified resistance based on an anode that is 90% consumed. Anode (and current) Distribution
It is evident that a greater number of anodes distributed over the structure will improve current distribution. However, aside from the unacceptable cost incurred by attaching excessive numbers of anodes, an anode must continue to function throughout the life of the structure and must, therefore, be of sufficient size to meet the design life. A very large number of heavy anodes is clearly impracticable and uneconomic. It is essential to ensure adequate current distribution such that all of the exposed structure remains protected; particularly important, for example, for the nodes of an offshore steel structure. Similarly, over-protection should be avoided. Thus, sacrificial anodes need to be distributed to ensure that the protection potential over the whole structure is achieved. A degree of flexibility in output to weight ratio from anodes can be achieved by varying the anode shape (as discussed above). This may, for example, provide a greater number of anodes with reduced output, whilst maintaining the desired anode life. Hence improved current distribution can be achieved. The proximity of the anodes to structures is also important. For example, if the sacrificial anodes are placed on, or very close to, steel pipework in soil then the output from the face of the anodes next to the steelwork can be severely limited. Alternatively, in high conductivity environments, corrosion products may build up and wedge between the anode and the structure. The resulting stresses can lead to mechanical failure of the anode. On the other hand, when anodes are located at an appreciable distance from the steelwork, part of the potential difference will be consumed in overcoming the environmental resistance between the anode and cathode. Complex computer models are now available to assist in defining the optimum anode distribution2’. The Anode Insert
The anode insert must be strong enough to support the weight of the anode and must be capable of being welded, or mechanically fixed to the cathode.
10:52
SACRIFICIAL ANODES
It should be appreciated that the attachment may be required to withstand the launching and pile driving of a steel jacket for offshore applications. Consideration must be given to the ease and speed of anode fixing, as this is a significant part of the total installation cost. The methods of fixing anodes to flat, vertical or horizontal surfaces are relatively well known and simple. The methods of fixing anodes to curved surfaces of pipelines and immersed structures are more complex, and generally require more steel insert. Figure 10.20 shows some methods of attaching anodes to curved surfaces. Figure 10.2Oe shows a pipeline coated with concrete with the anodes attached and with the anode thickness the same as that of the concrete. In practice, the coating would be brought up to the edge of the anode and cover the whole of the steel pipework.
Concrete coating,,-
,
Anode bracelet as
razed cable connection for electrical contact (e)
Fig. 10.20 Typical anode shapes and fixing methods. ( a ) Offshore stand-off anode; ( b ) Standoff anode - types of bowed core; (c) Stand-off anode, clamp fixings; ( d ) Typical tank fixing for shipping; ( e ) Bracelet anode assembly
SACRIFICIAL ANODES
10:53
Backfills for Anodes
When zinc or magnesium anodes are used for cathodic protection o n ~ h o r e ~they ~ ' , are usually surrounded by a backfill, which decreases the electrical resistance of the anode. Small anodes are usually surrounded with backfill in bags and large anodes are usually surrounded with a loose backfill during installation. The backfill prevents the anode coming into contact with the soil and suffering local corrosion thus reducing the capacity. By surrounding the anodes with a backfiI1, the combination of the anode with soil salts is reduced and this helps prevent the formation of passive films on the anode surface. The effect of the backfill is to lower the circuit resistance and thus reduce potential loss due to the environment. The additive resistances of the anode/backfill and backfilVsoi1 are lower than the single anode/soil resistance. Backfills attract soil moisture and reduce the resistivity in the area immediately round the anode. Dry backfill expands on wetting, and the package expands to fill the hole in the soil and eliminate voids. For use in high resistivity soils, the most common mixture is 75% gypsum, 20% bentonite and 5% sodium sulphate. This has a resistivity of approximately 50 ohm cm when saturated with moisture. It is important to realise that carbonaceous backfills are relevant to impressed current anode systems and must not be used with sacrificial anodes. A carbonaceous backfill is an electronic conductor and noble to both sacrificial anodes and steel. A galvanic cell would therefore be created causing enhanced dissolution of the anode, and eventually corrosion of the structure.
Other Considerations Calcareous Scale
A consequence of cathodic protection in seawater is the formation of a protective calcareous scale3'. The increased local pH at the steel surface caused by hydroxyl production (a product of the cathodic reaction) favours the deposition of a mixed scale of CaCO, and Mg(OH),. This scale is beneficial since it is protective and non-conducting, thus reducing the cathodic current density. Ensuring a high current density in the early period of operation will encourage calcareous scale deposition and thus reduce the current requirements in the long term (see Section 10.1 'Principles of Cathodic Protection'). The build-up of calcareous deposits is a complex topic. Very high current densities will not necessarily result in the most protective scale. In the extreme, hydrogen evolution may rupture the scale resulting in reduced protection. An optimum current density will exist, and this should be recognised.
10: 54
SACRIFICIAL ANODES
Combined Alloy Anodes for Rapid Structure Polarisation
New combined (or binary) alloy sacrificial anodes have been developed 32. An aluminium anode, for example, might have attached to it a short-life supplementary magnesium anode, or anodes, for quick polarisation of the structure. The overall reduction in structure current requirements is claimed to result in an anode weight saving of 35-50V0~~. Flame Sprayed Aluminium
The use of flame sprayed aluminium (FSA) with a silicon sealer paint has been applied to protect high-strength steel tension legs of a North Sea production facility”. The FSA system primarily acts as a very effective barrier coating. In addition the coating has significant anodic capability and aluminium corrosion products serve to plug coating defects. The sealer, although reducing the anodic current output, serves to increase the service life of the FSA coating. This coating system is subject to strict control of application procedures. Protection of High-Alloy Steels
High-alloy pipeline steels (e.g. austenitic-ferritic or duplex) have been used where the product stream demands materials with better corrosion resistance than carbon steel. In practice the external corrosion resistance of these materials cannot be guaranteed, so cathodic protection is employed to protect areas which may be subject to corrosion. Concern about hydrogen damage has lead to much debate regarding limits for protection potentials of high-alloy steels. However, it is thought that under normal seawater service and cathodic protection conditions, these materials will not be adversely affected provided that the microstructure has at least 40% austenite present34. This latter point is of particular importance to welds and their heat affected zone where careful control of heat input is necessary to maintain a favourable microstructure. The latter part of this chapter has dealt with the design considerations for a sacrificial anode cathodic protection system. It has outlined the important parameters and how each contributes to the overall design. This is only an introduction and guide to the basic principles cathodic protection design using; - sacrificial anodes and should be viewed as such. In practice the design of these systems can be complex and can require experienced personnel. L. SHERWOOD REFERENCES 1. Logan, A., ‘Corrosion Control in Tankers’, Transactions of the Institute of Marine
Engineers, No. 5 (1958) 2. Hanson, H. R . , ‘Current Practices of Cathodic Protection on Offshore Structures’, NACE Conference, Shreveport, Louisiana (1966)
SACRIFICIAL ANODES
10: 55
3. Compton, K. G.,Reece, A. M., Rice, R. H. and Snodgrass, J. S., ‘Cathodic Protection of Offshore Structures’, Paper No. 71, Corrosion/71. NACE. Houston (1971) 4. Tipps, C. W., ‘Protection Specifications for Old Gas Main Replacements’, Paper No. 27, Corrosion/70, NACE, Houston (1970) 5 . Nakagawa, M., ‘Cathodic Protection of Berths, Platforms and Pipelines’, Europe and Oil, July (1971) 6. Cherry, B., ‘Cathodic Protection of Buried Pre-Stressed Concrete Pipes’. In Cathodic Protection Theory and Practice, 2nd International Conference, Stratford upon Avon, June (1989) 7. Salleh. M. M. B. H., ‘Sacrificial Anodes for Cathodic Protection in Sea Water’, Ph.D. Thesis, pp 30-33, University of Manchester (1978) 8. Klinghoffer, 0. and Linder. B., ‘A New High Performance Aluminium Anode Alloy with High Iron Content’, Paper No. 59, Corrosion/87, San Francisco, USA, March (1987) 9. Crundwell, R. F., ‘SacrificialAnodes- Old and New’. In Cathodic Protection Theory and Pructice, 2nd International Conference, Stratford upon Avon, June (1989) 10. Zamin, M., ‘The Role of Mn in the Corrosion Behaviour of AI-Mn Alloys’, Corrosion, 32 (11). 627 (1981) 11. Jensen, F. 0. and Torleif, J., ‘Development of a New Zinc Anode Alloy for Marine Application’, Paper No. 72, Corrosion/87, San Francisco, USA, March (1987) 12. US Military Specification MIL-A-18001 J 1983 13. Det norske Veritas Recommended Practice, Cathodic Protection Design, RP B401, March (1986). This document has been superseded by RP B401 (1993) 14. Ahmed, D. S., Ashworth, V., Scantlebury, J. D. and Wyatt, B. S., British CorrosionJournul, 24. 149 (1989) 15. Ashworth, V.. private Communication 16. Reding, J. T. and Newport, J. J., Materials Protection, 5 (12). 15 (1966) 17. Wroe. S. P. and May, R. F.,‘Development and Testing of a New Improved Aluminium Anode Alloy’, UK Corrosion ’87, Brighton, October (1987) 18. Jacob, W. R.. private Communication 19. Lennox, T.J., Peterson, M. H. and Groover, R. E., Materialshotection, 7 (2), 33 (1%8) 20. NACE Standard Recommended Practice RP0387-87, Metallurgical and Inspection Requirements for Cost Sacr@cial Anodes for Offshore Applications, NACE, Houston (1990) 21. Schreiber, C. F. and Murray, R. W., ‘Effect of Hostile Marine Environment on the Al-ZnIn-Si Sacrificial Anode’, Paper 32, Corrosion/88, St. Louis, USA, March (1988) 22. Schreiber, C. F. and Murray, R. W., Materials Performance, 20 (3), 19 (1981) 23. Houghton, C. J., Ashworth. V., Materials Performance. 21 (7), 20 (1982) 24. Ashworth, V., Googan, C. G., Scantlebury, J. D., British Corrosion Journal, 14 (l), 46 (1979) 25. Ashworth, V., Googan, C. G., Jacob, W. R., Proceedings of Australasian Corrosion Association Znc., Conference 26, Adelaide (1986) 26. Morgan, J. H.. Cathodic Protection, 2nd edn., NACE. Houston 27. Nisancioglu, K.,‘Modelling for Cathodic Protection’. In Cathodic Protection Theory and Practice, 2nd International Conference, Stratford upon Avon. June (1989) 28. Osborn, 0 . and Robinson, H.A., ‘Performance of Magnesium Galvanic Anodes in Underground Service’, Corrosion, April (1952) 29. Craven, D., T h e Protected Gas Service’, Institution of Gas Engineers, Cardiff, March (1969) 30. Peabody, A. W., Control of Pipeline Corrosion, NACE, Houston (1971) 31. Evans, T. E., ‘Mechanisms of Cathodic Protection in Seawater’. In Cathodic Protection Theory and Practice, 2nd International Conference, Stratford-upon-Avon, June (1989) 32. Choate, D.L., Kochanczyk, R. W. and Lunden, K. C., ‘Developments in Cathodic Protection Design and Maintenance for Marine Structures and Pipelines’, NACE Conference on Engineering Solutions for Corrosion in Oil and Gus Applications, Milan, Italy, November (1989); not included in Proceedings 33. Fischer, K. P., Thomason, W. H. and Finnegan, J. E., ‘Electrochemical Performance of Flame Sprayed Aluminium Coatings of Steel in Water’, Paper No. 360,Corrosion/87, San Fransisco, USA, March (1987) 34. Procter. R. P. M., private communication
10.3 Impressed-current Anodes
Impressed-current Anodes for the Application of Cathodic Protection Numerous materials fall into the category of electronic conductors and hence may be utilised as impressed-current anode material. That only a small number of these materials have a practical application is a function of their cost per unit of energy emitted and their electrochemical inertness and mechanical durability. These major factors are interrelated and- as with any field of practical engineering-the choice of a particular material can only be related to total cost. Within this cost must be considered the initial cost of the cathodic protection system and maintenance, operation and refurbishment costs during the required life of both the structure to be protected and the cathodic protection system. There are obviously situations which demand considerable over-design of a cathodic protection system, in particular where regular and efficient maintenance of anodes is not practical, or where temporary failure of the system could cause costly damage to plant or product. Furthermore, contamination of potable waters by chromium-containing or lead-based alloy anodes must lead to the choice of the more expensive, but more inert, precious metal-coated anodes. The choice of material is then not unusual in being one of economics coupled with practicability. Although it is not possible in all cases to be specific regarding the choice of anode material, it is possible to make a choice based upon the comparative data which are at present available. Necessary factors of safety would be added to ensure suitability where lack of long-time experience or quantitative data necessitate extrapolation or even interpolation of an indefinite nature. The manufacture. processing and application of a particular material as an impressed-current anode requires knowledge of several physical characteristics. Knowledge and attention to these characteristics is necessary to design for anode longevity with maximum freedom from electrical and mechanical defects. The various types of materials used as anodes in impressed-current systems may be classified as follows: 10: 56
IMPRESSED-CURRENT ANODES
10:57
1. Precious metals and oxides: platinised titanium, platinised niobium,
2. 3. 4.
5.
platinised tantalum, platinised silver, solid platinum metals, mixed metal oxide-coated titanium, titanium oxide-based ceramics. Ferrous materials: steel, cast iron, iron, stainless steel, high-silicon iron, high-silicon molybdenum iron, high-silicon chromium iron, magnetite, ferrite. Lead materials: lead-antimony-silver, lead with platinum alloy microelectrodes, lead/magnetite, lead dioxide/titanium, lead dioxide/ graphite. Carbonaceousmaterials: graphite, carbon, graphite chips, coke breeze, conductive polymer, conductive paint. Consumable non-ferrous metals: aluminium, zinc.
Combination Anodes
These are anodes that, to reduce costs, use a combination of materials, sometimes coaxially, to extend the life of the primary anode, reduce resistance to earth, improve current distribution, facilitate installation and improve mechanical properties. Often the so-called ‘anode’ is primarily a means of conducting the current to the more rapidly consumable anode material. These can be classified as follows: 1. Canister anodes: consist of a spirally wound galvanised steel outer
casing containing a carbonaceous based extender which surrounds the primary anode element which may be graphite, silicon iron, magnetite, platinised titanium, mixed metal oxide-coated titanium or platinised niobium, etc. 2. Groundbeds: consist of a carbonaceous extender generally coke breeze and graphite, silicon-iron scrap steel, platinised titanium or niobium anodes. 3. Co-axial anodes: These are copper-cored anodes of lead silver, platinised titanium and platinised niobium. For long lengths of anode it is sometimes necessary to extrude one material over another to improve a particular characteristic. Thus titanium may be extruded over a copper rod to improve the longitudinal conductivity and current attenuation characteristics of the former; lead alloys may be treated similarly to compensate for their poor mechanical properties. It should he noted that these anodes have the disadvantage that, should the core metal be exposed to the electrolyte by damage to the surrounding metal, rapid corrosion of the former will occur. In flowing water enviroments a tubular rather than a solid rod cantilever anode may be used to give improved resistance to fatigue failure, since the anode design may result in fatigue failure by vortex shedding at high water velocities. Failures of impressed-current systems may occur not because of anode failure in a specific environment but because of poor integrity of the anode/cable connection or the use of an inferior cable insulation. Particular
’
10:58
IMPRESSED-CURRENT ANODES
attention must therefore be paid to these aspects of anode construction or rapid failure could take place.
Platinum and Platinum-coated Anodes The properties of platinum as an inert electrode in a variety of electrolytic processes are well known, and in cathodic protection it is utilised as a thin coating on a suitable substrate. In this way a small mass of Pt can provide a very large surface area and thus anodes of this type can be operated at high current densities in certain electrolyte solutions, such as seawater, and can be economical to use. When platinum is made the anode in an aqueous solution, a protective electron-conducting oxide film is formed by the following reaction: Pt
+ 2H20
Pt(OH),
+ 2H+ + 2e
E" = +0-98V VS. SHE
Once the protective oxide film is formed current flow may then only occur by oxygen evolution, which in pure aqueous solutions may be represented as HzOFt2H' + f 0 2 + 2 e
E"
=
+1-23Vvs.SHE
This anode half reaction is highly irreversible and is accompanied by an appreciable overvoltage"; usually the potential of oxygen evolution is about 0.5 to 0 - 7 V higher than Ee. In chloride-containing solutions evolution of chlorine will also occur and is usually the predominant anodic reaction even at low C1- concentrations, e.g. brackish waters: 2CI-+Cl2+2e
E" = +1*36Vvs. SHE
The relative proportions of oxygen and chlorine evolved will be dependent upon the chloride concentration, solution pH, overpotential, degree of agitation and nature of the electrode surface, with only a fraction of the current being used to maintain the passive platinum oxide film'. This will result in a very low platinum consumption rate. Tests carried out in the USA and initiated in 19533indicate the following consumption rates of precious metals and their alloys: Pt, Pt-l2Pd, Pt-SRu, Pt-lORu, Pt-SRh, Pt-lORh, Pt-5Ir and Pt-IOIr, 6-7 mg A-' y-'; Pt-ZOPd, Pt-SOPd, Pt-20Rh and Pt-25Ir, slight increase in rate; Pt-SOPd, greater increase in rate although the cost of Pd may offset this; Pd, Ag, Pd-40Ag, Pd-1ORu and Pd-lORh, excessively high rates. The tests were carried out for periods of some months in seawater at current densities ranging from 540 to 5 400 Am-', and the results appeared to be independent of current density and duration of test. The dissolution rate of solid rods of high purity platinum over the current density range 1 180 to 4600Am-' has also been investigated. Values of 17.5 to 26.3 mg A-' year-' were reported over the first year, but the rate decreased to a limiting value of 2.6 to 4.4mg A'ly-' over a 5-year period4. The high initial rate was attributed to preferential dissolution at grain boundaries and other high free energy sites. Tests carried out in the
IMPRESSED-CURRENT ANODES
10: 59
UK”” on electrodeposited platinum on a titanium substrate indicate a consumption rate in seawater of 8.8mg A-ly-’, although values of up to 15 mg A-’ y-’ have been quoted elsewhere6. Platinised Titanium
Titanium, which was in commercial production in 19507, is thermodynamically a very reactive metal (machining swarf can be ignited in a similar fashion to that of magnesium ribbon) but this is offset by its strong tendency to passivate Le. to form a highly stable protective oxide film. It is a valve metal and when made anodic in a chloride-containing solution it forms an anodic oxide film of Ti02 (rutile form), that thickens with an increase in voltage up to 8-12 V, when Iocalised film breakdown occurs with subsequent pitting. The TiOl film has a high electrical resistivity, and this coupled with the fact that breakdown can occur at the e.m.f.’s produced by the transformer rectifiers used in cathodic protection makes it unsuitable for use as an anode material. Nevertheless, it forms a most valuable substrate for platinum, which may be applied to titanium in the form of a thin coating. The composite anode is characterised by the fact that the titanium exposed at discontinuities is protected by the anodicaliy formed dielectric TiOz film. Platinised titanium therefore provides an economical method of utilising the inertness and electronic conductivity of platinum on a relatively inexpensive, yet inert substrate. Titanium can be forged, bent, cut, stamped, rolled, extruded and successfully welded under argon, making possible a large variety of electrode shapes, i.e. rod, sheet, tube, wire or mesh. It is a very light yet strong material with a high resistance to abrasion. Cotton8s9was the first to publish results on platinised titanium as an anode material, and the first commercial installation utilising platinised titanium anodes was completed in 1958 at Thameshaven for the protection of a Thames-side jetty. Manufacture of Platinised Titanium Anodes Platinised titanium anodes are mainly produced by the electrodeposition of a thin coating of Pt from aqueous solutions lo on to preroughened titanium. Warne” states that electrodepositing coatings from aqueous plating solutions has the advantage that control of thickness is easily achieved, irregularly shaped substrates can be plated, and the electrodeposited coatings are hard and abrasion resistant, by virtue of the interstitial hydrogen co-deposited in the plating process. Titanium is a very difficult metal to electroplate because of the presence of an oxide film. Sophisticated pretreatments with acids to remove the oxide film are necessary to achieve good adhesion. Improvements in the level of adhesion can, however, be obtained by heat treatment of the resultant Pt/Ti composites ”. Electrodeposits of Pt can only be applied as relatively thin coatings that are porous. Although the porosity decreases with increase in deposit thickness, so does the internal stress and if the platinum adhesion is poor the coating may exfoliate. As a consequence, thicknesses of 2.5 to 7.5 pm Pt
1o:m
IMPRESSED-CURRENT ANODES
are normally used, although it is possible to apply coatings of 12.5 pm in one operation and still achieve good adhesion'. However, 7.5 pm is generally considered the maximum thickness from one plating operation. Thicker deposits may be obtained by deposition in a number of stages, with interstage anneals. Pt electrodeposits may also be produced from molten salt electrolytes. Such a high-temperature process has the advantage that the deposits are diffusion bonded to the titanium substrate and thus have good adhesion, and, if necessary, thick deposits can be produced. However, they have the disadvantage that because of the complexity of the process there is a limitation on the size and shape of the object to be plated, and the resultant deposits are softer and less wear resistant than those from aqueous solutions 1 3 . Metallurgically bonded coatings may also be produced. These have the advantage that thick, low porosity, ductile platinum coatings can be produced. These are achieved by electrodeposition of platinum or, more often, by wrapping a thin platinum sheet over a cylindrical billet of titanium, vacuum encapsulating within a copper can, and then extruding it into the required shape". The copper sheath, which is used as a lubricant, has the advantage that it prevents fouling of the anode prior to energising. Platinum coatings may also be thermally sprayed or sputtered onto the titanium, to provide uniform well-bonded coatings. Titanium rod may also be spiral wound with platinum wireI4. However, the use of these techniques is limited.
The Operational Characterisics of Platinised-Titanium Anodes Platinisedtitanium anodes have the disadvantage that the protective passive film formed when titanium is made anodic in certain solutions can breakdown. This could result in rapid pitting of the titanium substrate, leading ultimately to anode failure. The potential at which breakdown of titanium occurs is dependent upon the solution composition, as is evident from Table 10.16. Table 10.16 Breakdown potentials of commercially pure titanium in various environments ~~~~~~~~~
Reference 15 15 16 17,18 19 19 19 19 18 18 18 18 18 20 21 22
Electrolyte and conditions Tests in pure seawater at ambient temperatures Tests in NaCl from 5 g / l to saturated below 60°C Seawater Sulphuric acid Chloride Sulphate Carbonate Phosphate Fluoride Bromide Iodide Sulphate Phosphate Ratio of sulphate plus carbonate to chloride ions, 4: I River water Tap water
Breakdown potential of commerciolly pure titanium (V) 8.5-15 8.5-15 9-14 80-100 8 60 60 60 50 2-3 2-3 > 80 z 80 > 35 50 80
10:61
IMPRESSED-CURRENT ANODES
In seawater the breakdown potential of titanium is often considered to be -9.5 V vs. SHE', whilst values as low as 6 V in 5 . 8 % NaCl solutions have been reportedz3. The value of the breakdown potential for titanium is dependent upon the C1-concentration and in high purity waters may be relatively high22;in the case of seawater certain, anions present, for example SO:-, favour passivation. It is also dependent upon the level of purity of the titanium and generally decreases with the addition of certain alloying elements. The presence of bromides and iodides will significantly reduce the pitting potential for titanium whilst other ions, notably fluorides and sulphates will tend to increase it. Temperature also has a significant effect on the anodic breakdown voltage of titanium, with an increase in temperature decreasing the breakdown potential. Platinised titanium anodes may be operated at current densities as high as 5 400Am-' ', however at these current densities there is the possibility that the breakdown potential of titanium may be exceeded. The normal operating current density range in seawater is 250-750 Am-' whilst that in brackish waters is given as 100-300 Am-2 24 with values within the range 100-150 Am-' being favoured lo. The consumption rate for platinised titanium anodes in seawater over the current density range 300 to 5000Am-' has been found to be directly related to the charge passed, with values of 8.7 to 17-4mg A - ' y - ' being generally used as the basis for system design. The consumption rate is also dependent upon solution composition, the rate increasing with decreasing chloride concentration and may reach a peak value of 435 mg A-l y - ' at a salt concentration of 2-5 gl lo. The reason for the increased corrosion rates is thought to be associated with the concurrent evolution of oxygen and chlorine, the rates of which are about equal in a neutral solution containing 2 . 5 g l - ' NaCl'. In brackish waters the platinum consumption rate may be as high as 174mg A-ly-', Le. more than ten times the rate in seawater, and increases with increase in current density24.Notwithstanding this, values of approximately 45 mg A-' y-I have often been used as the basis for design calculations in brackish waters. Baboian2' reports consumption rates of approximately 13 mg A-'y-l in seawater over the current density range 11.8 to 185Am-l, whilst those in 350ohm cm water (brackish river water) he reports as 92.3 mg A-' y-l at 11 - 8 Am-', increasing to 117.8 mg A-' y - ' at 185 Arnl2. The effect of temperature on the consumption rate of platinised titanium anodes has not been found to be significant over the ranges normally encountered in cathodic protection installations, although at elevated temperatures of 90-95"C, consumption rates of 570 mg A-' y-l in 0.02% Na2S0, and 12% NaCl solutions have been reported". Early failures of platinised titanium anodes have been found to occur for reasons other than increased consumption of platinum or attack on the titanium substrate caused by voltages incompatible with a particular electrolyte. The following are examples:
',
1. Attack on the substrate in low pH conditions, e.g. when covered in mud or marine growth, prior to energising, has been found to be a possible cause of failure2'*26.A commercial guarantee requires that the period in which anodes remain unenergised must not be longer than 8
10 :62
2.
3.
4.
5.
IMPRESSED-CURRENT ANODES
weeksI9. Indeed, if anodes are to be installed for extended periods prior to energising, they can be coated with a copper anti-fouling paint or with a flash of copper electrodeposit’. The copper coating will dissolve when the anode is energised and will not affect the anode’s subsequent performance or operation. Attack on the substrate by contact with Mg(OH), and Ca(OH), (calcareous scale) can also cause deplatinisation to occur, Anodes located close to the cathode or operating at high current densities can lead to a rapid build up of calcareous deposit, the major constituents of which are Mg(OH), and Ca(OH),”. The alkaline conditions so generated can lead to rapid dissolution of the platinum. The calcareous deposit can be removed by washing with dilute nitric acid. The formation of deposits on platinised anales can cause anode degradation 12s2’. Thus dissolved impurities present in water which are liable to oxidation to insoluble oxides, namely Mn, Fe, Pb and Sn, can have a detrimental effect on anode life. In the case of MnO, films it has been stated that MnO, may alter the relative proportions of Cl, and 0,produced and thus increase the Pt dissolution rate”. Fe salts may be incorporated into the TiO, oxide film and decrease the breakdown potential” or form thick sludgy deposits. The latter may limit electrolyte access and lead to the development of localised acidity, at concentrations sufficient to attack the underlying substrate Io. The superimposition of a.c. ripple on the d.c. output from a transformer rectifier can under certain circumstances lead to increased platinum consumption rates and has been the subject of considerable r e ~ e a r c h ~ Indeed, ~ ~ ~ ~ -when ~ ~ . platinised titanium anodes were first used it was recommended that the ax. component was limited to 5% of the d.c. voltage”. The frequency of the superimposed ax. voltage signal has also been shown to affect the consumption rate of platinum, which increases with decrease in frequency to 50 Hz and less. It was observed that at 100 Hz (the frequency of the a.c. component signal from a full-wave single phase transformer rectifier) and above, the a.c. signal had a negligible effect on consumption rate, provided of course that the a.c. component did not allow the electrode to become negative. In this case, even at 100 Hz a considerable increase in platinum dissolution can occur3’. This could be the case with a thyristor-controlled transformer rectifier operating at a relatively low current output. At low-frequency a.c. (2 Hz)an increase in platinum dissolution rate of two to three times has been reported, whilst negative current spikes of a few milliseconds duration at this frequency can cause dissolution rates of approximately 190mg A-ly-’. It is therefore recommended that all spurious waveforms on the d.c. supply to platinised anodes be avoided. Organic impurities in the electrolyte have also been quoted as increasing the rate of platinum dissolution when the metal is used as an anode in electroplating”. Saccharose was observed to increase the anodic dissolution of platinum by a factor of ten, in a 3% brine solution”, yet it did not affect the anodic breakdown voltage for titanium. Other organic compounds that may also have an effect are brightening agents
IMPRESSED-CURRENT ANODES
10 :63
for Ni plating solutions of the naphthalene trisulphonic acid type, detergents or wetting agents. 6. Fatigue failure due to underdesign or changes in plant operation of cantilever anodes in flowing electrolytes can occur as a result of vortex shedding”. However, with proper design and adequate safety factors these failures can be a ~ o i d e d ~ ’ . ~ ~ . 7. Attention must be paid to field end effects, particularly on cantilever anodes, e.g. on long anodes that extend away from the cathode surface. Under these circumstances the anode surface close to the cathode may be operating at a considerably higher current density than the mean value, with the exact values dependent upon the system geometry. The life of the platinising in this region would then be reduced in inverse proportion to the current density. Platinised-titanium installations have now been in use for 30 years for jetties, ships and submarines and for internal protection, particularly of cooling-water systems36. For the protection of heat exchangers an extruded anode of approximately 6 mm in diameter (copper-cored titaniumplatinum) has shown a reduction in current requirement (together with improved longitudinal current spread) over cantilever anodes of some This ‘continuous’ or coaxial anode is usually fitted around the water box periphery a few centimetres away from the tubeplate. Platinised-titanium anodes may also be used in soils when surrounded by a carbonaceous backfill. Warne and Berkeley4’have investigated the performance of platinised-titanium anodes in carbonaceous backfills and conclude that the anodes may be successfully operated in this environment at a current density of up to 200Am-’. This also supplements the findings of Lewis4’, who states that platinised-titanium anodes may be used in carbonaceous backfill without breakdown of the titanium oxide film. Success with platinised-titanium anodes has been reported with anodes operating at a few tens of Am-’ and failures of anodes have often been attributed to operation at high current densities”. Furthermore, the restrictions on operating voltage that apply to titanium in a marine enviroment are not always relevant to titanium in soils free of chloride contamination. Coke breeze is, however, an integral part of the groundbed construction and ensures a lower platinum consumption rate. However, for some borehole groundbeds, platinised niobium is preferred, particularly in the absence of carbonaceous backfill or in situations where the water chemistry within a borehole can be complex and may, in certain circumstances, contain contaminants which favour breakdown of the anodic TiO, film on titanium. In particular, the pH of a chloride solution in a confined space will tend to decrease owing to the formation of HOC1 and HCl, and this will result in an increase in the corrosion rate of the platinum. The high cost of platinised materials for use in borehole groundbeds as opposed to conventional silicon-iron anodes may also be offset by the reduction in required borehole diameter, hence lower installation cost, with the relative economics between the different systems dependent upon a combination of both material and installation costs.
10:64
IMPRESSED-CURRENT ANODES
Platinised Niobium and PIatinised Tantalum
The principle of these anodes is similar to that of platinised titanium since they are all valve metals that form an insulating dielectric film under anodic polarisation. Platinum electrodeposition on to tantalum had been carried out as early as 191343and the use of platinised tantalum as an anode suggested in 192244,whilst platinum electrodeposition on to niobium was first successfully carried out in 19504’. These anodes are considerably more expensive than platinised titanium, especially when expressed in terms of price per unit volume4. Indeed, since niobium is cheaper than tantalum the use of the latter has become rare. The extra cost of Nb anodes may be offset in certain application by their superior electrical conductivity and higher breakdown voltages, Table 10.17 gives the comparitive breakdown potentials of Ti, Nb and Ta in various solutions under laboratory conditions. Table 10.17
Comparison of breakdown potential* ~
Solution Seawater Sulphate/Carbonate Phosphate/Borate Drinking water Bromides
Ti
Nb
Ta
9
120 255 250 250
120 280 280 280
60 80
37.5 2-3
There have been instances reported in the literature where the breakdown potential for Nb and Ta in seawater has been found to be lower than the generally accepted value of 120 V, with reported values in extreme instances This has been attributed to contamination of the as low as 20-40V47*48. niobium surface from machining operations, grit blasting or traces of copper lubricant used in anode manufacture. These traces of impurities, by becoming incorporated in the oxide film, decrease its dielectric properties and thus account for the lower breakdown voltage. Careful control of surface contamination in the manufacture of platinised niobium is therefore essential to minimise the lowering of the breakdown potential of niobium. Platinised niobium anodes are prepared by electrodepositing platinum onto grit-blasted niobium, metallurgical co-processing (cladding) or by welding platinum or platinum/iridium wire to niobium rod4*. They are not prepared by thermal deposition because niobium oxidises at 350°C, and good adhesion cannot be obtained. Both materials may be welded under argon, utilising butt or plasma welding techniques. Platinised niobium and tantalum anodes have found use in applications where their high breakdown voltages and hence higher operational current densities can be utilised, e.g. ship and cooling system anodes, which may be used in estuarine waters and thus require higher driving voltages, offshore structures where high reliability in service is required, domestic water tanks49and deep well groundbeds”. Because of their higher breakdown voltages niobium and tantalum anodes can be operated at higher current densities than platinised titanium. Efird 32 found the consumption rate of
IMPRESSED-CURRENT ANODES
10 :65
platinised niobium in seawater over the range 5 OOO to 10000Am-' to be similar to that of platinised titanium, i.e. 7-8mg A-' y-'. However, at a current density of 30000Am-' he observed an increase in the platinum consumption rate to 15.6mg A-'y-' and concluded that this was the limiting current density for operating these anodes. Warne and Berkeley4' report that the maximum current density for these anodes in seawater is 2000Am-2, with a working current density of 500 to 1000Am-'. The operating current density selected should, however, be commensurate with the desired anode life, platinum coating thickness and platinum consumption rate in a given environment. In open-hole deep-well groundbeds, platinised niobium anodes have been successfully operated at current density of 215 Am-2 5 1 and in the range 100 to 267 Am-* Toncre and H a ~ f i e l dhave ~ ~ conducted work on the operating parameters of platinised niobium anodes in brackish waters and simulated groundbed environments. In an open-hole groundbed they concluded that operational current densities of 400Am-2 or higher were the most economical, since this leads to a lower consumption rate in sulphatecontaining soils. The platinum consumption rate in a deep-well environment may well alter because of variations in the environmental conditions. On extruded platinised niobium anodes a consumption rate of 175mg A-' y-' was considered for design purposes, whilst that for electroplated platinised niobium was taken as 8 7 - 6mg A-' y-'. In a backfilled deep-well groundbed, evidence of dissolution rates comparable with those for an open-hole environment were reported, i.e. 87-6mg A-ly-' at 200Am-'. At lower current densities, namely 100Am-2, it seems likely that the electrochemical processes would be limited to oxidation reactions involving coke alone and no electrochemical wear * on platinised niobium would occur. Indeed, Baboian '" reports negligible Pt consumption rates in carbonaceous backfill at current densities from 11- 8 to 29 Am-', increasing to 11- 9mg A-' at 57-9Amd2and 13.5 mg A-'y-l at 185 AmI2. The wear rates at 57.9 and 185 Am-' were comparable to those Baboian observed in seawater. The relative merits of platinised titanium and niobium in a deep-well environment, in comparison with those of other anode materials, have been given by StephensS3.
Pla tinised Silver This material can be used only in seawater or similar chloride-containing electrolytes. This is because the passivation of the silver at discontinuities in the platinum is dependent upon the formation of a film of silver chloride, the low solubility of which, in seawater, inhibits corrosion of the silver. This anode, consisting of Pt-1OPd on Ag, was tried as a substitute for rapidly consumed aluminium, for use as a trailing wire anode for the cathodic protection of ships hulls, and has been operated at current densities as high as 1900Am-'. However, the use of trailing anodes has been found inconvenient with regard to ships' manoeuvrability. In the case of the platinum metals the term 'wear' is frequently used in place of corrosion attack.
10:66
I M PRESSED-CUR RE NT A NODES
With the advent of hull mounted anodes this material has been replaced by the superior platinised titanium and niobium anodes and is now seldom used. Mixed Metal Oxide Coated Titmiurn
The material was originally developed by and its major application has been in the production of chlorine and chlorates s7.s8. It has now gained acceptance as an impressed current anode for cathodic protection and has been in use for this purpose since 1971. The anode consists of a thin film of valve and precious metal oxides baked onto a titanium substrate and when first developed was given the proprietary name 'dimensionally stable anode', sometimes shortened to DSA. Developments on the composition of the oxide film have taken place since Beer's patent, and this type of anode is now marketed under a number of different trade names. The anodes are produced by applying a paint containing precious metal salts or organic compounds in an organic solvent to the titanium surface and then allowing the solvent to evaporate. The completed assembly is then heated in a controlled atomosphere to a temperature at which the paint decomposes (between 350 and 600°C)s9to give the metal. Platinum does not form an oxide under the conditions selected, but other precious metals namely iridium, ruthenium, rhodium and palladium do. A number of paint coatings may be necessary to obtain the required deposit thickness, which is typically 2-12-5fim, although deposits up to 25pm thick have been obtained. Deposits thicker than this become brittle and poorly adherentm. At present only titanium substrates are coated in this way because at the temperatures encountered in the anode manufacturing process, niobium would oxidise. Tantalum can be coated with a mixed oxide but this is a relatively expensive process. The composition of the mixed metal oxide may well vary over wide limits depending on the environment in which the anode will operate, with the precious metal composition of the mixed metal oxide coating adjusted to favour either oxygen or chlorine evolution by varying the relative proportions of iridium and ruthenium. For chlorine production Ru0,-rich coatings are preferred, whilst for oxygen evolution Ir0,-rich coatings are utilised6'. The mixed metal oxide coatings consist of a platinum group metal (usually ruthenium, although in some cases two or three metals are used) and an oxide of a non-platinum group metal (usually Ti, Sn or Zr)s9. The precise composition of the coating is generally considered proprietary information and not divulged by the various anode manufacturers. Indeed, recent studies have shown that a proprietary electroactive mixed metal oxide anode coating used for the cathodic protection of steel in concrete contains Ta, Ir, Ti and Ru". The metal oxide coating has a low electrical resistivity of approximately lo-' ohm m, and a very low consumption rate. In seawater the consumption rate is 0.5 to 1 mg A - l y - ' , whilst in fresh waters and saline muds the rate is 6mg A - ' Y - ' ~ ,at current densities of 600 and 100 Am12, respectively@.
IMPRESSED-CURRENT ANODES
10:67
In fresh waters, where there is a limited chloride concentration, the predominant reaction is oxygen evolution. This process gives rise to a high level of acidity which may account for the increased oxide consumption rates observed in this environment. The current densities normally used for design purposes are 600Am-* for seawater, and 100Arn-, for fresh water, saline mud and coke breeze backfill63. Higher current densities may be utilised in certain circumstances but this reduces the anode life for a given coating thickness. The relationship between current density and life for the mixed metal oxide coated electrodes is non-linear and higher current densities increase the consumption rate of the oxide. The approximate value for the oxide coating dissolution rate in relation to current density6’ in soils and fresh water is: log,,L = 3 - 3 - log,,i and in seawater is: log,,L = 2-3 - O.4logl,i where i is the current density (Am-’) and L is the coating life (years). Indeed, with current densities of 4000Am-* in 3% NaCl rapid consumption rates of RuO, coatings have been reported6’. The oxide coatings are porous and therefore the limitations on operating voltage for platinised titanium anodes apply as well to the oxide-coated titanium electrodes. It has been reported that breakdown of mixed metal oxide anodes may occur at 50-60 V in low-chloride concentration water but at only 10 V in chloride-rich environments@. These anodes, like platinised Ti may be supplied in different forms e.g. rod, tube, mesh, wire, etc. They may be used for the cathodic protection of offshore structures, heat exchangers, or even pipelines as they can be installed in the soil surrounded by carbonaceous backfill, and are comparable in cost to platinised titaniumw. Conductive Titanium Oxide Based Ceramics
An electrically conductive titanium oxide based ceramic material has been developed recently, and is marketed under the trade name ‘Ebonex*. This material consists principally of Ti,O, but may also contain some higher oxides. It is black in colour, has an electrical resistivity of less than 2x ohm m and can be operated at current densities up to 100Am-2in 1% NaCI, however, if coated with a precious metal it can be operated at considerably higher current densities up to 400 Am - 2 67. However, no information is given by the manufacturers on the consumption rate of this particular material other than that it is inert. It is both porous and brittle, although its mechanical strength can be improved and porosity reduced by resin impregnation, preferably with inorganic fillers. It has a high overpotential for oxygen evolution, is not affected by current reversal, has no restriction on operating voltage and the makers claim it has an excellent resistance to both acid and alkali.
10 :68
IMPRESSED-CURRENT ANODES
To date the material has been used as an electrode in electro-winning, electro-chlorination, batteries and electrostatic precipitators, but only to a very limited extent in cathodic protection.
Ferrous Materials Steel
One of the earliest materials to be used in power-impressed cathodic protection was steel. Its economy lies in situations where steel scrap is available in suitable quantities and geometry and it is only in such situations where its use would now be considered. The anode tends to give rise to a high resistance polarisation due to the formation of a voluminous corrosion product, particularly when buried as opposed to immersed. This can be alleviated by closely surrounding the scrap with carbonaceous backfill; this of course increases the cost if the backfill is not also a local by-product. It is necessary under conditions of burial to ensure compactness and homogeneity of backfill (earth or carbon) at all areas on the steel, otherwise particularly rapid loss of metal at the better compacted areas could lead to decimation of the groundbed capacity. The problem of the high resistance polarisation decreases with increasing water content and salinity, such as prevails during immersion in seawater, where these anodes are particularly useful. Since no problems of burial arise in that environment an endless variety of disused iron-ware has been utilised for anodes, e.g. pipes, piling, machinery, rails and even obsolete shipping which has not been economic to salvage. Consumption rates in excess of the theoretical value have been reported for steel in different water^^^'^^. Experimental installations have been established from time to time to selected demonstrate the possibility of using ferrous metals in anolyte~’”-~* to minimise polarisation and to reduce metal ionisation by making the metal passive. The use of carbonaceous extender is of value if segregation of a steel anode in soil might be expected to result from high localised corrosion rates. Continuity of the anode is facilitated by the bridging effect of the extender. One example of this is in deep-well groundbeds, installed in stratified soils of widely differing ground resistivities, where a well casing may be filled with coke breeze. Commercial examples of these are known to be working well after periods of 28 years73. One advantage of steel as an anode is the low gassing at the electrode during operation, since the predominant reaction is the corrosion of iron. Thus, the problem of resistive polarisation due to gas blocking, as may be the case with more inert materials, does not occur. Iron compounds do, of course, form but these d o not appreciably affect the anodelsoil resistivity. Furthermore, the introduction of metallic ions, by anode corrosion, into the adjacent high resistivity soil is beneficial in lowering the resistivity. It is necessary to ensure the integrity of anode cable connections and to give consideration to the number of such connections related to longitudinal resistance of the anode and current attenuation, if early failure is to be avoided.
IMPRESSED-CURRENT ANODES
10 :69
Cast Iron
Cast iron may be used under similar circumstances, but has inferior mechanical properties. It has been used, although not in current practice, for internal cathodic protection, where it has been demonstrated that the presence of ferrous ions in water is of benefit in reducing sulphide-induced attack on Cu alloy tube plate and t u b e ~ ’ ~Water . treatment has now been found to be a more practical method. Iron
Swedish iron is sometimes used as galvanic wastage plates in heat exchangers, particularly for marine applications. This is possibly based on tradition, since it cannot be the most economical method in the light of current cathodic-protection practice. The material is not currently used as an impressed-current anode. Stainless Steel
Stainless steel has been tried as an inert anode, mainly under laboratory conditions and with only partial success. Even at low current densities in fresh water the majority of alloys pit rapidly, although others show the ability to remain passive at a low current However, at practical current densities, the presence of chloride ions, deposits on the anode or crevice corrosion at the anode support lead to rapid failure77,but it may be possible that stainless steel could give useful service under certain conditions and with particular alloys7*. High Silicon Iron IHSI)
These are iron alloys that contain 14-18% Si and are reported as first being developed in 191279,although it was not until 1954 that they were first evaluated for use as impressed-current anode material in cathodic protection6*. Its major disadvantage is that it is a hard brittle material unable to sustain thermal or mechanical shock. The only practical method of machining is by grinding, and to obviate machining it is cast into fairly standard sizes to suit the general requirements of industry. HSI has a long successful history as a corrosion-resistant material in the chemical industry for such items as acid storage vessels and has been used in this application for more than 60 years. A typical analysis for HSI anodes is 14-5%Si, 0-75%Mn, 0 * 9 5 % C , remainder Fe. The anodes manufactured in the UK conform to BS 1591:1975 which lists the permissible Si content range as between 14.25 and 15.25070, whilst the maximum content of other elements is given as 1% C, 0 -1 To S and 0.25% P. Used anodically it readily forms a protective film which is reformed if removed mechanically. This is grey-white in appearance and has a tendency to flake under the compressive stress produced at thickened areas. The film
10: 70
IMPRESSED-CURRENT ANODES
is 50% porous and contains 72-78070 The film, formed in this way, is a fairly good electron conductor, even though SiO, in its natural state is a dielectric. The mechanism whereby the SiO, becomes a conducting oxide has been reviewed in some detail by Shreir and Hayfield", and is probably associated with doping of the SiO, with Fe ions. A coke-breeze backfill can be installed around the anodes buried in soil, so as to reduce the groundbed resistance to earth, anode current density and concentration of oxidising gases around the anode, thus improving the operational life. There is a tendency, when buried as opposed to immersed, for the surface resistance of such an anode to increase, but not to an extent that affects performance. The large resistance changes sometimes reported are usually due to gaseous polarisation (gas blocking) caused by poor venting or inadequate compaction and quantities of backfill. HSI anodes are subject to severe pitting by halide ions and this precludes their use in seawater or other environments in which these ions may be present in quantity. They are ideal for fresh-water applications (below 200 p.p.m. C1 -), although not for temperatures above 38°C. The addition of Mo or Cr to the alloy can improve performance under these conditions, with an upper limit of temperature of 56"C80,which may be affected by the composition of the water and operating conditions. The wastage rate of HSI depends upon the current density and the nature of the soil or water in which the anode is used. HSI is superior to graphite in waters of resistivity greater than 10 ohm m, but in waters of 0.5 ohm m and below HSI is susceptible to pitting. From collated experience in fresh water in the pH range 3 to 10 a nominal consumption rate of approximately 0.1 kg A - ' y-' at 20°C has been observed. This is of course dependent upon solution composition and temperature". A number of reports on the performance of HSI anodes in different environments have been produced8'-84. The consumption rate of HSI anodes buried directly in soils will vary depending upon the soil composition and will be excessive in chloridecontaining soils. In quicksands consumption rates of approximately 0.35 kg A y -' have been reported8', whilst in other soils consumption rates in the region of 1 kg A - ' y - ' are possible. A lower consumption rate in the region of 0.1 to 0-25 kg A - ' y - l would be expected in a carbonaceous backfill correctly installed, in a soil of insignificant chloride content and the anode operating at a current density of 20Am-*. Very much higher apparent consumption rates would most likely be due to high local current densities caused when the anode is inadequately backfilled, partially submerged, or where it has become partially silted up. The anode effectiveness is only as good as the anode connection and loss of insulation at this point by deep pitting of the HSI or penetration of the anode cable seal will bring about rapid failure. Hydrostatic pressure should be borne in mind when considering the seal required for any depth of water. The useful life of HSI anodes is usually considered at an end after a 33% reduction in diameter, but this depends upon the original diameter, the amount of pitting sustained and the mechanical stresses to be withstood. Thus doubling the cross-sectional area may more than double the effective life of the anode.
10:71
IMPRESSED-CURRENT ANODES
High Silicon/Molybdenum Iron
The addition of 1-3'70 Mo to HSI results in an improvement in maintaining a conducting oxide film in chloride-containing solutions above 200 p.p.m. or at temperatures of 38°C or above. However, the addition of Cr has resulted in even greater improvements. In seawater at 10 Am the addition of Mo reduced the consumption rate from 0.22 to 0.15 kg A - l y - l at ambient temperatures and from 0.63 to 0.21 kg A - ' y - ' at 5loCE5.Yet a considerably higher wastage rate of 0.9kg A - l y - ' at 10.8Am-' has been reported for the molybdenum-containing silicon iron in chloride-containing waters 'I.
-'
High-Silicon/Chromium Iron (HSCI)
This alloy was first put into commercial use around 195985.Chromium, together with silicon, results in a film that has a high resistance to pitting in waters containing halide ions, and these alloys can be used in seawater or chloride-containing soils with confidence. A typical analysis is 14.5% Si, 0.75% Mn, 1.0% C, 4.5% Cr, remainder Fe. In the UK this anode is manufactured to BS 1591:1975 which permits a variation in silicon content of 14.25 to 15.25%, andthatofchromiumof4to5V0,withamaximumcarbon content of 1 -40%. The equivalent US standard is ASTM A 5 18-64 Grade 2 with a silicon content of 14.2 to 14.75% and a chromium content of 3-25 to 5 .00%. Neglecting possible mechanical damage and anode/cable joint failure, it is possible, in view of the very minor pitting sustained in free suspension, for the anode to continue operating until totally consumed. Comparative tests between HSI and HSCI in seawater at 93" C and 10-8Arn-' showed consumption rates of 8.4kg A - ' y - ' and 0.43 kg A - ' y-', respectivelyg6. These figures show that the consumption rate of HSI when used in seawater without the addition of chromium may approach that of steel, but because of the very deep pitting and its fragility, it is in most cases inferior to steel. However, in fresh waters HSI has a far lower corrosion rate than steel. The consumption rate of HSCI freely suspended in seawater in the current density range 10.8 to 53.8Am-2 increases from 0.33kg A - ' y - ' at 1 0 - 8 A m - ' t o 0.48kg A - ' y - ' at 53*8Am-'. Direct burial in seawater silt or mud will also increase the consumption rate, with values of 0 - 7 k g A - I y - l at 8.5Am-* increasing to 0.94kg A - ' y - l at 23 4 Am-' ". HSCI anodes cannot be used in potable waters because of the possibility of chromium contamination. A recent evaluation of HSCI anodes in different soil conditions has been conducted by Jakobs and HewesE8.They report a consumption rate for different HSCI alloys in 3% NaCI, at a current density of 21.5 Am-', of between 0.32 and 0.87 kg A - ' y - ' depending upon the alloy composition; whilst in soils containing 2% SO:- consumption rates varied between 0.29 and 0.53 kg A - ' y - I , again depending upon the alloy composition. Improvements in anode construction have also been carried out to reduce the non-uniform material loss along the length of the HSCI anode, the SOcalled 'end effect' phenomenon. This involves the use of hollow, centrifugally
-
10:72
IMPRESSED-CURRENT ANODES
cast anodes of uniform wall thickness with a centrally located interior electrical connectionE9.Care should be taken to ensure that the cable insulation and sheathing are adequate for use in an oxidising environment, in which chlorine evolution occurs.
Magnetite Anodes Magnetite (Fe,O,) has been in use since the 1970s as a cathodic protection anode, although its use as anode material has been known for some timew. Magnetite has a melting point of 1540°C and can be cast using special techniques and with the addition of certain alloying elements9‘. The anodes are constructed from a cast alloyed magnetite shell, the centre of which is hollow. The internal surfaces of the magnetite shell are then lined with an electronic conductor so as to ensure a uniform distribution of current density over the external surface. This technique overcomes the longitudial current attenuation that would occur because of the relatively high resistivity of magnetite (3-3 ohm m quoted by Linder” and 0 . 8 ohm m by Kofstad9’). In early magnetite anodes the internal lining consisted of a thin copper layer, but the poor electrical contact between the copper layer and the magnetite, together with the fact that the cable-to-anode connection was made at the anode head, resulted in a non-uniform current density on the external magnetite surface, which contributed, in part, to the poor performance reported for some of the early magnetite anodes. Subsequently, the manufacturers perfected a method of electrodepositing a lead alloy lining onto the internal magnetite surface with the cable-to-anode connection made at the mid point of the anode. The central portion of the anode is filled with polystyrene and the anode cable attachment, whilst the remainder is filled with an insulating resin. Magnetite anodes exhibit a relatively low consumption rate when compared with other anode materials, namely graphite, silicon iron and lead and can be used in seawater, fresh water and soils. This low consumption rate enables a light-weight anode construction to be utilised. For example, the anode described by Linder” is 800 mm in length 60 mm in diameter, 10 mm wall thickness and 6 kg in weight. Tests carried out in seawater over the current density range 30 to 190 Am showed the consumption rate to be dependent upon current density, increasing from 1.4 to 4 g A - l y - ’ over the current density range studied (with the recommendation that to achieve the required life, the current density should not exceed 115 Am -2)93. Later work by Jakobs and Hewes” indicated the consumption rate in seawater to be less than 1 g A - ’ y - ’ at 21*6Am-’, whilst at 32-4Am-’ a consumption rate of 12 to 41 g A - ’ y - ’ was observed. Higher consumption rates were reported for magnetite anodes in soils containing 2% SO:-; namely 75 g A - ’ y - ‘ at current densities of 21.6 and 32.4Am-2, respectively. Jakobs% also conducted a survey of different anode systems in soils and found magnetite anodes after 2 years exposure and operating at a current density of 43 Am-’ to be in good condition with little evidence of attack.
-’
10:73
IMPRESSED-CURRENT ANODES
Magnetite anodes can be operated at elevated temperatures up to 90°C. with the limitation in temperature being failure of the anode cable connection and not the magnetite itself. The main disadvantages of magnetite anodes are that they are brittle, and susceptible to high-impact shocks, as is the case with silicon iron anodes, whilst some of the earlier anodes were subject to failure from thermal cycling'"*9'.Indeed, one evaluation of magnetite anodes reports a high incidence of failurew, and a more recent lists a number of failures of magnetite anodes when compared with other more conventional anode materials. Although these failures were mainly associated with poor installation practice and operation at current densities in excess of the manufacturers' maximum recommended values which are 77 Am for seawater and 30Arn-' for soils. Improvements in the anode design have now led to a more reliable anode with a decrease in the level of reported failures's8. Magnetite may also be used in combination with lead or electrodeposited onto a titanium substrate%. The latter anode system has been shown to exhibit good operating characteristics in seawater but at present it is only of academic interest.
-'
Ferrite Anodes
Sintered and sprayed ceramic anodes have been developed for cathodic protection applications. The ceramic anodes are composed of a group of materials classified as ferrites with iron oxide as the principal component. The electrochemical properties of divalent metal oxide ferrites in the composition range 0.1MO-O.9Fe,O3 where M represents a divalent metal, e.g. Mg, Zn, Mn, Co or Ni, have been examined by Wakabayashi and Akoi9'. They found that nickel ferrite exhibited the lowest consumption rate in 3% NaCl (of 1 56 g A - y - I at 500 Am -') and that an increase in the NiO content to 40mol%, Le. 0.4Ni0-0.6Fe20, reduced the dissolution rate to 0 . 4 g A - ' y -' at the expense of an increase in the material resistivity from 0.02 to 0.3 ohmcm. Ceramic anodes may be cast or sintered around a central steel core which acts as the electrical conductor. However, anodes produced in this form are brittle and susceptible t o mechanical shock. Ceramic anodes based on a plasma-sprayed ferrite coating on a titanium or niobium substrate have also been developed. These consist of plasmasprayed lithium, nickel or cobalt ferrite on a machined Ti or Nb buttonshaped substrate fitted into a plastic electrode holder98. This method of anode construction is durable, and not as prone to mechanical damage as the sintered ceramic anode whilst the ceramic coating is abrasion resistant report a dissolution rate and has a long operational life. Kumar et for a sprayed lithium ferrite of 1 . 7 g A - l y - ' at a current density of 2 OOO Am -* in seawater. The anode exhibited good performance with no damage on the ceramic coating observed during a two-month trial. However, the normal restrictions on operating voltages for titanium electrodes were still found to apply, with pitting* of the titanium substrate reported at 9-66V VS. SCE. 7
'
10 :74
IMPRESSED-CURRENT ANODES
Le8d Materi8Is
Investigations into the use of lead alloys for cathodic protection were made in the early 1 9 5 0 ~ ~ " and ' ~ ~ a practical material had been developed by 1954. The general use of lead alloys in seawater had previously been established IO3* IO4. The anodic behaviour of Pb varies depending upon the electrolyte composition and the electrode potential and has been the subject of a number of reviews 10,104,105 . In NO;, CH,COO- and B F f - solutions, lead will form highly soluble lead salts whilst in C1- and SOf- solutions, insoluble lead salts are formed when Pb is anodically polarised, In using metallic Pb as an anode the formation and maintenance of a hard layer of PbO, is essential, since it is the P b 0 2 that is the actual inert anode, the P b acting both as a source of Pb0, and an electrical conductor. PbO, is relatively insoluble in seawater and its dissipation is more usually associated with mechanical wear and stress than electrochemical action. In alkaline solutions approaching pH 10, PbO, is unsuitable for use, and for this reason it should be mounted clear of any calcareous deposit which may be formed on a cathodic area close to the anode: this deposit indicates the formation of alkali which may have a detrimental effect on the PbO, deposit. Lead has found considerable use as an anode in a wide variety of electrochemical applications, with studies dating back to 1924i05-107.Pure Pb has been tried as an anode in seawater but fails to passivate, since PbCl, forms beneath the Pb0, and insulates the Pb0, from the Pb substrate. The anodic behaviour of Pb in C1- solutions depends upon C1- ion concentration, the solution pH and the presence of passivating anions such as CO;,, HCO; and SO:-. At low current densities and low C1- concentrations, dissolution of Pb will occur and a PbCl, deposit will not be formed at the anode. In high C1- concentrations and at high current densities the rate of formation of Pb2+ will be high enough for the solubility product for PbCI, to be exceeded, and PbCl,, not PbO,, will be deposited at the anode. The formation of a PbO, coating on Pb when it is anodically polarised in C1- is achieved more readily by alloying lead with silver or other metals, or by incorporating inert conducting microelectrodes in the Pb surface. Pb alloys have been investigated to determine their suitability as anodes for cathodic protection. Crennel and Wheelerimcarried out tests on Pb-Ag alloys and found that Pb-1Ag was suitable for use in seawater providing that the current density did not exceed 100-200 Am -*, since at high current densities an insulating film formed. Other Pb alloys have been investigated, notably Pb-6-8Sb which required more than 200 Am - 2 to passivate, whilst Pb-6Sb-1OSn exhibited a high corrosion rate IO3. However, Morgan IO9 found that Pb-6Sb-1Ag alloy gave a lower consumption rate and exhibited a harder PbO, film than Pb-6Sb or Pb-1Ag. Pb-6Sb-2Ag alloys are slightly better. but about 50% more expensive. More recent work'" has also shown that additions of Mn to Pb-2Ag alloys may have a beneficial effect on anode performance in seawater. The Pb-6Sb-1Ag alloy is commonly used where Pb-Ag anodes are specified. The results of tests on Pb-6Sb-1Ag given in Table 10.18 are of interest in recognising the scope
10:75
IMPRESSED-CURRENT ANODES
Table 10.18 Behaviour of PbdSb-IAg anodes Resktivity of electrolyte at
Average wastage rate at
35°C (ohmm)
(kg A - I y - l )
0.163 0.163 0.5 0.5 10 10
50 50
(sea) (NaCI) (sea) (NaCI) (sea) (NaCI) (sea) (NaCI)
108 Am-2 0.086 1-99
0.0145 0.654 23.80 23.70 0.10 11.64
Length of trial (days)
Note
236 1.75
234 1.75 5.75 1.75 236 1.75
Notes: I . Service indicates a practical consumption of between 0-057 and 0.114 kg A - ' y - ' . Under laboratory conditions PbO has been formed at current densities as low as 21.6 Am-'. Typical operating current densities are 54-270 Am" at wastage rates o f 0.045""' to 0,082 kg A - ' y ' ""'_ 2. Similar performance between 0.7 and 270 Am-'; formation o f thin adherent film o f PbO,""'. 3. Similar gerformance between 2 . 7 and 160 A K 2 ; thick nodules of PbO, in some areas; &re deterioration at 270 Am-' ,I 31
Pbo,; rapid deterioration. although at I 0 0 Am-' it slows down after several weeks. Increasing silver content results in some improvement"'". Anode passivated in 0.163 ohm m water continues to operate whilst PbO, is undamaged""". 5 . Above 22 Am-' deterioration rate may be low. but PhO, coating is poor and interspersed with PbCI,. 4. Tests have indicated failure to form
of practical lead alloys in waters of differing resistivities. In electrolytes containing both sulphate and chloride ions, the sulphate ion favours the formation of lead sulphate which is rapidly transformed to lead dioxide. The continuing satisfactory operation of the anode depends upon the initial conditions of polarisation. The lead dioxide is of better quality and more adherent when formed below 108 Am -2, in solutions containing higher sulphate concentrations or when the water is agitated '''. It should be remembered that a minimum current density is necessary to ensure passivation of the anode and that anodes operating below this current density may experience rapid consumption rates. A minimum value of 32*3Am-' is quoted by Barnard et a1.1°3.The consumption rate of lead silver is high in the initial stages of operation as can be seen from Table 10.18. However, the rate in seawater, taken over an extended period, is generally taken as 0.06 kg A - l y - ' . If a lead alloy is used as a ship's hull anode, consideration should be given both to the make-up of the water in which the anode is initially passivated and that in which it will normally operate. The same consideration will apply for static structures in estuarine waters. It should be noted that lead dioxide will discharge if electronically connected to a more base material, when in an unenergised state. The reverse current leakage of a rectifier will allow this to happen to a small extent if the rectifier is faulty, with the consequent formation of lead chloride and corrosion of the anode. Recent experience with Pb-6Sb-1Ag and Pb/Pt anodes operating in seawater at depths greater than 25 m has revealed a marked increase in consumption rate compared with that found on the surface. Hollandsworth and Littauer have calculated that on a fully formed anode at 400 Am - 2 , only 6x of the current is used to maintain the passive film, yet at a depth of l8Om this percentage increases to 2 x and results in a 30-fold increase in consumption rate. They propose that a combination of
10:76
IMPRESSED-CURRENT ANODES
the mechanical forces acting on the PbO, at increased depths, and the reduction in the evolution of chlorine, are responsible for the increased consumption rate. It is therefore recommended that lead anodes are not used at depths below 25 m. Leed/Patinum Bi-electrodes
The insertion of platinum microelectrodes into the surface of lead and some lead alloys has been found to promote the formation of lead dioxide in chloride solutions ' 16, I". Experiments with silver and titanium microelectrodes have shown that these do not result in this improvement ''O. Similar results to those when using platinum have been found with graphite and iridium, and although only a very small total surface area of microelectrodes is required to achieve benefit, the larger the ratio of platinum to lead surface, the faster the passivation ' 1 6 . Platinised titanium microelectrodes have also been utilised. Lead dioxide will readily form on lead with a platinum electrode as small as 0-076mm in diameter""'". It has been observed that the current density on the platinum is considerably less than on the lead dioxide once polarisation has been achieved, the proportion of current discharged from the platinum decreasing with increase in total current. Additions of antimony, bismuth and tin to the lead appear to be detrimental. There is an indication that the addition of 0.1 Yo Ag is almost as effective as 1070 and additions as low as 0.01070 has been utilised in practice. Dispersion-hardened lead alloys have been unsatisfactory, showing pronounced spalling in the direction of extrusion. Pb-O.1Te-0- 1Ag has been also used with apparent success 'I9. A typical anode for practical use would be in the order of 25 to 48 mm in diameter, with hard platinum alloy pins of 0-50mm diameter by 10mm length, spaced every 150 to 300 mm and progressively positioned around the circumference120.The pins are a press fit into holes in the lead or lead alloy (approximately 0.1 mm diametric interference) and lie flush with the surface. The lead is peened around the pins to improve the mechanical and electrical contact. The action of platinum microelectrodes has been extensively studied 10,105. Trials carried out by Peplow have shown that lead/ platinum bi-electrodes can be used in high velocity seawater at current densities up to 2000Am-' and that blister formation with corrosion under the blisters is decreased by the presence of platinum microelectrodes. The current density range in which the anode is normally operated is 200-750 Am -* with the maximum working current density quoted as 1000Am-'. The consumption rate of these anodes ranged from 0.0014 kg A - l y - ' to 0-002kg A - l y - ' at 500Arn-,, but increased to 0-003kg A - I y - l at 2000Am-'2'21 can be summarised as follows: The results of work in this field '"* I . Pt acts as a stable electrode for nucleation of PbOz and limits PbCl, formation. 2. In the case of a lead anode (without a platinum microelectrode), the
IMPRESSED-CURRENT ANODES
10:77
PbO, thickens during prolonged polarisation with the consequent development of stresses in the film. 3. The stresses result in microcracks in the PbO,, thus exposing the underlying lead, which corrodes with the formation of voluminous PbCI,, resulting in blisters; the resistance of the anode increases and high voltages are required t o maintain the current (if the voltage is maintained constant the current falls to a low value). The platinum microelectrode appears to act as a potentiostat and maintains the potential of the Pb-solution interface at a crack at a value that favours the re-formation of PbO,, rather than the continuous formation of PbCl, which would otherwise result in excessive corrosion. It is known that an increase in the resistance of the electrode indicates that corrosion is taking place with the formation of an insulating film of a lead compound, and this is confirmed in practice by observation of the anodes, which reveal localised areas coated with white corrosion products, although the PbO, remains intact at other areas. However, it is possible that an insulating film forms over the whole surface thus isolating the conducting PbO, from the lead. Wheeler’24suggested that the sole function of the platinum is to provide a conducting bridge between the lead and the PbO,. It has been demonstrated that, although initially the PbO, nucleates at the surface of the platinum, the initially formed PbCI, is rapidly converted into PbO, that is in direct contact with the leadLi6. The formation of PbO, is favoured in solutions containing passivating anions such as SO:- and in chloride solutions of intermediate concentrations; very high and very low concentrations of chloride inhibit the formation of PbO,. The platinum/lead bi-electrode performs best in seawater, and is not recommended for use in waters of high resistivity.
Lead/Magnetite Composites It has been demonstrated that particles of conducting Fe,O, in a P b matrix can produce results similar to that of platinum, in acting as stable nucleation sites for PbO, f ~ r m a t i o n ” ~Composite . Pb/Fe,04 anodes containing 10, 15 and 20% Fe,O, were prepared by mixing powders of the constituents (Pb 30 to 60 mesh, Fe,04 72 mesh), then compacting at a pressure of 300 MNm - 2 These anodes were found to operate successfully in both artificial seawater, resistivity 0.25 ohm m and in this water diluted with distilled water to give a higher resistivity of 10 ohm m. In seawater the anodes were found to operate in the current density range 100-1 000Am-’. with a weight loss of 50g A - ’ y -’ recorded for a 20% composite anode at 300Am-2. No initial rise in voltage at a constant current density was observed, as is the case with Pb/Pt electrodes where the potential increases due to the formation of PbCI,, with the steady-state potential of the anodes found to be dependent upon the Fe,O, content. In fresh water solutions, composite anodes were also able to form a passivating PbO, film. Although an induction period was necessary before stabilisation was complete, in the case of a Pb-10% Fe,O, composite, Hill reports that at current densities less than 150 Am - 2 , the anodes were unable to stabilise.
10:78
IMPRESSED-CURRENT ANODES
Consumption rates similar to those in artifical seawater were reported for the Pb-20% Fe,O, composites, which were found to give the optimum performance. However, in tap water with a high SO:- and CO3- concentration and low CI concentration (36 ppm), consumption rates of 100 g A y were recorded.
-’
-’ ’
Lead Dioxide On Other Substrates
Lead dioxide on graphite or titanium substrates has been utilised as an anode in the production of chlorate and hypochlorites’” and on nickel as an anode in lead-acid primary batteriesI2’. Lead dioxide on a titanium substrate has also been tested for use in the cathodic protection of heat exchangers*’ and in seawater may be operated However, this anode has not at current densities up to 1 OOOAm-’ gained general acceptance as a cathodic protection anode for seawater applications, since platinised Ti anodes are generally preferred.
Carbonaceous Materials Carbon
The corrosion product is predominantly carbon dioxide, but considerable amounts of free oxygen are produced at the anode surface, particularly in fresh-water applications, and can attack both the carbon and any organic binders used to reduce its porosity. For this reason carbon anodes for underground service are used in conjunction with a carbonaceous backfill. If all the oxygen produced were to combine with the carbon the maximum theoretical wastage rate would be of the order of 1 kg A - ’ y - ’ I3O. However, in practice the rate is usually of the order of 0 - 2 kg A - ’ y -’, and in coke breeze may be as low as 0.05 kg A - l y - ’ . In seawater, where chlorine is the predominant gas produced, to which carbon is immune, any oxygen formed will be quickly removed and the corrosion rate may be very low.
Graphite
Graphite is a denser crystalline form of carbon. Graphite anodes are prepared by heating calcined petroleum coke particles with a coal tar pitch binder. The mix is then shaped as required and heated to approximately 2 800°C to convert the amorphous carbon to graphite”’. Graphite has now superseded amorphous carbon as a less porous and more reliable anode material, particularly in saline conditions. The performance of graphite in seawater, where chlorine is the principal gas evolved, is considerably better than in fresh water where oxygen is produced. Graphite is immune to chlorine and has a long history in the chemical industry in this and similar applications 13’.
IMPRESSED-CURRENT ANODES
10:79
It is current practice to impregnate the graphite, traditionally with linseed oil, although synthetic resins are also successful. The concept behind impregnation is to reduce the porosity and hence inhibit subsurface gas evolution or carbon oxidation which would initiate spalling and early anode failure. Electrode processes occur to a depth of 0.5 mm below the surface of the anode and the true current density can be shown to be only 1/400th of the value indicated by the superficial geometrical area 133. Acidity has been found to increase the wear rate'34and so has the presence of sulphate ionsI3'. Indeed, when buried in soils containing 2% SO:- Jakobs and Hewes88 report graphite consumption rates of 1-56kg A-l y - ' at 21 e 6 Am -', which is considerably higher than the theoretical maximum consumption rate. These factors must be considered with regard to the operating environment and the chemical treatment of backfill. The material can be easily machined being a natural lubricant. It has a negligible contact resistance and it is relatively simple to make a sound cable joint, although the comments regarding cable/anode joints discussed under high-silicon iron also apply. The material can be d.c. welded under high pressure argon. It is brittle but a little more shock resistant than silicon iron, in that it can absorb energy by localised damage; it is of course a lighter material to handle. The anode is not recommended for use in water at above 5OoC, where the consumption rate increases rapidly and erratically. It is no longer the practice to use this material in cooling water plant where secondary attack from contact with the relatively noble pieces of anode may occur, should damage take place. The wastage rate of graphite is lower in seawater at higher current densities because of the preferential evolution of chlorine. Table 10.19 gives some results obtained with graphite under different conditions. Results obtained with one particular installation using a 100 mm diameter anode 1 m in length operating at 6 * 9 A m - *indicate a predicted life of 20 years'36. Table 10.19 Performance of graphite Environment
Backfill Hot Water Seawater Seawater Fresh water Fresh water Mud
Wastage rate (kg A - ' y-I) 0.9
0.9 0.045 Little 0.45 0.45 0.91-1.36
Current density (Am-')
Reference
10.8
so
-
I08 111 10s 108 111 173
4.5-115 10.8
3.5 2.7 71
Graphite anodes when used in soils are invariably placed in a carbonaceous backfill. This helps to compensate for the lower electrical resistivity of graphite when compared with silicon iron. In such an environment, no build-up of a film of high resistance between the anode and backfill occurs, unlike silicon-iron anodes where the resistance can increase with time 13'. Failures of graphite anodes can occur by corrosion of the anode connection, Le. high current densities at either end of the anode resulting in
10:80
IMPRESSED-CURRENT ANODES
excessive consumption rates often referred to as ‘end effect’ corrosion, sealant failure or surface contamination 13*. Conductive Polymers
A continuous polymer anode system has been developed specifically for the cathodic protection of buried pipelines and tanks. The anode, marketed under the trade name An~deflex’~’, consists of a continuous stranded copper conductor (6AWG) which is encased in a thick jacket of carbonloaded polymer, overall diameter 12.5 mm. To prevent unintentional short circuits an insulating braid is sometimes applied to the outer surface of the conductive polymer. The anode may be operated in the temperature range - 18°C to 65°C and at currents up to 0.05 A per linear metre in soil and 0.01 A per linear metre in water, which corresponds with an effective maximum current densities of 0.66 Am-’ in soil and 0.13 Am-* in water. No precise details on the anode consumption rate have been provided by the manufacturer, but since the electroactive material is carbon the consumption rate would be expected to be of a similar order to that exhibited by graphite anodes. The anode may be installed in conventional groundbeds or be laid in close proximity to the cathode, e.g. parallel to a pipeline route. The anode may be buried either directly in soil or in carbonaceous backfill. The major applications for this material are tank protection, internal protection, mitigation of poor current distribution and hot spot protection, i.e. to supplement conventional cathodic protection systems and provide increased levels of cathodic protection in areas that exhibit low levels of protection. The disadvantage of this anode system for the cathodic protection of pipelines is that the anode length provided by one single connection to the d.c. power source is limited by the ohmic losses along the copper conductors. Thus, the required current output per unit length and soil resistivity are limiting factors and a number of anode connections may be required to protect long lengths of pipeline. The anode has a poor chemical resistance to oils and should not be used in situations where oil spillage may occur.
Carbonaceous Backfills Coke breeze is used as an anode extender thus producing an anode with an enormous surface area, its main component being carbon. By virtue of its porosity it gives a large volume-to-weight ratio of conducting medium suitable for anodic conditions. This allows the economic extension of groundbed anodes both linearly, for decrease in resistance to ground, and volumetrically for longevity. The grading of the coke is of some importance in that too large a grade offers large local contact resistance, leading to uneven consumption, whilst an excessively fine coke leads to over-tight compaction and gas blocking (gaseous polarisation). Chemicals are sometimes added, e.g. slaked lime (5-10% by weight), to counteract the tendency to lose moisture by electro-osmosis, since it is essential that an aqueous electrolyte is present to replace water consumed in the anodic reaction and conduct the
10:81
IMPRESSED-CURRENT ANODES
current to the protected structure. The alkaline material also serves to neutralise the anodically formed acid. Calcium sulphate is sometimes used in very dry conditions. In using coke breeze the consumption of the primary anode is reduced, as the majority of the conduction from the anode to the coke breeze is electronic rather than electrolytic. The electrochemical and physical nature of the coke results in the dispersion of the anode reaction (formation of CO, and 0,)over a large surface area, thus reducing attack on the primary anode. The coke is oxidised primarily to carbon dioxide, which in a suitable groundbed will escape into the atmosphere together with any oxygen formed. If all the oxygen reacted, the coke consumption would be 1 -02 kg A - l y - ' , but in practice consumption can be of the order of 0.25 kg A - ' y -' '37, depending upon the environment. Some typical properties of coke breeze and similar materials are shown in Tables 10.20 to 10.22. The densities given in Table 10.20 are for bulk material and are dependent upon grading. Flake graphite is not recommended for use in groundbeds as it tends to conglomerate and prevent gas emission. Table 10.20
Densities of backfill
Backfill in bulk
Density range kg m-3
Coal coke breeze Calcined petroleum coke granules Natural graphite granules Man-made graphite, crushed
650-800 700-1 100 1 100-1 300 1 100-1 300
Typical density kg m-'
690 720-850
Table 10.21 Typical coal coke specification for cathodic protection('69) To pass 16 mm screen To pass between 16 mm and 8 mm screen To pass between 8 mm and I mm screen To pass 1 mm screen
100% 8.9-9.8% 78-90% 1-14%
Fixed carbon Volatile matter Ash Moisture Sulphur Phosphorus Resistivity (uncompacted)
82.7min to 91% max 0.1% 8.6% 5% max, typically 4% 1.2% rnax, typically 0.42-0.7% 0.55 ohm m max, (typically 0.35 ohm m) 1.4
Specific gravity
Table 10.22
Resistivities of carbonaceous backfills (ohm metre)
Material
Dry
Tamped
Wer
Coal coke Graphite granules
0.55
0.45
1.50
I .20
0.15 0.20
10: 82
IMPRESSED-CURRENT ANODES
When coke breeze is tamped down the correct pressure to aim for is approximately 15 Nm-*. This will ensure integrity of the groundbed whilst in operation, remembering that it will be reducing in volume by chemical oxidation. A pressure of this magnitude will reduce the initial bulk resistivity of the coke. The usual main object in using coke breeze is to lower the resistance of the anode to remote earth, with the coke cross-section, in a typical groundbed, normally about 300 x 300mm. In fresh-water soil conditions, the higher than average current density at the ends of the primary anodes can be prevented by not exceeding an anode spacing of twice its length. This depends entirely on the care taken in preparation of the groundbed assembly and ratio of the anode/coke resistance to the anode/electrolyte resistance. As the electrolyte resistance decreases, with a consequent increase in current density at the ends of the primary anodes, either a reduction in anode spacing or increase in backfill cross-section should be considered, particularly in foreshore groundbeds where the electrolyte resistivity will be of similar magnitude to that of the backfill. In practice resistivities between 0.08 and 0-29ohm m have been recorded on coke breeze samples used in typical groundbeds. The effect of pressure on the measured value for resistivity of different coke samples has also been 14'. The resistivity of bulk metallurgical coke is given reported elsewhere140. as 0.024 ohm m with a slightly lower value of 0.020ohm m at a pressure of 0.43 Nm-2, whilst in calcined fluid petroleum coke at zero applied pressure the resistivity was 0.02 ohm m, which decreased to 0.002 ohm m when tested at an applied pressure of 1 31 MNm -'. Calcined petroleum coke breeze with a high fixed carbon content of 99% is used in deepwell applications. The material has a low particle size and, with suitable additives, may be converted into a slurry and pumped into a borehole. The sulphur content of this material is high (1 -4Vo), yet moisture (0-2%), ash (0.4%) and volatiles (0.4%) are low. The typical resistivity of this material is 0.15 ohm m. A petroleum coke with round grains is available specifically for borehole cathodic protection applications '42. The round grains ensure high porosity and enable gas to escape, allowing the coke to sink to the base of the borehole. This material has a higher bulk density than petroleum coke (1 185 kg m - 3 ) which enables it to sink to the bottom of the borehole, yet a lower fixed carbon content (93Vo), with higher ash (2.06%) and sulphur (5.3%) contents. The resistivity of this material is quoted as 0.1 ohm m.
Anodes used for the Cathodic Protection of Reinforced Concrete Structures The corrosion of reinforcing steel due to chloride contamination in concrete is an increasingly serious problem, and interest in cathodic protection as a means of mitigating corrosion on reinforced steel has become of some importance in recent years. Reinforced concrete structures that are fully immersed or buried in a corrosive environment may generally be protected using conventional cathodic
IMPRESSED-CURRENT ANODES
10:83
protection groundbed design. However, for the cathodic protection of above-ground reinforced concrete structures, e.g. bridge decks, jetties, tunnel parking garages, and certain concrete buildings, a number of specific anode systems have been developed. These are applied directly on to the concrete surface and often consist of a primary and secondary anode. The various anode systems used specifically for reinforced concrete cathodic protection have been discussed in recent literature'". 14'* '66 and will now be summarised.
Conductive Overlay Systems
Some of the early systems were based on the use of silicon iron primary anodes and a coke breeze/asphaltic cement (85%/15%) mix as the secondary anode to ensure uniform current d i ~ t r i b u t i o n ' ~The ~ . silicon iron anodes were held in position using a non-conductive epoxy, then covered with a conductive cement. Fromm has investigated the performance of different coke breeze/asphalt mixes, and developed a mix containing only 45% coke breeze which had a resistivity of 0.03 ohm m and a voids content of 5%. This was reported to give good results. The conductive mix was then applied over the primary anodes, either silicon iron or graphite, to a total thickness of 50 mm, then given a protective top coat. Schutt '41 reported that the coke breeze specification and conditions in which the mix is prepared are important factors in determining the optimum operation of the conductive cement mix, whilst further details on the coke breeze asphalt mix composition are given by Anderson'48. Conductive concrete mixes, with a polymer binder have also been developed as an anode system specifically for reinforced concrete cathodic protection systems 149. Conductive overlay systems are not practical propositions on vertical surfaces or surfaces where weight restrictions are important. However, they are proven cathodic protection systems, and should be considered in conjunction with other reinforced concrete cathodic protection system anodes.
Conductive Polymers
A conductive polymer electrode has been designed specifically for the cathodic protection of steel reinforcing bars in concrete and is marketed under the trade name FerexI5'. The anode consists of a 16 AWG stranded copper conductor surrounded by a carbon-loaded polymeric coating similar to that used on the Anodeflex system'39)to provide a nominal anode diameter of 8 mmL5'.The manufacturer claims that at the maximum recommended current density of 0.08 Am-' the anode life in concrete will be 32 years with a proportionately longer life at lower current densities. The major electrochemical reaction at the anode surface is oxygen and chlorine evolution coupled with oxidation of the active carbon to carbon dioxide. Eventually all the carbon is removed from the anode coating and this allows perforation of the copper conductor leading to ultimate anode failure.
10: 84
IMPRESSED-CURRENT ANODES
The anode is fixed to the concrete using non-metallic fixings and may be supplied as a prefabricated mesh or more often as a continuous anode strand which is laid over the surface of the structure to be protected. The spacing between the anode strands may be adjusted to give the required current distribution and current density per unit area of concrete necessary to provide cathodic protection to a particular structure. A number of anode connections will be made to the d.c. power source using proprietary splice kits (approximately one for every 60-80 mz of concrete to be protected). This will provide redundancy for anode failure and reduce ohmic losses along the anode cable. Care must also be taken not to expose the copper conductor during installation or anode failure could take place. Once fitted to the concrete surface a 15 mm thick cementitious overlay is applied above the anode mesh, as recommended by the anode manufacturer, although thickness of up to 35-40 mm have been applied in some instances. Failures due to delamination of the gunite coating have been reported in the USA, but have not been observed to any significant extent in Europe'68, although some early failures of the anode system have been associated with high local current densities in areas of low concrete cover and high moisture or salt contentla. The major application of this anode system is therefore on structures that are relatively dry with a uniform current requirement. S/otted Anode Systems
These consist of a number of parallel slots cut into the concrete surface. Each slot is then filled with a secondary anode of carbodgraphite fibres embedded in a conductive polymer grout. The current to each of these secondary anode systems is provided by a primary anode of platinised niobium wire placed in slots filled with conductive polymer which acts as the primary anode, these slots intersecting each slot of graphite fibre/conductive polymer at right angles. These systems have not been installed to any significant extent and have now been superseded by conductive paints, conductive polymers or titanium mesh anode systems. Conductive Paints
Conductive paints (resins) have recently been used for the cathodic protection of steel reinforcing bars in concrete, but they are always used in conjunction with a primary anode material, e.g. platinised-niobium or platinised-titanium wire or a conductive polymer rod. Brown and Fessler Is* have conducted a laboratory evaluation of conductive mastics that can be brushed or sprayed onto the concrete surface to achieve the necessary thickness. However, the most extensive study on conductive paints for cathodic protection purposes has been undertaken by the Federal Highway Authority'49. A total of nine commercially available resins were evaluated in this work. It was shown that neither thermal cycling, freeze thawing nor the application of cathodic protection currents
IMPRESSED-CURRENT ANODES
10:85
resulted in any deterioration of the most successful paint system which was designated Porter XP90895,but now referred to as DAC-85, a solvent based acrylic mastic containing graphite. Minor failures with this system have been reported but only in localised areas with a high chloride contact. The anode system generally consists of platinised titanium or niobium wire laid in strips with the layers of carbon fibre interleaved between the strips. The paint is then mixed and applied on site. The paint consists of blends of resin and fine particles of coke. The performance of some paint systems is poor because of attempts to operate the anodes at currents in excess of 0- 1 Am - 2 . The advantages of conductive paints are that they are easy to apply and a concrete overlay is not required. They can be applied to complex shapes and are not a problem where weight restrictions are imposed. Mixed Metal Oxide Coated Titanium Mesh
The most recently developed anode for the cathodic protection of steel in concrete is mixed metal oxide coated titanium mesh'53-'s5.The anode mesh is made from commercially pure titanium sheet approximately 0-5-2 mm thick depending upon the manufacturer, expanded to provide a diamond shaped mesh in the range of 35 x 75 t o 100 x 200mm. The mesh size selected is dictated by the required cathode current density and the mesh manufacturer. The anode mesh is supplied in strips which may be joined on site using spot welded connections to a titanium strip or niobium crimps, whilst electrical connections to the d.c. power source are made at selected locations in a suitably encapsulated or crimped connection. The mesh is then fitted to the concrete using non-metallic fixings. The active coating consists of a thermally deposited mixed metal oxide coating, the composition of which is considered proprietary information, although it is known that certain filler materials, e.g. Ta, may be added to the mixed metal oxide to reduce the precious metal content of the coating, and hence the cost of the anode. The coating composition is iridium-rich to favour oxygen rather than chlorine evolution, and to assist in reducing the formation of acidic conditions at the anode-concrete interface. It has been shown that the evolution of chlorine can result in the formation of an equivalent quantity of acid as that generated by oxygen evolution, because of the reaction between chlorine and water to form hydrochloric acid and hypochlorite'56. The latter is a strong oxidising agent and may have a detrimental effect on the concrete surrounding the anode mesh. Recent work has shown that acid attack on the concrete surrounding the mesh is limited with 0 - 2mm recorded after 1 - 5 years at an anode current of 0-76 Am - z which corresponds to only 1 mm after 25 years at a current density of 0 . 2 Am -'156. The current density applied t o the electroactive coating has been set at 0.1 Am2, whilst for short-term polarisation current densities up to 0.2 AmZ may be applied. However, certain anode manufacturers now state that a maximum current density of 0 . 2 Am - 2 may be used for long-term polarisation and 0 - 4 A m - ' for short term use.'57 The current density range is
10 :86
IMPRESSED-CURRENT ANODES
limited by the concrete and the need to reduce the level of degradation at the anode-concrete interface. Indeed, for an anode current density of 0 - 2 Am2, the life of the coating would be 30-50years, based on a consumption rate of 87 mg A - ' y - ' and a mixed metal oxide coating thickness of 5gm-2.'57 The material once installed is then covered with a concrete coating, the minimum thickness of cover above the anode mesh is quoted as 10 mm, but 15 mm is preferred. The anode voltage must be limited to 10 V to avoid damage to the titanium mesh, whilst cementitious overlays with a fluoride or bromide content must be avoided. However, in practice because of the relatively large anode surface and low current required systems generally operate at approximately 2-3 V. Low iron levels in the aggregate must also be maintained to avoid staining and possible inclusion of iron in the titanium oxide film. The mesh is light, nominally in the range 0.1-0.25 kg m -2 dependent upon mesh type, so the only structural limitation is the weight of the cementitious overlay.
Reactive Metals Aturniniurn
This is not often considered for use as an impressed-current anode, although it has found limited use in fresh-water tank protection, particularly where weight is a problem 159-161. To reach the required circuit resistance in high resistivity waters, it is necessary to use long extrusions of the order of 20 mm in diameter. The alloys H14 and H15 have been used for this purpose, pure AI being preferred in seawater IO8. For tank protection the life of the anode is very much dependent on the extent of pitting. Necking can be a problem if the water level drops below part of the anode for long periods. The wastage rate in this area can be twice the normal. In fresh water, voluminous corrosion products (namely Al(OH),) can cause quite large increases (two-fold or more) above the initial anode/ electrolyte resistance. This product, whilst not toxic, could prove an embarrassment in potable waters. Resistive polarisation is negligible in seawater use. The extent of the scale formation is a function of the nature of the water under consideration. Theoretically pure aluminium would be expected to dissipate of the order of 2 - 9kg A - I y - I , although a reasonably large safety factor should be used when considering anode integrity. Aluminium has also been successfully utilised as a trailing wire anode for the protection of ships, but this is no longer considered a practical application. Zinc
Zinc is seldom used as a power-impressed anode. It may be a convenient way of achieving a high initial current density, particularly where descaling is involved, but it does, of course, require the anode to be locally insulated from the cathode. Used as a power-impressed anode the energy rate per
IMPRESSED-CURRENT ANODES
10 :87
unit energy tends towards the theoretical value of 10.7 kg A - ' y -' instead of the usual 90% efficiency when used as a galvanic anode. In recent years, there has been interest in using zinc as a power-impressed anode for the cathodic protection of steel in concrete. The zinc is flame sprayed onto a grit blasted concrete surface to a final film thickness of approximately 250 pm. A primary anode is necessary. Early systems used brass plates as the primary anode, but more recent systems used platinised titanium or niobium wire anodes as the primary current conductor. The reason for the use of zinc as a power-impressed rather than a sacrificial anode is that the high concrete resistivity limits the current output, and a higher driving voltage than that provided by the e.m.f. between zinc and steel in concrete is used to provide the necessary current output. No cementitious overlay is required, although it may be advisable to paint the top surface of the sprayed zinc to prevent atmospheric corrosion of the zinc anode.
Summary A comparison of typical properties of cathodic protection materials is given in Table 10.23, but is by no means comprehensive. It is obvious that the modification of an alloy, environment or other important factors will be reflected in the life and output characteristics. In some cases the maximum voltages and current densities recommended can be vastly exceeded. In others, particularly where abnormal levels of environmental dissolved solids are met, factors of safety should be applied to modify the proposed figures. Acceptance of a much reduced or uncertain life, weighed against a possible economy, may also influence the chosen working limits. For example, the life of ferrous alloy anodes may, in practice, be only two-thirds of that expected because of preferential attack eventually leading to disconnection of all or part of the anode from the source of e.m.f. Table 10.23 must be taken only as a guide and interpreted in the best manner available, preferably using experience in that particular environment or operational requirement. Table 10.23 should be consulted in conjunction with the text and references, specifically those covering the whole range of cathodic protection anodes 10*163*'64*167. Consideration must be given to practicability, the factor of safety required, the environment, physical and electromechanical hazards, maintenance, operation, installation, availability, cost and the economics of replacement 16*. J. W. L.F. BRAND P. LYDON REFERENCES 1 . Cathodic Corrosion Control Ltd. UK Patent No. 880 519 (1961) 2. Shreir, L. L., Platinum Metals Review, 21 No. 4, 110-121 (1977) 3. Anderson, D. B. and Vines, R. F., Extended extracts of the Second International Con-
gress on Metallic Corrosion, March (1963) 4. Warne, M. A. and Hayfield, P. C. S., Materials Performance, 15 No. 3, 39-42 (1776) 5. Hayfield, P. C. S. Warne, M. A. and Jacob, W. R., 'The Conditions for the Successful Use of Platinised Anodes', Harrogate, 24-28 November (1981) 6. Baboian, R., paper 183, Corrosion 76 (1976) 7. Lowe, R.A., Materials Protection, NACE, Houston, 23-24 April (1966)
10:88
IMPRESSED-CURRENT ANODES
Table 10.23 Comparison of typical properties of cathodic protection anode materials
Platinised tantalum
Plalinised niobium
Platinised titanium
Thermally deposited noble metal
oxide on titanium
Highsilicon/ chromium iron
Approximate consumption kg A - ’ y-’
See Pt
Suggested minimum factor of safety on cross-sectional area Max recommended current density (seawater Am-‘) Max recommended current density (fresh water Am-’) Max recommended current density (soil Am-’) Max recommended voltage (seawater) Max recommended voltage (fresh water) Specific resistivity 20°C ohmm x Density (kg m -’) Tensile strength (approx) (Nmm-’) Hardness (approx)
1.2
1-2
1.2
1.8
2000
2000
600
I 20
B
B
100
I20
B
B
B
60
B
B
E
D
B
B
B
D
123
15.2
48.2
72
16 600 1 260
HV
8 750 240 390 75-95 HV
YES YES YES NOW) YES
YES YES YES NOW) YES
See Pt
0.5 10 6 x
B
80-100
-
0.25-1.0
B
200
7000 I IO
6 Mohs
520 HB
YES YES YES NO(M) YES
YES(M) NOW) YES YES NO
General uses
Marine environment Potable waters In carbonaceous backfill Buried directly in soil High-purity liquids
Notes: A Used with carbonaceous backfill, see text -depends on water resisiivity/backfill resistivity B See text C Normal maximum longitudinally D Limited by local or environmental safety requirements regarding apparatus and/or earth voltage E Minimum current density io ensure passivation 50 Am-’ F Resistivity of PbO, 10-100 ohm m
gradient regulaiions
8. Cotton, J . B., Chem a n d f n d . ( R e v ) , 68 (1958) 9. Cotton, J. B., Williams, E. C. and Barber, A. H., UK Patent 877 901 (1958) 10. Shreir, L. L. and Hayfield, P. C. S., ‘Impressed Current Anodes’, Conference on Cathodic Protection Theory and Practice-The Present Status, Coventry, 28-30 April ( 1982) 11. Warne, M. A., Materials Performance, 18 No. 8, 32-38 (1979) 12. Jacob, W. R., paper 5 , ‘Substrate Materials for Platinised Anodes’, Proc. Symposium on Cathodic Protection, London, May (1975) 13. Baboian, R., Proc4th International Congress on Marine Corrosion and Fouling, Antibes, France (1976) 14. U.S. Patent 3 443 055 (1966) 15. Warne, M. A. and Hayfield, P. C. S.. ‘Durability Tests on Marine Impressed Current Anodes’, IM1 Marston/Excelsion Ltd., 15 March (1971) 16. Bibikov, N. N., Povarova, L. V . and Kashcheeva, E. A., Prof. Mer., 11 No. 1, (1975) 17. Hames, W. T., Aircraft Producfion, London, 20, p. 369 (1958) 18. Dagdale, I. and Cotton, J. B., Corrosion Science, 4, 397-41 1 (1964) 19. Commercial Guarantee, Martson Excelsior. (1972) 20. Czerny, M., 4th International Congress on Metallic Corrosion, Amsterdam (1969) 21. Hayfield, P. C. and Warne, M. A,, paper 38, Corrosion 82 (1982)
10 :89
IMPRESSED-CURRENT ANODES
Table 10.23 Comparison of typically properties of cathodic protection anode materials Highsilicon/ iron
Magnetite
Iron
Steel
pb-65b- Lead/ Graphire Aluminium IAg platinum
0.25-1.0 0.001-0.04 6.8-9.1 Approx 4.5-6.8 0.09 B B 9.5 B 1.8
1.2
N
77
-
I20
1.8
1.8
1.8
1.5
L
200 E
L
L
L
L
L
-
2.0
Zinc
breeze
0.09 B
0.1-1.0
1.5
1.5
2
500 E
30
20
20
-
2.5
20
20
-
10
2.5
2.5
D
D
D
D
27
6.2
B
2700 85
7400
B
-
60
30
5
5
5
D
D
D
D
D
D
D
D
D
D
72
8 x 10’
22 C 11300 7700 10900 7820 7100 500 300 150 30 25 25 130-160 120-170 140-170 13-10.7 4 H B HB HB HB HB
1560 28 28
YES NO YES NO(M) NO
YES YES YES NO(M) NO
-
10.8
0.5
1.2
1.5
B
2.6
A
7000 I30
4750
-
450 H B
NO YES YES YES NO
YES YES YES NO(M) NO
17
12
YES NO YES NO(M) NO
55
YES NO YES NO(M) NO
D
D
-
D -
D
25
700
F
C
YES NO NO NO NO
YES NO NO NO NO
80
85
-
-
-
~
-
SHORE NO(N) YES NO NO(M) NO(M)
YES YES NO NO NO NO(M) YES NO
NO
G Voltage on PI I .35 V minimum H Electrodeposited K 50% cold rolled L I f in free suspension in moving water. no limit. local effects under high current density may increase wastage rate‘ M May be used in the environment under special circumstances N High consumption rate in this environment
22. Segan, E. G . et al., Titanium Anodes in Cathodic Protection, Final Report, Army Construction Engineering Research Laboratory, Champaign, Illinois USA, Jan. (1982) 23. Baboian, R., Materials Performance, 16 No. 3, 20-22 (1977) 24. Toncre, A. C. and Hayfield, P. C. S., paper 148, Corrosion 83 (1983) 25. Baboian, R., paper 149, Corrosion 83 (1983) 26. Warne, M. A. and Hayfield, P. C., British Corrosion Journal, 6 , 192-195, Sept. (1971) 27. Pathmanaban, Phull, B., Proc UK corrosion 82, Hammersmith, London (1982) 28. Juchniewicz, R., Platinum Metals Review, 6 , 100 (1962) 29. Hoar, T. P., Electrochim Acta, 9, 599 (1964) 30. Juchniewicz, R. and Hayfield, P. C . S., 3rd International Congress on Metallic Corrosion, Moscow, Vol. 3, p. 73 (1969) 31. Juchniewicz, R., Walaskowski, J., Bohdanowkz, W. and Widuchowski, A,, 8th International Congress on Metallic Corrosion, Mainz (1981) 32. Efird, K. D., Materials Performance, 21 No. 6, 51-55 (1982) 33. Pompowski, J., Juchniewicz, R., Walaszkowski, E., Strelcki, H. and Sadowska, J., Marine Corrosion Conference, Gdansk, p. 87 (1967) 34. Love, T. J., Power U S A . 80-81, Feb (1981)
10:90
IMPRESSED-CURRENT ANODES
35. Dreyman, E. W., Moteriols Performonce, 11 No. 9, 17-20 (1972) 36. Nekosa, G. and Hanck, J., ‘Laboratory and Field Testing of Platinised Titanium and Niobium Anodes for Power Plant Applications’, The Electrochemical Society Meeting, Pittsburgh, Pensylvania, October (1978) 37. Proc. Symposium on Recent Advances in Cothodic Protection, Marston Excelsior, Wilton, Birmingham, May (1964) 38. Lowe, R. A. and Brand, J . W. L. F., Moteriols Protection and Performance, NACE, 9 NO. 1 1, 45-47 (1970) 39. C.W.E. UK Patent 40. Berkeley, K. G. C., ‘The Use of Platinised Anodes on Land Based Installations’, Proc. Symposium on Cothodic Protection, London, May (1975) 41. Warne, M. A. and Berkeley, K . E., paper 244, Corrosion 80 (1980) 42. Lewis, T . H., paper 144, Corrosion 79, Atlanta Georgia, USA (1979) 43. Stevens, R. H., US Patent 1077 894 and 1007 920 November (1913) 44. Baurn, E., US Patent I 4 7 7 0 0 0 August (1922) 45. Rosenblatt, E. F. and Cohen, J. G., ULS Patent 2 719 797 October (1955) 46. Private communication, CWEIMarston Excelsior (1966) 47. Hayfield, P . C. S., Materials Performance, 20 No. 11, 9-15 (1981) 48. Hayfield, P. C. S., paper 103, Corrosion 81, Toronto, Canada (1981) 49. Gleason. J . D., Materials Performance, 18 No. I. 9-15 (1979) 50. Sly, P . M., Platinum Met01 Review, 24 No. 2, 56-57, (1981) 51. Taturn, J. F., Moterials Performance, 18 No. 7, 30-34 (1979) 52. Toncre, A. C., Materiols Performance, 19 No, 3, 38-40 (1980) 53. Stephens, R. W., paper 144, Corrosion 83 (1983) 54. Preiser, H. S. and Cook, F. E., Corrosion, 13, 125-131 (1957) 55. Beer, H. B., UK Patent I 147 442 (1965) 56. Beer, H . B., UK Patent 1 195 871 (1967) 57. Bianchi, G., de Mora, V. and Gallone. P., US Patent 3 616 445 (1967) 58. De Nora, V. and Kuhn von Burgsdorf, J . W., Chem. Ing. Tech., 47, 125-128 (1975) 59. Katowski, S., ‘Chlorine’, in Ullmons Encyclopedio of Industrial Chemistry, Chapter 7 (1986) 60. Schrieber, C. F. and Mussinelli, G. L., paper 287, Corrosion 86 (1986) 61. Matusumoto, Y.,Tazawa, T., Muroi, N. and Sato, E., J. Electrochem. Soc., 133 No. I , 2257-2262 (1986) 62. Hook, V. F., Givens, J. H., Suarez, J . E. and Rigsbee, J . M., paper 230, Corrosion 88, St Louis USA (1988) 63. Reding, J . T., paper 9, Corrosion 97, San Francisco, USA (1987) 64. Kroon, D. H. and Schrieber, C. F., paper 44, Corrosion 84 (1984) 65. Takasu, M. and Sato, E., Corr. Eng., 26 No. 9, 499-502 (1977) 66. Trade name of Ebonex Technologies Inc., California USA, (European Patent Application 0047595; US Patent 4 422 917) 67. Product Information Sheet, Ebonex Technologies Inc (1988) 68. Kimmel, A. L., Corrosion, 12 No. I , 63 (1956) 69. Bernard, K. N., Chem. ond Ind. (1954) 70. McAnenny, A. W., PIEA News, July (1941) 71. McAnenny, A. W., PIEA News, 10 No. 3 , 11-29 (1940) 72. McAnneny, A. W., US Patent, 2 360244 (1944) 73. Peabody, A. W., Moteriols Protection, 9 No. 5 , 13-18 (1970) 74. Bengough and May, J. Inst. Met., 32 No. 2, Part 5 (1924) 75. Cotton, J . B., Chem and Ind. Review, 68, 492 (1958) 76. Brand, J . W. L. F. and Tullock, D. S., CWE International Report, London (1965) 77. Redden, J . C., Materials Protection, 5 No. 2 (1966) 78. Applegate, L. M.,Cothodic Protection, McGraw Hill, New York (l%O) 79. Bryan, W. T., Moteriols Protection and Performonce, 9 No. 9, 25-29 (1970) 80. Peabody, A. W., Control of Pipeline Corrosion, NACE (1967) 81. Tudor, S., Miller, W. L, Ticker, A. and Preiser, H. S., Corrosion, 14 No. 2,93t-99t (1958) 82. NACE T-2B Report, Corrosion, 16 No. 2, 651-691 (1960) 83. NACE T-2B Report, Corrosion, 13 No. 2, 103t-107t (1957) 84. NACE T-2B Report, Corrosion, 10 No. 12, 62-66 (1955). 85. Doremus, C. L. and Davis, J . G., Materials Protection, 6 No. I , (1967)
IMPRESSED-CURRENT ANODES
10:91
Bryan, W. T., Materials Protection and Performance, 9, No. 9, 25-29 (1970) Durion Company, Dayton, Ohio, USA, Technical Data Bulletin DA/7c (1984) Jakobs, J. A. and Hewes, F. W., paper 222, Corrosion 81 (1981) McKinney, J. W., Materials Performance, 18 No. 11, 34-39 (1979) 90. Allmand, A. J . and Ellingham, H. J. T., Applied Electrochemistry, Edward Arnold, London (I93 1) 91. Linder, B., Materials Performance, 18 No. 8, 17-22 (1979) 92. Ko fstad, P., Nonstoichiometry, Digusion and Electrical Conductivity in Binary Metal Oxides, Wiley Interscience (1972) 93. Miller, J., Danish Corrosion Centre Report, May (1977) 94. Jakobs, J . A., Materials Performance, 20 No.5 , 17-23 (1981) 95. Matlock, G. L., paper 340, Corrosion 84 (1984) 96. Kubicki, J. and Trzepierczynska, J., Ochrono Przed. Korozja, Nos. 11-12. 301-3 (1980) 91. Wakabayshi, S. and Aoki, T., Journal de Physique, 4, pC1 (1977) 98. Kumar, A., Segan, E. G. and Bukowski, J., Materials Performance, 23 No. 6, 24-28 86. 87. 88. 89.
( 1984)
99. Tefsuo Fujii et a/., Corrosion Eng. (Japan), 29 No. 4, 180-184 (1980) 100. Moller, G. E. et al., Materials Protection, 1 No. 2 (1962) 101. Littauer, E. and Shreir, L. L., Proc. 1st International Congress on Metallic Corrosion 102. Shreir. L. L.. Corrosion, 17, 1881 (1961) 103. Bernard, K. N., Christie, G. L. and Gage, D. E., Corrosion, 15 No. 11, 501t-586t (1959) 104. Von Fraunhoffer, J . A., Anti Corrosion, November, 9-14, and December, 4-7 (1986) 105. Kuhn, A. T., The Electrochemistry of Lead, Academic Press (1979) 106. Fink, C. G. and Pan, L. C., Trans. Electrochem. SOC., 46 No. 10, 349 (1924) 107. Fink, C. G. and Pan, L. C., Trans. Electrochem. SOC., 48 No. 4, 85 (1926) 108. Morgan, J . H., Cathodic Protection, Leonard Hill, London (1959) 109. Morgan, J. H.. Corrosion Technology, 5 No. 11, 347 (1958) 110. Private Report o n Films on Lead-Silver-Antimony Electrodes, Fulmer Research Institute to CWE Ltd. 111. Tudor, S. and Ticker, A., Materials Protection, 3 No. 1, 52-59 (1964) 112. Private report on lead alloy produced by CWE Ltd., Materials Laboratory of New York Naval Shipyard to CWE Ltd. (1957) 113. Kubicki, J. and Bujonek, B., Ochr. Przed. Koroz., 3, 41-45 (1982) 114. Dotemus, G. L. and Davis, J . E.,Materials Protection, NACE, 30-39, Jan (1967) 115. Hollandsworth, R. P. and Littauer, G. L., J . Electrochem. Soc., 119, 1521 (1972) 116. Shreir, L. L., Corrosion, I7 No. 3, 118t-124t (1961) 117. Metal and Pipeline Endurance Ltd., UK Patent 870 277 (1961) 118. Fleischmann. M. and Liler, M., Trans. Faraday SOC., 54, 1370 (1958) 119. Shreir, L. L., Platinum Metals Review, 12 No. 2, 42-45 (1968) 120. Shreir, L. L., Platinum Metals Review, 22 No. 1, 14-20 (1978) 121. Peplow, D. B., British Power Engineering, 1, 31-33, October (1960) 122. Shreir, L. L. and Weinraub, A., Chem. and Ind., No. 41, 1326. October (1958) 123. Shreir. L. L., Platinum Metals Review, 3 No. 2, 44-46 (1959) 124. Wheeler, W. C. E., Chem. and Ind., No. 75 (1959) 125. Shreir, L. L. and Metal and Pipeline Endurance Ltd., UK Patent 19823179 126. Hill, N. D. S., Materials Performance, 23 No. 10, 35-38, (1984) 127. Barak, M., Chem. Ind., UK, 20, 871-876 (1976) 128. Smith, J. F., Trans fMF, 53, 83, (1975) 129. Hamzah, H. and Kuhn, A. T., Corrosion J, 15 No. 3 (1980) 130. Palmquist, W. W., Petroleum Engineer, 2, D22-D24, January (1950) 131. Brady, G. D., Materials Performance, 10 No. IO, 20-23, (1971) 132. Heinks, H., fnd. Eng. Chem., 47, 684 (1955) 133. Bulygin, B. M., Ind. Eng. Chem., 32, 521 (1959) 134. Krishtalik, L. I., and Rotenberg, Z. A., Russian J . Phys. Chem., 39, 168 (1965) 135. Ksenzhek, 0. S. and Solovei, Z . V., J . Appl. Chem. USSR, 33, 279 (1960) 136. Oliver, J. P., AIEE Paper 52-506, September (1952) 137. Costanzo, F. E., Materials Protection, 9 No. 4. 26 (1970) 138. Tatum, J . F. and Tatum, Lady B., Replaceable Deep Groundbed-Anode Materials, International Conference of Marine Corrosion 139. Anode Flex - registered trade name, Raychem Lttl., USA
10 :92
IMPRESSED-CURRENT ANODES
140. Tatum, J . F. and Tatum, Lady B., Replaceable Deep Groundbed-Anode Materials, International Conference of Marine Corrosion 141. Espinolu, R. A., Mourente, P., Salles, M. R. and Pinto, R. R., Carbon, 24 No. 3, 337-341 (1986) 142. LORESCO, registered trade name, Cathodic Engineering Equipment Co Inc., Missouri USA 143. Stratful, R. F., ‘Experimental Cathodic Protection of Bridge Decks’, Transportation Research Record, 500 No. 1 (1974) 144. Wyatt, B. S. and Irvine, D. J., Materials Performance, 26 No. 12, 12-21 (1987) 145. Mudd, C. J., Mussinelli, G. L., Jettamanti, M., and Pedeferri, P., Materials Performance, 27 No. 9, 18-24 (1988) 146. Fromm, H. J., paper 19, Corrosion 76 (1976) 147. Schutt, W.R., paper 74, Corrosion 78 (1978) 148. Anderson, G., ‘Cathodic Protection of a Reinforced Concrete Bridge Deck’, American Concrete Institute Convention (1979) 149. Federal Highway Authority (USA). Cosf Eflective Concrete Construction and Rehabilitation in Adverse Environments, Project No. 4K, Annual Progress Report, Sept. (1981) 150. Registered trade mark Raychem USA (manufacturing and marketing rights purchased by Eltech Corporation, 1989) 151. Ferex Technical Data Sheet 152. Brown, R. P. and Fessler, H. J., paper 179, Corrosion 83 (1983) 153. Tvarusku, A. and Bennett, J . E., Proc. 2nd International Conference on Deterioration and Repair of Reinforced Concrete in Arabian Cuu, Bahrain, pp. 139-154 (1987) 154. Mudd, C. J., Mussinelli, G. L., Jettamanti, M., Pedeferri, P., paper 229, Corrosion 88, St Louis, USA (1988) 155. Hayfield, P. C., Platinum Metals Review, 30 No. 4, 158-166, October (1986) 156. Hayfield, P. C. S. and Warne, M. A., ‘Titanium Based Mesh Anodes in the Cathodic Protection of Concrete Reinforcing Bars’, presented at UK Corrosion, Brighton (1988) 157. Tectrode, registered trade mark, ICI, and Polymers, Technical Data Sheet 158. Technical Data Sheet, Bergsoe Anti Corrosion 159. Shepard, E. R. and Graeser, H. J., Corrosion, 6 No. 11, 360-375 (1950) 160. Jakobs, J. A., Materials Performance, 20 No. 5 , 17-23 (1981) 161. Russel, G. J. and Banach, J., Materials Performance, 12 No. 1, 18-24 (1973) 162. Brand, J. W. L. F., Cathodic Profection Electrical Rev., 781-783, December (1972) 163. Ferris, P., Corrosion, Australia, 11 No. 2, 14-17 (1986) 164. Berkley, K. E., paper 48, Corrosion 84, New Orleans, USA (1984) 165. Baboian, R., Materials Performance, 22 No. 12, 15-18 (1983) 166. Wyatt, B. S., ‘Anode Systems for Cathodic Protection of Steel in Concrete’, paper 23, Cathodic Protection Theory and Practice, 2nd International Conference, Stratford-uponAvon, UK, June (1989) 167. Moreland, P. J. and Howell, K. M., ‘Impressed Current Anodes Old and New’, paper 15, Cathodic Protection 2nd International Conference, Stratford-upon-Avon, UK, June (1989) 168. Wyatt, B. S. and Lothian, A., U K Corrosion 88, CEA, Brighton (1988) 169. Roxby Engineering Int. Ltd./Coal Products, private communication (1989)
10.4 Practical Applications of Cathodic Protection
The complexity of the systems to be protected and the variety of techniques available for cathodic protection are in direct contrast to the simplicity of the principles involved, and, at present the application of this method of corrosion control remains more of an art than a science. However, as shown by the potential-pH diagrams, the lowering of the potential of a metal into the region of immunity is one of the two fundamental methods of corrosion control. In principle, cathodic protection can be used for a variety of applications where a metal is immersed in an aqueous solution of an electrolyte, which can range from relatively pure water to soils and to dilute solutions of acids. Whether the method is applicable will depend on many factors and, in particular, economics - protection of steel immersed in a highly acid solution is theoretically feasible but too costly to be practicable. It should be emphasised that as the method is electrochemical both the structure to be protected and the anode used for protection must be in both metallic and electrolytic contact. Cathodic protection cannot therefore be applied for controlling atmospheric corrosion, since it is not feasible to immerse an anode in a thin condensed film of moisture or in droplets of rain water. The forms of corrosion which can be controlled by cathodic protection include all forms of general corrosion, pitting corrosion, graphitic corrosion, crevice corrosion, stress-corrosion cracking, corrosion fatigue, cavitation corrosion, bacterial corrosion, etc. This section deals exclusively with the practical application of cathodic protection principally using the impressed-current method. The application of cathodic protection using sacrificial anodes is dealt with in Section 10.2.
Structures that are Cathodically Protected The following structures are those which in given circumstances can benefit from the application of a cathodic-protection system:
Underground and underwater Underground fuel/oil tanks and pipelines; water, fire protection, gas and compressed underground air distribution schemes; underground metallic sewers and culverts; underground communication and power cables; deep wells; other buried tanks and tanks in 10 :93
Table 10.24
Methods of application of cathodic protection
n
..
0
Method Sacrificial anodes
Characteristics Metal protected by sacrificial wastage of more electronegative metal
Anode materiais*
Current source
Installation
Magnesium, aluminium or zinc (iron for capper and copper alloys)
Faradaic equivalent of sacrificial metal in practice the efficiency is seldom 100%
Extremely simple
Carbon, silicon-iron, lead-platinum, platinised titanium, platinised niobium, scrap iron, platinum metal oxides deposited on a titanium substrate
Source of lowvoltage d.c. This may be generated or drawn from transformerrectifier fed from main supplies
More complex
Bonded directly into stray d.c. supply
Drained from d.c. traction or straycurrent supply
Possibilities of secondary interaction in foreign structures Very improbable providing anodes properly located with respect t o surface being protected
-u
Fc, 2
0
$
%
v Impressed current (‘power impressed’)
Stray current (‘drained current’)
Impressed currents using transformerrectifiers, o r any other d.c. source
Buried structures bonded into traction system in such a way as to receive impressed-current protection
For protection of ferrous slruclures.
Very significant especially in built-up areas
t 2
P rA
8 c,
3z
0
E! 0
-a XI
2 Simple
Stray-current effects are basically associated with primary power supply
2
P
-0
Table 10.24
F 3
(continued)
; h
Method Sacrificial anodes
Impressed current (‘power impressed‘)
Stray current (‘drained current’)
Application for which scheme is economical
Major limitations
Poteniial disiribuiion
Small land based schemes and for avoidance of interaction problems. Marine structures, e.g. offshore platforms
High soiVwater resistivities and small driving e.m.f. may require a large number of anodes
Reasonably uniform
Especially suited to large schemes
Impracticable for small schemes o n account of high installation costs Requires an external power source
Varies- maximum at drainage point falling towards remote points, but not below the optimum potentials for protection, i.e. in most cases the potential -0.85 V
Applicable only in proximity to stray d.c. areas
Current limitation Cannot be applied in highresistivity environments
. Y r
c
0
2; 5
8 Can be used in high-resistivity environments
41 0
2; I
8 5
‘D
;1
9
$
10:96
PRACTICAL APPLICATIONS OF CATHODIC PROTECTION
contact with the ground; tower footings, sheet steel piling and ‘H’-piling; piers, wharfs and other mooring facilities; submarine pipelines; intake screens (condenser/circulating water); gates, locks and screens in irrigation and navigation canals; domestic oil distribution lines or central heating systems. Above ground (internalsurfaces only) Surface and elevated water storage tanks; condensers and heat exchangers; hot-water storage tanks, processing tanks and vessels; hot- and cold-water domestic storage tanks; breweries and dairies (pasturisers). Floating structures Ballast compartments of tankers; ships (active and in ‘mothballs’); drilling rigs; floating dry-docks; barges (interior and exterior); dredgers; caisson gates; steel mooring pontoons; navigation aids, e.g. buoys.
Type of System The use of an impressed-current system or sacrificial anodes will both provide satisfactory cathodic protection, but each has advantages and disadvantages with respect to the other (Table 10.24). Sacrificial anodes and power-impressed anodes have been dealt with in detail in the previous sections, but some further comment is relevant here in relation to the choice of a particular system for a specific environment. In this connection it should be noted that the conductivity of the environment and the nature of the anode reactions are of fundamental importance. The main anodic reactions may be summarised as follows: Sacrificial anodes primary reaction secondary reaction
+
M+M‘+ ze M z + + zH,O + M(OH),
Impressed-current anodes 3H,O
+
2H,O+
+ 2e + + 0,
+ zH
+
(10.18a) (10.18b) ( 1 0.19a)
and/or 2C1-
+
+ 2e
(10.19b)
CO,
(10.19~)
C1,
or, in the case of graphite anodes
c + 0,
-+
It should be noted that when metals like zinc and aluminium are used as sacrificial anodes the anode reaction will be predominantly 10.18a and 10.186, although self-corrosion may also occur to a greater or lesser extent. Whereas the e.m.f. between magnesium, the most negative sacrificial anode, and iron is ~ 0 . V, 7 the e.m.f. of power-impressed systems can range from 6 V to 50V or more, depending on the power source employed. Thus, whereas sacrificial anodes are normally restricted to environments having a resistivity of < 6 OOO fl cm there is no similar limitation in the use of powerimpressed systems. In the case of sacrificial anodes the electrons that are required to depress the potential of the structure to be protected are supplied by reaction 10.18a,
PRACTICAL APPLICATIONS OF CATHODIC PROTECTION
10 :97
and providing the metal ions can diffuse away from the structure before they react with water to form insoluble hydroxides the reaction will be unimpeded and will take place at a low overpotential. If, however, the metal hydroxide precipitates on the surface of the metal as a non-conducting passive film the anode reaction will be stifled and this situation must be avoided if the anode is to operate satisfactorily. On the other hand, in the case of non-reactive impressed-current anodes, rapid transport of the reactants (H,O and C1 -) to, and the reaction products (0, and Cl,) away from, the anode surface is essential if the anode reaction is to proceed at low overpotentials. This presents no problems in sea-water, and for this reason the surface areas of the anodes are comparatively small and the anode current densities correspondingly high. Thus, in sea-water inert anodes such as platinised titanium and lead-platinum can operate at = 500-1 OOO Am -*, since the anode reaction 10.19b occurs with little overpotential, and there is rapid transport of C1- to and C12 away from the anode surface. In this connection it should be noted that even in a water of high chlorinity such as seawater, oxygen evolution should occur in preference to oxygen evolution on thermodynamic grounds. This follows from the fact that the equilibrium potential of reaction 10.19a in neutral solutions is 0-84 V, whereas the corresponding value for 10.19b is 1 34 V, i.e. 0.5 V higher. However, whereas the chlorine evolution reaction occurs with only a small overpotential, very appreciable overpotentials are required for oxygen evolution, and this latter reaction will occur therefore only at high current densities. Even in waters of low salinity chlorine evolution will occur in preference to oxygen evolution at low overpotentials. In the protection of pipelines or other underground structures the anode reaction is dependent on diffusion of water to the anode surface and oxygen and CO, away from it, and since these processes do not occur with the same mobility as in water it is necessary to use a very large surface area of anode and a corresponding low current density. For this reason the actual anode is the carbonaceous backfill, and graphite or silicon-iron anodes are used primarily to make electrical contact between the cable and the backfill. It can also be seen from reaction 10.19a that the products of the oxidation of water are oxygen and the hydrated proton H,O +,which will migrate away from the anode surface under the influence of the field, thus removing two of the three water molecules that participate in the reaction, and this will tend to dehydrate the groundbed. This difficulty can be overcome, when feasible, by locating the groundbed below the water table.
-
Sacrificial Anode Systems
Advantages No external source of power is required; installation is relatively simple; the danger of cathodic protection interaction is minimised; more economic for small schemes; the danger of over protection is alleviated; even current distribution can be easily achieved; maintenance is not required apart from routine potential checks and replacement of anodes at the end of their useful life; no running costs. Disadvantages Maximum anode output when first installed decreasing with
10:98
PRACTICAL APPLICATIONS OF CATHODIC PROTECTION
time when additional current may be required to overcome coating deterioration; current output in high resistivity electrolytes might be too low and render anodes ineffective; large numbers of anodes may be required to protect large structures resulting in high anode installation and replacement costs; anodes may require replacement at frequent intervals when current output is high. Impressed-current Systems
Aduuntuges One installation can protect a large area of metal; systems can be designed with a reserve voltage and amperage to cater for increasing current requirement due to coating deterioration; current output can be easily varied to suit requirements; schemes can be designed for a life in excess of 20 years; current requirements can be readily monitored on the transformer-rectifier or other d.c. source; automatic control of current output or of the structure potential can be achieved. Disuduantages Possible interaction effects on other buried structures (Section 10.6); subject to the availability of a suitable a.c. supply source or other source of dx.; regular electrical maintenance checks and inspection required; running costs for electrical supply (usually not very high except in the case of bare marine structures and in power stations where structures are often bare and include bimetallic couples); subject to power shutdowns and failures. Hybrid Systems
Offshore structures are often protected by hybrid systems using both sacrificial anodes and impressed-current. These have the advantage that protection of the steel by the sacrificial anodes will be effective as soon as the platform enters the sea, which is particularly advantageous since some time may elapse before the d.c. generators required for the impressed-current system are operating. Further details are given in Section 9.4, Design in Marine and Offshore Engineering. Stray Current or Forced Drainage
Stray current schemes are relatively rare in occurrence in the UK as few localities now have widespread d.c. transport systems. Such systems are extensively used in overseas countries where d.c. transport systems are in use, i.e. Australia and South Africa. Where stray current can be employed it is normally the most economical method of applying cathodic protection since the power required is supplied gratis by the transport system. In such systems it is necessary to provide a metallic bond between the pipeline and the negative bus of the railway substation. By providing such a bond the equivalent of a cathodic-protection system is established whereby current discharged from the traction-system rails is picked up by all portions of the
PRACTICAL APPLICATIONS OF CATHODIC PROTECTION Load current required to operate train \
Overhead positive / feeder
10:99 Electrol sis switci
pick-up area
Fig. 10.21 Bond between pipeline and d.c. substation
pipeline and drained off via the bond. The bond must have sufficient carrying capacity to handle the maximum current drained without damage. In order to ensure that the direction of current flow in the bond does not reverse, it is normal to employ a reverse-current prevention device or ‘electrolysis switch’. This may take the form of a relay-actuated contactor which opens automatically when the current reverses. Diodes may also be used as blocking valves to accomplish the same purpose. They are wired into the circuit so as to ensure that current can flow to the negative busbar system only. Sufficient diodes must be used in parallel to handle the maximum amount of current anticipated. Also the inverse voltage rating of the diodes must be sufficient to resist the maximum reverse voltage between the negative busbar and pipeline (Fig. 10.21).
Design of a Cathodic-protection System To enable an engineer to design a cathodic-protection scheme, consideration should be given to the following points (see also Table 10.25). Good practice in modern underground or underwater structures involves the use of good coatings in combination with cathodic protection. With a well-coated structure the cathodic-protection system need only protect the minute areas of steel exposed to the corrosive environment rather than the whole surface of an uncoated structure. The effect of coatings can be demonstrated by comparing the current density of a bare steel pipeline in average soil conditions, which could be up to 30mA/m2, with that achieved on a well-coated and inspected line where a current density of only 0-01mA/mZ or even lower may be required to obtain satisfactory cathodic protection. In all cases the current density for protection is based on the superficial area of the whole structure. Surface area In the case of underground pipelines, calculation of the superficial surface area can be obtained from the diameter and length of line involved. The superficial surface area should include any offtakes and other
10: 100
PRACTICAL APPLICATIONS OF CATHODIC PROTECTION
Table 10.25 Steps in design of cathodic-protection installation Sacrt$cial and impressed-current anodes
1. Establish soil or water resistivity. 2 . Estimate total current requirements which will depend on aggressiveness of the
environment, nature of protective coating, area of structure, materials of construction. 3. Establish electrical continuity of structure. 4. Consider requirements for electrical isolation in order to restrict the spread of protective current. Alternatively assess extra current allowance for unrestricted spread. Sacrificial
5 . Selected suitable anode metal; calculate total mass of metal for required life.
6. Select individual anode shape to satisfy total current output and current distribution requirements. 7. Check that total anode weight as determined by (6) will satisfy the requirements of (5). 8. Consider facilities for monitoring performance.
trnDressed current
5 . Consider the number and disposition of
anodeslgroundbeds bearing in mind: (i) Uniformity of current distribution (ii) Proximity to available power supplies (iii) Avoidance of interaction (iv) Avoidance of mechanical damage (v) Desirability of low resistivity environment. 6. Select suitable anode material 7. Calculate anodelgroundbed size, shape,
configuration. 8. Calculate circuit resistance and system d.c. volts. 9. Consider facilities required for controllmonitoring.
metal structures in electrical contact with the main line. For marine structures the area should include all submerged steel work below full-tide level. In the case of power stations, details of the water boxes, number of passes on coolers and detailed drawings are required. In the case of ships, details of the full underwater submerged area at full load are needed.
Electrical continuity It is essential for any structure to be fully electrically continuous. In the case of pipelines, welded joints are obviously no problem but mechanical joints require bonding. For marine structures individual piles and fendering must be electrically connected either by the reinforcing bars in the concrete deckhead or separately by cable. In power stations and ships, rotating shafts must be bonded into the structure by means of brush gear or a suitable alternative. In the case of modern offshore mooring installations, it may be necessary to install a bonding cable to bring the outerlying dolphins, etc. into the system. Estimate of current required The surface area of the structure is calculated and the current density required for the particular environment is selected (Table 10.26). In the case of an existing structure the condition of the coating may be unknown and the application of a temporary cathodic-protection system may be necessary to determine the amount of current required for protection, as established by the potential. Such a test to determine the
PRACTICAL APPLICATIONS OF CATHODIC PROTECTION
10: 101
Table 10.26 Typical values of current requirements for steel free from adverse galvanic influences in various environments
Environment
Current density required for adequate cathodic protection * based on superficial area (mA/m2)
BARE STEEL
Sterile, neutral soil Well-aerated neutral soil Dry, well-aerated soil Wet soil, moderate/severe conditions Highly acid soil Soil supporting active sulphate-reducing bacteria Heated in soil (e.g. hot-water discharge line) Dry concrete Moist concrete Stationary fresh water Moving fresh water Fresh water highly turbulent and containing dissolved oxygen Hot water Polluted estuarine water Sea-water Chemicals, acid or alkaline solution in process tanks Heat-exchanger water boxes with non-ferrous tube plates and tubes WELL-COATED STEELS
4.3-16. I 2 I .5-32.3 5 4-16.1 26.9-64 - 6 53.8-161.4 451.9 53.8-269.0 5 '4- 16.1 53.8-269 ' 0 53.8 53.8-64.6 53.8-161.4 53.8-161 ' 4 538-0-1614.0 53.8-269 0 53.8-269 '0 1345.0 overall
(e.g. pipelines)
Soils
0.01
Higher current densities will be required if galvanic effects (i.e. dissimilar metals in contact) are present.
absolute amount of current required is known as a currenf drain test. Misleading information may, however, be obtained if the results from current drainage tests on bare or coated steel in sea-water are extrapolated, because long-term polarisation effects, together with the formation of a calcareous deposit on the structure, may considerably reduce eventual current requirements. On the other hand, in estuaries and polluted waters special care must be taken to allow for seasonal and other variable factors which may require higher current densities.
Establishing electrolyte resistivity To enable a satisfactory cathodic-protection'scheme to be designed, it is necessary to determine the resistivity of the electrolyte (soil or water). This information is necessary to enable the current output of anodes to be determined together with their position and power source voltage, and it also provides an indication of the aggressiveness of the environment; in general the lower the resistivity the more aggressive the environment. Economics After evaluating these variables, it must then be decided which type of system, i.e. sacrificial anode or impressed current, would be the most economical under the prevailing conditions. For instance, it would obviously be very expensive to install an impressed-current system on only 100 m of fire main. Similarly, it would be equally uneconomic to install a sacrificial-anode
10:102
PRACTICAL APPLICATIONS OF CATHODIC PROTECTION
system on hundreds of miles of high-pressure poorly coated gas main. Therefore, each system must be individually calculated taking note of all the factors involved.
Impressed-current Systems Cathodic-protection schemes utilising the impressed-current method fall into two basic groups, dictated by the anode material: 1. Graphite, silicon-iron and scrap-steel anodes used for buried structures
and landward faces of jetties, wharves, etc. platinised-niobium, lead and lead-platinum anodes used for submerged structures, ships and power stations.
2. Platinised-titanium,
These two groups will be discussed briefly since a more detailed account has been given in Section 10.3. Group 1 Anodes
Scrap steel In some fortunate instances a disused pipeline or other metal structure in close proximity to the project requiring cathodic protection may be used. However, it is essential in cases of scrap steel or iron groundbeds to ensure that the steelwork is completely electrically continuous, and multiple cable connections to various parts of the groundbed must be used to ensure a sufficient life. Preferential corrosion can take place in the vicinity of cable connections resulting in early electrical disconnection, hence the necessity for multiple connections. Graphite Graphite anodes are usually linseed oil or resin impregnated and supplied in standard lengths, e.g. 2.5 in (approx. 65 mm) dia. x 4 ft (1 - 2 m) long and 3 in (75 mm) dia. x 5 ft (1 5 m) long with a length of cable (called the anode tail)fixed in one end. Graphite anodes are still used and were particularly common in early cathodic-protection systems. However, they have tended to be replaced by silicon-iron, the main reasons being (a) graphite Table 10.27
Impressed-current anodes
Max. working capacify (A/m2) Material Scrap steel Scrap cast iron Silicon-iron Graphite Lead Lead-platinum Platinum Platinised titanium Platinised tantalum Aluminium
Approx. consumption (kg/A year)
Soil
Sea water
Soil
Sea water
5.4 5.4
5.4 5.4 32-43 21.5 107-2 15 1 080 < 10 800 < 10 800 icril.) only a small current density is required to maintain it, and that in the passive region the corrosion rate corresponds to the passive current density (ipass.).
Passivity of Metals Since anodic protection is intimately related to passivity of metals it is relevant to review certain aspects of the latter, before considering the practical aspects of the former. The relative tendency for passivation depends upon both the metal and the electrolyte; thus in a given electrolyte, titanium passivates more readily than iron, and Fe- 1SCr-lONi-3Mo steel passivates more readily than Fe-17Cr steel. The ability to sustain passivity increases as decreases, and as the total the current density to maintain passivity (ipass,) film resistance increases, as indicated for metals and alloys in 67 wt.% sulphuric acid (Table 10.30)’. The lower the potential at which a passive metal becomes active (i.e. the lower the Flade potential) the greater the stability of passivity, and the following are some typical values of E, (V): Table 10.30 Current density to maintain passivity and film resistance of some metals and
alloys in 67 wt% sulphuric acid (after Shock, Riggs and Sudburys) Current density to passivity
Metal or alloy
(LSs.. Am-*)
Total film resistance (Qcm)
~~
Mild steel Stainless steel (Fe-lBCr-8Ni) Stainless steel (Fe-24Cr-20Ni) Stainless steel (Fe- 18Cr- 10Ni-2Mo) Titanium Carpenter 20 (Fe-25Cr-20Ni-2.5Mo-3
1.5 x IO-’ 2 - 2 x 10-2 5 x IO-’
Xu)
1 x 10-~ 8 x 3 x
2.6 x 5.0 x 2.1 x 1.75 x 1.75x 4.6 x
104 io5 106 107 107
io7
10: 157
ANODIC PROTECTION
titanium -0.24, chromium -0.22, steel +0-10,nickel +0.36 and iron +0.589. These values are only approximate, since they depend upon the experimental conditions such as the pH of the solution". The Flade potential is given by
EF = E: - no-059pH where E: = the standard Flade potential at pH = 0, and n = a number between 1 and 2 depending upon the metal and its condition. Table 10.30 gives the current density to maintain passivity of certain metals and also the total film resistance. Only those metals which have a Flade potential below the standard reversible hydrogen potential (0.00V at a"+ = 1 ) can be passivated by non-oxidising acids, e.g. titanium can be passivated by hydrogen ions which are sufficiently oxidising, whereas mild steel requires an oxidising agent with the power of fuming HNO,. The addition of a more passive metal to a less passive metal normally increases the ease of passivation and lowers the Flade potential, as in the alloying of iron and chromium in 10wt. 070 sulphuric acid (Table 10.31)9. Tramp copper levels in carbon steels have been found to reduce the corrosion in sulphuric acid. Similarly 0.1070 palladium in titanium was beneficial in protecting crevices", but the alloy dissolved much faster than commercial grade titanium when both were anodically protected. The addition of 2% nickel in titanium has also improved the resistance to intense local attack Table 10.31 Effect on critical current density and Flade potential o f chromium content for iron-chromium alloys in 10 wt.% sulphuric acid (after West9)
Chromium
Critical current density (iCrit.,Am-')
Vu)
Flade potential ( E F ,V)
0
1.0 x
104
+0*58
2-8 6.1
3 - 6 x io3
+0-58
3-4x 2.1 x 1-9x
9-5 14-0
Id Id Id
+0.35 +0*15 -0.03
Table 10.32 Effect on critical current density and passivation potential on alloy~ (after ing nickel with chromium in 1~ and ION H,SO, both containing 0 . 5K2S04 Myers, Beck and Fontana") Critical current density
VO)
100
91 77 49 21 10
1 0
Passivation potential (Epp.V)
(iceit..Am-')
Nickel
I N acid 1 . 0x 9.5 1.1 2x 1.2 x 1.3 x 1.0 x 1.5 x
103
ION acid 2-3 x
Id
3.9 x 10 lo-: 10-
IO-' 10 10
8-2 2.0
4.1 x 10-1 1.1 x 10-1 5.0 x IO 8.0 x 10
I N acid
ION acid
+0-36 +O.M
4-0.47
+om07
+0.08
+0*03 +0.02 +044 -0.32
+O*M
-0.30
+0.14 +0*05 +0*08 -0.20 -0.20
10: 158
ANODIC PROTECTION
Table 10.33 Critical current density and current density to maintain passivity of stainless steel (Fe-18to 200-8 to 12Ni) in different electrolytes (after Shock, Riggs and Sudbury’) ~
Electrolyte
20% sodium hydroxide 67% sulphuric acid (24°C) Lithium hydroxide (PH= 9-5) 80% nitric acid (24°C) 115% phosphoric acid (24OC)
Critical current density G,.,Am-2)
Current density to maintain passivity (iwSs.. ~m-’)
4.65 x 10
9.9 x 10-2 9-3 x 2.2 x IO-^
5-1 8.0 x 10-1
2.5 x lo-*
3 - 1 x 10-4 1.5 x 10-6
1.5 x
Table 10.34 Effect of concentration of sulphuric acid at 24OC on corrosion rate and critical current density of stainless steel (after Sudbury, Riggs and Shock 14)
Sulphuric ocid
Corrosion rote
(TO)
(gm-2 d-I)
Criticol current density (iflit ,Am-’) ~
0
40 45 55 65
75 105
0 48
41
I20 I92
14 IO 7
168
144 0
16
4 1
in neutral and alkaline solutions’’. Since exceptions may exist, each system should be considered separately, as indicated by the fact that both the additions of nickel to chromium and also chromium to nickel decrease the critical current density in a mixture of sulphuric acid and 0.5 N K,S04 (Table 10.32)”. These parameters depend upon the composition, concentration, purity, temperature and agitation of the electrolyte. The current densities, required to obtain passivity icri,,,and to maintain passivity i,,,,,, for a 304 stainless steel (Fe-18 to 20Cr-8 to 12Ni) in different electrolytes, are given in Table 10.33*. From the data in this table, it can be seen that it is about 1OOOOO times easier to passivate instantaneously large areas of this steel in contact with 115% orthophosphoric acid than in 20% sodium hydroxide. The concentration of the electrolyte is also important and for a 316 stainless steel (Fe-16 to 18Cr-10 to 14Ni-2 to 3Mo) in sulphuric acid, although there is a maximum corrosion rate at about 5 5 % , the critical current density decreases progressively as the concentration of acid increases (Table 10.34)14. When the optimum conditions are used for anodic protection the rate of corrosion can be reduced to an acceptable value”. Thus for 40% nitric acid the rate of corrosion of mild steel was approximately lo5mm y - I , but with anodic protection it fell to less than 20 mm y - I . The presence of impurities, particularly halogen ions, that retard the formation of a passive film, is often detrimental as illustrated by the fact that the addition of 3.2% hydrochloric acid to 67% sulphuric acid raises the critical current density for the passivation of a 316 stainless stee1I6from 5 to 400 A m - 2 and the current. density to maintain passivity from 0.001 to 0.6Am -*.This is potentially dangerous,
ANODIC PROTECTION
10: 159
and the effect of the chloride ion on the passivation of iron has been studied by Pourbaix " who has produced a modified potential-pH diagram for the Fe-H,O system. Therefore, the use of the calomel electrode in anodicprotection systems is not recommended because of the possible leakage of chloride ions into the electrolyte, and metal/metal and other electrodes20*2' are often preferred. Because of this chloride effect the storage of hydrochloric acid requires a more passive metal than mild steel, and titanium anodically protected by an external source of current or galvanic coupling although even this oxide film has has been reported to be sometimes been found to be unstableu. Other additions, such as chromous chloride to chromic chloride, may result in the breakdown of passivity on titanium, but fortunately in this application, anodic protection gives repassivation and increases the corrosion resistance in the new solution by a factor of thirty25. An increase in the temperature of an electrolyte may have several effects: it may make passivation more difficult, reduce the potential range in which a metal is passive and increase the current density or corrosion rate during passivity as indicated in Fig. 10.54 for mild steel in 10% H,SO,. These changes are illustrated for several steels in different acids in Table 10.3526 and it may be noted that, whereas the critical current density for the 316 steel increases with the temperature of the sulphuric acid, the opposite effect is observed with the 304 steel. During the storage of an acid, changes in the ambient temperature between day and night or summer and winter may double the current required for protection, and the increase may be even higher during manufacture or heat-transfer processes, so these should be considered at the design stage. Agitation or stirring of an electrolyte in +2-0
+1.5
.1.0
0
.-
I
c
+ 0.5
Q
c 0
a
0
-05 Anodic current density (A/m*) Fig. 10.54 Potentiostatic anodic polarisstion curves for mild steel in 10% sulphuric acid. Note the magnitude of the critical current density which is ]@-IO3 A/m2; this creates a problem in practical anodic protection since very high currents are required to exceed L i t . and therefore to passivate the mild steel
L
..
0
Table 10.35 Effect of temperature on different acids on the operating variables for anodic protection of different steels (after Walker and Wardz6)
Alloy
Acid concentration
Temp. ("C)
Critical current density
Current density to maintain passivity
Upas.,
(icri,., Am-2)
Stainless steel 3048 (Fe-18 to 2 0 0 8 to 12Ni)
Phosphoric, 115%
177
Nitric, 80'70 Sulphuric. 67% Stainless steel 316 (Fe-16 to 18Cr10 to 14Ni-2 to 3MO)
Sulphuric, 67% Phosphoric, 115% Phosphoric, 75-80%
Carbon steet5'
24 82
Sulphuric. 96%
24 82 24 82 24 66 93 93 117 104 121 135 27 49 93
1.5 x 3.1 x 6-5 x 2-5 X 1-2 x 5-1 4-6 x
Am-')
1.5 x 1.5 X 2.2 x 3.1 x 1.1 x 9.3 x 2.9 x 1x 3x 9x
10-4 10-~ IO-' 10-1 lo-'
5.0 4.0 x IO' 1.1 x I d
9 1.5 3.8
X X X
Corrosion rate (mm y-1) Unprotected
Anodically protected
Passive potential range
(VI
IOT6 10-2
10-4
IO-^
W
10-3 10-~
0-26-1.09 0.27-1.04 0.32-0.72 0.26-1.14 0'26-0.94
IO-^
10-2-1*4 x 10-'-3.5 x 10-'-4.4 x 1.1 x 1.16 x 1-16
IO-' lo-'
IO-' 10-2
IO-'
1.5 2.2 5.3
0.12 0.12 0.85
0.IS 0.8 2.8
0.01 0-11 0-8
4
5
10: 161
ANODIC PROTECTION
Table 10.36 Effect of electrolyte agitation on corrosion rate and the current density to
maintain passivity of mild steel in acid solutions at 27°C (after Walker and Ward26) Corrosion rate (mm y - ’ )
Acid
Spent alkylation acid (sulphuric acid and organic matter)27 Sulphuric acid, 93VP
Condition Unprotected
$zz:cf
Stirred quiescent
3.0 1.4
0.15
Stirred quiescent
3.3 0.9
0.28 0.07
0.12
Current density to maintain passivity
.
(i,,,,. Am-2) 2.48 x IO-’ 3.2 x
certain conditions may increase the rate of corrosion of immersed metals and raise the passivation current (Table 10.36)26.Finley and Myers have found that both the temperature of the electrolyte29and cold working of the metal3’ have a marked effect on the anodic polarisation of iron in sulphuric acid. Because these variables have a very pronounced effect on the current density required to produce and also maintain passivity, it is necessary to know the exact operating conditions of the electrolyte before designing a system of anodic protection. In the paper and pulp industry a current of 4 O00 A was required for 3 min to passivate the steel surfaces: after passivation with thiosulphates etc. in the black liquor the current was reduced to 2 700 A for 12 min and then only 600 A was necessary for the remainder of the process”. From an economic aspect, it is normal, in the first instance, to consider anodically protecting a cheap metal or alloy, such as mild steel. If this is not satisfactory, the alloying of mild steel with a small percentage of a more passive metal, such as chromium, molybdenum or nickel, may decrease both the critical and passivation current densities to a sufficiently low value. It is fortunate that the effect of these alloying additions can be determined by laboratory experiments before application on an industrial scale is undertaken.
Practical Aspects It is essential that the throwing power of the system (the ability for the applied current to reach the required value over long distances) is good and that the potential of the whole of the protected surface is maintained in the passive region. This can normally be achieved with commercial potentiostats, providing the range of the potential over which the metal or alloy exhibits passivity is greater than 50mV. In the case of stainless steels the corrosion rate will increase if the potential rises into the transpassive zone (or if it falls into the active zone). However, titanium does not show transpassivity and, therefore, has a large potential range over which it is passive. In general, a uniform distribution of potential over a regular-shaped passivated surface can be readily obtained by anodic protection. It is much more difficult to protect surface irregularities, such as the recessions around sharp slots, grooves or crevice^^^-^' since the required current density will not be
10: 162
ANODIC PROTECTION
obtained in these areas; therefore, a local cell is set up and corrosion occurs within the recess. This incomplete passivation can have catastrophic consequences, in the form of intergranular corrosion 39, stress-corrosion cracking40,41 corrosion fatigue3' or Calculations have been made of the variation of potential and current distribution down anodically protected narrow passages, and these are important because they may result in local intense pitting". The distribution of current and potential as well as the electrochemical and design parameters for anodic protection systems have been discussed e l s e ~ h e r e ~ ~Stress - ~ ' . corrosion cracking of welded structural steel containing alkali-aluminate solutions was a problem at the Bayer plant but was overcome by anodic protection4'. This difficulty can be overcome by designing the surface to avoid these irregularities around bolt and rivet holes, threaded pipe sections and imperfect welds, or by using a metal or alloy which is very easily passivated having as low a critical current density as possible. In the rayon industry, crevice corrosion in titanium has been overcome by alloying it with 0-1% palladium". The throwing power of a system is particularly important in the anodic protection of pipelines and, therefore, has been widely ~ t u d i e d ~ ' *The ~~-~~. length of the pipe that can be protected by a single cathode placed at one end depends upon the metal, electrolyte and the pipe diameter; the larger the diameter the longer the length that can be protected. Thus, for mild steel in 93% sulphuric acid the length protected (or made passive) is 2.9m for 0-025 m diameter, 4.8 m for 0.05 m diameter and possibly about 9 m for 0-15 m diameter, whereas for mild steel in a nitrogen fertiliser (a less aggressive medium than sulphuric acid) the protected length can be as much as 60 m with one cathode. As a result of recent field tests with an 0.30m diameter carbon-steel pipe and 93% sulphuric acid at ambient temperatures, it is proposed to install anodic-protection systems for 650m of pipeline53. The actual passivation of a surface is very rapid, if the applied current density is greater than the critical value. However, because of the high current requirements, it has been found to be neither technically nor economically practical to consider initially passivating the whole surface of a large vessel at the same time. This can be illustrated by the fact that for a storage vessel with an area of 1 OOO m2 a current of 5 OOO A is necessary for some metal-environment systems, so it is therefore essential to use some other technique to avoid these very high currents. It may be possible to lower the temperature of the electrolyte to reduce the critical current density before passivating the metal. The feasibility of this is indicated from the values given for some acids in Table 10.35, but generally the reduction in the total current obtained by this method is insufficient. If a vessel has a very small floor area, it may be treated in a stepwise manneru*'' by passivating the base, then the lower areas of the walls and finally the upper areas of the walls, but this technique is not practical for very large storage tanks with a considerable floor area. A carbon steel carbonation tower, 18 m high and diameter 2 - 2 m with 80 plug-in heat exchangers, for the production of ammonium hydrogen carbonate has been satisfactorily passivated by this method 54. This was necessary because the critical current density was high, 280-480Am -'. With protection the corrosion rate decreased to less than ommmy-'. Another method which has been successful is to passivate the metal by
ANODIC PROTECTION
10: 163
using a solution with a low critical current density (such as phosphoric acid), which is then replaced with the more aggressive acid (such as sulphuric acid) that has to be contained in the vessel (see Table 10.33). Tsinrnan et ui.” passivated a large tank of 10000m3capacity (12m high and 33.4m internal diameter) with a dilute solution of ammonia and then gradually increased the concentration to 25%. This was necessary because the 25% solution was much more corrosive and required a much higher current to passivate than was available. The critical current density can be minimised by pretreating the metal surface with a passivating inhibitor; for example, chromate solution has been applied to the floor and lower walls of a carbon-steel storage tank, which was then used to contain 37% nitrogen fertiliser solutions3.
Applications and Economic Considerations The majority of the applications of anodic protection involve the manufacture, storage and transport of sulphuric acid, more of which is produced world-wide than any other chemical. Oleum is 100% sulphuric acid containing additional dissolved sulphur trioxide. The corrosion rate of steel in 77-100% sulphuric acid is 500-1 000pmy-’ at 24°C and up to 5 000pmy-’ at 100°C which indicates the necessity for additional protection. Anodic protection can be applied to metals and alloys in mild electrolytes as well as in very corrosive environments, including strong acids and alkalis, in which cathodic protection is not normally suitable. The operating conditions, Flade potential, critical and passive current densities can be accurately determined by laboratory experiments and the current density during passivity is often a direct measure of the actual rate of corrosion in practice. The very low corrosion rate of a passive metal or alloy results in very little metal pick-up and solution contamination or discoloration. Corrosion of unprotected steel in sulphuric acid can give an iron concentration of 5-20ppm per day. Hence, it is almost impossible to produce electrolytic grade sulphuric acid in bare or unprotected steel, because this grade should not contain more than 50 ppm iron. In general a higher purity chemical commands a higher price. However, special care should be taken in the selection of the metal or alloy and in the design if there is a possibility of crevice or intergranular corrosion. A portable form of anodic protection is available that can be applied to rail and road tanker^^'*'^-'^- It can also be used for old vessels as well as new, so that a container designed for one liquid can be protected and used to hold a more corrosive solution. Because a system with a good throwing power can be designed, anodic-protection systems have been applied to pipelines53and spiral heat exchanger^''.^'. It has been found possible to maintain protection of the vapour space above a liquid, once it has been completely immersed and passivated4’, and this is particularly important when the liquid level may rise and fall during storage and use. One consequence of reducing the rate of corrosion of steel in an acid is to decrease the formation of hydrogen, which has been reported as the cause of explosions in phosphoric acid systemss9.Hydrogen may also form
Table 10.37 Summary of anodic protection application (after N.A.C.E. &)
Application
vessel metal
%:''Tv
~
~
~
Vessel size range~$ (m)
Number
$
of ~ systems
eType of controller
Power supply size range (kW)
Date of first installation
Oleum
Mild steel
>50
Storage.
ED x 6H12D x 6H
4
On-off and proportional
0.5-5.0
Oct. 1960
100% H2S04
Mild steel
3-50
Storage*
4D x 4H
1
On-off
0.5
March 1960
99% H2S04
Mild steel
>50
Storage*
2D x 9L9D x 10H
5
On-off
2.5-5.0
Nov. 1962
98% H2SO4
Cast iron
293
Mix tank*?
3D x 2D
1
On-off
1.0
Aug. 1964
98% H2SO4
Mild Steel
250
Storage*
ED x 6H9D x 9H
2
On-off
2.5
Sept. 1963
96% H2S04
Mild steel
>50
Storage*
4D x 6H
2
On-off
1.0-2.5
Dec. 1961
On-off
0.5-5.0
Oct. 1962
Proportional
1.0
Sept. 19669
5E! 0 V
2
93% H$O4
Mild steel
>50
Storage
2D x 5L15D x 7H
93% H2S04
304 stainless steel
M6
Heat exchanger*t
61 m2
60"BC H2S04
Mild steel
x56
Storage*
8D x 6H12D x 6H
3
On-off
5.0
June 1965
60"Bt H2S04
430 stainless steel
250
Storage*
2D x 1OL
1
On-off
0.5
June 1965
14
9 $
Table 10.37
Application
Vessel metal
Temperature range ("C)
m
Vessel type and purpose
(continued)
Vessel size range
Number
(m)
systems
of
4D x 6L12D x 12H
6
Storage.
3D x 3H
1
Storage'
24D x 10H
1
On-offand
Power supply size range
(kW)
Date of first installation
Black and spent HSOA
Mild steel
Black H,SO,
304 stainless steel
>163
75% H3P04
304 stainless steel
>50
Mild steel
>so
Storage't
30D X 5H27D x 12H
7
On-off
5-0
Dec. 1963
Kraft cooking liquor in digester
Mild steel
p177
Reactor.
3D x 14H
1
Time
30
1%111
ClO, bleach
317 stainless steel
Washer wires*
Not applicable
I
Proportional
2-5
Nov. 1%3
N, fertiliser
Storage.
Type of controller
2.5-10.0
NOV.1962(
Proportional
5.0
Oct. 1966
Proportional
5.0
proportional
5 0 7)
solutions
Mo
Chemical purity o f product. of vessel. t D = diameter, H = height. I Exchangers failed alter two weeks operation due t o high chloride content of cooling water, causing stress-corrosion cracking of 304 stainless steel. l o n e application failed due to unanticipated composition variations. Relatively low unprotected rate did not provide incentive for further work. I(There are both successful and unsuccessfulpulp digester installations.
t Corrosion mntrol
Sept. 1963
2-
20
=!
2!
10: 166
ANODIC PROTECTION
blisters at inclusions in the metal surface and can also produce grooving on vertical surfaces. Anodic protection, which has been found to reduce@the formation of hydrogen by 97%, can therefore prevent this effects. The limitations of anodic protection arise from the inability to form a stable, continuous, protective, passive film on the metal to be protected. Thus metals and alloys which neither form passive films nor non-conducting solutions cannot be used. For strongly aggressive acids, such as hydrochloric acid, a very stable anode film is required and, while steel is not satisfactory, may be suitable. The performance of titanium in strongly aggressive conditions can be improved by anodic protection, the use of inhibitors and by alloying elements62. The stable oxide film on protected titanium is particularly useful for reactor tanks for electroless nickel deposition. In these solutions, based on nickel sulphate and hypophosphate, no nickel plating occurs on the tank walls and there is no self-decomposition of the hypophosphate. Because of the high throwing power a single cathode can be used to protect large A power failure may be a considerable danger, since it can result in a drop in potential from the passive region to the active region with a considerable increase in the current density. Following the mechanical breakdown of the oxide layer on a 304L stainless steel in potassium hydroxide solution a current density as high as lOAcrn-’ has been measured65. This failure may be rectified by the use of a 100% effective ‘fail-safe’ back-up current source or by the selection of a basically more corrosion-resistant alloy, which would be marginally satisfactory in an unprotected state. Table 10.37& gives some of the applications of anodic-protection systems which have been used in the USA. Most of these are for chemical purity and corrosion control in storage vessels and details are given of the size of the tank and the operating conditions. Economically the installation of anodic protection is often very good. The advantages include a reducton in capital investment, lower maintenance and replacement costs and an improvement in the product quantity and value. The use of a potentiostat and its associated equipment involves a high installation cost but low operating costs, because only very small current densities are required to maintain passivity. In some circumstances the passive condition may persist for several hours after the current has been switched off. If this is the situation6’ it is possible to use a relatively inexpensive switching mechanism with one control and power-supply system to anodically protect three separate tanks, and therefore reduced the high initial cost. Four storage vessels, volume 160 m3, containing aqueous ammonia have been protected simultaneously by one system by switching the current on for 2 min and off for 6 mina. The rate of corrosion with this technique decreased from about 0 * 1 8 6 m m y - ’to less than 0.001 mmy-’. It can be seen from Table 10.38 that it is more economical to anodically protect mild steel than to use mild steel with a P.V.C. lining or to use a more resistant and expensive metal or alloy such as aluminium or stainless steel. It is worth noting that because most of the expense of an anodic-protection system is due to the cost of the potentiostat, it is more economical per unit volume to use a larger instrument and a bigger tank. Not only is the anodic protection of a mild-steel tank cheaper than one with a glass or phenolic lining”, but, because the steel conducts heat, it can be used for heat exchangers, and in addition it may be more stable at high
10: 167
ANODIC PROTECTION
Table 10.38 Comparison of the relative cost of protecting tanks by various methods (after Reference 18:1967) ~~
Tank cost Anodic-protection system P.V.C. lining Power Maintenance Total cost
Mild steel protected
Mild steel lined
Stainless steel
Mild steel protected
Aluminium
1.70
1.70
4.57
1.70
3.5
0.04
0.94
4.05
0.04 0.54 3.22
0.81 6.56 95 OOO litre
0.04 0.54 4.57
2.32
3.5
3 800 OOO litre
temperatures for long time periods. The rate of corrosion of shell and tube heat exchangers, with 93-99'70 H,SO, on the shell side and water on the tube side, decreased from 5-10 mm/y to 25 pm/y when anodically protected. Hence protection enabled the use of either higher temperatures and velocities 7 ' , giving better heat transfer, or thinner metal sections or smaller exchangers, all of which are financially beneficial. A further advantage is that a considerable reduction in the corrosion may increase the life of chemical plant until it becomes obsolete instead of the need for repair and replacement. In very corrosive conditions it may be necessary to use a very resistant alloy together with anodic protection, e.g. Corronel 230 was employed in the extraction of uranium using hot acidic solutions, which were too aggressive to be contained by Neoprene, Karbate and Teflon coatings7,. The reduced rate of corrosion can also improve safety by maintaining the thickness and strength of the supporting metal as well as minimising the possibility of perforation. This is particularly important if poisonous, combustible, explosive or hot liquids are involved.
Conclusion Although the first industrial application of anodic protection was as recent as 1954, it is now widely used, particularly in the USA and USSR.This has been made possible by the recent development of equipment capable of the control of precise potentials at high current outputs. It has been applied to protect mild-steel vessels containing sulphuric acid as large as 49 m in diameter and 15 m high, and commercial equipment is available for use with tanks of capacities from 38 000 to 7 600000 litre". A properly designed anodic-protection system has been shown to be both effective and economically viable, but care must be taken to avoid power failure or the formation of local active-passive cells which lead to the breakdown of passivity and intense corrosion. The extent of the interest in the application of anodic protection is
10: 168
ANODIC PROTECTION
indicated in the following list of recent publications. These include the application of anodic protection to titanium in the rayon industry’*, heating coils73and chromic chloride/chromous chloride solutions”, electroless nickel plating”*64,hydrochloric7*and sulphuri~’~ acid and neutral and alkaline chloride solutions”. Different steels have been used with a wide range of solutions of fertilisers56*76,77, ammonia and ammonium salts7’-’’, bicarbonate^^^*^^, sodiums3and potassiums4hydroxides, formics5 and nitric’j acids and alkali-al~minate~~. Protected steel has also been used in the manufacture of acrylamideS6and electrolytic manganese dioxide” as well as biosewage treatment plant”. Other work on steel for storage tanks and boiling for sulphuric acid has been reported as well as heat exchangers for the a~id~’-~’. Further details on the application and theory of anodic protection are given in the excellent book by Riggs and Locke9’. R. WALKER
REFERENCES 1. Fontana, M. G. and Green, N. D., Corrosion Engineering, McGraw-Hill, New York, p. 214 (1967) 2. Edeleanu, C., Nature, 173, 739 (1954) 3 . Edeleanu, C., Metallurgia, Manchr., 50, 113 (1954) 4. Cherrova, G.P., Dissertation, Akad. Nauk., Moscow Institute of Physical Chemistry, SSSR (1953) 5. Novakovskii. V. M. and Levin. A. I., Dokl. Akad. Nauk., SSSR. 99 No. I , 129 (1954) 6. Palmer, J. D.. Canad- Chem. Process., 60 No. 8 , 35 (1976) 7. Flade, F., Z. Phy. Chem.. 76, 513 (1911) 8. Shock, D. A., Riggs, 0.L. and Sudbury, J. G.,Corrosion. 16 No. 2. 99 (1960) 9. West, J. M.. Electrodeposition and Corrosion Processes. Van Nostrand, New York. p. 81 (1965) IO. Uhlig, H.H.,Corrosion and Corrosion Control, Wiley, New York, p. 61 (1967) 1 1 . Myers, J. R., Beck, F. H.and Fontana, M. G., Corrosion, 21 No. 9, 277 (1965) 12. Evans, L. S., Hayfield, P. C. S. and Morris, M. C., Werkst. u . Korrosion, 21,499 (1970), and also Proc. of the Fourth International Congress on Metallic Corrosion, Amsterdam, p. 625 (1%9) 13. Riskin, 1. V. and Timonin, I., Prot. Metals, 15 No. 4, 45 (1979) 14. Sudbury, J . G., Riggs, 0. L. and Shock, D. A., Corrosion, 16 No. 2, 91 (1960) I S . Sastry, T. P. and Rao, V. V., Corrosion, 39 No. 2, 55 (1983) 16. Shock, D. A., Sudbury, J. D. and Riggs, 0. L., Proc. of the First International Congress on Metallic Corrosion, London 1961, Butterworths, London, p. 363 (1962) 17. Pourbaix, M.. Corrosion. 25 No. 6, 267 (1969) IS. Corrosion Control Systems Bulletin, No. 773, Magna Corporation, USA (1%7) 19. Sudbury, J. D.. Locke. C. E. and Coldiron, D., Chemical Processing. 11 Feb (1%3) 20. Togano, H..J . Japan. Inst. of Metals, 33 No, 2, 265 (1969) 21. Kuzub, V. S.,Tsinrnan, A. I.. Sokolov, V. K. and Makarov, V. A., Prot. Metals, 5 No. 1, 45 (1969) 22. Cotton, J. B., Chem. Ind.. Lond., 18 No. 3, 68 (1958) 23. Stern, M. and Wissenberg, H., J. Electrochem. SOC., 106, 755 (1959) 24. Togano, H., Sasaki, H. and Kanda, Y., J. Japan. Inst. Metals, 33 No. 11, 1280 (1969) 25. Letskikh, E. S., Komornokova, A. G., Kryasheva, V. M.and Kolotyrkin, Ya. M., Prot. Metuls, 6 No. 6, 635 (1970) 26. Walker, R. and Ward, A., Metallurgical Review No. 137, in MetalsandMuterials, 3 No. 9, 143 (1969) 27. Locke, C. E., Banks, W. P. and French, E. C., Mat. Prot., 3 No. 6, SO (1964) 28. Sudbury, J . D. and Locke, C. E., Oil Gas J . , 61, 63 (1963) 29. Finley, T. C. and Myers, J. R., Corrosion, 26 No. 12. 544 (1970) 30. Finley, T. C. and Myers, J. R., Corrosion, 26 No. 4, I50 (1970)
ANODIC PROTECTION
10: 169
31. Watson, T. R. B., Mater. Prof., 3 No. 6, 54 (1964) 32. France, W. D. and Greene, N. D., Corrosion, 24 No. 8, 247 (1968) and also Report No. AD665, 788 (1968) 33. Cowley, W. C., Robinson, F. P. A. and Kerrich. J . E., Brit. Corrosion J., 3 No. 5 , 223 ( 1968) 34. Makarov. V. A. and Kolotyrkin, Ya. M..Media for Prevention of Corrosion, Moscow, pp. 5-15 (1966) 35. Anon., Anticorrosion Methods and Mat., 15 No. 4, 5 (1968) 36. Karlberg, G. and Wranglen, G., Corros. Sci., 11 No. 7, 499 (1971) 37. Ruskol, Y. S. and Klinov, I. Y., J. Appl. Chem., USSR, 41 No. 10, 2084 (1968) 38. France, W. D. and Greene, N. D., 24th Annual NACE Conference, Cleveland, Ohio, p. I , March (1968) 39. France, W. D. and Greene, N. D., Corros. Sci., 8, 9 (1968) 40. Greene, J. A. S. and Haney. E. G., Corrosion, 23, 5 (1967) 41. Stammen, J. M..25th Annual NACE Conference, Houston, p. 688, March (1969) 42. Schwenk, W., Corrosion, 20, 129t (1964) 43. Banks, W. P. and Hutchison, M., Mat. Prof., 7 No. 9, 37 (1968) 44. Reingeverts, M. D., Kots, V. D. and Sukhotin, A. M., Elektrokhimiya, 16 No. 1, 41, 46; No. 3, 386 (1980) 45. Helle, H. P . E., Beck, G. H.M. and Ligtelyn, J. Th., Corrosion, 37 No. 9, 522 (1981) 46. Foroulis, 2. A.. Anticorros. Methods Mater.. 27 No. 11. 5 , 10, 13 (1982) 47. ValdCs, L. N. and A d n , A. C.. Rev. Cienc. Quim.. 14 No. 2. 335, 343 (1983) 48. Artem’ev, V. I.. Prot. Metuls, 19 No. 4, 500 (1983) 49. Edeleanu, C. and Gibson, J. G., Chem. Ind., Lond., 21 No. 10, 301 (1961) 50. Mueller, W. A., J. Electrochem. Soc., 110, 699 (1963) 5 1 . Timonin, V. A. and Fokin, M. N., Prot. Metals, 2 No. 3, 257 (1966) 52. Makarov, V. A. Kolotyrkin, Ya. M., Kryazheva, V. M. and Mamin, E . B., Prot. Metals, 1 No. 6, 592 (1965) 53. Stammen, J. M., private communication 54. Xiaoguang, L. and Tingfang, L., Met. Corros., 8th Int. Cong., Vol. 11, p. 1139 (1981) 5 5 . Tsinman, A. I., Danielyan, L. A. and Kuzub, V. S., Prof. Metals. 9 No. 2, 143; No.4,492 (1973) 56. Banks, W. P. and Hutchison, M., Mat. Prot., 8 No. 2 , 31 (1969) 57. Sudbury, J. D. and Locke, C. E., Chem. Eng., 70 No. 11, 268 (1963) 58. Locke, C. E., Mat. Prof., 4 No. 3, 59 (1965) 59. Lowe, J. B., Corrosion, 17 No. 3, 30 (1961) 60. Riggs, 0. L., Mat. Prot., 2 No. 8. 63 (1963) 61. Jaffee, R. I. and Promisel, N. E., The Science, Technology and Application of Titanium, Pergamon Press, London, p. 155 (1970) 62. Bavay, J. C., Metaux-Corros. Ind.. 57 No. 683-4, 241 (1982) 63. Sadakov, G. A. and Kolchevskii, A. K., Prot. Metals, 19 No. 2, 267 (1983) 64. Makarov, V. F., Prusov, Yu V. and Flerov, V. N., Prof. Metals, 18 No. 6, 732 (1982) 65. Burstein, G. T. and Marshall, P. I., Corros. Sci., 23 No. 4, 125 (1983) 66. Report compiled by NACE Task Group T-3L-2 (1968) 67. Hays, L. R., Mat. Pror., 5 No. 9, 46 (1966) 68. Li, T.,Yuan. M.. Wen, M., Pan, Y. and Wei, G., Chem. Engng. and Mach., 10 No. I , 49 (1983) 69. Fisher, A. 0.and Brady, J. F., Corrosion, 19, 37t (1963) 70. Anon., Mat. Prof., 2 No. 9, 69 (1963) 71. Fyfe, D., Sanz, D., Jones, F. W. S. and Cameron, G. M., paper No. 63, Corrosion ’75, Int. Corr. Forum, NACE, p. 22 (1975) 72. Robinson, F. P. A. and Golante, L., Corrosion, 20 No. 8, 239t (1964) 73. Schmidt, W., Hampel, H. and Grabinski, J., J . Chem. Tech., Berlin, 22 No. 5 , 296 (1970) 74. Krikum. S. I., Prot. Metals, 17 No. 2, 167 (1981) 75. Seagle, S. R., AIME Symp. on Corrosion and Biomed. Appl. for Ti., p. 21 (1974) 76. Kuzub, V. S . er a / . , Prot. Merals, 19 No. 1. 13 (1983) 77. Kuzub, V. S., Novitskii, V. S., Golovneva, L. B. and Rebrunov, V. P., Khim. Prom. (Moscow), No. 8, 609 (1974) 78. Moisa, V. G. and Kuzub, V. S., Prot. Metals, 16 No. 1, 83 (1980) 79. Szymanski, W. A., Mater. Perform., 16 No. 11, 16 (1977)
10: 170
ANODIC PROTECTION
80. Danielyan, L. A., Tsinman, A. I., Kuzub, V. Mefals, 9 No. 4, 457 (1973)
S.,Moisa, V. G. and Slatsenko, N. N., Prot.
81. Chen, C-C and Hsueh, H.. Tung Pao, No. 2. 117 (1974)
82. 83. 84. 85. 86. 87.
Davia, D.H.and Burstein, G. T.. Corrosion, 36 No. 8, 416 (1980) Tsinman, A. 1. and Danielyan. L. A., Prot. Metok. 12 No. 4, 450 (1976) Stypula, 9. and Piotrowski, A., Zesz. Nauk. Akad. Zesz. Spec., 45, 26 (1973) Novak, P., Bystriansky, J., Franz, F. and Bartel, V., Chem. Prum., 28 No. 9,461 (1978) Makarov, V. A. and Egorova, K. A., Prof. Metals, 6 No. 3, 302 (1970) Gelagutashvite, Sh. L., et a/., Prot. Melds, 18 No. 2, 258, 260; No. 6, 729; 19 No. 2, 304,
307 (1983) 88. Dohnalik. K. and Golec, J., Ochrono P e e d Korozjo. NR3, 63 (1983) 89. Fyfe, D.. Vanderland, R. and Rodda. J.. Chem. Eng. Progr., 73 No. 3. 65 (1977) 90. Novak, P., Stefac, R. and Franz, F., EUROCOR 77.6th European Congress on Metallic Corrosion, 89 (1977) 91. Ashby, W. A., Lewis, L. S. and Shepherd, W., ibid., 85 92. Antoniuk, A., Ochr. Przed. Koroz., 18 No. I , 19 (1976) 93. Kuzub. V. S.. Statsenko. N. N., Kuzub, L. G. and Moisa, V. G., Khim. Tekhnol. (Kiev), No. 1, 63 (1974) 94. Mallett, S. E.. Chem. Ind., 104 No. 9. I l l 1 (1971) 95. Makarov, V. A. el ai., Prot. Mefals, 13 No. 2. 143 (1977) 96. Novak, P., Vesela, L., Franz, F. and Bartil, V., Chem. Prum., 29 No. 2 , 342 (1979) 97. Paulekat, F., Grafen, H. and Kuron, D., Werkst. Korros., 33 No. 5 , 254 (1982) 98. Riggs, 0.L. and Locke, C. E., Anodic Protection: Theory and Practice in the Prevention of Corrosion. Plenum, New York (1981)
11
PRETREATMENT AND DESIGN FOR METAL FINISHING
1 1 . l Surface Treatment Prior to Applying Coatings
11:3
1 1 .2 Pickling in Acid 1 1 .3 Chemical and Electrolytic Polishing
11:14 11:24
11.4 Design for Corrosion Protection
11:40
by Electroplated Coatings 1 1 . 5 Design for Corrosion Protection by Paint Coatings
11:48
11: 1
1 I . 1 Surface Treatment Prior to Applying Coatings
The attainment of a clean surface prior to the application of any subsequent treatment or coating is essential, whether this subsequent operation is electroplating, anodising, chemical treatment or organic coating. The standard of cleanliness which must be achieved has been stated to be ‘that which will allow the subsequent process to be carried out satisfactorily’. As an example, the degree of cleanliness required to satisfactorily zinc plate from an acid solution is somewhat higher than that required prior to zinc plate from a high-cyanide alkali zinc solution. This should never be taken as a licence to skimp on surface preparation. However, the arguments over ‘surface-tolerant’ paint coatings abound and will probably continue. It is to a very large extent true that problems of early failure in metal finishing are traceable to incorrect or insufficient surface preparation. Although many standards exist for cleaning treatments for metal surfaces, for example Defence Standard DEF STAN 03-211, these are often fairly general guides which in some cases may be regarded as somewhat outdated due to recent advances in treatment technology and changes in industrial practice. In general, there are two types of surface contamination: (1) organic contamination -such as oils, greases, paint coatings etc.; and (2) inorganic contamination - such as rust, oxide films, corrosion products, scale, anodic films etc. Although these two types of contaminant can be removed simultaneously, it is simpler to consider the cases separately.
Removal of Organic Contamination As previously stated, this consists of oils, greases, preservatives or old paint coatings which must be removed prior to further finishing. The removal of paint coatings with chemical paint strippers is outside the scope of this section, and readers are referred to specialist publications on the subject. Sources of the remaining organic contamination are cutting and machining fluids, preservatives, tramp oils from, for example, rolling operations, press lubricants and mechanical or manual handling operations. Four means of soil removal have been proposed: mechanical action; 11:3
11:4
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
solvency; detergency; and chemical reaction. In all cleaning operations one or more of these mechanisms will contribute more or less t o the overall cleaning procedure, dependent upon the cleaning method and solution employed. Virtually 100% mechanical action is employed in abrasive blast cleaning. With chemical cleaning, performance will be enhanced by the use of mechanical action, such as brushing, air agitation. spraying, electrolysis or ultrasonics. Solvency is where the soil to be removed dissolves in the cleaning medium, for example mineral oil in chlorinated solvents. Detergency is the ‘lifting’ action attributed to some alkalis and to special surface-active agents -commonly referred to as surfactants or originally, ‘syndets’, short for ‘synthetic detergents’. Chemical reaction is characterised by, for example, the saponification of some oils in strong alkali, or the reaction of rust with acid solutions. The main types of cleaners used for the removal of organic contaminants are: solvent cleaners, neutral cleaners, acid cleaners and alkali cleaners. Solvent Cleaning
This area can be split into four major categories of cleaner type: cold solvent, hot/vapour solvent, emulsifiable and emulsion. Cold Solvents Solvents, for example, white spirit or paraffin, used either by immersion or by manual application are not to be recommended as effective, or particularly safe methods, of degreasing. When used by immersion, the holding tank can became heavily contaminated with soil, which will remain on the work after the solvent has evaporated. The use of solventsoaked rags, although a time-honoured procedure, is now being frowned upon on the grounds of operator safety; aqueous based pre-wipes are available.
Vapour Degreasing The use of hot/boiling solvents, with both immersion of the articles to be cleaned in the bulk solvent and/or in the overlying vapour, using specially designed installations, is a far more effective use of solvents for cleaning purposes. A simplified diagram of a typical small installation is shown in Fig. 11.1. The solvent, which is generally of a halogenated hydrocarbon type, is held in a sump at the base, which is heated by any suitable means and under thermostatic control. Above this may be a wire mesh on which the workpieces are rested. Some way above this, there are condenser coils, often water-filled. Between the mesh and the coils, therefore, is created a region where the solvent is in vapour form. When cold workpieces are introduced, the vapour condenses on the work, the liquid solvent flows off, taking the oil with it. To a large extent, only clean solvent is vaporised, thus ensuring that only fresh solvent is used to clean the workpiece until the sump becomes overcontaminated, when the solvent must be cleaned or replaced. Care should be exercised with some metals, notably aluminium, that solvent with free chloride is not used, as this could lead to pitting of the metal surface. The advice of the manufacturers of the installation and the suppliers of the solvent should always be heeded in the operation of these installations
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
, ’
CONDENSER
PIPES
11:5
-
COIL THERMOSTAT
LlOUlD
-
Fig. I I . I
SUM? THERMOSTAT
Cut-away drawing of vapour degreasing plant
to ensure their trouble-free running. Effective fume extraction must be available above the installation and the work must be removed slowly enough to ensure that all the solvent has evaporated from the work before it leaves the extracted area. The rules governing exposure limits are frequently changed, and up-to-date advice must be sought.
Emulsifiable Cleaners (Water Rinsable Cold Solvent Cleaning) Emulsifiable cleaners (sometimes incorrectly referred to as emulsion cleaners) are blends of organic solvent, often kerosene, with surface active agents. The work is immersed in the unheated solution for a sufficient time for the cleaner to penetrate the soil thoroughly. The articles are then removed and water rinsed. Additives in the cleaner allow the solvent, with its accompanying soil, to emulsify in the water thus removing the contamination. Spray rinsing or agitation in an immersion rinse will aid removal of the residues. The disposal of the rinse water is dependent on local effluent restrictions. In some areas, mere dilution will be required before discharge, but in others, the water may have to be stored and the emulsion broken before discharge of the water layer and approved disposal of the organic material. As with all solvent-based materials, the need to observe TLV limits and the need for the work to be carried out only under effective fume extraction must be taken into account when considering this type of cleaning product. The cleanliness of the surface produced by emulsifiable cleaners is rarely of a very high standard, and additional cleaning may well be necessary before further finishing operations. Success has been achieved, however, in the use of these products prior to some immersion phosphating operations, where the crystal growth can be quite refined due to the absence of the passivation effect often encountered with some heavy-duty alkali cleaners. The supplier of the phosphating solution should be asked to advise on the suitability of any particular cleaning/pretreatment combination.
11:6
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
Another benefit gained from the use of emulsifiable cleaners is that the surface produced is usually hydrophobic and thus, to an extent, resistant to tarnishing and corrosion in storage. Emulsion Cleaners These are materials, containing blends of organic solvents and surfactants, which are added to water to form an emulsion. Typical concentrations are in the range 0.5-5%. Such emulsions are normally used by spray, as either a pre-clean in a multi-stage pretreatment line, or as the cleaner in an industrial washing machine. Such washing machines are often used to clean parts which are contaminated with cutting oils etc., and which require inspection before storage. Like the emulsifiable cleaners, the emulsion cleaners, after rinsing, often leave a hydrophobic surface which is resistant to short-term corrosion. Emulsion cleaners can be used hot or cold. Heat generally improves the cleaning action but, in most cases, leads t o an objectionable increase in the smell associated with solvent products. Neutral Cleaners
These are rapidly replacing emulsion and alkali products for use in industrial washing machines. They are generally used at pH 7 5-9, considerably lower than corresponding alkali products. Neutral cleaners have a soap-type hydrotroping base, with additions of surfactant (to improve cleaning, wetting, penetration and defoaming), inhibitors (which may be nitrite or organic) and a bactericide. The bactericide was often formaldehyde, but this is now being superseded by formaldehyde-free materials, based on quaternary ammonium salts. Neutral cleaners provide the benefits of generally lower operating temperatures, reduced odour, easier effluent treatment and improved health and safety considerations over the alkali or emulsion products. Due to the inhibited nature of the surface produced, such products are used for interstage cleaning and prior to assembly. The surface is generally not suitable for immediate painting. Acid Cleaners
The vast majority of acid-based cleaning products are for the removal of scale, rust and other oxide films (see Section 11.2). These products may contain solvents and surfactants to degrease and derust simultaneously. There are, however, certain acid-based materials which can primarily be construed as cleaners. One such type of material is used in the cleaning of aluminium cans prior to treating and lacquering. Such cleaners are normally based on sulphuric or phosphoric acid, with, generally, additions of hydrofluoric acid and surfactants. These materials are sprayed on to pre-formed cans to remove the lubricant used during the can-forming operation. The fluoride is present to enhance the removal of ‘fines’ of metal swarf in the cans as well as t o remove the oxide film. Fluoride-free acid cleaners are finding their way into the general pretreatment cleaning of aluminium as an alternative to strong alkali materials.
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
11:7
Although more expensive in terms of initial make-up and plant requirements, the rate of loss through cleaning and etching can be less. Furthermore, the need for a desmut material, required after alkali etching, is obviated. As the surface smoothing and levelling effects are somewhat limited, the use of acid cleaners prior to anodising or electropainting, where surface defects can be enhanced, is not common. Care must be taken here not to confuse acid cleaners with the highstrength, phosphoric acid-based chemical polishes and chemical brighteners, which are used specifically to obtain the surface finish which such materials produce. Also in the category of acid cleaners could be considered the lightweight alkali-metal phosphating cleaner-coater solutions, but a discussion on such materials is best left to specialist publications on metal pretreatment chemicals. Alkali Cleaners
This category is without doubt the largest among cleaner types. Alkali cleaners can be used before almost every conceivable metal finishing operation at one stage or another. There is a bewildering array of products available in the market place. There are alkali cleaners which can be used by spray, by immersion, by manual application or by all three, or maybe by two out of the three methods. They can come as powders or as built liquids. They may be single or multi-pack, to be used as supplied or at a range of dilutions. They may require high temperatures or work successfully at ambient temperature. They may be suitable for cleaning one metal only or have multimetal capability. The user thus has an immense range from which to choose. Consideration will first be given to the inorganic builders used to produce the base material. The pH values of several commonly used materials are shown in Table 11.1. Hydroxides are the simplest, strongest alkalis and most commonly used. A major effect of hydroxides in cleaning is saponification: the conversion of certain oils and greases to water-soluble soap-type materials. Hydroxides also produce solutions of high conductivity, as required for electrocleaning. Beside the benefits of hydroxides must be placed certain disadvantages: 1. The possible passivation of iron and steel surfaces; this can be a pro-
blem prior to chemical conversion coatings. Table 11.1 pH values of certain alkalis as 1% w/v solution at 50°C Alkali
PU
NaOH Na2COB Na2SiO, Na3PO4
12.7 11.3 12.2 11.8
Na2 p2 O7
10.6
Na5P3010 Na2 B4 O7 NaC7 13 OS
9.8 9.3 7.8
11:8
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
2. The soaps produced by saponification may give excessive foam during spray cleaning or react with hard-water salts to form scum and scale. 3. Light metals, such as zinc and aluminium, can be attacked more than is desirable. 4. Powder products formulated with too much hydroxide can be hygroscopic and thus tend to go solid in storage rather than remaining as freeflowing powders. 5 . Spray cleaners based on hydroxide can pick up carbon dioxide from the atmosphere, and such carbonated solutions become less effective. Carbonates and bicarbonates are used as lower alkalinity adjuncts or substitutes for hydroxide. It has been suggested that hydroxide/carbonate systems are more resistant to carbonation during spraying than hydroxideonly solutions. Powder products blended with light sodium carbonate are much less hygroscopic, and the carbonate can be a useful ‘carrier’ for liquid additives, such as surfactants and solvents. Silicates can offer an almost complete cleaning system on their own. Sodium metasilicate, the most commonly used of these materials, has a high enough pH value to cause saponification, and the structure of the polysilicate anion formed gives degrees of detergency, peptisation and inhibition. Thus, silicates are often found in multi-metal cleaners. Light metals, such as zinc and aluminium will not be attacked if the silicate level is sufficiently high and the free caustic level sufficiently low. Cleaners containing silicate can cause problems. They should not be used prior to an alkaline process on aluminium, owing to the formation on the surface of alkali-insoluble aluminium silicate. Silicated cleaners can also cause problems before some surface-sensitive zinc phosphating solutions, especially the more modern low-zinc type. Phosphates are another common ingredient of alkali cleaners. They have both detergent and peptisation properties. The pyro- and polyphosphates in particular have water softening capabilities. Borates are often the base for light-duty cleaners associated with the cleaning of light metals, due to their inhibiting action and mild pH. They can also be used, to a certain extent, as a substitute for phosphates when a phosphatefree product is required. The organic acid salts, such as EDTA and heptonate, are included for water softening properties, and to assist in the removal of solid particles. Gluconate and heptonate, in particular, are effective in the highly alkaline solutions used for etching aluminium and prevent the precipitation of aluminium hydroxide scale and sludge. Surfactants are probably the materials which most affect the performance of alkali cleaners. Surfactants are complex chemicals which modify the solubility of various materials in, and their surface affinity for, oil and water. The diverse composite which makes up the surface of a metal object must be fully wetted out if the cleaner is to perform properly. Surfactants lower the surface tension to allow wetting out to occur. Oils and greases must either be dissolved off the surface or lifted from it; surfactants assist in both areas. There are four broad categories of surfactant, dependent on the charge associated with the active part of the molecule:
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
11:9
Cetyl trimethyl ammonium chloride (cationic)
Sodium dodecyl benzene sulphonate (anionic)
c~~H~~-A~cH~coo-
I
CH, Myristyl dimethyl betaine (amphoteric)
Polyethoxylated nonyl phenol (non-ionic)
Fig. 1 I .2 Typical surfactants
1. cationic, where the residual charge is positive; 2. anionic, where the residual charge is negative; 3. amphoteric, where there exists both positive and negative charge centres; 4. non-ionic, where there is no residual change.
Typical examples are given in Fig. 11.2. It is the job of the formulating chemist to invent a blend from amongst all of the foregoing materials to produce a cleaner suitable for use in a particular application. Care must be taken with some surfactant-containing cleaners not to exceed certain temperature and concentration limits. It is an old adage that a cleaning solution can be improved by making it hotter and stronger. This remains generally true, but with some surfactant-containing cleaners there are restrictions. Many commonly used surfactants have limited solubility in alkali. They become less soluble as the alkalinity, ionic strength and temperature rise. A point can, therefore, be reached when the surfactants come out of solution and, in immersion cleaning especially, performance will suffer drastically. Similarly, some spray cleaners are designed to work above a
11: 10
SURFACE TREATMENT PRIOR TO APPLYING COATINGS Table 11.2 Typical alkali cleaners
Constituent
NaOH Na2SiO, Na, CO, /NaHCO, Na, PO4 EDTANa, Na2 B 4 0 7 Surfactant Substrate Application
Composition (%)
20 50 20
5 -
5
60 20 10
5 5
-
-
0 0 20 20 8 50
3
5
2
20 40 20 I2 5
0 60 10
20 5
Steel Steel Zinc Multi-metal Aluminium Immersion Electrocleaning Electrocleaning Immersion Spray
certain minimum temperature and strength. In this case, a surfactant is designed to come out of solution to act as a defoamer for the system. Examples of typical simple formulations for various types of alkali cleaners are give in Table 11.2. Electrocleaning
As is mentioned above, a significant increase in immersion cleaning performance can be achieved by the use of an applied voltage, as in electrocleaning. Hydrogen is evolved at the cathode and oxygen at the anode. These gases act with a ‘scrubbing’ action, greatly enhancing the cleaning process. Where possible, work will generally be cleaned cathodically, as this results in twice as much hydrogen being evolved than oxygen. However, during cathodic cleaning any dissolved metal ions have a tendency t o plate-out on the metal surface, so the work will normally be given a short anodic cycle at the end of the cleaning time to dissolve this film. Periodic reverse cathodidanodic cycling is most commonly used for articles which are oxidised and corroded as well as oily and greasy. Alkaline products containing cyanide were commonplace for this purpose, but more recent, cyanide-free solutions are being increasingly used. For electrocleaning, care must be taken that a cleaner of sufficiently high conductivity is used to prevent solution voltage drops and ‘burning’ of workpieces in high current density areas. With brass and zinc, the cleaner must not be so alkaline as to cause chemical attack of the substrate before the cleaning period is completed. High anode current densities should be avoided. Care must also be taken when electrocleaning high-strength steel alloys. Hydrogen embrittlement which can occur during cathodic cleaning must be either avoided or catered for. Ultrasonics
Another method for introducing mechanical action into immersion cleaning is by the use of ultrasonics. Here, a high frequency vibration is imparted to the solution. At the surface to be cleaned, minute bubbles are formed and
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
11: 11
collapsed, scrubbing off the soil. Such installations are generally quite small and used for special purposes, although the overall applicability of the system is wide.
Removal of scale and rust from mild steel The hot rolling of steel produces a surface layer of complex oxides known as ‘millscale’. It is unstable, losing adhesion upon weathering, and must be removed prior to painting if predictable paint performance is to be obtained. In rusting, the initial corrosion product of iron is ferrous hydroxide. Reacting with oxygen and water, it forms higher oxides, mainly hydrated ferric oxide and magnetite. Rust formed in industrial or marine environments contains corrosion-promoting salts and is particularly dangerous. Rust is not considered a satisfactory base over which to paint and it too must be removed. The possible methods of surface preparation before painting hot rolled steel are discussed in the following sections.
Weathering In aggressive environments millscale detachment is likely to be complete within a year, while in a benign atmosphere descaling has taken more than five years. Depending upon the severity of exposure, steel can rust and pit during this period. The rust may be contaminated with soluble salts, making effective protective painting difficult if not impossible. For these reasons natural weathering is no longer considered an acceptable part of surface preparation. Manual Cleaning The term encompasses all manual and mechanical methods of cleaning other than blast-cleaning. Abrasive discs, wirebrushes, scrapers, vibratory needle guns and chipping hammers are available. Manual cleaning removes neither tightly adhering millscale nor deep-seated rust from pits. None the less, it is often used for maintenance work or for the preparation of new steelwork to be exposed in non-aggressive conditions. Manual cleaning is rarely used in conjunction with high-performance long-life systems, e.g. two-pack chemically curing coatings which require a high standard of blast cleaning. Swedish standard SIS 055900 contains two pictorial standards for manual cleaning, St2 and St3. Both require the removal of loose millscale, surface rust and foreign matter. The second and higher standard describes the prepared and dusted surface as having a pronounced metallic sheen. The St2 preparation is described as ‘a faint metallic sheen’. Both are expected to correspond with their respective coloured prints in the standard. These relate to four grades of new unpainted steel: Grade A : the surface is covered with adherent millscale; little or no rust is visible. Grade B: The surface has started to rust and the millscale has begun to flake. Grade C: Most of the millscale has flaked and what remains can be scraped off; the surface has rusted but there is little pitting visible to the naked eye. Grade D : The millscale has rusted away and considerable pitting is visible to the naked eye.
11 :12
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
Caution should be exercised when using this standard because the colours of the print may vary from one copy to another. Acid Pickling This does not refer to the site application of weak acid solution; such treatments are of dubious merit. Acid pickling is a factory process during which steel is immersed in hot acid, removing millscale and rust. In the ‘Footner’ or ‘Duplex’processes the steel receives a final treatment in 2% phosphoric acid, leaving a thin phosphate coating on the warm steel surface, to which the paint should be applied immediately. Once very popular for preparing steel plate, the process has been largely superseded by blast cleaning. It is still used in the pipe industry, but finding firms to deal with ad hoc quantities of structural steel or plate is now very difficult.
Flame Cleaning Now little used as a preparatory method, flame cleaning is a process whereby an intensely hot oxyacetylene flame is played on the surface of the steel. In theory, differential expansion causes millscale to detach. In practice, there is evidence that the treatment may not remove thin, tightly adhering millscale. Also, steel less than 5 mm thick can buckle. Finally, the process can ‘burn in’ chemicals deposited on the surface, causing premature paint failure. Blast cleaning To remove millscale and rust, abrasive particles are directed at high velocity against the metal surface. They may be carried by compressed air, high-pressure water, or thrown by centrifugal force from an impeller wheel. For some open blasting, e.g. maintenance work, highpressure water without abrasives may be used although this will not remove heavy corrosion products. Common abrasives for cleaning steel are chilled iron shot and grit, steel shot and grit, iron and copper slags. BS2451: 1963 (1988) covers chilled iron products. BS5493 deals with blast cleaning in the context of protective painting in some detail. There are three standards controlling surface finish in common use. They are issued by the Steel Structures Painting Council (USA), the Swedish Standards Organisation and the British Standards Institution. They are roughly equivalent. Light Thorough Very thorough To pure metal
SSPC
BS7079
SZS 05 5900
SP7 SP6
-
Sal Sa2 Sa2t Sa3
SPlO SPS
Sa2 Sa2t Sa3
Surface ‘finish’ is increasingly referred to as ‘surface cleanliness’. This can be misleading because the standards refer to the appearance of the blasted steel and do not deal with chemical contamination. Site tests for assessing the level of soluble salts on freshly blast-cleaned surfaces, and which allow the semi-quantitative determination of the chlorides, soluble sulphates and soluble iron salts, are urgently needed. Blast-cleaning produces a roughened surface and the profile of that surface is important. The size and nature of the profile is largely determined by the type and size of the abrasive used. To identify and control surface roughness, comparators are available conforming to I S 0 8503/ 1 specifica-
SURFACE TREATMENT PRIOR TO APPLYING COATINGS
11:13
tion. Type G is for use with grits and Type S with shot. The comparators are intended for visual and tactile assessment of surfaces blast-cleaned to S a 2 i or Sa3 only. G . L. HIGGINS R.S . HULLCOOP BIBLIOGRAPHY Spring, S.. Metal Cleaning, Reinhold (1%3) Freeman,D. B., Phosphating and Metal Pre-treutment, Woodhead-Faulkner (1986) Lorin, G., Phosphating of Metal, Finishing Publications (1974) Metal Finishing, Guidebook and directory, Metals and Plastics Publications Inc. (1988) Plaster, H. S., BImt Cleaning and Allied Processes, Vol. I (1972), Vol. I1 (1973) Good Painting Practice, SSPC Painting Manual, Vol. I, c 2.0-2.9,Vol. 11, c 2
Standards
1. Svensk Standard SIS OS 5900-1967,Pictorial surface preparation standards for painting steel surfaces
2. BS2451:1963 (1988), ‘Chilled Iron Shot and Grit’ 3. BS7079:1989, ‘Preparation of steel substrates before application of paint and related products’
4. BS5493:1977, Code of Practice for Protective Coating of Iron and Steel Structures Against Corrosion
5. DIN 8201. ’Synthetic Mineral Solid Abrasives’
11.2 Pickling in Acid
Mechanism of Scale Removal from Steel with Acid When mild steel is heated in air at between 575 and 1 370°C an oxide or scale forms on the steel surface. This scale consists of three well-defined layers, whose thickness and composition depend on the duration and temperature of heating. In general, the layers, from the steel base outwards, comprise a thick layer of wustite, the composition of which approximates to the formula FeO, a layer of magnetite (Fe,O,), and a thin layer of haematite (Fe,O,). When the steel is rapidly cooled, the thickness and composition of these layers remain more or less unchanged, but when it is slowly cooled through 575°C the scale becomes enriched in oxygen and the remaining wustite layer breaks down to some extent into an intimate mixture of finely divided iron and magnetite I . Holding of the temperature between 400 and 575°C causes the iron particles to coagulate and the scale becomes further enriched in oxygen. Since wustite is unstable below 575"C, scales produced at temperatures lower than this contain magnetite and haematite only'. In addition, the scales are often cracked and porous. This is due to the difference in contraction
-
ACID
Fe20)
t -Fc,
0,
PARTIALLY -DECOMPOSED WUSTITE Fe
-
Fc2 0 3
-
Fc, O4
Fe
Fig. 11.3
Mechanism of scale removal with acid. (u) High-temperature scale and ( b ) lowtemperature scale
11: 14
PICKLING IN ACID
11: 15
between scale and metal on cooling and t o the change in voIume when the metal is oxidising. When a steel which has been slowly cooled through 575°C is immersed in mineral acid, the acid penetrates through the cracks and pores in the upper layers of scale and rapidly attacks the decomposed wustite layer, thus releasing the relatively insoluble magnetite and haematite layers (Fig. 11.3). This rapid dissolution of the wustite layer is due to the setting up of many minute electrolytic cells between the finely divided iron particles, magnetite and acid. The iron, being anodic, dissolves to form ferrous ions, and the magnetite, being cathodic, is reduced, forming more ferrous ions. Since the three constituents of these cells are good electrical conductors, the resistance of the cells is so small that the rate of dissolution of the decomposed wustite layer is largely governed by the rate at which acid diffuses through the cracks to it, and the rate at which spent acid diffuses from it. A similar but slower action occurs between the exposed metal and the magnetite and haematite layers which have not been detached2. The pickling rate of steels which have been rapidly cooled or held between 400 and 575°C is slower. This is due in the former case to the absence of the irodmagnetite cell action, and in the latter to the increased cell resistance resulting from coagulation of the iron. Similarly, the pickling rate of steels scaled at temperatures below 575OC is slow, because the resistance of the few larger cells formed between the magnetite and the base metal is high. Apart from this cell mechanism in the scale, and between metal and scale, another cell action occurs on the exposed steel surface. In this ferrous ions are produced at the anodic areas and hydrogen at the cathodic areas.
Hydrogen Embrittlement Although the majority of the hydrogen produced on the cathodic areas is evolved as gas and assists the removal of scale, some of it diffuses into the steel in the atomic form and can render it brittle. With hardened or highcarbon steels the brittleness may be so pronounced that cracks appear during pickling. Austenitic steels, however, are not so subject to ernbrittlement. If the acid contains certain impurities such as arsenic, the arsenic raises the overvoltage for the hydrogen evolution reaction. Consequently, the amount of atomic hydrogen diffusing into the steel, and the brittleness, increase. As well as causing brittleness, the absorbed gas combines to form molecular hydrogen on the surface of inclusions and voids within the steel. Thus a gas pressure is set up in the voids and this may be sufficient to cause blisters to appear either during pickling or during subsequent processing such as hot-dip galvanising. The embrittlement effect can be largely removed by ageing the steel at about 150"C, but even then the original ductility is not entirely restored. In the estimation of the degree of embrittlement, the temperature and rate of testing have an important effect. Thus the embrittlement tends to disappear at very low and very high temperatures, and it is reduced at high strain rates. Several theories of the mechanism of embrittlement have been put forward3-' and further details are given in Section 8.4.
11: 16
PICKLING IN ACID
Acids Used for Pickling Before steel strip or rod can be cold rolled, tinned, galvanised, or enamelled, etc. any scale formed on it by previous heat treatment must be removed. This can be done by mechanical and other special methods, but if a perfectly clean surface is to be produced, acid pickling is preferred, either alone or in conjunction with other pretreatment processes.
Sulphuric Acid
Sulphuric acid is used to a very large extent for pickling low-alloy steels. The rate at which it removes the scale depends on (a) the porosity and number of cracks in the scale, (6)the relative amounts of wiistite, decomposed wiistite, magnetite and haematite in the scale, and (c) factors affecting the activity of the pickle. Temperature is the most important of the factors affecting pickle activity. In general, an increase of 10°C causes an increase in pickling speed of about 70%. Agitation of the pickle increases the speed since it assists the removal of the insoluble scale and rapidly renews the acid at the scale surface. Increase in acid concentration up to about 40% w/w in ferrous sulphate-free solutions, and up to lower concentrations in solutions containing ferrous sulphate, increases the activity. Increase in the ferrous sulphate content at low acid concentrations reduces the activity, but at 90-95°C and at acid concentrations of about 30% w/w it has no effect. For economic reasons, continuous wide mild steel strip must be pickled within 0 -5-1 -0 min. To achieve this, the strip is flexed to increase the number of cracks in the scale and then passed through four or five long tanks. Acid of 25% w/w strength enters the last tank and flows countercurrent to the strip, and finally emerges from the first tank as waste pickle liquor containing about 5% w/w acid and almost saturated with ferrous sulphate. To increase the activity of the pickle to a maximum, live steam is injected to agitate the pickle and to raise the temperature to about 95°C. The scale on the edges of the strip and on the leading and trailing ends is usually more difficult to pickle than that in the centre. Consequently, whereas the centre scale is removed in the first or second tank, the remainder is removed only in the last tank. After pickling, the strip is thoroughly rinsed and dried. For the batch pickling of rod, sheet, tube or strip in coil form, short pickling times are not so important, and pickling times of several minutes at 60-80°C in 5-10% w/w acid are common. The acidity is maintained by the addition of fresh strong acid, until the pickle is nearly saturated with ferrous sulphate, and then the acidity is worked down to 1 or 2%. Electrolytic pickling Anodic pickling Sulphuric acid is also used in electrolytic pickling. Anodic pickling, which is suitable only for lightly scaled steel, has the advantage that no hydrogen embrittlement is produced, but the base metal is attacked as the scale is being removed. Under ‘active’conditions the steel is rapidly attacked, and the surface is left rough and covered with smut. Under ‘passive’ conditions the attack on the steel is reduced, and the evolved oxygen mechanically
PICKLING IN ACID
11: 17
removes the smut and other surface contaminants, leaving the steel with a clean satin finish which provides good adhesion for electrodeposits.
Cathodic pickling Cathodic pickling protects the base metal from acid attack while the scale is being reduced to spongy iron, but there is a danger of hydrogen embrittlement, and particularly if the acid contains arsenic. In the Bullard-Dunn' process the steel is made cathodic at 0.065 A/cm2 in hot 10% w/w acid containing a trace of tin or lead. The tin or lead plates out on the descaled areas as a very thin coating, and owing to the high hydrogen overvoltage of these metals the formation of hydrogen ceases and so the current is diverted to the remaining scaled areas. The tin or lead can be removed by means of anodic alkali treatment, or may be left on and used as a base for subsequent painting. A.c. pickling Alternating current pickling can be used where it is difficult to feed current into the steel by direct electrical connection, e.g. in the case of strip moving at high speed. In this process the electrodes are placed above and below the strip and so while one face of the strip is anodic the other is cathodic, the polarity being reversed during each cycle of alternation. The application of 0.11-0-16A/cm2 to strip in 10% v/v H,SO, at 88°C has been claimed to increase the pickling speed by 35%'. Alternatively, cathodic/anodic pickling may be employed on moving strip without the use of contacts. The moving strip is made cathodic with anodes in one tank and anodic with cathodes in the following tank, the strip itself carrying the current from tank to tank. Hydrochl$ric Acid
Although hydrochloric acid is more expensive than sulphuric acid, it is gradually replacing the latter for pickling mild-steel strip, because the waste liquor can be recovered more economically. It is more active than sulphuric acid at an equivalent concentration and temperature, probably because the rates of diffusion of acid to, and ferrous ions from, the steel surface are greater. Consequently it is used cold for pickling in open tanks and for highspeed pickling of mild steel strip it is used hot in covered tanks to prevent loss of acid by volatilisation. It is more suitable than sulphuric acid for pickling articles which have to be tinned or galvanised since it gives less smut on the steel. In addition, any residual iron chloride left on the steel can be rinsed off more readily than residual iron sulphate deposits. Hydrochloric acid, however, readily dissolves the detached magnetite and haematite and, consequently, the ferric ion produced increases the rate of attack on the steel and thus increases the acid consumption. Phosphoric Acid
Although phosphoric acid can be used for pickling steel, it is seldom used simply for scale removal since it is so expensive and slow in action. Steel plates are often initially descaled in sulphuric acid and then, after rinsing, immersed in 2% phosphoric acid containing 0-3-0.5% iron at 85°C for 3-5mins. The plates are then allowed to drain and dry without further
11: 18
PICKLING IN ACID
rinsing. This treatment produces a grey film of iron phosphates on the steel surface, which provides a good base for subsequent painting. Nitric Acid
Nitric acid does not dissolve scale so readily as mineral acid. A cold 5% w/v nitric acid solution is used to etch bright mild-steel strip when the smut resulting from the acid attack is easily and completely removed with a light brushing. It is also used in conjunction with sulphuric acid for cleaning bright annealed strip, which is difficult to pickle in mineral acid. This difficulty arises when certain types of rolling lubricant have not been thoroughly removed before annealing. During annealing, these lubricants polymerise to gum-like materials which are unattacked by mineral acid but are oxidised and removed with nitric acid. For this type of steel, pickling in a bath containing 20% w/v H,SO, and 4% w/v HNO,, with a trace of HC1, at 30°C for 4-6min has been recommended'.
Pickling of Alloy Steels The furnace scales which form on alloy steels are thin, adherent, complex in composition, and more difficult to remove than scale from non-alloy steels. Several mixed acid pickles have been recommended for stainless steel, the type of pickle depending on the composition and thickness of the scale". For lightly-scaled stainless steel, a nitric/hydrofluoric acid mixture is suitable, the ratio of the acids being varied to suit the type of scale. An increase in the ratio of hydrofluoric acid to nitric acid increases the whitening effect, but also increases the metal loss. Strict chemical control of this mixture is necessary, since it tends to pit the steel when the acid is nearing exhaustion. For heavy scale, two separate pickles are often used. The first conditions the scale and the second removes it. For example, a sulphuric/hydrochloric mixture is recommended as a scale conditioner on heavily scaled chromium steels, and a nitric/hydrochloric mixture for scale removal. A ferric sulphate/ hydrofluoric acid mixture has advantages over a nitric/hydrofluoric acid mixture in that the loss of metal is reduced and the pickling time is shorter, but strict chemical control of the bath is necessary. Electrolytic pickling of stainless steel in 5-10% w/v sulphuric acid at 5OoC can be used for removing the majority of the scale. The strip is first made anodic, when a little metal dissolves, and then cathodic, when the evolved hydrogen removes the loosened scale. To complete the pickling, a nitric plus hydrofluoric acid dip is given for austenitic steels and a nitric acid dip for ferritic steels. Austenitic and ferritic stainless steels are not subject to hydrogen embrittlement with reducing acids, but steels of relatively high carbon content in the hardened state may be.
Organic Inhibitors During the pickling of scaled steel the thinner and more soluble scale is removed before the thicker and less soluble scale. Consequently, some
PICKLING IN ACID
11: 19
exposed base metal is attacked before the pickling operation is complete. In order to reduce this acid attack to a minimum, organic inhibitors are used. Their use also leads to less acid being consumed and less smut and carbonaceous matter is left on the steel. Because of the reduced hydrogen evolution, the amount of acid spray and steam consumption are also reduced. Although a good inhibitor reduces the acid attack, it does not prevent the attack of oxidising agents on the exposed base metal. Thus the ferric ions resulting from the gradual dissolution of the detached magnetite and haematite attack the exposed steel even in the presence of an inhibitor, and are reduced to ferrous ions. The inhibitor should not decompose during the life of the pickle nor decrease the rate of scale removal appreciably. Some highly efficient inhibitors, however, do reduce pickling speed a little. It would be expected that since the hydrogen evolution is reduced the amount of hydrogen absorption and embrittlement would also be reduced. This is not always the case; thiocyanate inhibitors, for example, actually increase the absorption of hydrogen. Since inhibitors form insulating films on the steel, they interfere with any subsequent electroplating. In many cases, however, the films can be removed prior to plating by anodic cleaning or by a nitric acid dip. Surface-active agents are often added t o the pickle if the inhibitor has no surface-active properties. They assist the penetration of the acid into the scale, reduce drag-out losses, and form a foam blanket on the pickle. This blanket reduces heat losses and cuts down the acid spray caused by the hydrogen evolution. Many organic substances soluble in acid or colloidally dispersible have been shown to have inhibiting properties. The most effective types contain a non-polar group such as a hydrocarbon chain and a polar group such as an amine. They contain oxygen, nitrogen, sulphur, or other elements of the fifth and sixth groups of the Periodic Table. They include alcohols, aldehydes, ketones, amines, proteins, amino acids, heterocyclic nitrogen compounds, mercaptans, sulphoxides, sulphides, substituted ureas, thioureas and thioazoles. The efficiency of an inhibitor under a given set of conditions is expressed by the formula
A - B
I=-x 100 A
where Z is the per cent inhibition efficiency. A the corrosion rate in uninhibited acid, and B the corrosion rate in acid containing a certain concentration of inhibitor. In general the efficiency increases with an increase in inhibitor concentration-a typical good inhibitor gives 95% inhibition at a concentration of 0.008% and 90% at 0.004%. Provided the inhibitor is stable, increase in temperature usually increases the efficiency although the actual acid attack may be greater. A change in acid concentration, or in type of steel, may also alter the efficiency. Thus, the conditions of a laboratory determination of efficiency should closely simulate the conditions expected in commercial practice.
11:20
PICKLING IN ACID 100
4w u
95
U
Y z 0 c m
5
90
P
85 '/o
0.01 0.02 0.03 INHIBITOR CONCENTRATION IN PICKLE
0.04
Fig. 1 I .4 Relationship between '70inhibitor efficiency and inhibitor concentration in 6% w/w H2SO4 Curve (a) di-o-tolyl thiourea; (b) mono-0-tolyl thiourea; (c) commercial inhibitor containing 20% di-o-tolyl thiourea; ( d ) commercial inhibitor containing 20% di-phenyl thiourea; (e) gelatin ~
Table 11.3
Pickling solutions for non-ferrous metals
Metal
Temperature Time
Acid
Copper and brass (60-!30% c u , 10-40%
zn)
7-25%
W/W
HZSO,
I5-60"C
15-25%
W/W
HCI
Aluminium bronze Scale conditioned with 10% w/w NaOH (82-95% Cu, 5-1070 AI, followed by HzS04 or HCI as above @-5% Fe, &5% Ni) Copper-silicon alloys
1-10 min
or
7-25% w/w H,SO,
+ 1-3% w/w
HF
15°C
1-3min
75°C
2-5min
15°C
1-5min
80°C
30min
(%-97% Cu. 1-3% Si)
Nickel-copper alloys
10% wlw HCI
+ 1 . 5 % w/w CuCI,
(55-!30% Cu, 10-30% Ni,
0-27% Zn)
Nickel-chromium alloys Scale conditioned with (35-80% Ni, 16-20% 20% NaOH 5% w/w KMnO, Cr, 0-45% Fe, 0-2070 Si) followed by 20% w/w HNO, + 4% w/w HF
+
Aluminium alloys 25% (0-10% CU,&IO% Mg, 40% 0-6% Zn, &12% si) Magnesium alloys (0-10% AI, 0-3%
0-0.2070 Mn)
W/W W/W
HzSO, + 5% W/W C r 0 3 HNO3 + 1 - 5 C W/W H F
10-20% w/w C r 0 3
Zn,
+ 3% w/w HzS04
100°C 50°C
1-2h 5-30min
65°C 15°C
20min I-5min
100°C 25°C
1-30min 15s
11:21
PICKLING IN ACID
Figure 1 1.4 shows the relationship between efficiency and concentration of some thiourea derivatives and gelatin in the pickling of cold-reduced and annealed strip in 6% w/w sulphuric acid at 85°C. The thiourea derivatives, diluted with sodium chloride, gelatin and a wetting agent, are used commercially. Mono- and di-o-tolyl thioureas are stable in this pickle for at least 50 h, but diphenyl thiourea and gelatin decompose after four or five hours.
Inorganic Inhibitors Inorganic inhibitors are salts of metals having a high hydrogen overvoltage, e.g. antimony and arsenic. The inhibiting action is associated with the formation of a coating of the metal, which, being cathodic to the steel and having a high hydrogen overvoltage, prevents the discharge of hydrogen ions and so stops the dissolution of the steel. These inhibitors are seldom used in commercial practice, but antimony chloride dissolved in concentrated hydrochloric acid is used in the laboratory for stripping deposits of zinc, cadmium, tin and chromium from steel, and with the addition of stannous chloride for removing scale and rust". Further details of inhibitors for acid solution are given in Section 17.2.
Acid Pickling of Non-ferrous Metals Table 1 1.3 summarises the pickling conditions for removing oxide and scale from some of the more important non-ferrous metals and alloys.
Recent Developments Mechanisms of Scale Removal from Steel with Acid
The mechanisms of oxide dissolution and scale removal have been widely studied in recent years. This work has been thoroughly reviewed by Frenier and Growcock'*, who concluded, in agreement with others ", that oxide removal from the surface of steel occurs predominantly by a process of reductive dissolution, rather than by chemical dissolution, which is slow in mineral acids. In this process the reduction of the ferric components of the scale is coupled to oxidation of the base metal, both reactions yielding ferrous species readily soluble in the acid. For magnetite the processes are as shown in equations 11.1 and 11.2. Fe,O, + 8 H + + 2e- = 3Fe2++ 4H20 cathode (11.1) Fe = Fez+
+ 2e-
anode
(11.2)
Scale removal is also assisted by the dissolution of the underlying metal by normal acid corrosion processes, which undermines the scale, and by the physical effect of hydrogen gas evolved in this latter reaction. Some authors l4 attribute major effects to the latter.
11 :22
PICKLING IN ACID
In general there does not appear to be any direct correlation between the rate of the chemical dissolution of oxides and the rate of scale removal, although most work on oxide dissolution has concentrated on magnetite. For example, Gorichev and co-workers have studied the kinetics and mechanisms of dissolution of magnetite in acids and found that it is faster in phosphoric acid than in hydrochloric, whereas scale removal is slower. Also, ferrous ions accelerate the dissolution of magnetite in sulphuric, phosphoric and hydrochloric acidI6, whereas the scale removal rate is reduced by the addition of ferrous ions. These observations appear to emphasise the importance of reductive dissolution and undermining in scale removal, as opposed to direct chemical dissolution. As further confirmation of this Rozenfeld ” has reviewed Russian work on this subject and reports that in pickling with sulphuric acid the amount of acid used in scale dissolution is only about one-tenth that consumed by the dissolution (corrosion) of the underlying metal. However, in hydrochloric acid the direct scale dissolution occurs to a much greater degree, and is responsible for about 40% of the acid consumption. A mechanism such as that given above provides explanations for the known effects of many process variables 14. The reductive dissolution and undermining processes require access of the acid to the metal surface, hence the benefits obtained by the deliberate introduction of cracks in the oxide by cold-working prior to pickling. Also the increase in pickling rate with agitation or strip velocity can be explained in terms of the avoidance of acid depletion at the oxide-solution interface. Acids Used for Pickling
Currently the importance of hydrochloric acid is increasing, with sulphuric acid still widely used, and with some applications for other mineral acids. Pickling of Alloy Steels
The chromium-containing oxides on stainless steels are more resistant to reductive dissolution and harder to remove than oxides on mild steel. Typically mixed acids and multistage treatments are used and many formulations have been reported” Scale conditioning can be carried out in acids, in molten salts (e.g. sodium hydroxide plus sodium nitrate) or in alkaline solutions (e.g. alkaline permanganate). Scaleremoval can be obtained with a variety of acids, the commonest being a nitric/hydrofluoric mixture. Rozenfeldl3 also reports effective pickling with ferric sulphate plus sulphuric acid mixtures and considers that the effect of the ferric ions is to speed up the dissolution of the underlying metal. Organic acids, such as citric acid, also have a role in the cleaning of lightly corroded alloy steelsI8. Organic Inhibitors
The principles behind the selection of effective inhibitors for steel in the various acids have been reviewed by Schmitt l 9 and Gardner *O. The selection
PICKLING IN ACID
11 :23
of an inhibitor is dependent on both the metal and the acid. For steel, in general, nitrogen-based inhibitors (e.g. amines and heterocylic compounds) are used in hydrochloric acid, whereas sulphur-containing ones (e.g. thiourea and its derivatives) find more favour in sulphuric acid. Given the reductive dissolution process involved and the contributions from undermining and hydrogen evolution in scale removal, inhibitors might be expected to affect the rate of this removal. Also, if the inhibitor adsorbs on the oxide surface then the rate of chemical dissolution of the oxide may be affected. Experimental evidence suggests that these effects may occur, depending on the acid and the inhibitor. Cumper2' has shown that pyrrole and indole can increase the rate of dissolution of magnetite in hydrochloric acid. It has been reported that commercial amine-based inhibitors can either increase or decrease the rate of scale removal in the same acid. Other reports suggest that the presence of inhibitor has little effect on scale removal rate in hydrochloric acid but markedly decreases it in sulphuric acid. One area that has not been widely studied is the effectiveness of inhibitors on scaled surfaces, but there is experimental evidence that the presence of magnetite scales can significantly affect the performance of nitrogen-based inhibitors in alkaline solutions used for chemical cleaning. S. TURGOOSE W. BULLOUGH
REFERENCES 1. 2. 3. 4.
5. 6. 7. 8.
Pfeil, L. B., J. Iron St. Inst., 123, 237 (1931) Winterbottom, A. B. and Reed, J. P., J. Iron St. Inst., 126, 159 (1932) Zapffe, C., Trans. Amer. Soc.Metals, 39, 191 (1947) Petch, N. J. and Stables, P., Nature, Lond.. 169,842 (1952) Morlet, J. G.. Johnson, H. H. and Troiano, A. R., J. Iron St. Inst., 189, 37 (1958) Fink, C. G. and Wilber, T. H., Trans. Electrochem. Soc., 66, 381 (1934) Neblett. H. W., Iron St. Engr., 16 No. 4, 12 (1939) Footner, H. B., Iron and Steel Institute, 5th Report of the Corrosion Committee, London (1938)
9. 10. 11. 12. 13. 14.
15. 16. 17. 18. 19.
20. 21. 22.
Liddiard, P. D., Sheer Metal Ind.. 22, 1731 (1945) Spencer, L. F., Metal Finish., 52 No. 2. 54 (1954) Clarke. S. G.. Tram. Elecirochem. Soc., 69, 131 (1936) Freiner, W. W. and Growcock, F. B., Corrosion, 40. 663 (1984) Rozenfeld, I. L., Corrosion Inhibitom, McGraw-Hill(l981) Metals Handbook, 9th ed.. Vol. 5 , p. 68 (1984) Gorichev, I. G.. Klyuchnikov, N. G., Bibikova, 2.P. and Boltovskaya, I. G., Russ. J. Phys. Chem., 50, (1976) Gorichev, I. G., Gorsheneva. V. F. and Boltovskaya, I. G.. RELFS. J. Phys. Chem..53,1293 (I 979) Roberts, W. J., in Cleaning Stainless Steel, ASTM Special publ. 538. p. 77 (1972) Blume. W. J., in Cleaning Stoinless Steel. ASTM Special publ. 538. p. 43 (1972) Schmitt, G.. Br. Corros. J., 19, 145 (1984) Gardner, G., in Corrosion Inhibitom {ed. Nathan. C . C.), NACE p. 156 (1973) Cumper. C. W. N.. Grzeskowiak, R. and Newton, P., Corrosion Science. 22, 551 (1982) Riggs, 0. L. and Hurd, R. M., Corrosion, 24, 45 (1%8)
11.3
Chemical and Electrolytic Polishing
Introduction The choice of polishing method for the finishing of metal components depends not only on the type of finish required but also on the state of the metal surface at that stage of the production route. The bulk of surface soil -both physically and chemically attached - will have been removed in primary cleaning and pickling stages, and so polishing is concerned with removal of last traces of soil, but more particularly the removal of surface roughness, blemishes and burning arising as the result of prior fabrication processing. Mechanical polishing and buffing can be a very effective method, but has the disadvantage that it work-hardens the surface, thereby inducing a degree of residual compressivestress, and may also promote the incorporation of soil, oxide, polishing compound etc. into a soft surface and which may consequentlybecome contaminated to a significant depth. Nevertheless, for small components mass finishing in polishing barrels has become an acceptable alternative production method. In contrast chemical and electrolytic polishing enables a smooth level surface to be produced without any residual stress being developed in the surface because the surface is removed by dissolution at relatively low chemical potential and at relatively low rates is such a way that metallic surface asperities are preferentially removed. For this to be most effective the solution properties must be optimised and the pretreatment must leave an essentially bare metal surface for attack by the electrolyte. A variety of synonyms have been used for these processes but they can quickly be placed in two categories.
1. Chemicalpolishing which relies entirely on a solution without any externally applied current. Other processes have been termed ‘bright dipping’, and these are very similar but usually produce lustre without good levelling and the high level of specular reflectivity implied by the term polishing. 2. Electropolishing which exploits a generally similar type of solution, but introduces anodic currents as an additional means of dissolution thereby providing better control of rapid processing. ‘Electrosmoothing’ and ‘electrobrightening’are terms used to describe inferior finishes which may have lustre but have lower specular reflectivity. 11:24
CHEMICAL AND ELECTROLYTIC POLISHING
11:25
Either process can only be successful if good quality metal is supplied, for example with uniform grain size and freedom from non-metallic inclusions, and the pretreatments are applied conscientiously. The choice between chemical and electrolytic polishing is often governed by the quality of finish and that obtainable from existing processes. But those being equal, it is generally true to say that initial and operating costs are less for chemical polishing because of the electrical requirements for electropolishing, but that capital costs can be greater in view of the greater corrosivity of the solutions themselves and the fact that in use they produce considerable problems with corrosive acid fumes. The characteristics of a polished surface are that it should be level on a macroscopic scale related, for example, t o machine and grinding marks of 1-5 Fm depth, and be smooth and bright on a microscopic scale typically 1-100nm size for fine grained metal. To achieve dual levelling and smoothing a solution must satisfy three requirements by including three types of constituent: 1. an oxidising agent capable of dissolving metal in solution through a
surface film which smooths any preferential dissolution on an atomic level; 2. a contaminating agent which controls the thickness of that oxidant film - if it is too thick the metal passivates and polishes extremely slowly, if it is too thin or absent preferential etching can occur; 3. a diffusion layer promoter which provides a viscous liquid adjacent to the surface and which promotes macroscopic levelling. The common defects arising in processing include etching -preferential attack of grain boundaries - which occurs if the film has not fully formed; it may be exploited in certain circumstances because the finish can be artistically attractive and the surface area may be increased. Pitting occurs if the film is disrupted at local sites, either by incorrect balance of film former/contaminant or by gas evolution on the surface. Although most solutions satisfy the three-component criterion they have usually been established by empirical methods and their compositions can be found by referring to tables on a ‘recipe-book’ basis’-’. Many have been extensively explored by metallographers in search of improved preparation techniques, notably for electron m i c r o ~ c o p y ~ - ~ . Industrial application of these processes whilst being widespread only involves large proportions of metal being processed in a few specialised instances, for example anodised aluminium for reflectors. Industrial use is justified by virtue of improved reflectivity and brightness, appearance and occasionally corrosion resistance. It is normally an expensive pretreatment for anodising and electroplating unless the high level of reflectivity is essential, but it does provide a most effective means of removing excessive mechanical burring and roughness, highly stressed surface layers of metals sensitive to stress-cracking failure and thin alloyed layers arising incidentally from prior processing. In some of these respects it may be considered a sister process to electrochemical machining.
11:26
CHEMICAL AND ELECTROLYTIC POLISHING
Bright Dipping and Chemical Polishing As already indicated, bright dipping is essentially a simpler and cheaper process than chemical polishing carried out by simple immersion in a strong, often hot, acid solution for a time of 0-5-5 min. Proprietary forms of these solutions may have inhibitive additives to reduce pitting and gas-evolving tendencies. The rate of metal attack is high, substantial acid fume arises and the reduction of rate of attack as the solution becomes depleted may be combatted to some extent by increasing the temperature. The surface should be rinsed quickly after removal from the bright-dip solution, otherwise ‘transfer etch’ will occur; this is essentially attack by the residual acid film and may be accelerated by atmospheric oxygen absorbed by the film during transfer. A typical bright-dip solution for copper might be an aqueous solution containing 40% H2S04and 10% HNO,. In this solution nitric acid is the oxidant ( A ) and sulphuric acid the contaminant (B); there is no diffusion layer promoter. It is used cold and gives a good bright surface in 0-5-2 min accompanied by substantial fuming. Improved performance can be achieved by adding 5-1 5 g/l hydrochloric acid, and increasing the nitric acid content t o 20%. Solutions of this type are widely used in industry’. Chemical polishing, yieiding a surface of high specular reflectivity, exploits fully optimised bright dip solutions often achieved by the further addition of phosphoric acid at the expense of the residual water. Because phosphoric acid is relatively viscous at lower temperatures (e.g. less than 40°C) it can act as diffusion layer promoter (C),but its presence increases the chemical costs considerably. By invoking these principles such solutions may now be designed for most metals, and in the case of atmospheric metals a range of alkaline solutions may also be considered. For the more reactive metals such solutions become increasingly strong, hazardous and expensive. A typical, solution for the chemical polishing of titanium could contain 40% v/v HNO,, 30% v/v H2S0, and 30% v/v of 40% HF. The titanium would be exposed to this solution for 30s at 70-80°C. A list of recommended solutions for the commoner metals is given in Table 11.4; for other metals reference should be made elsewhere’-6. A notable recent achievement has been the development of a first class chemical polishing solution for aluminium. The older solutions are essentially bright dips, requiring a ‘desmutting’ post-treatment which may arise either from intermetailic compounds within the aluminium alloy or from excessive loose oxide on the surface, based on nitric-sulphuric-phosphoric acid mixtures’. However, the addition of small amounts of a noble metal to the solution improves the degree of brightening obtained substantially and the resulting process has proportional features as Phosbrite (trade mark of Albright and Wilson plc). One of the best compositions is 77.5% v/v H,PO,, 16.5% v/v H,SO,, 6.0% v/v HNO,, and 0-5-2.0g/l CuSO,. Copper is deposited on the surface of the aluminium as a fine colloidal precipitate and may be washed or wiped off leaving a highly specular surface. The mechanism of its behaviour has been the subject of many investigat i o n ~ ~ -and ’ ~ ,the process may be regarded as highly successful such that its use for the industrial production of reflectors and mirrors is now widespread, replacing electroplating.
11 :27
CHEMICAL AND ELECTROLYTIC POLISHING
Table 11.4
Metal Aluminium
Cadmium
Chemical polishing solutions
Solution I . Phosphoric acid Nitric acid Acetic acid Water 2. Phosphoric acid Hydrogen peroxide Water 3. Phosphoric acid Sulphuric acid Nitric acid Copper nitrate 1. Chromium trioxide
Sulphuric acid 2. Sulphuric acid Hydrogen peroxide (30 vol)
Copper
1. Sulphuric acid
Nitric acid Hydrochloric acid Water Mild and Carbon steels
Nickel
Conditions
80% v/v 5 v/v 5 10 v/v 75% w/w 3.5 w/w rem w/w 70% v/v 20 v/v 10 v/v 5 g/l
Temp. 90-1 10°C Time 0.5-5 min Temp. 90°C Temp. W ' C Time 1-4 min
85 g/l 1-2 g/l 0.3% v / v 7 v/v
45% v / v 22 v/v 1-2 v / v 33 v/v
Temp. 20°C Time 5-45 s
I . Oxalic acid 25 g/ Hydrogen peroxide (100 vol) 13 g/l 0.1 g/l Sulphuric acid 40% v / v 2. Nitric acid Hydrofluoric acid (40010) I O v/v Water 50 3. Hydrogen peroxide (30 Vol) 80% v/v Hydrofluoric acid (40Vo) 5 v/v Water 15 v/v
Temp. 20°C
50% v/v
Temp. 90°C
Glacial acetic acid Nitric acid Phosphoric acid Sulphuric acid
Temp. ~ 2 0 ° C Dissolves 30-50 pm/min
30 v / v 10 v/v 10 v/v
Titanium
Nitric acid 40% v / v Sulphuric acid 30 v /v Hydrofluoric acid (40W)30 v/v
Temp. 70-80°C Time 30 s
Zinc
Chromium trioxide 220 g/l Sodium sulphate 12-30 g/l (followed by dip in 5 g/l sulphuric acid for 1-10 s)
Temp. 20°C Time 5-30 s
All acids used are the most concentrated forms available. Solutions should be made up by using water or the acid SoIution containing most acid as the base to which other acids are added. All solutions should be mlxed with care using cooling and continuous mixing.
11:28
CHEMICAL AND ELECTROLYTIC WLISHING
Electropolishing Electropolishing techniques utilise anodic potentials and currents to aid dissolution and passivation and thus to promote the polishing process in solutions akin to those used in chemical polishing. The solutions have the same basic constitution with three mechanistic requirements -oxidant (A), contaminater (B) and diffusion layer promoter (0-but, by using anodic currents, less concentrated acid solutions can be used and an additional variable for process flexibility and control is available. The electrochemical characteristics of electropolishing can be seen by referring to a typical polarisation (potential versus current density) diagram (Fig. 113).The aim is to provide a ‘polishing plateau’ at constant current over a substantial range of potential, but the value of that constant current can be fairly critical. Thus in (a) the metal is passivated and in (c) it dissolves under solution diffusion control, neither condition giving effective electropolishing. A wide potential range is desirable in order to provide process flexibility, but does indicate the need to use potential control as a means of controlling the process (Fig. 11.6). Potentiometric control, or potentiostatic if exercised by a potentiostat instrument, is clearly preferred, but demands the use of a good reference electrode to be effective. The series, or galvanostatic, technique is most generally used in industry because of its convenience in using conventional transformer/rectifier equipment, but can only be considered equivalent in the fortuitous circumstance of Fig. 1 1.5b where the current ‘fall back’ is slight. The all-too-common habit of quoting a cell voltage for electropolishingconditions is consequently rather meaningless, dependent as it is on electrolyte conductivity and inter-electrode spacing in the process cell used. Referring to Fig. 11.Sb, the initial rise in current corresponds to simple metal dissolution, expressed quantitatively through the Tafel equation relating potential and current logarithmically, and for multi-grained metals
E
€
E
Log I
(a) Fig. 11.5 Anodic polarisation (potential-current density) curves for nickel in (a) dilute sulphuric acid, (b) cold 10 M sulphuric acid, and (c) hot or agitated 10 M sulphuric acid
11 :29
CHEMICAL AND ELECTROLYTIC POLISHING
50-100 V
6-12
v
(b)
Fig. 11.6 Simple electrical circuitry for electropolishing
can be used to electro-etch the surface (see 'Electrolytic Etching' below). At a certain critical potential film formation can occur and dissolution is limited by that critical metal/electrolyte interface whose stability depends both on metal film forming tendencies and solution viscosity: if the film is too stable, the current is very low and any polishing takes days or weeks to be accomplished. If mass transfer in the viscous layer is increased by agitation or increased temperature the requisite levelling action may be lost. At higher anodic potentials the current rises sharply and, while polishing may still occur, it is accompanied by pitting and is therefore an unacceptable condition. Such behaviour is usually associated with oxygen evolution becoming thermodynamically possible (the overpotential being over 1 V corresponding to a reaction potential of at least 1 5 V (SHE).The oxygen bubbles evolve at discrete favoured sites causing local film breakdown and stirring which increases local dissolution rates resulting in pit formation. A flexible process has a polishing plateau over a range of 1 V, and -0.3 V is a minimal requirement. The simplest and most thoroughly studied solutions are those based on phosphoric acid at low temperatures ( O
-0.378
-0.388
-0.71
-0.24
-
-0.68
+0.042 -0.293 +0,275 approx. -0.35 -0.378 -0.353
Cuprocyanide, pH 12, 55°C Argentocyanide, pH 1 1 . 5 , 20°C
Zinc sulphate, pH 4.0, 20°C 18% w/v HCl pickle for ferrous metal, pH < 0 3% w/v HCI pickle for copper alloys, pH 0
immersion deposit impedes measurement) +0.122 -0.578 -
-1.213 ( 162)
-0.64
-0.17
Plated Hydrogen steel evolved below:
-0.65
-0.24 > O
0
+0.17
Electroplating aluminium and its alloys requires a similar technique. In aqueous solutions it is impossible to lower the potential sufficiently to reduce an alumina film, so the substrate is immersed in a strongly alkaline solution capable of dissolving it: A120,
+ 20H-
= 2A10;+
H20
The solution also contains a high concentration of zinc (as zincate), which is noble relative to aluminium. As metallic aluminium is exposed, it corrodes, reducing zincate ions and forming a coating of zinc:
+
+ +
A1 40H- = A10;+ 2H,O 3eZnO$- 2 H 2 0 2e- = Zn 40H-
+
+
The immersion deposit is necessarily somewhat defective, for the reasons already mentioned, though immersion deposits from complex ions are finer grained and more satisfactory than those reduced from aquocations. The zinc coating is, under the best conditions, an acceptable basis for a copper undercoat from the cuprocyanide bath, on which other coatings can be plated, but there is usually a fair proportion of rejects in commercial operation. Other processes similar in principle use tin or bronze immersion coatings. Service corrosion effects Undercoats, 'flash' deposits produced by strike baths, and immersion deposits are potential sources of weakness. If their structure is faulty it affects the subsequent layers built on the faulty foundation. The greater the number of stages, the higher the probability of faults.
ELECTROPLATING
12 :23
Additional metal layers can create bimetallic corrosion cells if discontinuities appear in service. The layer of copper beneath cadmium plate on aluminium (using a zincate plus cuprocyanide deposit technique) can cause corrosion troubles. When aluminium is plated with nickel and chromium, rapid service corrosion in the zinc layer causes exfoliation.
Corrosion potentials in plating baths The standing potentials of steel and copper (before application of current) are shown in Table 12.2, together with the standing potential of the plated metal and the potential below which hydrogen should, in theory, be evolved. The potential of the cathode during deposition at a typical current density is also given.
Factors influencing Structure sa-61
Substrate effects: epitaxy and pseudomorphism Both the words epitaxy and pseudomorphism are derived from classical Greek, the former meaning literally close to or close upon an arrangement, row or series (technically an arrangement imposed upon a skin or layer, e.g. an electrodeposit, which is close upon a substrate) and the latter false form (technically a mineral or crystal displaying a form more characteristic of another material than its usual one). For many years the two terms were held to be synonyms for one phenomenon in electrodeposits. Since 1936 it has become clear that there are two related phenomena, on each of which one of the names is bestowed. Not all authors recognise this, nor is the usage employed here adopted uniformly. Both phenomena are of great practical importance. Pseudomorphism received methodical study from about 1905. A microsection taken across the interface between a substrate and an electrodeposit shows the grain boundaries of the former continue across the interface into the deposit (Fig. 12.5). As grain boundaries are internal faces of metal crystals, when they continue into the deposit the latter is displaying the form of the substrate. Hothersall’s 1935 paper contains numerous excellent illustrations with substrates and deposits chosen from six different metals, crystallising in different lattice systems and with different equilibrium spacing. Grain boundary continuation and hence pseudomorphism is evident despite the differences.
Fig. 12.5
Pseudomorphism; grain boundaries in the substrate (S)are continued in the electrodeposit (0)
12 :24
ELECTROPLATING
Epitaxy is a relation on the atomic scale between substrate and electrodeposit. Imagine that the interface of the micro-section were magnified about lo7times so that the rows of atoms in the metal lattice become visible. If the deposit shows epitaxy, there will be an ordered and regular relation between substrate and deposit atom positions (Fig. 12.60). A non-epitaxial deposit shows no such relation (Fig. 12.66). Direct experimental demonstration of epitaxy was first made in 1936 by Finch and Sun. Earlier, metdlographers argued that pseudomorphism (which they could see) meant there must be epitaxy (which they could not), as grain boundaries are surfaces where the direction of lattice rows of atoms changes; if epitaxy were assumed to exist, pseudomorphism should result. Reversing the argument, pseudomorphism was taken as evidence for epitaxy (Fig. 12.6~). Nan-epitaxial
Epitaxy
0 0 0 0 0 0 0 0 0 0
0
0
0
0
0
0
0
0 0
..... ..... .....
0
~
0 0 0 0 0 0.0..
.om..
0 . 0 . .
(a)
(b)
0
0
S
0-
0 0
S
0
0 0 0
a
Fig. 12.6 (a) Co-ordination across a substrate S-electrodeposit D interface on the atomic scale produces epitaxy. (b)a non-epitaxial deposit has no co-ordination and (c) epitaxy would be expected to produce grain boundary continuation at the interface, though in fact grain boundaries often continue to thicknesses far greater than those at which epitaxy disappears
ELECTROPLATING
12:25
Electron diffraction investigations showed that epitaxy did indeed exist when one metal was electrodeposited on another, but that it persisted for only tens or hundreds of atomic layers beyond the interface. Thereafter the atomic structure (or lattice) of the deposit altered to one characteristic of the plating conditions. Epitaxy ceased before an electrodeposit is thick enough to see with an optical microscope, and at thicknesses well below those at which pseudomorphism is observed. Epitaxy reflects the formation of metallic bonds between the dissimilar atoms at the interface. When the two metals crystallise in different systems, their relative orientation is that which promotes the maximum co-ordination and the maximum metallic bonding. The stability achieved by epitaxy overrides any lost due to the lattice strains imposed. These strains may be considerable; ‘stresses’ calculated from the bulk elastic moduli are correspondingly high, and sometimes puzzle the uninitiated if they exceed the bulk tensile strength. It is an oversimplification to regard the interface as being highly stressed; were the ‘stress’which seems to be parallel to the interface reduced by some means to zero, the energy that would have to be put into the bonds normal to the interface would be much greater than that released. The simple concept of stress in a homogeneous alloy is not applicable to the peculiar case of a substrate-electrodeposit interface. The latter is unique in having metallic bonds carried across a very sharp boundary. The practical result of epitaxy is a very high degree of adhesion between coating and substrate. The force needed to separate the interface is similar to that needed to break the metals on either side. Where a true metallic bond forms at an epitaxial interface it is only possible to measure adhesion if the bond is the weakest of the three near the interface. An adhesion test based on breaking the joint indicates only which of the three is weakest. For practical purposes any epitaxial joint will have a strength more than adequate for service conditions. Non-epitaxial electrodeposition occurs when the substrate is a semiconductor. The metallic deposit cannot form strong bonds with the substrate lattice, and the stability conferred by co-ordination across the interface would be much less than that lost by straining the lattices. The case is the converse of the metal-metal interface; the stable arrangement is that in which each lattice maintains its equilibrium spacing, and there is consequently no epitaxy. The bonding between the metallic lattice of the electrodeposit and the ionic or covalent lattice of the substrate arises only from secondary or van der Waals’ forces. The force of adhesion is not more than a tenth of that to a metal substrate, and may be much less. Epitaxial growth is prevented if semiconducting films of grease, oxide, sulphide, etc. cover the cathode surface. These occur wlien pretreatment is inadequate, when plating baths are contaminated, or when, as with stainless steel, aluminium, titanium, etc. an oxide film reforms immediately after rinsing. Low adhesion resulting from non-epitaxial electrodeposition is used in electroforming to promote easy separation of deposit and substrate. When semiconductors or non-conductors are to be electroplated, a form of dovetail mechanical joint (achieved as outlined above) is essential. Means similar to those for stainless steel and aluminium have been devised to deal with other alloys which passivate readily. Sometimes, even with special methods, some oxide remains so that the electroplated coating is anchored
12 :26
ELECTROPLATING
only by small epitaxial areas. There is risk of failure. Thermal stress or relatively mild abrasion may part the interface and cause the unanchored areas to blister. Adhesion is improved by post-plating annealing. The oxide at the interface is dissolved in one or other metal, or diffuses to grain boundaries, etc. and alloying at the interface produces the desired metallic bond. Pseudomorphism has less desirable consequences, and usually means are sought to suppress it. If the substrate has been scratched, ground or abrasively polished, or if it has been cold rolled or cold formed, the surface is left in a peculiar state. Cold working reduces the surface grain size, and produces deformed, shattered and partly reoriented metal. It may produce microcrevices between the deformed grains, and, with some processes, nonmetallic impurities and oxides are embedded in the surface. The disturbed state of the substrate is copied by a pseudomorphic electrodeposit with several consequences (Fig. 12.7). One is aesthetic; it has often been noted that almost invisible abrasion of the substrate develops as more prominent
Fig. 12.7 The disturbed structure of a scratch, with fragmented and distorted grains, is perpetuated by a strongly pseudomorphic electrodeposit
Fig. 12.8 A fairly strongly pseudomorphic bright tin deposit (left) has its brightness impaired by the shattered surface layer produced on steel by cold rolling. When this layer is removed, the deposit is mirror bright (right). Coating 5 pm thick
ELECTROPLATING
12 :27
Fig. 12.9 Corrosion resistance of tin-nickel electrodeposit impaired by pseudomorphic porosity originating on cold-rolled steel surface (left). Panel on right has had the shattered grain surface removed by chemical polishing (0.125 pm removed). Coating thickness 15 pm; panels exposed 6 months to marine atmospheric corrosion (Hayling Island)
markings in the deposit. A chalk mark on steel produces local abrasion, hardly noticeable when the chalk is wiped away. If a strongly pseudomorphic electrodeposit is applied the chalk mark reappears indelibly on its surface. A bright deposit may have its lustre greatly reduced by pseudomorphic growth on a deformed surface (Fig. 12.8). The corrosion protection is reduced if pseudomorphism with a deformed substrate leads to discontinuities at illfitting deposit grains (Fig. 12.9). A pseudomorphic coating usually presents a dull or rough crystalline appearance. When the crystal form of the substrate is copied in the deposit, growth generates faces of simple index. An artificial face of high index soon grows out when plated. Tradition demands a featureless mirror surface on metal coatings, and a way of producing this which has attracted much commercial effort is by using brightening addition agents. Micro-sections of electrodeposits from the more effective bright plating baths d o not exhibit pseudomorphism. The deposit usually shows no grain structure, but instead a series of light and dark bands parallel to the substrate (Fig. 12. IO). Pseudomorphism is suppressed by the addition agent adsorbing on and blocking areas taking part in pseudomorphic growth. In the initial stages of bright plating the addition agents adsorb at similar points on the substrate. Growth commences from fewer substrate nuclei when annealed nickel is plated in a bright nickel bath than in a dull (Watts’) bath without additions. In the earliest stages of deposition, replicas of the surface show evidence of pseudomorphism even in bright baths (the substrate grain boundaries are carried into the deposit) but this is suppressed rapidly as the thickness increases. The aim with bright plating baths is to inhibit growth sufficiently to suppress pseudomorphism, but not so much as to suppress epitaxy and adhesion. An excessive concentration of addition agent will also suppress epitaxy, so that deposition occurs on to an adsorbed layer of brightener. Brightener adsorption is often potential dependent and trouble may occur first at high current density (low potential) areas.
12 :28
ELECTROPLATING
Fig. 12.10 Banding often observed in micro-sections of bright electrodeposits. (a) Bright tin (courtesy of the Tin Research Institute), and (b)and (c) Bright gold
ELECTROPLATING
12:29
Electrolyte e f f e ~ t s ~ ’As - ~ ~a deposit becomes thicker, the influence of the substrate diminishes, and eventuallythe structure is characteristic only of the electrolyte composition, the temperature, current density and mode of agitation. A great variety of structure is observed; some are analogous to those seen in cast metals, but others are obtained only by electrodeposition. Crystalline deposits from baths containing little or no addition agent often develop a preferred orientation texture. Some bright deposits show a texture, but in general as growth processes are progressively inhibited by increasing addition agent concentration or by using more active materials, the deposit becomes progressively finer grained and loses preferred orientation textures. The compositions of baths chosen for practical use result in initial rates of lateral growth much greater than the rate of outward growth. This is a desirable feature; it causes the coating to become continuous at low thicknesses. The opposite condition of a faster rate of outward growth is undesirable, and results in a non-coherent deposit. Predominantly outward growth occurs when the transport of metal ions becomes slow compared with their rate of discharge, Le. it is favoured by high current density, low temperature and lack of agitation. Lateral growth processes are then starved of material to support them, but outward growth moves the deposit towards the supply, and the prominences formed benefit from greater diffusive flux. There are strong pressures in industrial production to increase electroplating rates, which carry a danger of using high current density and causing a shift to outward growth. In baths where the coating is electroplated from aquocations at high cathode efficiency, the onset of lateral growth is fairly sharp. Cathodes have a range of local current density, and the coating on the high current density areas becomes friable, dark coloured and rough as the transition is reached. Such coatings are termed burnt and the corrosion protection is degraded. With baths working in the acid pH range there is the complication that once an appreciable part of the current is used to reduce water, the pH at the cathode rises and insoluble hydroxides are precipitated and incorporated in the coating. With complex cyanide baths the onset of ‘burning’ is less sharp. There is normally considerable simultaneous hydrogen discharge, and as the current density rises there is no sharp limiting current density for metal discharge. Addition agents raise the lateral-outward transition to higher current densities, by inhibiting outward growth. Nevertheless all electroplated coatings show signs of deteriorating properties if the baths in which they are produced are worked at sufficiently high current density. Form of current passed through cell7w79 Commercial electroplating began with pure d.c. from galvanic cells. Later, for many years d.c. generators were used. Their current output is unidirectional but with a superimposed ripple. Part of the ripple stems from the angular motion of the armature coils during the period they supply current to a commutator segment, and part from variations of contact resistance at the commutator. Generators have been superseded by transformers and rectifiers. Copper-oxide, mercury-arc, selenium, germanium and silicon rectifiers have been used, and examples of each are to be found in service. These devices supply varying unidirectional current whose form depends on the number of phases in the input and the circuit used. A half-wave single-phase rectifier provides a pulsating current; a full-wave three-phase set has a much smoother output.
12:30
ELECTROPLATING
Alternating currents with asymmetric forms have been used, mainly for electroforming and thick engineering deposits. Where the cycles are slow, e.g. several seconds, the term periodic reverse current (p-r-c) is used. The benefit claimed for p-r-c plating is that smoother, thick deposits result from selective dissolution of peaks in the reverse part of the cycle. This assumes the electrode process reverses during the anodic period, which is not always the case. In chromium plating the coating becomes passive in anodic periods, while in acid gold baths based on aurocyanide, the process is also irreversible. More recently, asymmetric a.c. with a much higher frequency of 500 Hz was found to alter beneficially the properties of nickel from chloride baths. Pulses of unidirectional current have been used to modify coating properties. When plating starts it is possible, for a time, to use a current much higher than the steady state limit, drawing on the stock of ions near the cathode. Provided sufficient time is allowed between pulses, a coating can be built of layers plated at much higher current density than normal. Improved gold coatings were produced by relatively rapid pulses. The technique of barrel p/afing results in pulse plating of an irregular sort, with pulse durations of the order of a second and inactive periods rather longer. Chromium plating from chromic acid baths is more sensitive to the source of current than most other processes, sufficiently so for commercial operators to use at least three-phase rectifiers as a rule, and to take precautions against any temporary break of current during voltage regulation. A recent investigation showed that the ripple introduced by thyristor control of rectifiers was detrimental to chromium electrodeposits. Industrial Electroplating Techniques
Electroplating is usually a finishing technique applied after an article has been completely fabricated. Fairly large articles, from cutlery to motorcar bumpers, are dealt with by vat plating. They are suspended by a conducting connection in a rectangular tank or vat of electrolyte. The anodes are arranged about the periphery of the tank. For small runs the cathodes may be suspended by copper wire wrapped round a suitable part, but for longer runs a plating jig is used. This is a copper frame with phosphor bronze spring contacts to hold the work, and insulated, usually with a P.V.C. coating, on all but the contact points. The point of contact between wire or jig and the article becomes a weak part in the coating, and some thought should be given to providing or selecting contact points in insignificant areas. Vat plating is used sometimes with articles too large for complete immersion. Printing, calendering, drying and similar rolls are part-immersed and revolved continuously during plating. However, it is much more difficult to plate half an object, reverse it, and complete the other half later; the ‘join’ between the two deposits is rarely satisfactory. Small objects, nuts, bolts, screws and small electrical parts are plated in a revolving barrel. Electrical connection is made by a conductor immersed in the tumbling mass, and electrodeposition, which is confined to the outer layer of the mass at any instant, takes place in intermittent stages for any individual object. The coating is abraded during the process. The peculiarities of chromium deposition set it apart, and the normal barrel-plating
12:31 processes are not used. In so-called chromium barrels the small parts travel and tumble along a helix inside a rotating cylinder during deposition, and are electroplated for a much greater proportion of the time than are parts in normal barrels. Brush plating is a special technique which dispenses with a container and uses a swab soaked in electrolyte applied to the work. In jet plating a stream of electrolyteis applied to the cathode. Both are methods of selective plating, applying an electrodeposit to only a part of an article. Little has been published about the techniques or the properties of coatings they produce. Continuous plating of wire and strip is, unlike the preceding techniques, a prefabrication process. The production of tinplate is the largest scale continuous operation, but any electrodeposit may be applied this way. Subsequent fabrication processes are likely to damage the coating, so that pre-coating is best reserved for ductile coatings which are anodic to the substrate in service, as is the case for tin. ELECTROPLATING
Between all stages of immersion (cleaning, pickling, plating, post-plating treatment) work has to be rinsed. Once the hydrophobic solid has been removed, metal surfaces withdrawn from solutions carry a film of liquid. The solution lost this way is known as drag-out. A film lOpm thick is the minimum retained by smooth, well-drained, vertical surfaces. On rough or horizontal surfaces and in recesses it is much thicker, as it is also with viscous solutions. During rinsing the film is diluted, and the ratio of the final concentration to that present initially is the dilution ratio. The dilute material is carried forward to the next process, and clearly the highest concentration of impurity permissible before the subsequent process is affected adversely determines the maximum dilution ratio which can be allowed. Sometimes there is a minimum dilution ratio; between nickel plating and chromium plating it is essential that the rinsed metal surface does not become passive, and prolonged rinsing carries a danger of eliminating the slight but important amount of rinse water corrosion which keeps the surface active between stages. Usually rinsing troubles are caused by a dilution ratio that is too high. If incoming work passes through a process stage, and the drag-out from that stage is in turn discarded in a subsequent rinse, the maximum concentration of material carried into the bath is equal to that in the film carried over. However, there is an increasing tendency to conserve materials and steps are taken to return drag-out losses. In so doing the impurities are also returned, so conservation measures require a reduction in the dilution ratio of the preceding rinse. Inadequate intermediate rinses are detrimental to the corrosion resistance of the coating because carried-over impurities impair the functioning of plating baths. Inadequate final rinsing leads to increased corrosion of the coating, and to staining. Staining, which is a serious aesthetic problem with decorative coatings, may itself arise from corrosion. Some stains are caused by the precipitation of dissolved solids when rinse water evaporates, but in other cases they are caused by corrosion supported by the presence of an electrolyte in the rinse water.
12:32
ELECTROPLATING
Post-pleting T r e a t r n e n t ~ ~ ~ - ~
Where the corrosion resistance of a coating depends upon its passivity, it is common to follow plating with a conversion coating process to strengthen the passive film. Zinc, cadmium and tin in particular are treated with chromate solutions which thicken their protective oxides and also incorporate in it complex chromates (see Section 15.3). There are many proprietary processes, especially for zinc and cadmium. Simple immersion processes are used for all three coatings, while electrolytic passivation is used on tinplate lines. Chromate immersion processes are known to benefit copper, brass and silver electrodeposits, and electrolytic chromate treatments improve the performance of nickel and chromium coatings, but they are not used to the extent common for the three first named. The tin coatings as deposited in tinplate manufacture are not bright. Until comparatively recently bright tin electrodeposition was not practised commercially, there being no reliable addition agents. To produce bright tin on tinplate and other products, the process offlow melting or flow brightening is used; tinplate is heated by induction or resistance, and plated articles by immersion in hot oil to melt the tin, which flows under surface tension to develop a bright surface. While the tin is molten it reacts to form an alloy layer with the substrate. The alloy layer alters the corrosion behaviour. Other electroplated articles are heated after plating to expel hydrogen which has entered the substrate during cleaning, pickling and plating, and which embrittles some metals, mainly high-strength steels. Generally speaking alteration of the deposit structure and properties is not desired. Another use of post-plating heat treatment is to improve adhesion, as already mentioned (p. 12:26). Mechanical polishing, formerly the principal means of producing bright coatings, has become less important with the extension of the use of brightening addition agents. Mechanical polishing reduces the thickness of a coating, and may cut through to the substrate. As corrosion resistance is related to thickness, mechanical polishing can be detrimental. It may also increase porosity.
Properties of Electrodeposits
’’
Thickness
Coating thickness is one of the most important quantities connected with corrosion resistance, and its measurement and control is a feature common to all electroplating operations and in all quality specifications. In some cases coating thickness has functional importance, e.g. where there are fitting tolerances, as with screw threads. In most cases however it is the connection with corrosion resistance that makes thickness important. Where the coating is anodic to an area of substrate exposed at a discontinuity the coating is slowly consumed by corrosion, but the criterion of failure is the appearance of substrate corrosion product. This does not form until almost all the coating is consumed. Coatings which are cathodic to the substrate must have no discontinuities if substrate corrosion is to be suppressed.
ELECTROPLATING
12:33
The criterion of failure is usually the same. Freedom from discontinuity is also related to thickness. Discontinuities have three origins: spontaneous cracking to relieve internal stress, pores formed during the growth of the coating (see p. 12:41), and abrasion and wear. The last two causes, i.e. porosity and wear, both exhibit diminishing incidences as thickness rises. Apart from the peculiar case of electrodeposited chromium, internal stress cracking is a sign of incorrect plating conditions. Broadly speaking, thickness and corrosion resistance increase together, The thickness of an electroplated coating is never uniform. On the significant area (Le. that on which corrosion resistance and other special properties are important) of a plated surface there are two important thicknesses, Le. (a)average thickness, which determines the production rate and plating costs; and (b) minimum local thickness, which, as the weakest link in the chain, determines the corrosion resistance. The ideal is to make these equal; the larger the difference the greater the waste of metal. The difference can be reduced by special procedures, but at a cost. When the cathode is being plated, the electrical field is not uniform. Both electrodesare equipotential surfaces, so that prominent parts of the cathode, e.g. corners, edges, protuberances, etc. which are relatively nearer the anode are plated at a higher average current density, resulting in a thicker coating. Recesses and more distant parts are more thinly plated. The distribution of thickness tends to be the reverse of that found with paints, hot-dipped and other coatings which are applied as liquids. Liquid-applied coatings are thin on sharp edges, and thick in recesses because of the effects of surface tension and radii of curvature. The numerous factors which contribute to the thickness distribution can be divided into two groups, i.e. (a)those connected with the nature of the plating bath (see below) and (b) those to do with the geometry of current paths in the bath, including the shapes of the electrodes. Throwing Power98-'05
In a given plating cell, thickness distribution is found to vary with bath composition, current density, temperature and agitation. It is common to speak of the throwing power of a plating bath. The throwing power of chromic acid baths ispoor, i.e. there is a relatively large difference between maximum and minimum local thickness; conversely, the throwing power of alkaline stannate baths is good, i.e. there is much less difference in the local thicknesses. Strictly speaking the bath composition should be qualified by the conditions of use, as they affect throwing power. Otherwise, the usual conditions are implied. A numerical throwing index can be calculated from the performance of a plating bath in a cell of standard geometry. Two widely used cells are (a) the Haring-Blum cell and (b) the Hull cell (Fig. 12.11). The Haring-Blum cell was devised for throwing index measurement; the Hull cell is used mainly to study the effects of varying bath composition. The Haring-Blum cathode is divided into two equal plane areas, distant PI and P2 from a common anode, and a quantity called the primary current density ratio P is defined as P = P,/O,
12:34
Fig. 12. I 1
ELECTROPLATING
(0) Haring-Blumcell for throwing index measurement, in elevation and (6)Hull cell
(plan view) which can also be used for measuring throwing indices
This is the ratio in which the current would divide, if electrolytic resistance were to control its flow entirely. The metal distribution ratio M is the ratio of the thicknesses of the coating actually deposited during a measurement. There are several numerical scales of throwing index T, but Field's is widely adopted:
T = 100
P-M P+M-2
(12.11)
VO
On this scale, zero represents the case when M = P, and electrolyte resistance is the main factor. Throwing power can be worse, down to a limit T = - 100% when M = 00, Le. no deposit at all on the far cathode. Conversely, when A4 < P , T is positive. Were M to reach 1.0 despite the difference in position, T = 100%. At one time 100% was regarded as an unrealisable limit, but conditions have been found for which T = 150% in a Haring-Blum cell. The Hull cell cathode has a continuous variation of current density along its length, and there are equations which give the primary current density at any point not too near the end. If the local thickness is measured at two points for which P is known, Tcan be calculated. The real current distribution is a function of cathode and anode polarisation as well as of the resistance of the electrolyte. The metal distribution ratio will be
+
+
+
( 12.12)
where V = cell potential difference between anode and cathode, AE =total potential difference caused by polarisation (anode and cathode) on the cathode area indicated by the subscript and e = cathode efficiency as indicated by the subscript. As AEwill be a function of current density, Twill be a function of electrode area, and comparisons should therefore be made with cells of standard size. Equation 12.12shows that high throwing indices will result when polarisation rises steeply with current (A,!?,>> A&) and cathode efficiency falls steeply (E;? > > e , ) . The primary current ratio, P = &'*/t,, affects the result because
12: 35
ELECTROPLATING
by altering the currents the polarisation terms are altered. For example, with an acid copper bath in a Haring-Blum cell, 194A/m2 average c.d.:
P= 2 7Y'o=+7
5 +11
11 +22
23 +41
An increase in conductivity usually increases T because it increases the proportion of polarisation in the total cell potential difference and lowers the ( V - SZ). Changing the conductivity of an acid copper ratio ( V - SI)/ bath with sulphuric acid produced the following result (291 A/mz average c.d., P = 5 ) : Conductivity (S/cm) T ( OJO 1
0.08 +5
0.15 +11
0.26 +13
0.30 +27
where S is the SI unit of conductance (siemens). Many baths in which metal is reduced from complex anions (e.g. cyanide baths, stannate baths) give high throwing indices because both polarisation and cathode efficiency variation favour a low value of M. The cathode efficiency for a typical copper cyanide bath (40°C) was: Current density (A/mZ) 32 65 Cathode efficiency 76 68
129 258 388 56 34 21
The throwing index for the cyanide bath is usually about +40% and rises as the cell current is increased to as high as + 85%. Aquocation baths give values near T = 0, though conditions may be selected which give much higher figures if there is a steeply rising section of the polarisation curve. Chromium plating baths invariably have large negative throwing indices, despite deposition from a complex ion. The cause is the anomalous rising trend of cathode efficiency with current density and the existence of a minimum current density below which the efficiency is zero. A typical bath (400g/l CrO,, 4g/l H,SO,, 38°C) gave: Cathode current density (A/mz) 199 253 384 763 1785 5 130 30800 Cathode efficiency (To) 0 5.9 11.9 13.9 18.8 22.7 24.4 If the current density on the far cathode in a Haring-Blum cell was 199 A/m2 or less, T = - 100%. Throwing indices measured in a Hull cell differ from those in a HaringBlum cell because of the differences in geometry. In a Hull cell several pairs of points can be found which have the same primary current ratio, but for which M and hence T a r e found to vary because of polarisation changes. Current Path Geometry'06-110
The polarisation and cathode efficiency terms in equation 12.12 cannot be altered in practice to improve thickness distribution, as they tend to be decided by overriding considerations. It is usual to accept the distribution obtained without special precautions as being the best commercial solution, although the average thickness needed to achieve the necessary minimum
12:36
ELECTROPLATING
local thickness may be high. Where this approach does not serve there are a number of methods of altering the term !,/PI in equation 12.12: (a) By using shaped (conforming) anodes, additional (auxiliary) anodes
or ‘bipolar’ anodes to bring anode areas nearer to cathode recesses. Insoluble anodes are better where they are applicable as, once made, they do not alter shape during use. (6) By using non-conducting shields of plastic or glass to equalise the current path lengths. (c) By placing auxiliary cathodes (‘robbers’ or ‘thieves’) near high-currentdensity points t o divert deposition. This does not save metal, but has the merit that auxiliary cathodes can be incorporated into jigs for long runs in automatic plating machines. Auxiliary cathodes are used in heavy chromium deposition where metal waste is secondary t o the cost of removing excess chromium when grinding t o precise dimensions. Where a number of small parts are plated together on a jig, it is usually possible to dispose them so that they serve as ‘robbers’ for each other. (d) By attention to certain ‘rules’ when designing articles which will be finished by electroplating. Many external contours are chosen for reasons of style. It helps to avoid features like sharp recesses, which are bound to cause trouble. A simple rule is the ‘1 in ball test’ or perhaps the ‘25 mm ball test’: if there is any part of a surface which a ball of this diameter cannot touch when rolled over it, there will be difficulties. There are other design aspects, covered in specialist publications, attention t o which improves the corrosion resistance which can be imparted by plating (see also Section 11.4).
Structure-dependent Properties”’-‘‘‘
Composition of the electrodeposit Attention has been drawn to the dependence of structure on both substrate and plating conditions, and the transition in properties which occurs across the section of a deposit. Most commercial electrodeposits have a high purity, yet in a sense impurities are vital to their successful application. Alloy electrodeposition possesses a literature whose bulk attests the subject’s fascination for research (which the author shares), but is out of proportion to the extremely limited commercial applications. Alloys in general metallurgical practice provide a variety of mechanical properties; in electroplating the range of properties desired is narrower, and it can generally be achieved by altering the structure of a single metal deposit through changes in the plating bath composition or plating conditions. The microstructure of an electrodeposit can be altered much more than that of a cast and worked metal. This is because the deposit forms well below its melting point, where crystallisation processes are hindered by the virtual absence of solid-state diffusion. Consequently, very small amounts of ‘impurity’ absorbed at important growth sites on the surface cause large changes in the structure of what is, chemically, almost pure metal. The structure is metastable, but permanent as long as the electrodeposit is not heated. A variety of mechanical and physical properties are a
ELECTROPLATING
12:37
reflection of the structure: hardness, ductility, tensile strength, internal stress, electrical and thermal conductivity, etc. As the structure of an electrodeposited metal is altered by changing the plating conditions, the mechanical and physical properties also alter. A plot of structure-dependent properties against the plating variable usually shows the various properties moving in parallel or inverse motion, and over ranges not accessible in cast and worked metal of the same composition. However, if electrodeposits are heated to temperatures where moderate mobility of the atoms is possible, their properties rapidly revert to ‘normal’. The corrosion resistance of electrodeposits depends much more on chemical composition rather than on structure, so that the corrosion resistance of a particular metal is retained for a wide range of mechanical and physical properties. The ‘impurities’ responsible for modifying the structure may originate from: water (dispersed oxides); adsorbing ions, especially cyanides; organic addition agents parts of which are incorporated; or ions of a second metal which are co-deposited. Some regard deposits in which the impurity is a small amount of a second metal as an alloy, but generally they have the same sort of metastable structures as are obtained with non-metallic impurities, rather than those of stable alloys of the same composition. The ‘alloying’ metal serves to cause and perpetuate a non-equilibrium structure whose real basis is the low temperature of the electrocrystallisation process. Generally, the corrosion properties of the various different structures of a given metal are much the same, with the notable exception of nickel containing sulphur from addition agents, which has already been mentioned.
intemai Electrodeposits are usually in a state of internal stress. Two types of stress are recognised. First order, or macro-stress, is manifest when the deposit as a whole would, when released from the substrate, either contract (tensile stress) or expand (compressive stress) (Fig. 12.12). Second order or microstress, occurs when individual grains or localities in the metal are stressed, but the signs and directions of the micro-stresses cancel on the larger scale. The effects of first order stress are easily observed by a variety of techniques. Second-order stress is difficult to observe and much less extensively studied. The causes of internal stress are still a matter for investigation. There are broad generalisations, e.g. ‘frozen-in excess surface energy’ and ‘a combination of edge dislocations of similar orientation’, and more detailed mechanisms advanced to explain specific examples. Tensile first-order stress is a corrosion hazard in coatings cathodic to the substrate. Compressive stress is not usually troublesome, nor is stress of either sign in anodic coatings. Less can be said about high second-order stress, though it may well cause brittleness. If tensile stress is large enough, the coating cracks and a cathodic coating will fail to protect, as illustrated in Fig. 12.13. Tensile stress below the level needed for spontaneous cracking lowers the fatigue limit of a substrate. Tensile stress can in several cases be reduced to safe values by fairly minor changes in microstructure and plating conditions, insufficient to upset other desirable properties. Saccharin is an addition agent for reducing stress in nickel; additions of ammonium chloride
12:38
ELECTROPLATING
Fig. 12.12 An electrodeposit showing unusually high compressive stress. A 150 x 150 mm copper sheet was insulated with lacquer on one side and electroplated with Sn-35 Ni alloy. The high compressive stress has caused the sheet, originally flat, to coil in the manner shown, with the electrodeposit outside
reduce stress in tin-nickel alloy, and small changes in bath temperature and CrO,: H2S0, ratio reduce stress in chromium. The effects of tensile stress in the various layers of nickel plus chromimn coatings are complex, and internal stress in both chromium and nickel (postnickel strike or PNS)layers can be harnessed to produce beneficial cracking (‘microcracking’). Ductility, Hardness, Wear, Strength 121-1z4
The mechanical properties reflect very closely the structures of electrodeposits. The softest, most ductile, weakest form of a particular metal is that with a large crystal size, deposited with minimum polarisation from baths which have no addition agents. This is the type of deposit in which pseudomorphism is strongest. In terms of the accepted deposition mechanism, there is the least inhibition of adion mobility as the deposit grows, and least inhibition of those sites at which equilibrium growth would occur. This electrodeposit has properties the nearest to those for the annealed metal, but even so tends to be somewhat harder. Because of pseudomorphism the properties near the substrate interface may be greatly modified if the latter has a metastable structure, especially one with very small grains produced by mechanical working. The deposit in turn becomes ‘work hardened’ by pseudomorphic growth. When electrodeposition is inhibited the metal becomes harder, less ductile and increases in tensile strength. Metals deposited from acidic solutions of
’I
I
i I
I i
i I
I
I
12:40
ELECTROPLATING
aquocations become harder when the pH is raised to near the value at which the hydroxide precipitates. Co-deposited oxide acts as an addition agent, giving small grained, hard deposits. Hard nickel is produced for engineering surfacing from high pH baths. Many metals can be electrodeposited in extremely hard forms from inhibited baths, but they tend to become brittle, with high internal stress, so that the true tensile strength is hard to establish. Ductility necessarily falls as hardness rises, and coatings become more susceptible to damage by impact, reducing their protective value if they are cathodic to the substrate. Some applications of electroplating depend on the production of unusually hard and wear-resistant forms of corrosionresistant metals. Thick coatings of chromium and nickel are applied to numerous steel parts to combine wear resistance with corrosion resistance. Thick or engineering chromium electrodeposits crack repeatedly during deposition, but the cracks are subsequently sealed and none should traverse the entire coating. Thick chromium coatings have practically no ductility, and because of their defective structure they have a low effective strength. They serve best on stiff substrates. Gold coatings on separable electric contacts and slip rings make use of the high hardness possible with electrodeposition to resist wear. Rhodium is another metal which can be exceptionally hard. Thick coatings have a cracked-sealed structure similar to that of chromium. Interdiffusion with the Substrate’25-’26
A thin metal coating on a metal substrate is not a stable entity; greater stability would be attained if the coating were to diffuse evenly throughout the substrate. Fortunately, at ambient temperatures most of the usual combinations interdiffuse so slowly as to present no practical problem. At high temperatures however, many coatings diffuse quickly. Diffusion in a few systems at moderate temperatures causes corrosion problems. Difficulties can occur with tin, which, with its low melting point of 231 OC, is relatively ‘hot’ at room temperature. On copper and copper-alloy substrates diffusion transforms the tin into the intermetallic phases Cu6Sn, and Cu,Sn. At 100°C the transformation is accelerated, and 5 pm of tin may become wholly alloyed within a year. The alloy coating may pass as tin having a silvery colour, but it is much harder and has a very stable passivity. One use of tin on copper is to facilitate easy joining by soldering, but the alloy has a high melting point and is not easily wet by solder. Thin ‘tin’ coatings on copper which have become wholly alloyed in storage are difficult to solder. Sometimes extremely thin coatings (0-25 pm) used purely for solderability become wholly alloyed in a few weeks. Parts should not be stored too long, and very thin coatings are a false economy (Section 9.5). Tin will protect copper from corrosion by neutral water. Pure tin is anodic to copper, and protects discontinuities by sacrificial corrosion. Both intermetallic phases are strongly cathodic to copper, and corrosion is stimulated at gaps in wholly alloyed coatings. An adequate thickness of tin is needed for long service, e.g. 25-50pm. Another diffusion problem occurs with tin-plated brass. Zinc passes very quickly to the tin surface, where under conditions of damp storage zinc corrosion products produce a film
ELECTROPLATING
12 :41
which greatly impairs solderability. An underplate of copper, or better still nickel, usually cures this trouble. A similar problem, i.e. diffusion of the substrate through the coating to corrode at the surface, arises with gold-plated copper. Many gold coatings are used to ensure a low electric contact in electrical connectors. Gold is pre-eminent because of the absence of stable-corrosion-product films under most service conditions, but it is expensive, so coatings are kept as thin as possible. Electronic devices may operate at fairly high temperatures (100-150°C), and significant amounts of copper may diffuse through the coating to produce a film of oxide on the surface, nullifying the contact value of the gold. Nickel underplate mitigates this trouble (though increasing plating difficulties). To reduce costs, attempts have been made to dilute gold with cheaper metals, while retaining gold-like corrosion properties. Cadmium has been used as a diluent, but while quite high cadmium-golds are gold-like at 25"C, at higher temperatures cadmium oxidises at the surface. Pure gold is preferred for high-temperature contacts.
Porosity In the very earliest stages of electroplating the substrate carries discontinuous areas of deposit growing around nuclei. Lateral growth causes the great majority of growing edges to coalesce with sufficient perfection to be impervious to corrosive gases and liquids. On normal metallic substrates a few edges do not grow together, and a gap remains in the coating. As the coating thickens the gap is propagated as a channel through the coating, to form a pore. Under the conditions chosen for practical electroplating, pores diminish in cross section as deposition continues, and pore density (pores per unit area) falls as thickness increases. The corrosion which occurs when pores allow liquid and gaseous corrosive agents to reach the substrate varies in importance according to the relation between the corrosion potentials of deposit and substrate, the corrosive environment and the function of the coating. If the environment favours wet corrosion processes, relative polarity is the main consideration. If the coating is anodic, porosity is seldom of any serious consequence. The cathode is the very small area of substrate exposed at the base of the pore, and the restricted channel limits the diffusion of reactants and products. The large anode area provided by the coating reduces the bimetallic corrosion current density thereon. Two important examples of this type are zinc coatings on steel in cold waters or the atmosphere and tin coatings on steel on the inside (but not the outside) of a sealed, air-free can of wet food. In the first case oxygen is the cathodic reactant; in the second it is hydrogen ions (or water). Where the coating is cathodic, porosity enables the exposed substrate to corrode. In most cases this is detrimental; the exception is found in some multi-layer nickel plus chromium coatings where certain forms of porosity in the chromium layer are harnessed to divert the direction of corrosion to the overall benefit of coating life. In other cases corrosion at pores causes trouble. In wet atmospheric corrosion, substrate corrosion product, if coloured and insoluble, spoils decorative appearance. In immersed conditions or condensing atmospheres,
12:42
ELECTROPLATlNG
if the corrosion product is soluble intense pore corrosion will perforate sheet metals. Here a porous coating may accelerate corrosion when compared with the uncoated substrate. Porosity causes little trouble when corrosion is restricted to dry processes (oxidation). Corrosion products block the pores and stifle the reaction. There was much research into the causes of porosity in nickel deposits when it was thought to be the main cause of failure in nickel and chromium plate. Much was discounted as it became clear that nickel pitting at discontinuities in the chromium was the factor determining service life. Porosity remains relevant to the corrosion resistance of simpler cathodic coatings, and especially for gold. The use of gold for contact surfacing since about 1950 has revived the importance of studies of porosity. Pores in gold coatings allow films of substrate corrosion product to contaminate the surface and to destroy the low contact resistance of the gold. Sulphides, which are one of the products of corrosion by service atmospheres, have a particularly high rate of spreading over gold in the solid state (Fig. 12.14). Pores originate on substrate areas known as precursors, which are of at least three types. Firstly, an obvious cause is an inclusion of foreign material which is a semiconductor or insulator - particles of oxide, sulphide, slag, polishing abrasive, etc. When electrodeposition starts, inclusions will not be nucleation sites, and they will impede the lateral growth and coalescence of crystals from neighbouring nuclei. Secondly, substrates whose surface grain structure has been severely disturbed by cold working (abrasion, cold rolling, drawing, etc.) have precursors whose physical state (rather than chemical difference as in the first type) precludes coalescence of the electrodeposit. This is probably an effect of pseudomorphic growth. Relatively low-temperature annealing (as low as 210°C for steel) greatly reduces the effect, and further cold work increases it again (Fig. 12.15). The third type of precursor
Fig. 12.14 Spread of silver sulphide from discontinuities in gold electrodeposits on silver substrates. The gold was deliberately scratched and the specimen exposed for 24h to an atmosphere containing 10% SO2. Immediately after this the sulphide stain extended 0 . 2 mm. Five years later, the stain extends to about 13 mm, after storage in a normal indoors atmosphere
ELECTROPLATING
.. \t. .
Y.
* I
12 :43
. :
Fig. 12.15 Porosity caused by a cold-worked substrate. Left (EQE 76) cold-rolled steel as received; centre (EDE 52) steel bright-annealed in vacuum before plating, 2.5 h at 700'C; right, annealed steel, further cold-rolled (0.914 mm to 0.864 mm) produces porosity again. No steel was removed from the surface; 5 pm tin-nickel electrodeposit
is a crevice in the substrate. If the depth is great relative to the width, the electric field is excluded and deposition does not occur within the crevice. Lateral growth is impeded once the edges from neighbouring nuclei reach it in much the same way as with a non-conducting inclusion. A pore caused by any type of precursor in one electrodeposit becomes in turn a precursor for a second deposit plated over it. There may be other forms of precursor. In a particular area of substrate there will be a number of precursors, distributed over a range of sizes, and reflecting the nature, composition and
Fig. 12.16 Increase in porosity of an electrodeposit caused by mechanical polishing. Left, 7.5 pm unpolished coating; right, polished with lime finishing compound. The average thickness removed by abrasian was 0.1 pm
12:44
ELECTROPLATING
history of the metal. In principle anything affecting the substrate surface will affect porosity in an electroplated coating. As deposition continues growth gradually diminishes the surface opening of a pore, and if continued to a sufficient thickness, closes it, leaving a sealed cavity filled with solution. Small precursors will generate pores which seal relatively early, large ones will require greater thicknesses. The total pore density revealed by a test which renders pore sites visible falls as thickness increases. The minimum thickness required to seal a precursor of fixed size will depend on the rate of narrowing of the surface opening, and, as a growth process this will reflect the plating conditions. Because of this, the density of pores still open at a fixed thickness is a function of all the plating conditions, i.e. of the composition of the plating bath, of temperature, current density, agitation and anything affecting deposit growth. Post-plating treatments affect pore density, either by closing pores which are still open or opening sealed ones. It has been asserted that mechanical polishing in general, or flow-melting for tin, are both processes which could seal pores and reduce porosity. It is also conceivable that polishing might cut-open sealed pores, and likewise under flow-melting conditions the vaporisation of solution trapped in sealed pores could disrupt the coating and recreate discontinuities. The author has come across no convincing demonstration of porosity reduction by either treatment, but has found experimental evidence for porosity increases (Fig. 12.16).
Recent Developments Although the basic principles of electroplating remain unchanged, the extent of development and variety of application have widened substantiallyl'. In this section some notable developments will be cited. The development of new solutions and processes continues unabated, driven as ever by commercial and proprietorial needs as well as pressure from pollution and effluent control demands and simply for the need to supersede some less-than-satisfactory solutions. Non-cyanide solutions are continually being sought for metals such as gold, copper, cadmium and zinc, but cyanide remains pre-eminent as the most effective and best understood complexant available and few competitors have been discovered. The other ecological Mfe-noireis hexavalent chromium, and several commercial bodies offer non-toxic trivalent chromium plating solutions, both aqueous and organic based although only the former is believed to be industrially viable. The solutions are based upon chromic sulphate or chloride salts, a complexant such as hypophosphite, glycollate, thiocyanate etc. and a depolarising anode reactant which could include ammonium ions or a separated anode compartment. The cathode efficiency is still below 50% and only thin coatings can be reproducibly produced (10 pm max.), but pollution difficulties are largely eliminated. This has proved to be a difficult area, but the number of successes is expected to i n c r e a ~ e ' ~ ' - ' ~ . A separate problem is the establishment of a good process for electroplating aluminium which must necessarily be based upon a non-aqueous electrolyte. This field is a history of many discoveries, but few developed processes have been claimed, although recent work suggests that at least
ELECTROPLATING
12:45
two good possibilities exist which may make inroads in the electronics field rather than in the other important area of wide steel strip a l ~ m i n i s i n g ' ~ ~ - ' ~ . In his classic treatise, Brenner' reported that over 500 alloy electrodeposition systems had then been studied in depth-that number has now been substantially increased-yet barely 10 to 20 have any real degree of industrial exploitation. The list continues to grow and the present type of work on alloys can be divided into three classes: 1. The development of new alloys in new fields; for example the development of molybdenum and tungsten with iron, cobalt or nickel for coating of dies and nozzles, or the development of palladium-nickel alloy as an alternative to gold for connectors. 2. The development of new alloys as a means of modifying existing electrodeposits; for example the production of hard gold by alloy codeposition of copper, cadmium etc. to yield 23 or 18 carat alloys, or the use of zinc alloys for improved electrogalvanised coatings. 3. The development of new solutions for established alloys; for example the replacement of fluoborate for lead-zinc brasses.
With industry proving to be so conservative about binary alloys it is hardly surprising that ternary alloys receive little attention. Nevertheless, two ternary alloys at least have become commercially available: ironchromium-nickel (so-called stainless steel) for both functional and domestic markets and an electronic connector and solderable alloy based on copperzinc-tin. The field of composite materials has been the major growth area of materials engineering in the last twenty years, based mainly on ceramic and polymer materials. While electroplated (and electroless) composites show more modest growth this is attributable to the necessary limitation of metal matrices. Thus the principle is to take a well-established metal deposition process (gold, cobalt, copper, nickel, tin) and to induce co-deposition of second-phase particles, thereby enhancing coating properties such as hardness, wear and oxidation resistance. The key to successful co-deposition is having particles of appropriate size and density, typically 0.1-10 pm size, suitably suspended in solution by a non-swirling agitation technique, codeposition occurring by physical entrapment or electrophoretic attraction. Such particles include oxides (e.g. A1,03, TiOz, ZrO, etc.) or refractory hard compounds (e.g. Cr,C,, WC, M0,C etc.), abrasives such as diamond, lubricants such as MoS, or graphite and low-friction material such as p.t.f.e. A substantial literature exists, relating to both process and product characteristics and reference should be made to two notable review^'^'.'^^. Several obvious applications have to a large extent been achieved; e.g. second-phase hardening by A120, of gold without serious loss of electrical conductivity, high-temperature erosion or wear resistance of nickel or nickel-cobalt gas turbine or jet engine alloys improved by using carbide incorporation, and improved surface lubricating of nickel by incorporation of p.t.f.e. particles. The use of current or voltage pulsing during electroplating has long been known to have a beneficial effect on the deposition process rate and on the deposit itself in terms of grain size variation, internal stress, levelling etc. Periodic reverse techniques (cycle time of IO-lOZ s) are widely employed in electrowinningand electrorefining operations while pulse plating (cycle time
12:46
ELECTROPLATING
of to 1 s), which requires more sophisticated electronics, is now of considerable interest for metal finishing. The basic theory has been discussed by Ibl'49who has defined the parameters involved. Claims for improved brightening, levelling and throwing power are of especial interest in electronics, but are not yet fully substantiated in many instances"'. The cooperation of industrial and engineering designers with the metal finishers, who are frequently required to perform the near-impossible as a consequence of poor communication, is notoriously bad largely as a consequence of the nature of sub-contract industrial relationships. To meet this need an important new standard -BS 4479- has been issued; although it is ostensibly a revision of the old standard it is in reality a new standard written essentially as a code of practice. Invaluable advice is given to finishers and designers alike: the challenge now is to have it widely read and appreciated! Increasing awareness of the cost-effectivenessof electroplating processes has led to critical appraisals being made of cell design, not only to improve the product through improved efficiency and economics of the process itself, typically through the costs of electricity. Thus the use of more conductive solutions, combined with minimisation of the anode-cathode spacing can yield a 40% saving in electrical power. However, not all of this saving is necessarily desirable if chemical costs thereby increase and the peripheral cost of solution heating has also to be increased. Similarly, improved agitation and filtration may also be considered for optimisation studies. This 'chemical engineering' approach has found increasing v a l ~ e ' ~ ~not - ' ~least ~, in the development of new types of plating cell specifically for metal recovery from trade effluent, dragouts and In fact the number of new designs far outnumbers the number of optimising and independent assessment studies so that it is not possible to name a 'best-buy', and time is needed for commercial realities to eventually declare a winner, albeit not on entirely objective terms. The largest-scale electroplating activities have always been carried out by the steel industry in an atmosphere largely divorced from traditional metal finishing. Upwards of 20% of all steel produced may be coated, the products of relevance to this chapter being tinplate, its alternative for packaging 'tinfree steel' and zinc electrogalvanisedsteel in the form of sheet, strip and wire. During the last twenty years little advance has been made in the electroplating stage of tinplate production, the electrolytes and additives have changed little and the plant design remains essentially the same -marked changes have occurred in other aspects of tinplate production, however. The alternative 'tin-free steel' or TFS, has settled into a well-established sector of the market, largely for lacquered beer and beverage cans and non-critical container applications such as oil, polish, some paint etc. Its invention is attributed to Japan in the period 1958-1965 and it has been widely exploited. The technology is based on that of tinplate as a fast cathodic process (1 -20 s) in a chromic acid-based solutions yielding a coating (70% the presence of moisture encourages the reactions SOz --t SO, + sulphuric acid and oxide film breakdown occurs. A new film of sulphate plus hydrated oxide eventually stifles corrosion reactions in the industrial atmosphere. Longer term tests in a marine atmosphere (15 years) can cause perforation of the coated sheet from the ground-facing side. This is explained in terms of longer periods of wetness and the greater propensity to retain salt on that side. A more constant corrosion rate results from the presence of a high-conductivity film with sea-salt incorporated. Aluminised steel produced by hot dipping is used in the construction of parts of many exhaust systems of road vehicles. Failure of some of these exhausts does take place well within the expected two-year average life. This arises in the rear end of the exhaust where dew point corrosion occurs on the inside of the system. ‘Acid dew’ of pH 2.7-3.1 is produced in the exhaust gases at temperatures below 48°C and this concentrates as the system eventually heats up towards 1 0 0 O C . The aluminised coating is attacked at weak positions, e.g. where holes have been punched and the aluminium does not completely coat the steel. Eventually, the aluminium coating is undermined and the steel severely attacked. It is estimated that the use of aluminium coatings can increase the life of unprotected steel by at least 12%. Aluminium does provide a protective barrier to corrosive attack on steel but its ability to provide sacrificial protection is limited. The use of Al-Zn
ALUMINIUM COATINGS
13 :33
hot-dipped coatings does result in greater sacrificial action. It is claimed that such action is gained without too much loss on the general corrosive attack on aluminium. This has promoted the use of AI-Zn coatings on steel roofing panels, and exposure tests to date point towards very good service in this field. With the recently discovered Al-Zn-In-Sn alloys, sacrificial protection is dependent on the indium content. S. J. HARRIS E. W. SKERREY BIBLIOGRAPHY General Bailey, J. C., Porter, F. C. and Round, M.. ‘Metal Spraying of Zinc and Aluminium in the United Kingdom’, in 12th Int. Conf. on Thermal Spraying, Welding Institute. London, paper 8, pp. 1-8 (1989) Barton, H., ‘Ivadizing: Ion Vapour Deposition of Aluminium’, in Ion Assisted Surface Treatments, Techniques and Processes, The Metal Society, London, pp. 1-6 (1982) Blickwede, D. J., ‘New Sheet Steel Product with 55% Al-Zn Alloy Coating’, ibid., pp. 44-53 Boden, P. J. and Harris, S . J., ‘The Strategic Replacement of Mild Steel in Car Exhaust Systems’, in Dewpoint Corrosion, edited by D. R. Holmes, Ellis Horwood Ltd., Chichester, pp. 256-275 (1985) Burns, R. M. andBradley, W. W., ProtectiveCoatings forMetals, Reinhold, New York (1955) James, D. H., ‘Thermal Spraying by the Electric Arc Process’, Metallurgist and Materials Technologist, 15, 85-90 (1983) Jones, R. D. and Thomas R. J ., ‘Production of Hot-Dip Alluminised Steel Strip’, in Production and Use of Coil-Coated Strip, The Metals Society, London, pp. 55-63 (1981) Legault, R. A. and Pearson, V. P., ‘The Kinetics of Atmospheric Corrosion of Aluminised Steel’, Corrosion, 34, 344-349 (1978) Lowenheim. F. A., Modern Electroplating, Wiley/American Electrochemical Society, 2nd edn. (1%3) Restall, J. E., in Development of Gas Turbine Materials, edited by G . W. Meetham, Applied Science Publishers, London, p. 280 (1981) Suchentrunk, R., ‘Corrosion Protection by Electrodeposited Aluminium’. Z. Werkstoftechnik, 12, 190-206 (1981) Wernick, S . and Pinner, R., Surface Treatment of Aluminium, Robert Draper, London, 4th edn. (1972) Spraying Andrews, D. R., ‘The Protection of Iron and Steel by Aluminium Coatings’, Metallurgia, Manchr., 62, 153 (1960) Ballard, W. E., Metal Spraying and the Flame Deposition of Ceramics and Plastics, Griffin, London, 4th edn. (1963) Birley, S. S., Hepples, W. and Holroyd, N. J. H., ‘The Corrosion Protection of Weldable 7xxx Aluminium Alloys by Aluminium-based Arc Spray Coatings’, in 3rd Int. Conf. on Aluminium Alloys Vol 11. Sintef, Trondheim, 497-502 (1992) Carter, V. E. and Campbell. H. S.. ‘Protecting Strong Aluminium Alloys Against Stress Corror sion with Sprayed Metal’. British Corrosion Journd, 4, No. 1, 15-20, Jan. (1%9) and 4 NO. 4, 1%-198, July (1%9) Franklin, J. R., ‘Metallized Coatings for Heat Corrosion Protection’, Corrosion Technol., 2, 326 (1956) Harris, S. J., Green, P. D. and Cobb, R. C., Thermally Sprayed AI-Zn-In-Sn Alloys in 3rd Int. Conf. on Advances in Surface Engineering for Corrosion and Wear Resistance, Newcastle-upon-Tyne, 1-10 (1992) Hoar. T. P. and Radovici, 0..‘Zinc-Aluminium Sprayed - . Coatings’, Truns. Inst. Met. Fin., 42, 21 1-222 (1964) Hudson, J. C., Sixth Report of the Corrosion Committee of BISRA, Special Report NO. 66, I.S.I. (1959) Mansford, R. E., ‘Sprayed and Diffused Metal Coatings’, Metal Ind., London, 93, 413 (1958)
13:34
ALUMINIUM COATINGS
Mansford, R. E.,‘Sprayed Metal Coatings in the Gas Industry’, Chem. and Znd. (Rev.), 150 (1961) Porter, F.C., ‘Aluminium Coatings on Iron and Steel’, Metal Finishing Journal, 9 No. 104. 303-312, AUg. (1963) Porter, F. C., ‘Aluminium Sprayed Steel in Marine Conditions’, Engineer, Lond., 211, 906 (1961) Reininga, H., ‘Further Developmentsin Metal Spraying Technique’. Metallobe@che, 15,52. 88, 118, 148 (1961). (Translation available as TM460, Aluminium Federation, Brimingham) Scott, D. J., ‘Aluminium Sprayed Coatings: Their Use for the Protection of Aluminium Alloys and Steel’, Trans. Znst. Met. Fin., 49 No. 3 , 111-122 and 49 No. 4, 173-175 (1971) Sprowl. J. D., Aluminized Steel-A Description. Report of the Department of Metallurgical Research, Kaiser Aluminium and Chemical Corporation, April (1958) Stanners, J. F. and Watkins, K. 0.. ‘Painting of Metal Sprayed Structural Steelwork’, British Corrosion Journal, 4 No. 1, 7-14, Jan (1969)
Aluminizing, DTD 907B Defence Guide DG-8, Part 2, H.M.S.O., April (1971) How to Prevent Rusting, BISRA (1963) Metallizing: Aluminium and Zinc Spraying. DTD 906B Metho& of Protection Against Corrosion for Light Gauge Steel Used in Building. PD 420 (1953) Painting of Metal Sprayed Structural Steel, BISRA Corrosion Advice Bureau (1966) Proceedings of the Second International Metal Sprayers’ Corlference, Birmingham, Association of Metal Sprayers(1958). (Specificreferencesto aluminium notes in TM 420. Aluminium Federation, London) Protection by Sprayed Metal Coatings, The Welding Institute, London (1968) ‘Protection of Iron and Steel against Corrosion and Oxidation at Elevated Temperatures’, Sprayed Metul Coatings. BS 2569: Part 2 (1965) ‘Protection of Iron and Steel by Aluminium and Zinc against AtmosphericCorrosion’, Sprayed Metal Coatings, BS 2569: Part 1 (1964) The Protection of Steel by Metal Coatings, No. 5 of Series by Corrosion Advice Bureau of BISRA (1976) The Protection of Iron and Steel Structuresfrom Corrosion, CP 2008 (1966) Hot-Dip Aluminising Coburn, K. G., ‘Aluminized Steel. Its Properties and Uses’, Metallurgia, Manchr., 60, 17 (1959) Drewett, R., ‘Diffusion Coatings for the Protection of Iron and Steel’, Part I, Anti-Corrosion, 16 No. 4, 11-16, April (1969) Edwards, J. A., ‘Coated Engine Valves’ Auto. Engr., 45, 441 (1955) Gittings, D. 0..Rowland, D. H. and Mack, J. 0.. ‘Effect of Bath Composition on Aluminium Coatings on Steel’. Trans. Amer. SOC.Metals, 43, 587 (1951) Hughes, M. L., ‘Hot Dipped Aluminized Steel: Its Preparation, Properties and Uses’, Sheet Metal Industry, 33, 87 (1956) Schmitt, R. J. and Rigo, J. H., ‘Corrosion and Heat Resistance of Aluminium-coated Steel’, Materials Protection, 5, No. 4. 46-52, April (1966) Serra, M., ‘ConsiderationsArising from Experiments on a New Process of Metallic Protection’, Rev. Cienc, Appl., 12,222 (1958). (Translation available as TM398, Aluminium Federation, Birmingham) Whitfield, M. G.,‘Rolling of Hot Dipped Aluminized Steels to Make Them More Durable’, Anti-Corrosive Matenals and Processes, 2, 31, Oct. (1963) and US Pat. 2 170 361 Anon., ‘Largest Aluminium Line in UK’, Product Finishing, 21 No. 7, 61-65, July (1968) Anon.,‘Welded Aluminized Steel Sheet’, Anti-Corrosive Materials and Processes, No. 2 , 4-7, Oct. (1963) ‘A.S.T.M. Field Tests and Inspection of Hardware’, Proc. Amer. Soc. Test. Mater., 52, 118 (1952) Calorising (cementation) Drewett, R.. ‘Diffusion Coatings for the Protection of Iron and Steel., Part I. Anti-Corrosion, 16 No. 4, 11-16, April (1969)
ALUMINIUM COATINGS
13:35
Porter, F. C., ‘Aluminium Coatings on Iron and Steel’, Metal’Finishing Journal, 9 No. 104, 303-312, Aug. (1963)
Vacuum Deposition Holland, L., Vacuum Deposition of Thin Films, Chapman and Hall, London (1956) Porter, F. C., ‘Aluminium Coatings on Iron and Steel’, Metal Finishing Journal, 9 No. 104, 303-312, AUg. (1%3) Remond, 0.B. and Johnson, A. R., ‘Vacuum Deposition of Metals’, MetalFinishing JournaI, 4, 393 (1958)
Weil, F. C., ‘Recent Developments in Vacuum Metal Coatings’, Electroplating Metal Finish, 9, 6, 25 (1956)
Anon., Vacuum Coating of Formed Metal Parts, National Research Corporation, Cambridge, Massachusetts (1959)
Electrodeposition (electroplating) Couch, D. E. and Brenner, A., ‘Hydride Bath for the Electrodeposition of Aluminium’, J. Electrochem. Soc., 99, 234 (1952) Couch, D. E. and Connor, J. H., ‘Nickel-Aluminium Alloy Coatings Produced by Electrodeposition and Diffusion’, J. Electrochem. Soc., 107, 272 (1960) Honand, B. O., ‘Aluminium Coating of Steel with Special Reference to Electrodeposition’, Broken Hill Pty, Technical Bulletin, 3 No. 1, 29 (1959) Lowenheim, F. A., Modem Electroplating. Wiley/American Electrochemical Society, 2nd edn. (1%3) Menzies, I. A. and Salt, D. B., ‘The Electrdeposition of Aluminium’, Trans. Inst. Met. Fin., 43, 186-191 (1965)
Safranek. E. H., Schiekner, W. C. and Faust, C. L., ‘Electroplating Aluminium Wave Guides Using Organo-Aluminium Plating Baths’, J. Electrochem. SOC., 99, 53 (1952) Utz, J. J. and Kritzer, J., ‘New High Purity Aluminium Coatings’, Muter. Design Engn., 49, 88 (1959)
Electrophoretic Coatings and Other Compacted Coatings Bright, A. W. and Coffee, R. A.. ‘Electrostatic Powder Coatings’, Trans. Inst. Met. Fin. 41, 69-73 (1964)
Brown, D. R. and Jackson, A. E.,‘The Elphal Strip-Aluminizing Process’, Sheet Metal Ind., 39. 249 (1%2)
Sugano, G . , Mari, K. and Inoue, K., ‘A New Aluminium Coating Process for Steel’, Electrochemical Technology. 6 Nos. 9 and 10. 326-329, Sept.-Oct. (1%8) Process for Forming Sintered Metal Coatings. Texas Instruments Inc., UK Pat. 1 163 766 (10.9.69)
Chemical Deposition. Gar or Vapour Plating Dow Chemical Co., ‘Catalytic Plating of Aluminium’, Financial Times, Oct. 21 (1%9) Hiler, M. J. and Jenkins, W.C., Development of a Method to Accomplish Aluminium Deposition by Gas Plating, US Air Force WADC Tech. Report, 59-88, June (1959) and US Pat. 2 929 739 Powell, C. F.. Campbell, I. E. and Gonser. B. W., Vapour-Plating, Chapman and Hall, London (1955) Anon, ‘Aluminium Plating Via Alkyd Gas’, The Iron Age, 52-53. Dec. 23 (1965) Method of Aluminium Plating with Diethylaluminium Hydride, Continental Oil Co., UK Pat. 1 178 954 (28.1.70)
Mechanical Bonding BIOS Final Reports. Nos. 1467 and 1567; H.M.S.O., London (1974) Bonded Aluminium Steel Composites and Methods of Making Same, Du Pont de Nernours amd Co., UK Pat. 1 248 794 (6.10.71) Casting Little, M. V., ‘Bonding Aluminium To Ferrous Alloys’, Machinery, N. Y.,56, 173 (1950) Drewett, R., ‘Diffusion Coatings for the Protection of Iron and Steel‘, Part I, Anti-Corrosion, 16 No. 4, 11-16, April (1969)
13.3 Cadmium Coatings
In some environments cadmium gives better protection than zinc (Section 13.4); it is, however, considerably more costly. It does not compete with zinc on articles on which a high degree of protection can be achieved by the use of a thick film deposited by hot dipping (immersion in molten zinc, i.e. galvanising) or metal spraying. Where only thin coatings of 25 pm or less are tolerable, the greater protection of cadmium in some environments is worth while, and as uniform thin coatings must be deposited by relatively expensive processes such as electroplating, the greater cost of cadmium then has little effect on the cost of the finished article. However, because of the toxic nature of both the metal and its compounds, the use of cadmium is generally limited to those applications that demand the unique combination of properties that cadmium possesses.
Coatings of both cadmium and zinc protect steel mainly by simple physical exclusion. At gaps in the coating, however, whether these are in the form of porosity, pits, scratches or cut edges, protection is by sacrificial action of the coating followed probably by the plugging of gaps with sparingly soluble corrosion product. It is not at once clear why cadmium should be sacrificially protective to steel. Standard equilibrium electrode potentials of iron and cadmium in contact with solutions of their salts of normal activity, given in Table 13.3, suggest that iron should be sacrificial to cadmium, but Hoar' has shown by means of E/I curves that the mixed potential of corroding cadmium will be more electronegative than the mixed potential of corroding iron. This follows from the higher exchange current for Cd Cdz' + 2e. Under these circumstances iron will be sacrificially protected by cadmium (see also Section 1.4). Whatever the explanation, the fact of sacrificial protection is reflected in the potentials, also given in Table 13.3, found for the two metals in sea-water. The degree of protection given in practice by zinc and cadmium, whether by physical exclusion or by sacrificial action at gaps, depends on the durability of the coatings themselves against corrosive attack. It is now well established that, thickness for thickness, cadmium is more resistant to
+.
13:36
13:37
CADMIUM COATINGS
Table 13.3 Potentials of iron, cadmium and zinc Metal Iron
Cadmium Zinc
Standard electrode potential, hydrogen scale
frowing sea-wafer, hydrogen scale
Corrosion potential' in
(V)
(VI
4-44 -0.40
-0-36 -0.45 -0.76
-0.76
The corrosion potential will vary with aeration and velocity of the sea-water.
Table 13.4 Corrosion rates of zinc and cadmium coatings in various atmospheres ~
Location
Industrial Suburban Marine
~~~~
Rate of corrosion of electrodeposited coating bmh)
Zinc
Cadmium
5.1 1.8 2.5
10.2
2-3 1-3
marine and tropical atmospheres and zinc more resistant to industrial atmospheres. This is well demonstrated by comparative tests made by Biestek' in various laboratory conditions, and by Clarke and Longhurst' in practical tropical exposure tests. Table 13.4 gives the order of corrosion rates, based on the results of these and other4 tests. It must be emphasised that these figures give only a broad comparison; actual corrosion rates will be much affected by the exact environment. If the corrosion mechanism in an industrial atmosphere is mainly a straight chemical dissolution in sulphur acids, then the relative chemical equivalents present in a given thickness of the two metals account for a large part of the difference in corrosion rate. In an unpolluted humid atmosphere the slightly greater corrosion resistance of cadmium compared with zinc at unit thickness (and therefore much greater resistance per unit chemical equivalent) is likely to be due to a greater insolubility and protective power of the first corrosion product. The solubility data in Table 13.5 (quoted from the Handbook of Chembtry and Physics, 40th edition) show that cadmium hydroxide is more soluble in water than zinc hydroxide, but that the cadmium carbonate is the less soluble, it is concluded therefore that the protective films formed are carbonates or possibly basic carbonates. The considerably greater comparative corrosion resistance of cadmium in a marine atmosphere must be postulated as being due to a greater insolubility of the basic chloride of cadmium compared with that of zinc. In conclusion, relative cost and relative behaviour towards different conditions of exposure lead to the use of zinc on parts on which thick films can be tolerated and for general industrial use, and of cadmium for fine-tolerance special applications, such as aircraft and instrument parts, required to withstand conditions include humid and marine atmospheres.
13:38
CADMIUM COATINGS Table 13.5 Solubility in water of cadmium
and zinc carbonates and hydroxides Solubility
(g/lOO ml)
Metal
Cadmium Zinc
Carbonate
Hydroxide
insoluble
0-OOO26 0-OOOOOO 26
0.001
Other Factors Governing the Choice between Zinc and Cadmium As well as the reasons already given, other considerations influence the
choice between zinc and cadmium. Cadmium is easier to solder and has a lower contact resistance than zinc, and for such reasons it may be selected for certain applications. However, account must be taken of the toxic nature of cadmium and cadmium vapour. On very strong steels cadmium is also preferred because it appears that cadmium electroplating from a given type of electroplating solution, e.g. a specially formulated cyanide solution, causes less hydrogen embrittlement than zinc plating from the same type of solution5. On the other hand, on steels subject to elevated temperatures in use, the possibility of intergranular penetration of stressed steel which occurs above (and even, if the steel is highly stressed, somewhat below) the melting point of the coating, may iead to the choice of zinc (m.p. 419-5°C) in preference to cadmium (m.p. 321"C). Acid vapours emitted by wood, oleoresinous paints and some plastics (cf. Section 18.8-10) attack both zinc and cadmium. The relative behaviour varies, and appears to depend on the nature and concentration of the acid vapours and on the relative humidity. For these conditions of exposure, therefore, no advice can be given as to which metal should be used. It should not be assumed, therefore, that because one metal has failed therefore the other would be better. Both metals are applied to copper-base alloys, stainless steels and titanium to stop bimetallic corrosion at contacts between these metals and aluminium and magnesium alloys, and their application to non-stainless steel can serve this purpose as well as protecting the steel. In spite of their different potentials, zinc and cadmium appear to be equally effective for this purpose', even for contacts with magnesium alloys'. Choice between the two metals will therefore be made on the other grounds previously discussed. Protection of Cadmium Coatings
Full chromate passivation (Section 15.3) improves the corrosion resistance of both zinc and cadmium towards all environments and is applied for a wide
range of applications. Clear and olive-coloured chromate coatings can also be applied for certain purposes. The highest degrees of corrosion protection
13:39
CADMIUM COATINGS
are generally obtained from olive-coloured chromate coatings. Passivation improves the adhesion of normal types of priming paints, but for best adhesion and protection an etch primer should be used. Methods of Deposition
Electroplating Cadmium is usually electroplated from a cyanide solution. Zinc is also deposited from cyanide electrolyte, but for some applications mildly acidic and alkaline non-cyanide electrolytes are increasingly being used. Typical cyanide-based electrolyte formulations for both metals taken from Specifications DTD 903 and 904 are given in Table 13.6. Table 13.6 Typical plating solutions for zinc and cadmium Constituents Zn or Cd Total cyanide Caustic (as metal) (as NaCN) soda (min-max) (min-max) (min-max) (s/l) (s4
Temperature of operation (win-max) I" C)
Current density (A /dm ')
25-50
56-112
40-80
32 (optimum)
1-2 (vat) 0.3-0.7 (barrel)
Cadmium plating Vat 14-17 Barrel 23-27
56-63 78-84
11-14 17-20
15-35 15-35
1.0 1 .o
Solution
Zinc plating.
For barrel plating, solution concentrationstowards the maximum are recommended It is also important to maintain the constituents in the ratios recommended in the specifications
Other solutions, some based on cyanide, some on sulphate, fluoborate, etc. will be found in textbooks and handbooks of electroplating, and a comprehensive review of methods and of their relative advantages has been published by Such'. Much work has been devoted to the development of cadmium plating processes which cause little or no hydrogen embrittlement to very strong steels; references are given in Section 8.4 and in a paper by Yaniv and Shreir'. However, hydrogen removal can be effected by baking the steel at 200°C after plating.
Vapour deposition Hydrogen embrittlement can be avoided by depositing cadmium by vacuum evaporation. Vapour plating is carried out in a N/m2. Cadmium metal is placed in chamber evacuated to below 2 . 7 x mild-steel boats arranged along the chamber and heated to about 200°C. The evaporating metal moves in straight lines, so the parts to be coated are held in jigs that rotate on their own axes and revolve round the chamber, thus presenting all surfaces to the moving vapour. Before evaporation is begun, the parts must be cleaned by ion bombardment in a high-tension (- 1 kV) glow discharge at a pressure of approx. 4N/m2. Formerly, the glow discharge was stopped before the chamber was pumped down to evaporation pressure, but adhesion of the coating was poor unless the parts had first been roughened by fine abrasive blasting. In an improved process" the glow discharge is maintained concurrently with pumping down and the start of evaporation; under these conditions there is no interval during which oxide
13 :40
CADMIUM COATINGS
reforms on the steel by reaction with residual oxygen in the chamber, cadmium atoms arrive on a surface still under bombardment, and adhesion of the coating is good even on smooth machined surfaces. Specifications for Cadmium Plating
Cadmium plating for general engineering use is covered by BS 1706: 1960 and BS 3382: 1961, and for aircraft parts by Ministry of Aviation Supply Specification DTD 904. Special requirements for very strong steels are given in Defence Standard 03-4 (Directorate of Standardisation, Ministry of Defence). Health and Safety Cadmium metal and its compounds are toxic and are injurious to health, and for this reason, cadmium is being replaced by other forms of coating wherever possible. For a number of important applications, however, no suitable alternatives have yet been identified. Where cadmium plating continues to be used, it is essential to comply with the regulations covering the use of cadmium. H.G. COLE M. ROPER
REFERENCES 1. Hoar, T. P., J. Electrodep. Tech. Soc., 14, 33 (1938) 2. Biestek, T., Proceedings of the First International Congress on Metallic Corrosion, London, 1961, Butterworths, London, 269 (1962)
3. Clarke, S. G. and Longhurst, E. E., Proceedings of the First International Congress on Metallic Corrosion, London, 1961, Butterworths, London, 254 (1962) 4. Uhlig, H. H. (Ed.), The Corrosion Handbook, Wiley, New York; Chapman and Hall, London, 803 and 837 (1948) 5. Maroz, I. I. et al., see Domnikov, L., Metal Finish., 59 No. 9, 52 (1%1) 6. Evans, U. R. and Rance, V. E., Corrosion and its Prevention at Bimetallic Contacts, H.M.S.O., London (1958) 7. Higgins, W. F., Corrosion Technol., 6 , 313 (1959) 8. Such, T. E., Electroplating Met. Finish., 14, 115 (1961) 9. Yaniv, A. E. and Shreir, L. L. Trans. Inst. Met. Finish., 45, I (1967) 10. U K Pat. 1 109 316 and specification DTD 940
13.4 Zinc Coatings It is estimated that approximately 40% of the world production of zinc is consumed in hot-dip galvanising of iron and steel, and this adequately demonstrates the world-wide use of zinc as a protective coating. The success of zinc can be largely attributed to ease of application, low cost and high corrosion resistance. Metbods of Application
The principal method for applying zinc coatings to iron and steel is hot-dip galvanising. There are four other important methods, each of which has its own particular applications; these methods are spraying, plating, sherardising and painting with zinc-rich paints. The choice of method in any given case is determined by the application envisaged, and the five processes may be said to be complementary rather than competitive, for there is usually little doubt as to which is the best for any particular purpose. Processes of applying coatings by various methods are discussed in detail elsewhere ' and are considered in Chapter 12, and will not, therefore be considered here. The reactions inherent in galvanising tend to ensure a thick and even coating but guides to the inspection of galvanising, sherardising and zinc spraying are available
'.
Char8CteriStiCS of Zinc Coatings
In practice the thickest zinc coatings can be obtained by hot-dip galvanising or spraying. Table 13.7 compares the essential aspects of each coating. Plated coatings can also be produced mechanically in a wide range of thicknesses as well as electrochemically. The thickness of hot-dip galvanised coatings depends on the nature of the steel and the dipping conditions. It can be controlled to a certain extent in practice. Heavier coatings are obtained on grit-blasted steel or on steels of high silicon content, and at higher operating temperatures and longer dipping times. In strip galvanising, aluminium is deliberately added to the bath to suppress the action between molten zinc and steel, with the result that lighter coatings are produced compared with those on fabricated assemblies galvanised after manufacture. Mechanical wiping of the surface on withdrawal from the bath, as employed in wire or sheet galvanising, also causes a reduction of coating thickness. 13:41
Table 13.7
Characteristics of coating
Hotdip galvanking
1. Process considerations Parts up t o 20 m long
and fabrications of 18 m x 2 m x 5 m can be treated. Care required at design stage for best results. Continuous galvanised wire and strip up to 1.4 m wide) in UK 2. Economics
Generally the most economic method of applying metallic zinc coatings 20-200 prn thick
Comparison of zinc coatings
Metal spraying
PIating
Sherardising
Zinc dust painting
No size or shape limitations. Access difficulties may limit its application. e.g. inside of tubes. Best method for applying very thick coatings. Little heating of the steel
Size of bath available. Process normally used for simple. fairly small components suitable for barrel plating or for continuous sheet and wire. No heating involved
Batch processing is mainly suitable for fairly small complex components. Semicontinuous process for rods, etc.
Can be brush, spray or dip applied on site when necessary. No heating involved. Performance varies with media used
Most economic for work with high weight to area ratio. Uneconomic on open mesh
Used, where a very thin zinc coating is sufficient. Thick coatings are expensive.
More expensive than galvanising for equivalent thicknesses. Generally used when control of tolerances is more important than thickness of coating
Low overheads but high labour element in total cost as with all paints. Thixotropic coatings reduce number of coats and hence labour costs
3. Adhesion
Process produces iron-zinc alloy layers, overcoated with zinc; thus coating integral with steel
Good mechanical inter- Good. comparable with locking provided the other electroplated abrasive grit-blasting coatings pretreatment is done correctly
Good-the diffused iron-zinc alloy coating provides a chemical bond
Good-abrasive grit blasting preparation of the steel gives best results
4. Thickness and
Normally about 75-125 pm on products, 25 pm on sheet. Coatings up to 250 pm on products by prior grit-blasting. Very uniform-any discontinuities due to poor preparation of the steel are readily visible as ‘black spots’
Thickness variable at will, generally 100-200 pm but coatings u p to 250 pm or more can be applied. Uniformity depends on operator skill. Coatings are porous but pores soon fill with zinc corrosion products; thereafter impermeable
Thickness variable at will generally 2-25 pm. Thicker layers are possible but generally uneconomic. Uniform coating within limitations of ’throwing power’ of bath. Pores not a problem as exposed steel protected by adjacent zinc
Usually about 12-40 pm closely controlled. Thicker coatings also possible. Continuous and very uniform even on threaded and irregular parts
Up to 40 pm of paint (and more with special formulations) can be applied in one coat. Good uniformity-any pores fill with reaction products
uniformity
Id
w
8
c! z 0
0
3
2 $
T8bk 13.7-(continued)
Characterktics of coating
Hot-dip galvanking
Metal spraying
Plating
Sherardising ~
5. Formability and
joinability
6. British standards
~
~
~~
Zinc dust painting ~~~~
Applied to finished articles, not formable: alloy laya is abrasion resistant but brittle. Sheet has little or no alloy and is readily formed. All coatings can be arc or resistance welded
When applied to finished articles. forming not required. Can weld through thin coating if necessary but preferable to mask edges to be welded and spray these afterwards
Electroplated steel has excellent formability and can be resistance spot welded. Small components are usually finished before electroplating or mechanical coating
Applied to finished articles: forming not required. Excellent abrasion resistance
Abrasion resistance better than conventional paints. Painted sheet can be formed and resistance welded with little damage
General work: BS 729 Continuous galvanised: plain sheet: BS 2989 Corrugated sheet:
BS 2569 Part I
BS 1706 Threaded components:
BS 4921
BS 4652 for the paint
t,
z
Wire: BS 443 Tubes: BS 1387 Conversion coatings; chromates used to prevent storage stains on new sheet; chromates or phosphates used as base for paints or powder coatings. Weathered coatings painted for long service
x
E
BS 3382
BS 3083
7. Extra treatments
E
8 Usually sealed; sealed surface is suitable base for paints on long-life structures. Alternatively sealer can be reapplied periodically
Conversion coatings, e.g. chromates used to prevent 'wet storage' stain. Chromates or phosphates used as a base for paints
Can be painted if required
Can be used alone, or as primer under other paints
n
w
P
w
13 :44
ZINC COATINGS
Properties and Nature of the Coatings
The actual structure and composition of zinc coatings depends upon the method of deposition. Zinc coatings produced by hot-dip galvanising and sherardising consist partly or wholly of iron-zinc alloys. Sprayed and plated zinc coatings contain no alloys, plated coatings consisting essentially of pure zinc. The characteristic properties of each type of coating are discussed below. Hot-dip galvanised coatings (see Section 12.2) Here the coating is not uniform in composition but is made up of layers of zinc-iron alloys becoming progressively richer in zinc towards the coating steel interface, so that the actual surface layer is composed of more or less pure zinc. Because of this alloy formation there is a strong bond between the coating and the steel. The alloy layers are harder than mild steel. The nature and thickness of the alloy layers are influenced greatly by the composition of the underlying steel, and also by the galvanising conditions. Notably the presence of silicon in the steel encourages the formation of iron-zinc alloys and thereby leads to the formation of heavier coatings, and indeed a steel with a high silicon content is often used intentionally when very heavy zinc coatings are required. In such cases the coating may consist entirely of iron-zinc alloys, and this is seen in the uniform grey spangle-free appearance of the galvanised surface obtained under these conditions. The addition of up to 0.2% of aluminium to the galvanising bath, on the other hand, depresses the formation of alloys and produces lighter and more ductile coatings, which are more suitable for galvanised sheet since they render it more amenable to bending and forming. Details of the method and the nature of the coatings are given in Section 12.4. In this method there is no alloy formation and the bond is primarily mechanical. Although porous, the coating is protective partly due to its sacrificial action and partly due to the zinc corrosion products which soon block up the pores, stifling further attack. Zinc plating Electroplating with zinc normally gives a dull-grey matt finish, but lustrous deposits can be obtained by adding brighteners to the electrolyte. Mechanical coating also gives a dull finish.The coating is of uniform composition throughout, containing no alloy layer, (Section 12.1). Plated coatings are very ductile and zinc-plated sheet can therefore be readily fabricated. Mechanical deposition of zinc by barrel-plating is also possible for small parts. Sherardising This is another alloy-forming process, a typical coating containing alloys with 8 or 9% of iron. Like galvanised coatings, the deposits are very hard (Section 12.3).
Sprayed zinc coatings
Relative Advantages of the Coating Methods
Each coating method has its own particular advantages, which are really the decisive factors in determining which one is used for a given purpose. Consideration must be given to the complexity and size of the work, the corrosion resistance, and hence the coating thickness needed, and the quality of finish required.
ZINC COATINGS
13 :45
Hot-dip galvanising produces a thick coating which thoroughly covers the work, sealing all edges, rivets, seams and welds when fabricated articles are treated. The size of the article which can be treated is limited to a certain extent by the size of the galvanising tank, but the technique of double dipping, i.e. dipping first one end and then the other, makes it possible to treat surprisinglylarge items, over 20 m long, successfully. Hot-dip galvanising is the most widely used method for coating with zinc. Zinc spraying possesses the important advantage that, since the equipment is essentially portable, it can be applied on site to either small or large structures. Thick coatings can be applied where desired, and it is possible to ensure that welds, edges, seams and rivets on finished articles receive sufficient coverage. It is not normally a suitable process for coating the inside of cavities or for coating open structures, such as wire meshes, since a large amount of zinc would be wasted. In hand spraying the uniformity of the coating depends largely on the skill of the operator. Zinc coatings produced by plating have the advantages that the thickness can be accurately controlled according to the protection desired. The acid zinc sulphate bath is used for applying coatings to uniform sections, e.g. box strapping, wire, strip, etc. The plating rates; in these solutions can be very high. The throwing power of this bath is poor, and for more intricate shapes and where appearance is important the cyanide bath is preferred. Bright deposits can be produced from the latter by the use of either addition agents or bright dips. Sherardising is distinguished particularly by the uniformity of the coatings which it produces. This makes it an ideal process for work, such as screw threads, where close tolerances are required and where complex or recessed parts are involved, since the inside surfaces of pipes and other hollow articles receive coatings comparable with those on the outside. The coating is very hard and offers a high resistance to abrasion. The maximum size of the articles which can be sherardised is limited by the size of the drums. In general, sherardising is best suited to the treatment of small castings, forgings and pressings, and fixings, such as wood screws, nuts, bolts, washers, chains, springs, etc. The outstanding virtue of zinc-rich paints is simplicity in application. No special equipment is required and the operation can, of course, be carried out on site, large or small structures being equally suitable for treatment. While there is some evidence that the zinc-rich paints will reduce iron oxides remaining on the steel surface, proper surface preparation is as important here as with traditional paints if the best results are to be achieved. The main use of zinc-rich paints is to protect structural steel-work, ships’ hulls, and vulnerable parts of car bodies, and to repair damage to other zinc coatings. Corrosion Resistance of Zinc Coatings
There are two main reasons why zinc is chosen as a protective coating for iron and steel. The first is the natural resistance of zinc itself against corrosion in most atmospheric conditions, and the second is the fact that zinc is electronegative to iron and can protect it sacrificially*. *The reversal of polarity of Zn and Fe which can occur in certain circumstancesis discussed in Section 1.3.
13 :46
ZINC COATINGS
The corrosion resistance of zinc is discussed in Section 4.7, and it is only necessary here to say that zinc is protected against further attack by a film of corrosion products. It is remarkably resistant to atmospheric corrosion except perhaps in the most heavily contaminated industrial areas, and even there its use as a protective coating is still a sound practical and economic proposition. The value of zinc coatings as a basis for painting under very aggressive conditions has been clearly demonstrated. The natural corrosion resistance of zinc is, therefore, its most important property in relation to zinc coatings. The electrochemical property becomes important when the zinc coating is damaged in any way to expose the steel, when sacrificial corrosion of the zinc occurs and the steel is thereby protected. Moreover, the corrosion product of the zinc normally fills the break in the coating and prevents or retards further corrosion of the exposed steel. Life of Zinc Coatings in the Atmosphere
As the protective value of the zinc coating depends largely on the corrosion resistance of zinc, the life of a coating is governed almost entirely by its thickness and by the severity of the corrosive conditions to which it is exposed. Extensive tests and field trials which have been carried out have shown that the life of a zinc coating is roughly proportional to its thickness in any particular environment3 and is independent of the method of application. The corrosion rates and lives of zinc coatings in UK atmospheres are given in Table 13.8. These are based on practical experience as well as exposure trials. The figures should be taken only as a guide because of the difficulty of defining atmospheres in a word or two (indeed there is now a tendency for research workers to define the corrosivity of an atmosphere in terms of the corrosion rate of zinc) because of unpredictable local variations from place to place and time to time. For example, moorland which is frequently covered with acid-laden mist can be very corrosive. Corrosion tests have also shown that there is a difference in the rates of corrosion throughout the year. This is partly because the sulphur content of the air is greater in winter than summer and partly because more of the
Table 13.8 Typical corrosion rates and lives of zinc coatings in the UK Corrosion rate A tmosphere gm-2
Rural Urban Marine Industrial
Y- I 14
40 40 80
W Y - I
2 5 5 10
Life of coating with overoge thickness (yeors)11 200pm*
100pm'
25pmt
5pms
50- 150 30-50 30-50 10-30
25-75
6-20 4-6 4-6
1-3
15-25 15-25 5-15
1-4
'Can bc produced either by grit-blasting before galvanising or by zinc spraying. 'Typical thickness o f coating on galvaniscd or zinc-sprayed structural sleel. :Typical thickness o f coating on galvanised shcct or sherardixd componenls. 'Typical thickness o f zinc plating. "Theliver given are additional to the life of the unprotected steel.
=I =1 0.25-1
ZINC COATINGS
13 :47
zinc corrosion products are dissolved under the wetter winter conditions. Thus if unpainted zinc coatings are first exposed to the atmosphere in spring or early summer a more protective film will be formed. Detailed test results for a 2-year exposure period4are given in Table 13.9. It should be remembered that test sites are sometimes chosen because they are believed to represent particularly corrosive examples of the type of atmosphere being studied. The ratios of steel: zinc losses are particularly interesting. It shows that zinc is far less affected than steel by many chloridecontaining atmospheres. Time of wetness and amount if atmospheric SO:’ are the most important factors with Zn. On a global view the single word description of the site is often misleading; particularly, it gives no indication of the times each year that objects remain wet, which varies considerably from country to country and also within countries. An extensive compilation of atmospheric exposure test data on zinc is now available5 and complements the slightly earlier critical study by Schikorr‘. Water Zinc-coated steel, like zinc, behaves less favourably in distilled and
soft waters than in hard waters, where the scale-forming ability of the hardness salts provides considerable protection. Hot-dip galvanised tanks, cisterns and pipes are very widely used for storing’ and carrying domestic water supplies throughout the world, and as a rule such equipment gives long and trouble-free service (Section 9.3) and is hygienically acceptable. Sea-wafer The protective properties of zinc coatings in sea-water have been shown to be very good, and zinc is widely used as a coating metal in the shipbuilding industry and for protecting structural steel work on docks and piers, etc. In BISRA tests at Gosport’, specimens of steel coated with aluminium, cadmium, lead, tin and zinc were immersed for two years. In this time all but the zinc-coated specimens had failed. The zinc-coated specimens were then transferred to Emsworth and immersed for a further four years-a total of six years- before the coatings ceased to give complete protection. The coating on these specimens was about 900g/m2, indicating a rate of attack of about 20pm/y in this sea-water. Other tests”* show corrosion rates of 10-25 pm. Conditions within a few hundred metres of the surf line on beaches are intermediate between total immersion in sea-water and normal exposure to a marine atmosphere. High corrosion rates can occur on some tropical surf beaches where the metal remains wet and where inhibiting magnesium salts are not present in the sea-water.
Soil Galvanised pipe is frequently used for underground water services. Table 13.10 gives results of tests’ carried out with galvanised pipes and plates buried at different sites. The specimens were removed after five years, when the only ones that had failed were some plates buried in made-up ground, consisting of ashes, at Corby and one pipe at Benfleet, At Corby no galvanised pipes were exposed and most of the coatings on the plates had corroded away. For this reason no figures are recorded for Corby in Table 13.10. The high rate of corrosion at Benfleet was attributed to the fact that the specimens were below the soilwater level for about half their life as the tide rose and fell. Similar tests have been carried out in the United States”; in these the
13:48 Table 13.9
ZINC COATINGS Average loss of zinc in two years and steel/zinc ratio for 45 test sites4
Location
Norman Wells, N.W.T., Canada Phoenix, Ariz., USA Saskatoon, Sask., Canada Esquimalt, Vancouver Is., Canada Fort Amidor Pier, Panama C.Z. Melbourne, Aust. Ottawa, Canada Miraflores, Panama C.Z. Cape Kennedy, 0.8 km from ocean, USA State College, Pa., USA Morenci, Mich., USA Middletown, Ohio, USA Potter County, Pa., USA Bethlehem, Pa., USA Detroit, Mich., USA Manila, Philippine Is. Point Reyes, Calif., USA Halifax (York Redoubt) N.S., Canada Durham, N.H., USA Trail, B.C., Canada South Bend, Pa., USA East Chicago, Ind., USA Brazos River, Texas, USA Monroeville, Pa., USA Daytona Beach, Fla., USA Kure Beach, N.C. 240-m lot, USA Columbus, Ohio Montreal, P.Q., Canada Pilsea Island, Hants., UK Waterbury, Conn., USA Pittsburgh, Pa., USA Limon Bay, Panama C.Z. Cleveland, Ohio, USA Dungeness, UK Newark, N.J.,USA Cape Kennedy, 55 m from ocean 9 m up, USA ditto, ground level ditto, 18 m up Bayonne, N.J.,USA Battersea, UK Kure Beach. N.C., 24 m lot. USA London (Stratford), UK Halifax (Federal Bldg.), N. S., Canada Widnes, UK Galeta Point Beach, Panama C.Z.
Described by authors as:
2-year SleeUzinc Life of 100 test: zinc loss ratio pm zinc lost per (by weight) coating (calc.) year without (Pm) mainrenonce (years)
Rural Rural Rural Rural/ marine Marine Industrial Urban Marine
0.2 0.3 0.3
10.3 17.0 21.0
500 300 300
0.5
200
1.2
31.0 25.2 37.4 19.5 41.8
Marine Rural Rural Semiindustrial Rural Industrial Industrial Marine Marine Urban Rural Industrial Semi-rural Industrial Industrial/ marine Semiindustrial Marine Marine Urban Urban Industrial/ marine Industrial Industrial Marine Industrial Marine Industrial
1.2 1.2 1.2
84.0 22.0 18.0
80 80 80
1.3 1.3 1.3 1.4
7s IS 75 70
1.6 1.6 1.6 1.6 1.8 1.8
26.0 18.3 32.4 12.2 39.8 364.0 18.5 19.0 24.2 20.8 52.1
1.9
56.0
50
2.0 2.1 2.1 2.2 2.5
28.4 164.0 80.0 16.8 10.9
50
2.5 2.6 2.7 2.7 2.8 3.7 3.8
21.6 9.8 13.1 25.9 15.7 148.0
4.1 4.3 4.5 4.9 5.8 6.5 7.1 7.6
45,s 117.0 33.0 17.9 20.0 93.0 17.8 17.0 39-0 49.4
Marine Marine Marine Industrial Industrial Marine Industrial Industrial Industrial Marine
0.7 0.8 1.1
1.5
10.5
IS.9
15.1
1 so
125 90 80
65
60 60 60 60 55 5s
45 4s 45
40 40 40 35 3s 35 21 2s 24 23 22 20 17 15
14 13 9 6
13:49
ZINC COATINGS
Table 13.10 Loss of coating thickness of galvanised specimens after five years in various soils Galvanised pipes 'Oil conditions
Alluvium or reclaimed salt marsh Gotham Keuper Marl (gypsum) Pitsea London clay Rothamstead Clay with glacial flints Corby Made up ground (ashes)
Galvanised plates
Initial Loss Initial Loss thickness of coating thickness of coating of coating thickness of coating thickness (am) ( em) ( em) ( em)
Ilncoared steel loss in lhickness
(ern)
Benfleet
82
47
90
52
200
77 77
17 17
90 90
17 17
50 160
82
13
95
13
120
-
-
-
*
300
Most of coating on plates corroded away
maximum depth of pitting was also measured. Except in the most corrosive soils the maximum depth of pitting in steel specimens exposed for about 124 years was more than 11 times that in zinc-coated specimens, even though the ratio of the rates of corrosion was only about half that figure. This resistance to pitting, combined with the fact that rusting appears to start only when nearly all the zinc and zinc-iron alloys layers have corroded away, reduces the risk of premature failure in galvanised piping. The coatings on galvanised specimens remained virtually intact during exposure for 24 years in about half the 15 soils in which they were buried. Their corrosion resistance was most marked in alkaline soils. In clays and loams, where little or no organic material was present, a 600g/m* coating could be expected to provide protection for 10 years or more. Protective Systems Applied Subsequently to Zinc Coatings
Chromating Chromating is considered in Section 15.3. The chromate film on zinc is adherent and can be drab, yellow-green or colourless in appearance; the colour varies considerably with the method of application. It retards 'white rust', the white deposit which sometimes forms on fresh zinc surfaces which are kept under humid conditions (see Section 4.7). A chromate film is damaged by heat and if used as a basis for paint adhesion, should preferably not be heated above 7OoC, nor for longer than 1 h. Painting In mildly corrosive conditions zinc coatings will probably have a life longer than that expected of the coated article, and no further treatment of the coating will be necessary. When, however, the coating is subjected to a more strongly corrosive environment one or more coats of paint can, with advantage, be applied over the zinc. Paint films used in conjunction with zinc coatings give systems whose lives are longer than the sum of the lives of the coatings used independently.
13 :50
ZINC COATINGS
Paint applied to a suitably prepared zinc coating will last longer than would be the case if it were applied direct to iron or steel, and the need for repainting thus becomes less frequent. With hot-dip galvanised or zinc-plated coatings, however, it is necessary either to use special primers or to prepare the surface before painting. This is primarily because most oil-based paints react with the unprepared zinc surface to form zinc soaps resulting in poor adhesion. Weathered galvanised steel surfaces give good adhesion for many paint systems, but where new galvanised or zinc-plated steel is painted or powder painted it is necessary to convert the surface into an adherent phosphate or chromate coating or to use specially developed primers". Many commercial phosphating processes are available, but all consist essentially of an etch in a phosphoric acid solution containing zinc salts and certain accelerators. These treatments produce uniform, fine-grained and strongly adherent phosphate films on the surface of the work. Many chromate finishes also give a satisfactory base for painting (see Section 15.3). Etch primers are widely used. They are mostly based on polyvinyl butyral and contain chromates and phosphoric acid. They are said to act both as primers and as etching solutions because it is believed that the chromates and phosphoric acid form an inorganic film, which provides adhesion, while oxidised polyvinyl butyral provides an organic film. For direct application to new galvanised steel, the best known primers are based on calcium orthoplumbate pigment and metallic lead, but these are now less used for environmental reasons. Zinc-dust paints and zinc-phosphate pigmented paints are also used, but the trend is to use pretreatments to assure good adhesion. Applications of Zinc Coatings
Zinc coatings are successfully used in a very wide field to protect iron and steel goods from corrosive attack. The building trade is one of the largest users of zinc-coated steel. The frameworks of large modern buildings can be either galvanised or zincsprayed before erection, or sprayed on site. Where a framework is accessible, zinc-rich paints provide an excellent way of renovating old buildings. Apart from the structural aspect, galvanised sheeting is used for roofing, ventilation ducts and gutters, as well as for water tanks and cisterns, and galvanised pipe is used for public and domestic services. In the Forth, Severn and many other suspension bridges, zinc coatings have an important function. The whole main structure is of steel and has been zinc-sprayed on the external surfaces, while the main cable and hanger ropes have been coated by continuous hot-dip galvanising. Case histories of galvanised multi-truss bridges cover more than 30 years. Zinc-coated structures are used in pylons carrying electrical transmission lines, in masts for radio, television and radar aerials, and for supports for overhead wires on electrified railways. An interesting new use of galvanised steel is for reinforcement of concrete. This reduces the risk of spalling and staining or can enable the depth of the concrete cover to be reduced leading to slimmer structures of lower cost.
ZINC COATINGS
13:51
Economics
All coatings cost money and a true appraisal of both the initial and maintenance costs is essential when specifying the protective scheme for a new structure. An example showing the comparative costs of galvanising and painting is given in Section 9.1.
Recent Developments Since the previous edition was written, the main development has been the introduction of a range of zinc alloy coatings designed to give increased corrosion resistance and sometimes with additional advantages such as increased formability and retention of paint adhesion with a wider range of paints. The zinc-aluminium alloys are most important. The zinc-55%aluminium- 1.5 %-silicon alloy hot-dip coating was initiated over 20 years ago by the steel industry and has recently become of major worldwide importance (known as Galvalume, Zincalume, Alugalva, Aluzink, Aluzinc, Zincalit or Zalutite). The coating usually has 100-400% more corrosion resistance than galvanising in the atmosphere, but less cathodic protection and also has the inherent problem of aluminium alloys when in contact with alkalis. Since 1980, the zinc-5%-aluminium (notably Galfan which has a mischmetal addition) alloys, which are essentially based on the eutectic structure, have been developed commercially. They give 30-200% increase in corrosion resistance in the atmosphere and are extremely flexible. They can be used for sheet, wire and some types of tube galvanising whereas the zincd5%-alurninium alloy is restricted to sheet. Intermediate alloy compositions include a zinc- 15070 -aluminium alloy for metal spraying (higher aluminium contents are unsuitable for spraying coating wire) and a zinc-30%-aluminium-0.2%-magnesium-0.2%-silicon (Lavegal) for sheet. A further range of alloys has been produced using nickel. The 13% nickel alloy has been adopted commercially in the USA for electroplating sheet destined for car-body manufacture. Other developments range from 5 to 14% nickel. Additions at a much lower level-less than 0.1% nickel in the operating bath - are used in the hot-dip galvanising of products to give more uniform growth of the iron-zinc alloy layers than would otherwise occur on steels containing silicon. Other alloy additions in commercial use include; iron (often a two-layer electroplated coating with less iron -typically 20% -in the under-layer to assist formability and more iron - often 80% -in the outer layer to assist paintability); cobalt (0.15-0.35%); similar amounts of chromium (the zinc/ chromium/chromium-oxide coating known as Zincrox); and a range of ternary alloys and of composite coatings. New coating techniques in commercial use are mechanical coating (now incorporated under ‘plating’ in Table 13.7)and adhesive-bonded or vapourdeposited coatings, although each of these represents less than 1% of the zinc coatings used.
13 :52
ZINC COATINGS
The mechanical coatings are primarily barrel processed on to parts up to about 150 mm long or 400 g mass. Adhesive bonding (with a conductive adhesive to maintain the electrochemical protection by zinc) is particularly suitable for wrapping pipes. Vapour deposition has some use in products but the newest development is application on continuous strip for car-body manufacture- the surface is smooth so that the subsequentlypainted surface has no unevenness. Duplex coatings of zinc followed by organic materials have increased and there has been a swing away from lead-bearing paints, One or two organic coats are sufficient. Some paints are formulated for direct application to zinc surfaces, but adhesion is most assured by pretreatment with phosphate-type solutions, particularly as most paints are not covered by national specifications and slight modificationsin formulation or method of application which may not be significant on steel can markedly affect adhesion, especially on untreated zinc surfaces. Sprayed powder coatings (see BS 6497)are applied over zinc for long-life decorative effects. Sprayed zinc coatings are often only sealed, i-e. a labile solution fills the pores in the coatings but provides no specified overlay. Zinc or zinc-alloy coatings, previously used on luxury cars, are now being used on more mass-produced cars to meet 10-year warranties against perforation corrosion. In general, the percentage of steel which is zinc-coated is increasing. Hot-dip galvanising after fabrication, in addition to giving long lives to first maintenance, enables hidden interior areas of tubular space frames and lamp posts to be safely protected against failure by rusting. A new use is for earth reinforcement in which galvanised tie-bars, embedded deeply into prepared earth walls, hold in position concrete slabs forming the near-vertical exterior of road and rail embankments. The bibliography covers some of the major developments. A. R. L. CHIVERS F.C. PORTER
REFERENCES I . Technical Notes on Zinc. Separate leaflets entitled Zinc Coatings, Galvanising,Zinc Spraying. Sherardising, Zinc Plating and Zinc Dust are available from Zinc Development Association, London 2 . General GalvanizingGuide, Galvanizers Association (1 988) and Inspection of Zinc Sprayed Coatings, Z.D.A. (I%@ and Inspection of Sherardized Coatings, Z.D.A. (1985) 3. Hudson, J. C.. Sixth Reportof theCorrosion Committee. I.S.I. Special Publication No. 66 (1959) 4. Metal Corrosion in the Atmosphere, A.S.T.M., STP435 (1967) 5. Slunder, C. J . and Boyd, W. K., Zinc: Its Corrosion Resistance, Z.D.A. (1983) 6 . Atmospheric Corrosion Resistance of Zinc, Z.D.A. (1964) 7. Hudson, J. C. and Banfield, T. A., J. Iron 8. Inst., 154, 229 (1946) 8. Wiederholt, W., Korrosionverhalten von Zink, Metall-Verlag GmbH, Berlin (translation available on loan from Z.D.A. Library) 9. Hudson, J. C.. Corrosion of Buried Metak, I.S.I. Special Publication No. 45 (1952) 10. Denison, I. A. and Romanoff, M., J. Res. Nut. Bur. Stand., 49, 299 (1952) 1 1 . Porter, F. C., Br. Corr. J . , 4, 179-186 (1969). (Reprint available from Z.D.A.) 12. Andrews, T. O . , Edge Protection by Zinc, BSC, Prod. Dev. Tech. Digest (1964) 13. Haynie. F. H. and Upham, J. B., Materials Protection and Performance, 35-40 (Aug. 1970) 14. Carters, V. E., Met. Finishing J . , 18, 304-306 (1972)
ZINC COATINGS
13:53
15. Bergmann, J. etal.. Corr. Prev. and Control, 25, No. I , 7-11 (Aug. 1978). (Reprints
available from Z.D.A.) 16. Cook, A. R.. Anti-CorrosionMethodsandMaferials, 23 No. 3.5-8 (Mar. 1976). (Reprints
available from Z.D.A.)
BIBLIOGRAPHY
Porter, F. C., Zinc Handbook, Marcel Dekker, New York (1991) ILZRO, ‘Zinc’ in Corrosion Resistance. 3rd ed (in press) BCSA/BSC/PRA/ZDA, SfeelworkCorrosionProfectionGuide: InferiorEnvironments ( I 982) and Exterior Environments (1986). Z.D.A., London BS 5493, ‘Code of Practice for Protective Coating of Iron and Steel Structures Against Corrosion, B.S.I., London (1977) Chandler, K. A. and Bayiiss. D. A., Corrosion Protection of Sfeel Structures, Elsevier Appl. Science Publ. LondonINew York (1985) Gabe. D. R. (ed.), Coafingsfor Protection. Inst. of Prod. Engs. (1983) Johnen, H. J. Zink fiir Sfahl, Zinkberatung Diisseldorf (1984). (In German) 25 years of GA V (the German galvanizers association), GAV Dusseldorf, (1977). 155 pp inc, extensive reference list. Updated by Bottcher. H. J. in annual reviews (In German). Galvanizing Guide, Galvanizers Association. London (1988) Thomas, R., Rust Prevenfionby Hot Dip Galvanizing,Nordic Galvanizers Association (1980). English edition available from Z.D.A., London Galvanizing-a Pructical Reference, Galvanizers Association of Australia, Melbourne (1985) Galvanizing Info The Nexf Cenfury. SAHDGA, NIMR, Pretoria (1986) Galvanizing Characterisria of structural steels and fheir weldments, ILZRO Res., Triangle Park (1975). (Includes data sheets for six countries’ steels) Gauan Manual ILZRO. 3rd edn. (1988) Zalutite Technical Manual, BSC Strip Mill Products (1986) Zinrec Technical Manual, British Steel Corporation, Newport, UK (1986) Aes. H. W., Handbuch der Galvanotechnik Carl Hanser Verlag, Munich, Chapter 17.06, Zinc (plating) (1966). (Typescript translation from German, in Z.D.A. library) Protection by Sprayed Metal Coatings, Welding Inst., Abingdon, UK (1968) Corrosion Tests of Flame-sprayed Coaled Sfeel- 19 Year Report, Amer. Weld. SOC.,Miami (1974)
Bibliography-Zinc Spraying, Z.D.A., London (1970). (Later items in Zinc Abstracts (to 1986) then Zincscan) Prorecling Sfeel With Zinc-Dust Painfs. Nos. 1,2 and 3 Z.D.A.. London (1%9, 1972 and 1975) Bibliography-Zinc Dust in Paints. Z.D.A., London (1969 and 1976) Sherardizing, Zinc Alloy Rust Proofing Co, Wolverhampton, UK (1975) Inspection of Sherardized Coatings, Z.D.A.. London (1985) ASTM 8695, Coatings of Zinc Mechanically Deposited on Iron and Zinc, ASTM (1985) James, D. G., ‘Mechanical Deposition -Hydrogen Embrittlement Study’. Paper to IMF Annual Conf., UK (1985) Maeda, M. et a/., ‘Newly Developed Continuous Zinc Vapour Deposition Process. Paper to SITEV 86 (1986) Euronorm 169: Confinuously Organic Coated Sfeel Flat Producfs, (1985) (Available from national standards organisations) Colorcoat and Stelvetite PreFinished Steel- Technical Manual. BSC Strip Mill Products, Newport, Gwent, UK (1984) van Eijnsbergen, J. F. H., ‘Duplex Coatings: a Review of Recent Developments, EGGA General Assembly Monaco. Z.D.A.. London (1986) Schmid, E. V., Painting of Zinc Surfaces and Zinc-Containing Anticorrosive Primers, Monograph No. 3, OCCA, Wembley, U K (1986) Hall, D. E. ‘Electrodeposited Zinc-Nickel Alloy Coatings- A Review’, Plafing and Surface Fin., 59-65 (Nov. 1983) Adaniya, T. et a/., ‘Iron-Zinc Electroplating on Strip’, AES Fourth Cont Strip Plating Symp, Amer. Electroplaters SOC.(1984) 55% AIZn, BlEC, Bethlehem (1987)
13.5 Tin and Tin Alloy Coatings
Tin Coatings Methods of application (Chapter 12) Tin coatings are applied by hotdipping, electrodeposition, spraying and chemical replacement. A variant of hot-dipping called wiping,in which the tin is applied either solid or melted to the fluxed and heated surface and is wiped over it, is also used for local application, e.g. to one face of a sheet or vessel. The hot-dipped and wiped coatings are bright; electrodeposited coatings may be matt or bright. The electrolytes used for electrodeposition of matt coatings for general purposes are: ( a ) an alkaline bath containing tin as stannate; ( b ) an acid bath containing stannous sulphate, free sulphuric acid, cresol-sulphonic acid, gelatin and /3 naphthol; and ( c ) an acid fluoroborate-bath containing organic addition agents. For high-speed plating on rapidly moving strip in the production of tinplate, the baths used are the ‘Ferrostan Bath’ based on stannous sulphate and phenolsulphonic acid, the ‘Halogen Bath’ based on chloride and fluoride (both with appropriate addition agents) and to a lesser extent, the alkaline stannate bath and the acid fluoroborate bath. Bright coatings are deposited from acid stannous sulphate baths containing combinations of organic addition agents. The electrodeposited matt coatings may be brightened by momentary fusion. This process of flowbrightening or flow-melting is achieved with most of the electrolytic tinplate production by conductive or inductive heating; for manufactured articles, it is usually carried out by immersion in a suitable hot oil2. The hot-dipped coatings-’are distinct from the others in having practical thickness limits and in possessing an inner layer of intermetallic compound, usually described as the alloy layer. The flow-melted electrodeposited coatings also have an alloy layer, which is somewhat thinner than that obtained in hot dipping. Coatings of tin produced from tin-containing aqueous solutions by chemical replacement may be used to provide special surface properties such as appearance or low friction, but protect from corrosion only in non-aggressive environments. Copper and brass may be tinned in alkaline cyanide solutions or in acid solutions containing organic addition agents such as thiourea. Steel may be first coated with copper and then treated
’
13: 54
TIN AND TIN ALLOY COATINGS
13:55
as copper, or it may be tinned in acid tin salt solutions with or without contact with zinc. Aluminium alloys may be tinned by immersion in alkaline stannate solutions. Articles of steel, copper or brass which require a thicker coating than is possible by chemical replacement, and which are difficult to tin by normal electrodeposition, may be coated by immersion in alkaline sodium stannate solution in contact with aluminium suitably placed to act as anode.
Thickness of tin coatings The thicknesses of the various types of tin coating are shown in Table 13.11. Table 13.1 1 Thicknesses of tin coatings Hot dipped Electrolytic tinplate Wiped coatings Electrodeposits other than tinplate Flow-melted electrodeposits Sprayed coatings Chemical replacement coatings
I .5-25 pm 0-4-2 pm
1-12 pm 2-5-75 pm 2-5-7.5 pm 75-350 pm trace-2.5 pm
Properties of tin coatings When the choice of coating is not governed by the size and geometry of the article to be coated, it depends upon appearance, the thickness required, and the degree of porosity which can be tolerated. Bright coatings, as produced by hot dipping, flow-melting or bright electrodeposition, have the advantage of smoothness, good appearance and resistance to finger-marking. The presence of an alloy layer in hotdipped and flow-melted coatings also confers some advantage in the making of soldered joints. On the other hand, with hot-dipped coatings it is rarely possible to ensure absence of coating porosity, whereas electrodeposition can build the coating up to the thickness, above 25 pm, at which pores are unlikely to penetrate the coating. Sprayed coatings have structures in which fine pores thread tortuous paths through the deposit, and it is necessary to apply a coating thickness of about 350pm if all these paths are to be closed. Scratch-brushing of the deposit, however, makes it possible to consolidate the surface and to achieve adequate continuity in thinner deposits, e.g. 200 pm. Tin coatings are ductile and are able to contribute a lubricating effect in the deep drawing of steel. The presence of the thin alloy layer in flow-melted tinplate coatings does not impair this property appreciably but bright electrodeposited coatings may be less ductile than others. Sometimes a spontaneous outgrowth of metal filaments about 1 pm diameter, commonly called whiskers, occurs on tin coatings in a time after application which may vary from days to several years. This growth does not affect the protective quality of coatings but the whiskers are able to shortcircuit compact electronic equipment. The character of the substrate is influential and tin coatings on brass should always be undercoated, e.g. with nickel or copper. The introduction of some impurities, e.g. 1 Yo lead, into the tin coating is some safeguard. Hot-dipped or flow-melted coatings are rarely affected49.
13 :56
TIN AND TIN ALLOY COATINGS
Corrosion Resistance of Tin Coatings
General considerations Influential factors in the behaviour of tin coatings are the variation according to environment in the relative polarity of coating and substrate, the nature of any intermetallic compound layers formed between coating and substrate and the extremely low rate of corrosion of tin in alkaline and mildly acid media in the absence of cathode depolarisers. The depression of the corroding potential of tin, when the tin ion concentration in solution is reduced by the formation of complex ions, has been referred to previously (Section 4.6). Iron may also be complexed and the potential of iron is affected by the presence of tin ions in solution. The extent of the potential shifts4depends on the complexing agents present, the solution concentration and pH. The electrochemical relationship of tin and iron is therefore a complicated one, but for practical purposes, tin can be regarded as being anodic to iron in contact with such products as fruit juices, meat and meat derivatives and milk, in solutions of citric, tartaric, oxalic and malic acids and their salts, and in alkaline solutions. In solutions of inorganic salts, natural waters or atmospheric water, tin is cathodic to iron. Supplementary protection can be given to tin coatings either by passivation treatments or by organic finishes. Passivation in chromate solutions gives some protection to the steel exposed at the base of pores as well as to the tin coating (Section 15.3). Electrolytic tinplate is passivated on the production line by rapid passage through acid solutions, usually dichromate, with applied cathodic current. Similar treatments may be employed on other forms of tinned cathodic metal, and a process of immersion in hot alkaline chromate’, which combines cleaning and passivation, is useful for treating metal coated by oil or other contaminants in manufacturing operations. Tinplate for containers and closures is often decorated by colour printing and protected by clear lacquers. No surface preparation is carried out and difficulties with wetting and adhesion, sometimes associated with the character of the oxide layers on the surface, are rare. The corrosion of tinned steel Atmospheric corrosion During full exposure to the weather, some rust at pores in the coating soon appears. In coatings of thicknesses less than about 5 pm the rust spreads out from the pores and in due course the whole surface becomes covered by rust. With thicker coatings, this spread of rust does not occur. In industrial atmospheres, penetration of the steel may cease, after a few weeks, the surface becoming covered by a growing grey layer of tin corrosion product with faint rust stains; tin coatings of upwards of 12 pm will outlast zinc coatings of comparable thickness6*’. In marine atmospheres, however, attack at pores persists even with the thicker coatings and pits are formed. In most of its uses, e-g. the external surfaces of tinplate cans, tinned steel has only to resist condensed moisture. In the absence of pollution of the atmosphere by unusually large amounts of sulphur dioxide or chlorides, or of several days of continuous wetting, tinned steel remains unrusted; even the thin porous coatings on the common grades of tinplate remain bright and unmarked over the periods involved in the commercial handling and domestic storage of cans, and the domestic use of kitchenware. When
TIN AND TIN ALLOY COATINGS
13:57
wetting persists for long periods, especially if pools of water collect, rusting at pores begins. This situation can easily arise in the holds of ships in transit through the tropics unless proper precautions are taken; shipment in large sealed containers seems likely to avoid most of the trouble'. The conditions needed to ensure complete absence of pore rust are similar to those needed to preserve uncoated steel, although with the tinned steel, rust-promoting conditions can be tolerated for a much longer period without the general appearance of the metal being spoilt. Condensed moisture rarely produces serious pitting of the steel at pore sites, but for many purposes maintenance of appearance is important. The change in aspect which takes place on rusting is much influenced by the degree of porosity of the coating, which is usually dependent on coating thickness. The thinnest coatings of electrolytic tinplate of 5 g/m2 of sheet (equivalent to a coating thickness of 0.4 pm) will develop a continuous rust coating in conditions where a hot-dipped coating of 30 g/m2 will show only inconspicuous rust spots. A coating heavier still may show no visible change. The oil film present on both types of tinplate and on newly hot-dipped tinware has a slight protective value. The passivation processes have much more effect but this is unlikely to compensate for a substantial reduction in coating weight. The effects of oil and passivation on the outside of tinplate cans may be reduced during can manufacture, filling and sterilisation. The resistance of tin to organic acid vapours emanating from wooden cases and from some insulating materials and paints gives tin an advantage over zinc and cadmium as a coating for equipment likely to be exposed to these vapours. There is, however, some risk of rust-spotting at pores in tin coatings; one method of trying to secure immunity of the coating from organic vapour corrosion and of the pores from rusting, consists in plating a layer of tin over a layer of cadmium or of zinc.
Immersion in aqueous media open to air Solutions in which tin is cathodic to steel cause corrosion at pores, with the possibility of serious pitting in electrolytes of high conductivity. Porous coatings may give satisfactory service when the corrosive medium deposits protective scale, as in hard waters, or when use is intermittent and is followed by cleaning, as for kitchen equipment, but otherwise coatings electrodeposited or sprayed to a sufficient thickness to be pore-free are usually required. Sometimes it is possible to add corrosion inhibitors to an aqueous product that is to remain in contact with tinned steel. The normal inhibitors used for protecting steel, e.g. benzoate, nitrite, chromate, etc. are suitable, provided that they are compatible with the product and that the pH is not raised above 10. In a closed container with an air-space, such inhibitors will not protect the zone above the water-line, and possibly not the water-line zone itself, against condensate. Volatile inhibitors have been used to give protection to these areas. Fruit juices, meat products, milk and milk products, fish and most vegetables, in which tin is likely to be anodic to steel, can be handled open to the air in tinned steel vessels. Some corrosion of the tin occurs at rates similar to those found for pure tin and in due course retinning may be necessary. The alloy layer in hot-dipped tin coatings is cathodic to both tin and steel and, under aerated conditions may stimulate the corrosion of both metals, but this effect appears to be unimportant in practice.
13:58
TIN AND TIN ALLOY COATINGS
Tin is anodic to steel in alkaline solutions, the corrosion rate for a continuous coating being similar to that of pure tin, and tinned articles that are washed in aerated alkaline detergents slowly lose their coating.
Tinplute containers The behaviour of tinplate is basically similar to that of other types of tinned steel, but performance requirements of tinplate containers are special. Containers are used in several forms: 1. Cans with replaceable closures for such products as dry foodstuffs,
pharmaceuticals, tobacco, solvents, liquid fuels and paint. These usually contain an appreciable amount of oxygen. Tinplate closures for bottles and jars made of non-metallic materials may also be considered in this category. 2. Sterilised sealed cans of foodstuffs, including fruit, vegetables, meat, fish and milk, which should contain only residual traces of oxygen. 3. Cans for beer and soft drinks. 4. Aerosol cans which may contain a propellant together with products such as paints, cleaners, cosmetic preparations and foodstuffs. These may also contain some oxygen. With all of these containers, both the can and its contents must reach the user in a visibly good condition. Cans must therefore resist external rusting, and methods of achieving this (e.g. adequate coating thickness, passivation treatment, attention to packaging and storage conditions and, if need be, lacquering)have already been mentioned. In other respects, both the requirements and the methods of achieving them may differ for the several classes. For categories 1 and 4, the relative polarity of tin and steel may be in either direction, depending on the product contained, but, more commonly, steel is anodic to tin and sufficient oxygen is present to make perforation by corrosion a possibility with water-containing products. Small quantities of water in nominally non-aqueous products can be seriously damaging because they are able to use all the oxygen present in the contents. Change of formulation, including addition of corrosion inhibitors, is possible for many non-food products and protection by lacquering is a generally available means of protection. Containers of foodstuffs should not be unduly stained or etched and must not be perforated or allowed to become distended by pressure due to evolution of hydrogen, and the contents must not suffer unacceptable changes of colour or flavour. Long storage periods, e.g. two years, may be required. Yellow-purple staining of can interiors may be produced by adherent films of tin sulphides formed by S2- and HS-compounds derived from proteins in meat and vegetable products. It may be prevented by suitable passivation treatment of the tinplate9 or by the use of appropriate lacquers. Loose iron sulphide, occasionally formed by sulphur-containing products at pHs above about 5 - 5 in the headspace of a can where there is some residual oxygen, is more objectionable and is not prevented by passivation or by normal lacquer since it occurs at breaks in the coating. Careful control of can-making and canning procedures is the best safeguard. Discolouration of products inside cans may follow the reduction of colouring matters or the formation of new coloured compounds with tin or iron. This is a problem with strongly coloured fruits and the remedy is to
13 :59
TIN AND TIN ALLOY COATINGS
use fully lacquered cans. Other than this effect, dissolved tin has no objectionable action on the quality of canned products, but very small amounts of dissolved iron have adverse effects on flavour. Except in special circumstances, the anodic relation of tin to steel and the inhibition of steel corrosion by dissolved tin ‘O-” protects the unlacquered tinplate can of food from risk of perforation or from taking up appreciable quantities of iron. The main hazards are excessive dissolution of tin, which may impair the appearance of the can and breach food regulations, and evolution of hydrogen which may distend the can and make it an unsalable ‘hydrogen swell’. The amount of hydrogen collecting in a can to produce a ‘swell’ is usually roughly proportional to the amount of iron dissolved, and the high rate of hydrogen evolution responsible for swells seems to arise from self-corrosion of the steel when protection by tin has been lost, and not from the combination of tin anode with steel cathode I 3 In general, tin dissolution inside an unlacquered can has a high but diminishing initial rate followed by a steady slow ratel4. The initial phase is associated with the reduction of cathodic depolarisers, including residual oxygen, and its duration and the corrosion rate reached depend on the nature of the product and on canning technology. In the second phase the cathode reaction is hydrogen ion reduction and the slow rate of tin dissolution, often equivalent to corrosion currents of the order of 10-9A/cmZ,is due to the scarcity of effective cathodes. The area of steel exposed at pores and scratches may be expected to have some influence on the corrosion rate, and small grain size of the tin coating has been considered to be associated with high rates. Many compounds capable of acting as cathodic depolarisers are naturally present in foodstuffs; they vary in character from product to product and, even in the same product, may vary in amount under such influences as season of growth, harvesting condition^'^ and sterilising The reduction of colouring matters in fruit has already been mentioned; other organic compounds in fruit and vegetables may be reduced and, in fish, trimethylamine oxide is a known large stimulator of corrosion. Inorganic nitrate, which is reduced to ammonia, is a most damaging promoter of corrosion in many vegetables and in fruit at pH values below 5 5 If cathodic depolarisers are present in amounts sufficient to promote dissolution of a substantial amount of tin coating then the best means of obtaining satisfactory can appearance and shelf-life is to use lacquered tin-plate. Passivation films are not a reliable means of preventing etching of the tin coating. In the more acid media they are removed wholly and, in some slightly corrosive products such as milk, the films break down locally where the surface has been slightly damaged in can manufacture and unsightly local corrosion then occurs. Although lacquering is used increasingly for can interiors, there are advantages in cost and in preserving the flavour and colour of some products for the use of plain tinplate. With plain cans, deferment of serious hydrogen evolution can be obtained by increasing the thickness of the tin coating but the preferred method is to control characteristics of the coating and steel base in manufacture, checking achievement by suitable tests. Control measures in use are:
- ”.
13:60
TIN AND TIN ALLOY COATINGS
1. To limit the content in the steel of phosphorus, sulphur and ‘tramp’ elements such as copper, nickel and chromium’’. 2. To avoid the slight oxidation of the steel surface during the brightannealing process that precedes tinning19*” The harmful, so called ‘pickle-lag layer’ so produced is detected by its resistance to 10 N HCI. 4. To limit the total porosity of the coating, checking by the Iron Solution Value (ISV)test in which samples are immersed under standard conditions in a solution of sulphuric acid, hydrogen peroxide and ammonium thiocyanate, and the amount of iron dissolved is measured19. 5 . To ensure maximum continuity of the tin-iron compound layer between tin and steel. This layer is itself corrosion resistant and appears to act as a nearly inert screen limiting the area of steel exposed as tin is removed by corrosion. Its effectiveness is measured by the Alloy-Tin Couple (A.T.C.) test, in which the current flowing is measured between a sample of tinplate from which the unalloyed tin layer has been removed, and a relatively large tin electrode immersed in an anaerobic fruit Tinplate that meets the rigid specifications imposed by these controls is sometimes supplied as special quality material and undoubtedly can give improved shelf-life, particularly with citrus fruits. The A.T.C. value has probably more effect on shelf-life determined by hydrogen swell than any other factor. A limited degree of control over the corrosivity of the product packed is possible. Minor pH adjustments may be helpful, especially in ensuring an anodic relation of tin to steel; corrosion promoters, like nitrate, sulphur and copper may be excluded from necessary additives, such as water and sugar, and from sprays applied to crops approaching harvest. The effect of sulphur compounds which may remain from spray residues is complexz4but often includes reversal of the tin-iron polarity. The use of lacquered tinplate does not automatically guarantee freedom from serious corrosion. The covering of the tin surface largely denies both corrosion inhibition by dissolved tin and cathodic protection to any steel exposed at coating discontinuities. Consequently, if discontinuities exist, perforation of cans and hydrogen swells are possible. Lacquer is applied to the tinplate before it is made into cans so that there is a risk, especially at seam areas, of scratching through or cracking of the coating. The dangers are minimised by suitable choice of lacquer and, for critical packs, by double coating and by applying a stripe of lacquer to the seam after can manufacture. The tin coating is not entirely without influence and coating thicknesses In general, the coating properties may still influence found desirable for the plain can are not likely to be so important for the lacquered can, although steel quality remains an important factor. Requirements for cans for beer and soft drinks differ from those for food cans in that (a) only low tin and iron contents can be tolerated in the product and (b) the anticipated shelf-lives are much shorter. Specialised lacquering techniques including striping the seams are used to give complete cover to the metal. For soft drinks it is sometimes possible to select colouring matters and acids least likely to give rise to corrosion troubles, and rapid methods of testing formulations have been devised”. Steel quality is also controlled by special tests.
.
TIN AND TIN ALLOY COATINGS
13:61
Tinned copper and copper alloys Copper itself has a fair corrosion resistance but traces of copper salts are often troublesome and a tin coating offers a convenient means of preventing their formation. Thus copper wire to receive rubber insulation is tinned to preserve the copper from sulphide tarnish and the rubber from copper-catalysed oxidation, and also to keep the wire easily solderable. Vessels to contain water or foodstuffs, including cooking vessels, water-heaters and heat exchangers, may all be tinned to avoid copper contamination accompanied by possible catalysis of the oxidation of such products as milk, and discolouration in the form of, for example, green stains in water and food. Tin is anodic to copper in water supplies and in all solutions except those in which copper is dissolved as a complex, e.g. strong ammonia solutions. In water supplies the corrosion of the tin coating is, like that of tin, localised, but once the copper is reached it may spread slowly. This simple behaviour can, however, be considerably altered by the action of tin-copper compound layers in the coating. A hot-dipped or wiped coating will have from the outset a layer of Cu,Sn, and perhaps also another, nearer the copper, of Cu,Sn. Even an electrolytic coating will in time develop a compound layer by diffusion, at a rate depending on temperature; in boiling water the formation of the compound consumes about 2 - 5 pm of the coating per year. The compounds are always cathodic to tin; in a wiped coating, which usually has streaks of compound in the surface, this has the effect of increasing the extent of local corrosion of tin with the production of unsightly black streaks. In addition, the compound Cu,Sn, can be cathodic to copper; this behaviour is favoured by mild oxidising conditions, which ennoble the compound, and water movement, which anodically depolarises the copper. So long as some tin coating remains it will protect the copper, and a complete coating of compound is protective, but if all the coating is converted to compound and if there is a break in it which exposes copper, then pitting can occur. Adequately thick tin coatings and re-tinning of equipment when necessary are the proper safeguards. The unfortunate action of the compound layer is observed only rarely, usually in hot water. In cooking vessels (domestic or industrial) the copper is protected satisfactorily at some sacrifice of tin, and occasional re-tinning ensures long service. In atmospheric corrosion the arrival of compounds at the surface of the coating results in some darkening and in loss of solderability. With tin coatings on brass, the interdiffusion of coating and substrate brings zinc to the surface of the tin: the action can be rapid even with electrodeposited coatings. The effect of zinc in the surface layers is to reduce the resistance of the coating to dulling in humid atmospheres, and the layer of zinc corrosion product formed makes soldering more difficult. An intermediate layer of copper or nickel between brass and tin restrains this interdiffusion*'. Since galvanic action (Section 1.7) between tin and aluminium alloys is slight, tin coatings are often applied to copper and copper alloys which are to be used in contact with these metals. Both direct galvanic action and corrosion resulting from copper dissolving and re-depositing on the light alloy are prevented by this means.
13 :62
TIN AND TIN ALLOY COATINGS
Applications of tin coatings The properties of tin coatings which are advantageous in most of their applications are: fair general resistance to corrosion except in strongly alkaline or acid environments, lack of colour, toxicity or catalytic activity of any corrosion product formed, and ease of soldering. The ready availability of coated steel sheets in the form of tinplate, which has a bright appearance and is easy to form and receptive to decoration and protective finishes, is also an advantage. The main application is in the form of tinplate. Apart from its use for containers mentioned earlier, tinplate is made into domestic and industrial kitchen equipment, light engineering products and toys. For most of these purposes, coatings in the thickness range 0.4-2-5 pm, with or without organic finishes, are used. For returnable containers and more permanent articles such as fuel tanks and gas-meter cases, heavier coatings of up to 15 pm may be necessary. Hot-dipped and electrolytic coatings are applied to vessels and equipment made of steel, cast iron, copper or copper alloys for use in the food industry, and to wire and components for the electrical and electronics industries, where ease of soldering is an essential property. Although tin coatings are not immune from damage by fretting corrosion, and fretting between tinplate sheets in transit sometimes produces patterns of black spots, tin coatings may be used to reduce the risk of fretting damage in press fits and splined joints of steel component^^^. The coating packs the joint and any movement takes place within the coating. An allied application is the tinning of aluminium alloy or iron pistons to provide a suitable working surface during the running-in period3'. Sprayed coatings find a use in large vessels and some equipment used in the food industry. The necessity for these coatings to be thick enough to be pore-free has already been mentioned. As a general guide to the thickness of coating desirable for various applications, the requirements of BS 1872: 1964 for electrodeposited tin coatings are shown in Tables 13.12 and 13.13. For many purposes involving contact with food and water, coatings Table 13.12 Thickness suggested for electrodeposited tin coatings on ferrous components Purpose
Contact with food or water where a complete cover of tin has to be maintained against corrosion and abrasion Protection in atmosphere Protection in moderate atmospheric conditions with only occasional condensation of moisture To provide solderability and protection in mild atmospheres Coatings flow-brightened by fusion (solderability and protection in mild atmospheres)
Minimum local thickness ( pm)
30 20 10
5
2.5 (maximum 8)
TIN AND TIN ALLOY COATINGS
13 :63
Table 13.13 Thickness suggested for electrodeposited tin coatings on copper and copper alloys with at least 50% copper
Purpose Contact with food or water where a complete cover of tin has to be maintained against corrosion and abrasion Protection in atmosphere and in less aggressive immersion conditions To provide solderability and protection in mild atmospheric conditions Coatings flow-brightenedby fusion (solderability and protection in mild atmospheres)
Minimum local thickness (ctm)
30 15
5, 2.5.
(maximum 8)
*On brass an undercoat of copper. nickel or bronze of thickness 2.5 rm is required.
thinner than those specified in the first category are sometimes sufficient; much depends on the expected amount of abrasion, or loss during cleaning processes. Hot-dipped coatings in the usual thickness range of 10-25 pm give good protection to water-heaters, dairy equipment and much industrial plant, and the thinner coatings of the tinplates in common use are usually sufficient with proper care to preserve appearance in storage and transport. On the other hand, on copper for hot-water service it may sometimes be desirable to use coatings thicker than those recommended in view of the risk in interdiffusion between tin and copper.
Tin Alloy Coatings Tin-lead
Tin-lead coatings with upwards of 5% lead may be applied by hot dipping to steel, copper and copper alloys. Steel sheets are commonly coated with alloys containing 7%, 10% or 25% tin; these are called terne-plate, with the name tin-terne sometimes applied to the higher tin-content coating. Tin-lead alloys may also be electrodeposited from a fluoroborate solution containing organic addition agents and bright deposits are possible. These alloy coatings have advantages over tin in atmospheric exposure where there is heavy pollution by oxides of sulphur. They are cathodic to steel and anodic to copper. In industrial atmospheres, however, formation of a layer of lead sulphate seals pores and produces a generally stable surface' and terne-plate has been used extensively as roofing sheet, especially in the USA. It is easily and effectivelypainted when additional protection is required. Copper heat exchangers in gas-fired water-heaters may be coated by hot dipping in 20% tin alloy". Tin-lead alloy coatings have some of the susceptibilityof lead to vapours
13:64
TIN AND TIN ALLOY COATINGS
of organic acids such as acetic acid, and may be attacked by vapours from wood and insulating materials when enclosed in wooden cases or in electrical apparatus. They are, however, widely and successfully used as protective and easily solderable coatings on wire, electronic components and printed circuit boards. Tin-lead can be substituted for tin for other purposes, although the toxicity of lead limits the field of application. The corrosion resistance is usually no better than that of unalloyed tin, but there may be some saving of cost in applications such as wash-boilersand other vessels for non-potable liquids and light engineering components formed from sheet metal. Heavily coated terne-plates may be used for the fuel tanks of stoves and vehicles.
Tin-zinc
Tin-zinc alloys of a wide range of composition can be electrodeposited from sodium stannate/zinc cyanide baths; only the coatings with 20-25% zinc have commercial There is no intermetallic compound formation and the electrodeposit behaves as a simple mixture of the two metals. It can be considered as basically a stable wick of tin through which zinc is fed to be consumed at a rate lower than its consumption from a wholly zinc surface. If the conditions are such that zinc is rapidly consumed, and no protective layer of corrosion products is formed, the coating may break down, but in mildly corrosive conditions some of the benefits of a zinc coating, without some of its disadvantages, are obtained. In condensed moisture, there is sufficient corrosion of zinc to give protection at pores in a coating on steel without the formation of as much zinc corrosion product as would develop on a wholly zinc surface. In solderability the coating is tin-like when new or stored dry, but the selective corrosion of zinc in humid conditions may produce a layer obstructive to easy soldering. In full weathering in industrial areas, the zinc is taken from the coating too quickly and the alloy coatings do not endure as long as either zinc or tin coatings of comparable thickness; they may, however, outlast cadmium 34. By the sea, the alloy coatings are somewhat better, and in more continuously wet conditions, such as at half-tide positions, they may outlast zinc coatings; possibly here the corrosion product is protective. It is, however, in sheltered conditions and special environments that tin-zinc is most useful. Its easy solderability combined with protection at pores makes it applicable in electrical and radio equipment and in components of tools and mechanisms. It is also used on the bodies of water-containing fire extinguishers, and on components exposed to hydraulic fluids. The coating is, in addition, useful in preventing galvanic corr~sion’~. Plated on steel which is to be used in contact with aluminium alloys, it protects the steel and does not stimulate the corrosion of the light alloy and is itself not consumed as rapidly as a 100% zinc coating.
TIN AND TIN ALLOY COATINGS
13 :65
Tin-cedmium
Tin-cadmium alloys of a range of compositions can be deposited from stannate/cyanide solutions or fluoride/fluorosilicate solutions 36. The behaviour of the coatings is rather similar to that of tin-zinc, but as cadmium is less effective than zinc in giving cathodic protection to steel, a 25% cadmium coating is barely able to protect pores and a 50% content is better for this purpose. The coatings in some conditions form an extremely dense layer of corrosion product, and give an outstanding performance in laboratory salt-spray tests”, but there has been no substantial practical application. Coatings of tin over cadmium, which combine an inert outer surface with protection from rusting at pores, have been used on containers of solvents and to protect electrical components against organic vapour corrosion.
Tin-copper
Tin-copper alloys may be electrodeposited from copper cyanide/sodium stannate baths3*or from cyanide/pyrophosphate baths” to give a range of compositions. Alloys with 10-20% tin have a pleasant golden colour but are not tarnish-resistant unless coated with lacquer. The alloy with 42% tin known as speculum is silver-like in colour and is resistant to some forms of corrosion. At this composition the deposit is formed as the intermetallic compound Cu,Sn. It has a useful hardness (about 520H,). The deposit becomes dull on exposure to atmospheres containing appreciable amounts of sulphur dioxide, but resists hydrogen sulphide, and remains bright in the more usual indoor atmospheres. Although out of doors it becomes dull and grey if not cleaned frequently, the coating is very suitable for metalwork used indoors; it resists the action of most foodstuffs and is suitable for tableware. Like many intermetallic compounds, the deposit shows a corroding potential which becomes increasingly noble with duration of immersion in electrolyte. It is strongly cathodic to steel, and pore-free deposits are desirable. Recommended minimum thicknesses are 12 pm on brass, copper, nickel silver, etc. and 25 pm on steel. The fact that the composition of the speculum deposit must be closely controlled to obtain the best results has been a serious drawback to development. The coating finds uses on decorative hollow-ware, oil lamps and tableware. The bronze deposits with 10 or 20% tin are used lacquered in decorative metal-ware for domestic and personal ornament and, in thick layers to protect hydraulic pit props against corrosion and abrasion. They have also been used with success as undercoatings for nickel-chromium 39*40 or tin-nickel alloy deposits.
Tin-nickel
Tin-nickel alloy coatings are deposited from a bath containing stannous chloride, nickel chloride, ammonium bifluoride and a m m ~ n i a ~ ’The .~~. useful deposit contains 65% tin and the conditions are maintained to obtain
13:66
TIN AND TIN ALLOY COATINGS
this composition only; control is, fortunately, easy. A special feature of the process is the good throw of deposit into recesses. The deposit plates out as the intermetallic compound NiSn, which is white with a faintly pink tinge, and has a hardness of about 710H,. Deposits from new baths are usually in tensile stress but those from baths used for some time are in compressive stress; the stress can be controlled if desired by adjustment of solution c o m p ~ s i t i o n ~The ~ . properties of the intermetallic compound differ from those of both the constituent metals. It is easily passivated, resisting concentrated nitric acid and becoming considerably ennobled during immersion in solutions of neutral salts, including sea-water. In a wide range of solutions, the potential of NiSn with reference to the standard hydrogen electrode was, on immediate immersion, +0-330-055 pH and, after some hours, +0*59-0.056pH4. Higher potentials are reached after long immersion or in oxidising conditions, but ennoblement occurs in solutions with extremely low oxygen concentrations; evidence for five oxidation states for the surface film has been obtained, one at the low potential of -0-42-0-06 pH45. The film thickening that accompanies this change to a more noble potential may become visible, and in hot water or steam a purple film may be produced. The deposit resists atmospheric tarnish even in the presence of high pollution by sulphur dioxide (in contrast to nickel) and hydrogen sulphide, and coatings exposed to the outdoor atmosphere remain bright indefinitely, sometimes taking on a slightly more pink colour as the oxide film thickens. The passivity at pH values above about 1 5 is maintained in a great variety of solutions, including fruit juices, vinegar, sea-water, alkalis, and even ferric chloridea. Hot caustic alkali solutions above about 10% attack the coating slowly, and the halogens etch it. The nobility of the coating brings with it the handicap that corrosion of base metal exposed at pores is stimulated. In an electrolyte of good conductivity, steel, brass or copper are attacked freely at pore sites; steel plates 1 mm thick were perforated after 12 months in the sea. In the outdoor atmosphere, the rate of penetration of the basis metal is slow, but disfigurement by the appearance of corrosion products at pore sites may O C C U ~ ~ , ~ ’ . Since the coating itself is not attacked, new pores do not develop during atmospheric exposure, so that the risks of corrosion at pores can be mitigated by attention to the original condition of the coating. Deposits more than 30 pm thick will usually be pore-free, and for deposits on steel for outdoor exposure an undercoating of copper is decidedly advantageo~s~’*~~. The copper undercoat, preferably about 12 pm thick, reduces the number of pores penetrating from the surface to the steel, and in industrial atmospheres tends to reduce corrosion at such pores as remain. Tin or tin-copper alloy undercoats may also be used and in marine environments are somewhat better than copper. Indoors, pore corrosion is troublesome only if there are prolonged periods of wetting by condensed moisture, and coating thicknesses may safely be much less than those desirable out of doors. The coating will not, however, withstand much deformation, and even with the thinner coatings plating should, if possible, be carried out after all forming operations are complete. The application of the tin-nickel coating for out of doors service has been restricted by fear of pore corrosion and of physical damage, and by
-
TIN AND TIN ALLOY COATINGS
13 :67
the poor colour match with chromium. For indoor use, the coating has many applications, e.g. laboratory instruments, balance weights, the valves of wind instruments, internal mechanism of watches, electrical instruments, lighting fittings, interiors of cooking vessels and decorative hollow-ware. Many of these are special applications in which, in addition to corrosion resistance, the hard, smooth surface, non-magnetic quality and the good covering power in deposition of the coatings, may have been required. These qualities have also lead to its use on printed circuit boards and on electrical connectors, although the persistent oxide film obstructs easy solderability and produces too high a contact resistance for satisfactory switching at low voltages.
Recent Developments Tin Coatiqs Recent has shown that tin may be deposited by an autocatalytic process using transition-metal ion reducing agents. Very thick coatings may also be economically applied to a variety of substrates by the process of roll bondings3. Tinplate still represents the largest use for tin, but continuing developments in can-making technology mean that coatings as thin as 0.1 pm are in use54.These may be non-reflowed, reflowed to produce a duplex tin and tin-iron alloy coating, or reflowed to convert all of the tin coating to tin-iron alloy. These products are almost exclusively used in the lacquered condition but the presence of tin still plays a significant role in controlling the corrosion of the steel basis material. In some cases, the properties of the coatings are modified significantly by the application of a passivation film consisting of a mixture of chromium metal and chromium oxides and much heavier than that used on tinplate with thicker tin coatings. The shelf-life of containers made from unlacquered tinplate is now dictated by national and international regulations governing the permitted tin content of foods. Since the onset of hydrogen swells is usually during the later stages of plate detinning during service, the value of the A.T.C. test in predicting container shelf-life is severely limited. General thickness requirements for electroplated tin coatings on ferrous and non-ferrous substrates are contained in BS 1872: 1984 and I S 0 2093 and these are essentially the same as those in Tables 13.12 and 13.13. Tin Alloy Coatings
The corrosion resistance of tin-lead alloy coatings on copper and copper alloys is directly relevant to their use in the electronics industry. The solderand ability of coatings as a function of storage time has been accelerated ageing techniques have been compared 56. The electrical contact although resistance of tin-lead coatings increases in some the film of corrosion products is easily ruptured when contacts are maetd@ .' General requirements for tin-lead coatings are contained in BS 6137: 1982.
13 :68
TIN AND TIN ALLOY COATINGS
There has been some renewed interest in the use of tin-zinc alloy electroIt has been pointed plate as a substitute for cadmium coatings on steel61s62. out that tin-zinc coatings produce less loose corrosion product than zinc during full outdoor exposure63. In view of the difficultiesin controlling the electroplating of speculum and bronze coatings, alternative preparation routes through the heat treatment of duplex tin and copper electroplated finishes have been proposedM. The resistance of tin-nickel alloy electroplate to corrosion has been the subject of recent studies using surface analytical techniques. Workers generally agree that the surface of NiSn electroplate exposed to the atmosphere is enriched with tin and that this probably also applies to other nickel-tin intermetallic compounds 70. The interaction of coatings of NiSn, Ni3Sn, and Ni,Sn, with SO,, H,S, NH,, NO,, sulphur vapour, salt-spray and synthetic dust have been monitored with particular reference to electrical resistance7'. Changes seen have been related to the structure of the passivation film7,, and the advantages in using tin-nickel alloy electroplate as an undercoat for the very thin gold deposits used in electrical contacts have been described73.The requirements for tin-nickel alloy electroplate are contained in BS 3597: 1984. Electroplated coatings based on deposits containing tin and cobalt are now available as substitutes for chromium plating. Deposits corresponding to COS^^^"' or CoSn mixed with CoSni6may be obtained which show a good colour match with chromium and a number of proprietary processes have been patented. Most studies of the corrosion of tin-cobalt alloy deposits have been concerned with the performance of thin coatings (0-5 pm) over nickel. It has been noted77 that the coating performs well in salt-spray and CASS tests and that it resists ammonia: comparisons have been made with a nickel-chromium finish and it was found that tin-cobalt alloy on nickel deposits perform as well as the conventional coating in all but the most severe exposure conditions 78,79, While the ductility of tin-cobalt coatings is greater than that for tin-nickel deposits7', the corrosion resistance of each finish is very similars0. As with tin-nickel, it has been shown that tin-cobalt deposits have surface enrichment by tin oxides". S.C. BRITTON
REFERENCES 1. Instructions for Electrodepositing Tin, Tin Research Institute, Greenford (1971) 2. Thwaites, C. J., The Flow-melting of Electrodeposited Tin Coatings, Tin Research Institute, Greenford (1959) 3. Thwaites, C. J. Hot-tinning, Tin Research Institute, Greenford (1965) 4. Willey, A. R., Br. Corros. J., 7, 29 (1972) 5 . Britten. S. C. and Angles, R. M., J . Appl. Chem., 4, 351 (1954) 6. Hudson, J. C. and Banfield, T. A., J . Iron St. Inst., 154, 229P (1946) 7. Britton, S.C. and Angles, R. M.. Metullurgiu, Munchr., 44.185 (1951) 8 . Middlehurst J. and Kefford. J. F.. Condemurion in Curgoes of Conned Foods, CSIRO, Tech. Paper No. 34 (1968)and Poc. Conference on the Protection of Metal in Storuge and in Transit, London, Brintex Exhibitions Ltd., 83 (1970) 9. Rocquet, P.and Aubrun, P., Br. Corros. J . , 5 , 193 (1970) 10. Hoar, T. P. and Havenhand, D., J . Iron St. Inst., 133, 253 (1936) 11. Buck, W. R. and Leidheiser, H. J., J . Electrochem. SOC., 108, 203 (1961) 12. Koehler, E. L., J . Electrothem. Soc., 103, 486 (1956)and Corrosion, 17, 93t (1961)
TIN AND TIN ALLOY COATINGS
13 :69
13. Liebmann, H., Proceedings 3- CongrPs International de la Conserve, Rome-Parmag 1956, ComitC International Permanent de la Conserve, Paris. 133 (1956) 14. Frankenthal, R. P., Carter, P. R. and Laubscher, A. N., J. Agr. Food Chem., 7, 441 (1959) 15. Cheftel, H., Monvoisin, J. and Swirski, M., J. Sci. Fd. Agric., 6 , 652 (1955) 16. Dickinson, D., J. Sci. Fd. Agric., 8, 721 (1957) 17. Strodtz, N. H. and Henry, R. E., Food Techn., 8, 93 (1954) 18. Hartwell, R. R., Surface Treatment of Metuls, Amer. SOC. Metals, Clweland, Ohio, 69 (1941) 19. Willey, A. R.. Krickl, J. L. and Hartwell, R. R.. Corrosion, 12, 433t (1956) 20. Koehier, E. L.. TMW. Amer. Soc. Metals, 44, 1 076 (1952) 21. Kamm, G. G. and Willey, A. R., Corrosion, 17, 77t (1961) 22. Kamm, G. G.. Willey, A. R., Beese, R. E. and Krickl, J. L., Corrosion, 17, 84t (1961) 23. Carter, P. R. and Butler, T. J., Corrosion, 17, 72t (1961) 24. Board, P. W., Holland, R. V. and Elbourne, R. G. P.. J . Sci. Fd. Agric., 18, 232 (1967) 25. Salt, F. W. and Thomas, J. G. N.. J . Iron St. Inst.. 188, 36 (1958) 26. Koehler. E. L.. Werht. u. Korrosion. 21, 354 (1970) 27. Koehier. E. L., Daly, J. J., Francis, H. T. and Johnson, H. T., Corrosion, 15.477t (1959) 28. Britton, S. C. and Clarke, M., Truns. Inst-Merul Finishing, 40, 205 (1963) 29. Wright, O., J. Electrodep. Tech. Soc., 25, 51 (1949) 30. Smith, D. M., Truns. SOC. Automot. Engrs., N . Y.. 53, 521 (1945) 31. Kerr, R. and Withers, S. M., J. Inst. Fuel., 22, 204 (1949) 32. Angles, R. M., J. Electrodep. Tech. SOC.,21, 45 (1946) 33. Cuthbertson, J. W. and Angles, R. M.. J. Electrochem. Soc.. 94.73 (1948) 34. Britton. S. C. and Angles, R. M.. Metallurgiu. Manchr.. 44, 185 (1951) 35. Britton, S. C. and de Vere Stacpoole. R. W., Metullurgio, Munchr., 52, 64 (1955) 36. Davies, A. E., Truns. Inst. Mer. Finishing, 33, 75, 85 (1956) 37. Britton, S. C. and de Vere Stacpoole, R. W., Truns. Insr. Met. Finishing, 32, 211 (1955) 38. Angles, R. M., Jones, F. V., Price, J. W. and Cuthbertson, J. W., J . Elecrrodep. Tech. Soc., 21, 19 (1946) 39. Safranek, W. H. and Faust, C. L.. Plating, 41, 1159 (1954) 40. Chadwick. J., Electropluting, 6, 451 (1953) 41. Parkinson, N., J. Electrodep. Tech. Soc.,27, 129 (1951) 42. Davies, A. E.. Truns. Insr. Mer. Finishing. 31, 401 (1954) 43. Clarke, M., Trans. Inst. Met. Finishing, 38, 186 (1961) 44. Clarke, M. and Britton, S. C., Corrosion Science, 3, 207 (1963) 45. Clarke, M. and Elbourne, R. G. P., Corrosion Science, 8, 29 (1968) 46. Britton, S. C. and Angles, R. M., J. Electrodep. Tech. Soc., 27, 293 (1951) 47. Britton, S. C. and Angles, R. M.. Truns. Insr. Mer. Finishing, 29, 26 (1953) 48. Lowenheim, F. A.. Sellers, W. W.and Carlin. F. X.,J . Electrochem. SOC., 105,339 (1958) 49. Britton, S. C.. Trans. Inst. Met. Finishing, 52, 95 (1974) 50. Molenaar. A., Heynent, G. H. C. and van der Meerakker. J. E. A. M., Phorogruphic Sci. und Eng., 20 No. 3, 135 (1976) 51. Jonker, H., Molenaar, A, and Dippel, C. J., Photographic Sci. and Eng., 13 No. 2, 38 ( 1969) 52. Warwick, M. E. and Shirley, B. J., Truns. Inst. Metul Finishing, 58, 9 (1980) 53. Butlin. 1. J. and MacKay, C. A.. Sheer Metul Ind., 56 No. 11. 1063 (1979) 54, Proc. 3rd Int. Tinplute Conf., London, Int. Tin Research Institute. Session 4 (1984) 55. Thwaites. C. J., Soft-Soldering Handbook, Int. Tin Research Inst., London, Publication No. 533 (1977) 56. Ackroyd, M. L., A Survey of Accelerated Ageing Techniques for Solderuble Substrures, Int. Tin Research Inst., London, Publication No. 531 (1976) 57. Antler, M., Graddick, W. F. and Tompkins, H. G., Proc. Holm Serninur on Electricol Contucrs, p. 3 1 (1974) 58. Hyland. A. J. and Dewolff, W. S., Proc, 6th Ann, Connector Symp., New Jersey, p. 59 (1973) 59. Waine. C. A.. Peddar, D.J.. Lewis J. C. and Souter. J. W., Proc. Holm Seminur on Electricul Conructs, p. 2 13 (1980) 60, Waine, C. A. and Sollars, P. M.A., Electr. Contucrs, 24, 159 (1978) 61. Raub., J.., Pfeiffer. W.and Vetter. M.. Galuunorechnik, 70 No. 1, 7 (1979) 62: Koeppen, H. J. and Runge, E., Gulvunotechnik, 73 No. 11, 1217 (1982)
13:70
TIN AND TIN ALLOY COATINGS
63. Phillips. S. L. and Johnson, C. E., J. Electrochem. Soc., 117. 827 (1970) 64. Denman, R. D.and Thwaites, C. J., Mer. Techn., 11 No. 8, 334 (1984) 65. Hoar, T. P., Taleran, M. and Trad, E., Nature (London) Phys. Sci., 244 No. 133, 41 (1973) 66. Tompkins. H. G. and Bennett, J. E.,J. Electrochem. Soc.,123 No. 7, 1003 (1976) 67. Thomas, G. H. and Sharma, S. P.,J. Vuc. Sci., Technol., 14 No. 5, 1168 (1977) 68. Sharma, S. P. and Thomas, G. H.. Anal. Chem.. 49 No. 7, 987 (1977) 69. Tompkins, H.G.. Wertheim, G. K. and Sharma. S. P., J. Vac. Sci., Technol., 15, 20 (1978) 70. Tompkins, H. G. and Bennett, J. E., J. Electrochem. Soc., 124 No. 4, 621 (1977) 71. Antler, M.. Feder, M.,Hornig. C. F. and Bohland. J., Pluring cmd Sur. Fin., 63 No. 7, 3 (1976) 72. Antler, M., Proc. Conf. on Corrosion Control by Coatings, Lehigh Univ., Bethlehem, Pennsylvania (1978) 73. Cowieson, D. R. and Warwick, M. E., Proc, Holm Seminar on Electrical Contacts. p. 53 (1 982) 74. Sree. V. and Rama Char, T. L.. Metallobefluche. 15, 301 (1961) 75. Clarke, M.,Elbourne, R.G. and MacKay, C. A., Trans. Inst. Metal Finishing, 50 No. 4, 160 (1972) 76. Clarke, M. and Elbourne. R. G., Electrochim. Actu.. 16 No. I1 1949 (1971) 17 Miyashita, H.and Kurihara, S., J. Met. Fin. SOC. Japan, 21, 79 (1970) 78. Hyner, J., Plating and Sur. Fin., 64 No. 2, 33 (1977) 79. Hemslev. J. D. C. and Rorxr. M. E., Truns. Inst. Metal Finishing, 57 No. 2, 77 (1979) 80. Tsuji, Y i and Ichikawa, M., Corrosion-NACE, 27 No. 4, 168 (1971) 81. Thomas, J. H. and Sharma, S. P.,J. Vac. Sci., Technol., 15 No. 5 , 1706 (1978) I
13.6 Copper and Copper Alloy Coatings Copper coatings are usually applied by electrodeposition (Section 12. l), although for more limited purposes ‘electroless’ or immersion deposits are used. Less frequently, copper may also be applied by flame spraying’.
Applications Copper deposits are applied predominantly for the following purposes: 1. As an undercoat for other metal coatings. The main use of copper plating is as an undercoating prior to nickel-chromium plating steel and zinc-base die castings. On steel, the primary purpose is to reduce polishing costs. Other advantages are that with a copper-plated undercoating, cleaning is less critical for achieving a well-adherent nickel deposit and metal distribution is frequently improved. Nickel-chromium plating standards of most countries permit some part of the nickel thickness to be replaced by coppe?v3. On zinc-base die castings a copper undercoat is almost universally used, as an adherent nickel deposit cannot be deposited directly from conventional baths. For a similar reason copper is deposited on aluminium which has been given an immersion zinc deposit4 before nickel plating is applied. Under micro-discontinuous chromium coatings, copper undercoats improve corrosion resistance. On non-conductors, especially on plastic substrates, copper is often applied before nickel-chromium plating over the initial ‘electroless’copper or nickel deposit in order to improve ductility and adhesion, e.g. as tested by the standard thermal-cycling test methods5. 2. As a decorative finish on steel and zinc-base alloys for a variety of domestic and ornamental articles. The finish may be protected by clear lacquers or may be coloured by metal colouring techniques for use on, for example, door handles, luggage trim, etc. 3. As a ‘stop-off‘ for nitriding or carburising of steel. The 10-40pm deposits, which are electroplated on selected areas, are removed after the heat treatment. 4. For protection of engineering parts against fretting corrosion, on electrical cables and on printing cylinders. Temporary protection allied with lubrication is provided by immersion deposits of copper on steel wire. 5. Chemical deposits of copper are applied to provide conducting surfaces on non-metallic materials. 6. Copper is plated on printed circuit boards to provide electrical conductors and for a variety of other electrical and electronic applications6. 13:71
13:72
COPPER AND COPPER ALLOY COATINGS
Plating Solutions Copper is electrodeposited commercially mainly from cyanide, sulphate and pyrophosphate baths. For rapid deposition in electro-forming, a fluoborate bath may also be used. The sulphate bath The sulphate bath, the earliest of electroplating solutions and the simplest in composition, contains typically 150-250 g/l of copper sulphate and 40-120g/l of sulphuric acid. The composition is not critical and the higher concentrations are used for plating at higher current densities, normally up to 6 A/dm2. Addition agents used to produce smooth and fine-grained (though dull) deposits include gelatin, glue, phenol sulphonic acid, hydroxylamine and triethanolamine. These are believed to inhibit crystal growth by forming colloids in the cathode layer, and, in some cases, to change the crystallographic orientation. Modern bright acid copper plating baths contain both organic and inorganic addition agents which act as brighteners and levellers. The two functions are largely distinct, the latter being the more important when copper is plated as an undercoat for decorative nickel-chromium coatings. Additives of this type include organic sulphur compounds, e.g. thiourea derivatives. Such solutions are sensitive to the chloride ion concentration which must be maintained at a low level. On ferrous metals immersion deposition in the copper sulphate bath produces non-adherent deposits, and a cyanide copper undercoat is therefore normally used. Where the use of a cyanide strike cannot be tolerated, an electroplated or immersion nickel deposit has been used Additions of surface-active agents, often preceded by a sulphuric acid pickle containing the same compound, form the basis of recent methods for plating from a copper sulphate bath directly on to steel9-”. While the sulphate bath has a high plating speed, its throwing power is poor, and this limits its application to articles of simple shapes. ’s8.
Cyanide baths Most general copper plating, other than that applied, for example, to wire and strip or for electroforming, is carried out in a cyanide bath. Its main advantages are (a)that it can be used to plate directly onto steel and zinc-base alloys, and (b) that it has good throwing power, which renders it suitable for plating a large variety of shapes. Modern solutions fall mainly into three types: (a) the plain cyanide bath which contains typically 20-25 g/l of copper cyanide, 25-30 g/l total sodium cyanide (6.2g/l ‘free’ sodium cyanide), and is operated at 21-38°C and 110-160A/m2; (b) the ‘Rochelle’ copper bath to which is added 35-50g/l of Rochelle salt and which is used at 66°C at up to 645 A/m2; and (c) the high-efficiency cyanide baths which may contain up to 125g/l of copper cyanide, 6-1 1 g/l of ‘free’ sodium or potassium cyanide, 15-30 g/l of sodium or potassium hydroxide, and are operated at up to 6-9A/dm2 and 6590°C. Most bright cyanide copper baths are of the high-efficiency type and, in addition, contain one or more of the many patented brightening and levelling agents available. Periodic reverse (p.r.) current is also sometimes used to produce smoother deposits. Plating speeds for the high-efficiency baths are high, partly because higher current densities can be used without ‘burning’, but mainly because the
COPPER AND COPPER ALLOY COATINGS
13 :73
cathode efficiency of the more concentrated solution is higher at higher current densities (e.g. 90-98% compared to 30-60Vo for the ‘plain’ and ‘Rochelle’ type solutions). However, a more dilute solution must generally be used as a ‘strike’ bath on steel and zinc-base alloys t o avoid immersion deposition.
Pyrophosphate bath The pyrophosphate bath is intermediate in throwing power between the sulphate and cyanide baths. A typical bath contains 80-105 g/l of copper pyrophosphate, 310-375 g/l of potassium pyrophosphate and 25 g/l of potassium citrate, pH 8-7-9.4. Similar baths containing nitrate, ammonia and oxalate are also employed. The solutions are used at 50-60°C with vigorous air agitation when current densities of up to 10A/dm2 are permissible. A proprietary bath is available with excellent brightening and good levelling characteristics. A more dilute strike bath is employed for obtaining the initial deposit on steel, while for strongly recessed parts, e.g. tubular work, an immersion nickel deposit has been used’. A short cyanide copper strike is used before plating on zinc-base die castings. Other electroplating solutions Other solutions ”, which are more rarely used for plating copper, include the fluoborate bath, the amine bath, the sulphamate bath and the alkane sulphonate bath. Chemical deposition Simple immersion deposits of copper may be obtained on iron and steel in a solution containing, for example, 15 g/l of copper sulphate and 8 g/l sulphuric acid, and on zinc-base alloy in a solution containing copper sulphate 300 g/l, tartaric acid 50 g/l and ammonium hydroxide 30ml/l”. Such deposits are thin and porous and are mainly plated for their colour, e.g. for identification, or for their lubricating properties, e.g. in wire drawing. Solutions containing tetrasodium E.D.T.A. have also been used for this purpose and give slightly superior coatings. On non-conductors, copper may be deposited by chemical reduction from a modified Fehling’s solution. Such solutions have gained wide application in the plating of ABS and other plastics which are ‘electrolessly’ copper plated before nickel-chromium plating. Pretreatment of the plastic is important in order to gain adequate adhesion and includes steps for etching the surface as well as for providing a conducting substrate by treatment in stannous chloride and palladium chloride solutions.
Properties of Copper Deposits Deposit uniformity The uniformity of a deposit is an important factor in its overall corrosion resistance and is a function of geometrical factors and the ‘throwing power’ of the plating solution. A distinction is made here between macro-throwing power, which refers to distribution over relatively large-scale profiles, and micro-throwing power, which relates to smaller irregularities 14. The copper cyanide bath has excellent macro-throwing power and is chosen whenever irregular-shaped parts are to be plated. The sulphate bath is not inferior when parts with very narrow recesses, Le. with width of opening less than 6 mm, are to be plated, although its macro-throwing power is
13 :74
COPPER AND COPPER ALLOY COATINGS
poor. Pyrophosphate baths are intermediate between the two in macrothrowing power. Porosity As is the case with all cathodic deposits, the corrosion resistance of a copper deposit is reduced in the presence of continuous porosity. Experience has shown that porosity is least when attention is paid to adequate cleaning, and the solution is kept free from solid or dissolved impurities (see Section 12.1). Porosity of copper deposits is also related to polarisation15. Corrosion resistance The corrosion resistance of a copper deposit varies with the conditions under which it is deposited and may be influenced by co-deposited addition agents (see, for example, Raub 16). Copper is, however, plated as a protective coating only in specialised applications, and the chief interest lies in its behaviour as an undercoating for nickel-chromium on steel and on zinc-base alloy. Its value for this purpose has long been a controversial issue. A thin copper deposit, e.g. 2.5pm, plated between steel and nickel, improves corrosion resistance during outdoor exposure”, and many platers also believe that a copper undercoating improves the covering power of nickel, particularly on rough steel. Where heavier copper coatings are plated as a partial replacement for nickel, as is permitted under most nickel plating specifications, the effects are not clearly established. According to Blum and Hogaboom the protective value of nickel on steel is reduced by the presence of a copper undercoating, but this does not apply when the nickel is chromium plated. This is largely corroborated by subsequent corrosion tests and the detrimental effect in the absence of chromium is probably due to attack on the nickel by the copper corrosion products. In the presence of conventional chromium plate, on the other hand, the fact that statistical evidence on many thousands of chromium-plated motor components has not established any difference in the behaviour of parts in which nickel formed respectively 95-100% and 50% of the copper-nickel coating”, bears out the view that after chromium plating the differences in protective value tend to disappear. Moreover, when, as frequently happens in practice, the copper coating is polished, the protective value of the copper-nickel coating is higher than that of nickel alone, owing to the pore-sealing effect of the polishing operation. The case is different again under micro-discontinuous (Le. micro-cracked or micro-porous) chromium, on which a definite improvement in corrosion resistance can be achieved when copper is present under the nickel coating 21. 22. As an undercoating for chromium, i.e. in place of nickel, copper is not to be recommended. On the other hand, both accelerated and outdoor corrosion tests have shown that a tin-bronze deposit, containing 80-90% copper, is considerably better for this purpose and it has been claimed to be approximately equal to nickel in this respect.
’*
Mechanical properties The hardness and strength of copper deposits may vary widely according to the type of bath used (see Table 13.14). In the
presence of addition agents which decompose in use, the hardness may, moreover, vary appreciably with the age of the bath23. In copper sulphate solutions, hardness and tensile strength are increased
COPPER AND COPPER ALLOY COATINGS
13:75
Table 13.14 Mechanical properties of electrodeposited copper I 2 Plating bath
Sulphate bath Sulphate bath with addition agent Fluoborate bath Cyanide bath Cyanide bath with p.r. current Pyrophosphate bath
Hardness
(H,) 40-65 80- 180 40-75 100-160
150-220 125- 165
Elongation
[Vo on
Tensile strength
50.8 mm (2 in)]
(MN/m2)
20-40 1-20 7-20 9-15 6-9
230-310 480-620 240-275 415-550 690-760 -
-
by raising the current density and reducing the temperature. As will be seen from Table 13.14, particularly high hardness values can be obtained in the cyanide bath by using p.r. current. Annealing of electrodeposited copper reduces the mechanical properties. As an example, the tensile strength has been reported to decrease from 275-330 MN/m2 to 180-255 MN/mz on heating at above 300"CZ4while the hardness of deposits obtained in the presence of addition agents may drop from as high a value as 300 HV to 80 HV after annealing at 200°C. Internal stress of copper deposits may vary between -3.4 MN/mZ (compressive) and 100MN/mz (tensile). In general, tensile stress is considerably lower in deposits from the sulphate bath than in those from cyanide s o l u t i ~ n s ~ while ~ - ~ ~ pyrophosphate , copper deposits give intermediate values. In cyanide solutions, tensile stress increases with metal concentration and temperature decreases if the free cyanide concentration is raised. P.r. current significantly lowers tensile stressz8. With some exceptions, inorganic impurities tend to increase tensile stress". Thiocyanate may produce compressive stress in cyanide bathsz5. In the sulphate bath the tensile stress increases if the temperature is reduced or the current density is increased, and gradually diminishes with increase in deposit thickness2'. Addition of thiourea (1 g/l) or gelatin to the acid bath results in compressively stressed deposits, though at higher concentrations of addition agent this effect may be reversed". Dextrose and gum arabic increase tensile stress3'. The effect of other organic compounds may similarly depend on the operating conditions3z833. The relationship between ductility and stress is complex, e.g. thiourea additions increase ductility over a wide range34. Despite the large differences in respect of other mechanical properties, it has been established3' that the wear resistance of copper deposits, which is markedly inferior to, for example, that of electrodeposited nickel, is not significantly affected by either type of bath or addition agents. Embrittlement by hydrogen absorbed by the substrate during pretreatment, e.g. in acid pickling baths or during plating, is generally important only on copper-plated wire or where copper is plated for lubrication before drawing operations on high-strength steels. For these purposes the acid copper bath is slightly preferable to the cyanide bath. Hydrogen may be removed and ductility restored by heat treatment in air (140-200°C for 0.5-1 h), in water (80-100°C for 0.5-2h), or in oil (175-230°C for 1.5-2h)". Other properties have been comprehensively summarised in the literature
+
13 :76
COPPER AND COPPER ALLOY COATINGS
Copper Alloy Deposits Copper-zinc Copper-zinc alloys are deposited for two main purposes: (a) as a decorative finish, e.g. on steel and (b) as a means of obtaining an
adhesive bond of rubber to other metals. Cyanide solutions are used almost exclusively. One typical solution contains copper cyanide 26 g/l, zinc cyanide 11 g/l, sodium cyanide (total) 45 g/l and sodium cyanide (‘free’) 7 g/1I2. This bath is operated at pH 10.3-1 1.O, 110 A/m2 and 27-35”C, with 75 Cu-25 Zn alloy anodes. Many other solutions are used 12, including a special rubber-bonding bath37and a high-speed bath which is capable of being used at up to 16A/dm2(2*38’. Brass deposits normally contain 70-80% copper and 30-20% zinc; the colour does not normally match solid brass of the same composition and may, moreover, vary with the operating conditions and solution composition. White brass deposits containing 85% zinc and 15% copper have also been plated to a limited extent ”, mainly as an undercoating for chromium during the nickel shortage, but they did not prove fully satisfactory. While brass deposits have a somewhat higher protective value on steel than the equivalent thickness of copper, the deposits tend to tarnish, and when used for decorative purposes bright deposits arc normally protected by a clear lacquer. Although a wide range of copper-zinc alloy deposits can be plateda, most experience has been gained with two compositions, i.e. the red copper-rich tin-bronze which contains %93% copper and 10-7% tin and the white speculum which contains 50-60% copper and 50-40’70 tin. While tin-bronze has been successfully plated as an undercoating for chromium during the nickel ~ h o r t a g e ~ ’its . ~ main ’ use now is as a decorative finish in its own right, because of its pleasing red-gold colour. As in the case of brass, however, the deposits must be protected against tarnishing by a clear lacquer. Speculum deposits are similar in appearance t o silver, but are harder and have good tarnish resistance. Alloys containing only 2% copper and 98% tin are plated on bearing surfaces. Copper-tin deposits can be plated from cyanide or pyrophosphate4’. 49 baths and deposits are of good corrosion resistance (approximately equivalent to the same thickness of nickel). Hardness values of up to 314 Hv are obtainable for the copper-rich alloys45,and up to 530HV for the tin-rich alloys can be obtained. (See also Section 13.5.) Copper-tin
Other alloys Other copper alloys can be plated, including copper-tin*~~, and zinc (Alballoy)&, copper-nickel 47, c o p p e r - ~ a d m i u m ~ *copper-gold copper-lead 50. R. PINNER
REFERENCES 1 . Ballard. W. E., Metal Spraying and. Sprayed Meld, Griffin, London (1948) 2. Electroplated Cootings of Nickel and Chromium,BS 1224 (1970) 3. A.S.T.M. 166
COPPER AND COPPER ALLOY COATINGS
13:77
4. Wernick, S. and Pinner. R., Surface Treatment and Finishing ofAluminium and its Alloys, Draper, Teddington, 2nd edn (1959) 5 . Crouch, P. C., Trans. Inst. Metal Finishing, 49 No. 4, 141 (1971) 6. Saubestre. E. B. and Khera, K. P.. ‘Plating in the Electronics Industry’, Symposium ofthe Am. Electroplaten’ Soc., 230 (1971) 7. Clauss, R. J. and Adamowict, N. C., Plating. SI No. 3, 236 (1970) 8. O’Dell, C. G . . Elecfroploring and Mefal Finishing, 24 No. 7, 14 (1970) 9. Dehydag Deutsche Hydrierwerke, UK Pats. 784091 (1957) and 811 773 (1959); US Pat. 2 903 403 (1959) 10. Antropov, L. I. and Popopov, S. Ya., Zh. Prikl. Khim., Leniner., 27, 5 5 , 527 (1954) 11. Pantshev, B. and Kosarev, C., Metalloberflache, 24 No. IO, 383 (1970) 12. Pinner, R., Copper and Copper Alloy Plating, Copper Development Assoc., London ( 1962) 13. Saubestre, E. B., Proc. Amer. Electropl. Soc., 46, 264 (1959) 14. Raub, E., Metalloberflache, 13 No. 10 (1959) 15. Kovaskii, N. Ya. and Golubev. V. N., Zh. Priklad. Khim.. 43 Nu. 2. 348 (1970) 16. Raub. E., 2. Metallk., 39, 33. 195 (1948) 17. Knapp. B. B. and Wesley, W. A.. Plating. 311, 36 (1951) 18. Blum. W. and Hogaboom, G. B.. Principles of Electroplating and Electroforming, McGraw-Hill, New York, 3rd edn, 136 (1949) 19. Pray, H. A., Rept. of Subcomm, I1 of Comm, B-8, Proc. Amer. SOC. Test. Mat., 49 (1949) 20. Phillips, W. M., Plating, 38, 56 (1951). 21. Turner, P. F. and Miller, A. G. B., Trans. Inst. Metal Finishing, 47 No. 2, 50 (1969) 22. Preprints, Discussion Session, Annual Conf. of the Inst. of Met. Fin. (1972) 23. Spahn, H. and Tippmann, H., Metalloberflache, 13 No. 2, 32 (1959) 24. Prater, T. A. and Read, H. J., Plating, 36, 1221 (1949) 25. Kushner, J. B., Metal Finish., 56 No. 4,46; No. 5,82; No. 6, 56 (1958) 26. Nishiharaud, K. and Tsuda, S., Suyokuro-shi, 12 No. 12,25 (1952) 27. Phillips, W. M. and Clifton. F. L.. Proc. Amer. Electropl. Soc.. 34, 97 (1947) 28. Bachvalov. G.T., All Union .%./Tech. Conference on Corrosion and Protection of Metals. Moscow. 1958 (cf. Plating. 46, I57 (1959)) 29. Fujino. T. and Yamomoto, J.. J. Metal Finishing Soc. Japan, 20 No. I , 18 (1%9) 30. Lizlov, Yo. V. and Samartsev, A. G., All Union SciJTech. Conference on Corrosion and Protection of Metals, Moscow, 1958 (cf. Pfafing, 46, 266 (1959) 31. Graham, A. K. and Lloyd, R., Plating, 35, 449, 506 (1948) 32. Walker, R. and Ward, A., Electrochimica Acta., 15 No. 5 , 673 (1970) 33. Walker, R., Plating, 57 No.6, 610 (1970) 34. Sard, R. and Weil, R., Plating, 56 No. 2, 157 (1969) 35. Ledford, R. F. and Dominik, E. A., Plating, 39. 360 (1952) 36. Lamb, V. A., Johnson, C. E. and Valentine, D. R., J. Electrochem. Soc.,117 No. 9,291C; No. 10, 341C; No. 11, 381C (1970) 37. Compton. K. G., Ehrhardt. R. A. and Bittrich, G., frm. Amer. Electropl. Soc., 41, 267 (1954) 38. Roehl, E. J. and Westbrook, L. R., Proc. Amer. Electropl. Soc.. 42, 3 (1955) and Roehl, E. J.. Electroplating and Metal Finish., 11, 299 (1958) 39. Saltonstall. R. B., Proc. Amer. Electropl. Soc., 39, 67 (1952) 40. Batten, H.M. and Welcome, C. J., US Pats. 1 970 548 and 1 970 549 (1934) 41. Schmerling, G., Electroplating and Metal Finish., 5, 115 (1952) 42. Lee, W. T., Trans. Insf. Metal Finish., 36, 2, 5 1 (1958-1959) 43. Faust, C. H. and Hespenheide, W. G., US Pat 2 658 032 (1953) 44. Safranek, W. H. and Faust, C. L., Proc. Amer. Electropl. SOC., 41, 201 (1954); Plating, 41, 1 159 (1954) 45. Rama Char, T., Elecfroplating and Metul Finish., 10, 347-9 (1957) 46. Diggin. M. B. and Jernstedt. G. W., Proc. Amer. Electropl. Soc., 31, 247 (1944) 47. Priscott, B. H.. Trans. Inst. Metal Finish., 36, 93 (1959) 48. Hogaboom. G. B.. Jr. and Hall, N.. Metal Finishing Guidebook and Directoty. Metal Finishing, Westwood. N. J., USA, 295-298 (1959) 49. German Pat. 876630 (1953) 50. Krasikov, B. S. and Grin, Yu. D., Zh. Prikl. Khim.. Leninger., 32, 387 (1959)
13.7 Nickel Coatings
Nickel coatings have long been applied to substrates of steel, zinc and other metals in order to provide a surface that is resistant to corrosion, erosion and abrasion. Most of the nickel is used as decorative coatings 5-40pm thick, usually under a top coat of chromium about 0-5pm thick so as to give a nontarnishing finish. Such coatings are applied to metal parts on cars, cycles, perambulators and a wide range of consumer items; they have also been applied increasingly to plastic components during recent years in order to Decorative nickel coatings are give an attractive metallic appearance also applied without chromium top coats to products such as spanners, screw-driver blades, keys and can-openers. About 3% of all nickel used in the form of coatings is employed in engineering applications where brightness is rarely needed and the deposits are relatively thick; these coatings are used for new parts and for reclamation. Most nickel electroplating is carried out in solutions based on the mixture of nickel sulphate, nickel chloride and boric acid proposed by 0. P. Watts Typical composition and operating conditions are: ‘9’.
’.
Composition Nickel sulphate (NiSO, -7H,O): 240-300g/l Nickel chloride (NiCl, -6H,0): 40-60 g/l Boric acid (H, BO,): 25-40 g/l Operating conditions Temperature: 25-50°C Air agitation pH: 4.0-5.0 Cathodic current density: 3-1 A/dm2 Mean deposition rate: 40-90 pm/h The Watts solution is a relatively cheap, simple solution which is easy to control and keep pure. The nickel sulphate acts as the main source of nickel ions, though nickel chloride is an additional source. Higher deposition rates can be used when the ratio of nickel chloride to nickel sulphate is raised and some proprietary bright nickel solutions are available in a ‘high-speed’ version which contains an increased concentration of nickel chloride. Chloride ions are also needed to ensure satisfactory dissolution of some nickel anodes at usual values of pH and solution temperature. Where sulphur is deliberately incorporated in the anode during manufacture however, 13 :78
13:79 anodic dissolution of the nickel is activated and the chloride in the solution may be reduced or entirely eliminated, depending upon the degree of anodic activation achieved and the maximum anodic current density required. Nickel anodes are usually either (a)bars or sheets fabricated by casting, rolling or extrusion, or (b) strips of electrolytic nickel, pieces of electrolyticnickel or carbonyl-nickel pellets contained in a basket of titanium mesh. The anodes are held in bags of cotton twill, polypropylene or Terylene in order to prevent metallic particles from entering the solution and causing deposit roughness. Accounts of the anodic dissolution of nickel are given by Raub and Disam4, and Sellers and Carlin'. (See also Section 12.1.) At normal current densities, about 96-98% of the cathodic current in a Watts solution is consumed in depositing nickel; the remainder gives rise to discharge of hydrogen ions. The boric acid in the solution buffers the loss of acidity arising in this way, and improves the appearance and quality of the deposit. Although phosphates, acetates, citrates and tartrates have been used, boric acid is the usual buffer for nickel solutions. A detailed discussion of the function of the constituents of the Watts bath is given by Saubestre6. In addition to inorganic constituents, organic wetting agents are often added to prevent pitting of the deposit that might otherwise arise from adhesion to the cathode of small bubbles of air' or hydrogen evolved cathodically. Elimination of pitting and other defects is discussed by Bouckley and Watson*. NICKEL COATINGS
Decorative Plating The majority of decorative nickel plating is carried out in solutions containing addition agents which modify growth of the nickel deposit so that a fully bright finish is obtained that is suitable for immediate chromium plating without mechanical finishing. At one time, wide use was made of deposits with brightness achieved through additions of cobalt salts plus formates and f~rmaldehyde~.'~, but the use of a mixture of organic addition agents enables deposits to be obtained which are smoother, more lustrous, give bright deposits over a wider range of current densities, and have lower internal stress. In consequence, the bulk of bright nickel plating is carried out in organic bright nickel solutions.
Organic bright nickel solutions Several organic substances are used at appropriate concentrations in these solutions in order to give brightness, levelling and control of deposit stress. Portions of the addition agent molecules are incorporated in the deposit, resulting in a hard, fine-grained coating which has a finely striated structure when etched in section and which usually contains incorporated sulphur. The sulphur causes the deposit to be electrochemically less noble than pure nickel deposits. Decomposition products of the additives form in the solution with use, and at one time they accumulated and impaired the mechanical properties of the plate, eventually necessitating batch purification. In modern solutions however, continuous carbon filtration can be used to remove deleterious organic substances without significant removal of the addition agents themselves. Brighteners Modern solutions contain a brightener system comprising
13:80
NICKEL COATINGS
several additives which together enable bright deposits to be obtained over a wide range of current densities such as that occurring over a component having a complicated shape with deeply recessed areas. Brighteners are broadly divided into primary brighteners and secondary brighteners, but the division is not sharp. Primary brighteners have a powerful effect on the deposit and are normally used at low concentrations which are carefully controlled. Metals such as cadmium and zinc act as primary brighteners, as do organic substances such as amino polyarylmethanes, quinoline and pyridine derivatives, and sulphonated aryl aldehydes. Primary brighteners often, especially at higher concentrations, affect adversely the mechanical properties of the deposit. Secondary or carrier brighteners have a milder effect on the deposit when used alone, and modify the effect of primary brighteners. Judicious combination of primary and secondary brighteners gives fully bright but relatively ductile deposits having low internal stress. Aryl sulphonic acids and sulphonates, sulphonamides and sulphimides frequently act as secondary brighteners. Combinations of brighteners often behave synergistically, so that the final brightening effect is greater than might have been expected from the individual effects. Stress reducers Many organic substances used as secondary brighteners also reduce the tendency for the internal stress in the deposit to become tensile. In the absence of primary brighteners, they are able to give zero or even compressive stress in nickel deposits, and thereby find wide application in electroforming where accurate control of deposit stress is vital. Saccharin, p-toluene sulphonamide, and mono-, di- and tri-sulphonates of benzene and naphthalene are common stress-reducing agents. Stress is usually measured in nickel deposits by observing the bending induced by plating one side only of a metal strip. Convenient and sensitive developments of this technique are available '-I4. Levelling agents A nickel plating solution is said to have levelling action if deposits from it, when applied to an uneven cathode surface, become increasingly smooth as plating proceeds. Levelling agents are therefore widely used to eliminate expensive final polishing of the nickel surface and to reduce the fineness of the surface finish needed on the substrate surface. Both features reduce the cost of producing a bright and smooth finish on a plated article. Levelling agents increase cathode polarisation and are consumed at the cathode by decomposition or incorporation in the deposit. They are used at a concentration sufficiently low that a diffusion layer is established at the cathode surface, and then the levelling agent is able to diffuse at a greater rate to peaks than to recesses on the surface. In order that the layer of solution adjacent to the cathode surface shall remain an equipotential surface, the current density at recesses rises above that at peaks, giving progressive smoothing of the deposit as deposition proceeds. The levelling action of an addition a g e d 5depends upon its concentration C in the solution, the rate of change of cathode potential with change of concentration (dE/dC) and the rate of change of cathode potential with current density (dE/dl). Levelling power (L.P.) at a given current density, defined
NICKEL COATINGS
13:81
as deposit thickness in recesses minus thickness at peaks divided by average thickness, may be expressed as
L.P. = KC( dE/dC) (dZ/dE) where K is a constant. Typical levelling agents include coumarin, quinoline ethiodide, butyne 1,4 diol and its derivatives, and thiourea and its derivatives at certain concentrations. Extensive studies of the mechanism of levelling have been carried out in the United States16, Britain” and the Soviet Union 17. Semi-bright solutions Maximum levelling action is often found in solutions which do not give a fully bright deposit, but the deposit is smooth and can easily be lightly buffed to give a lustrous finish; moreover, many levelling agents used are sulphur-free, so that the deposits are also free from sulphur and as noble as a Watts deposit when subjected to corrosive attack. This feature is exploited in double-layer nickel coatings (see below). Wetting agents As mentioned earlier, wetting agents are added to nickel solutions to prevent pitting. These wetting agents can be cationic, non-ionic or anionic in nature. In general, the best anti-pit agents tend to produce the most foaming, and a compromise must be struck. Where mechanical agitation of the solution is used, by stirrers or by cathode movement, a greater tendency to foaming can be tolerated than when air-agitation is employed. Interaction of addition agents The success of modem proprietary bright nickel solutions has resulted in large measure from the skill of the research departments of plating supply houses in balancing the effects of various additives to give optimum results. The detailed, findings are usually kept confidential, but the broad principles of addition agent action and interaction are discussed in published work 18-m. A commercial-scaleoperation with bright and semi-bright solutions based not on Watts but on a solution having nickel sulphamate as the main constituent (430-450 g/l), is described by Siegrist
Decorative Coating Systems that give Improved Resistance to Corrosion Double-layer nickel coatings These coatings have an undercoat of highlylevelled sulphur-free nickel covered with sufficient bright nickel to give a fully bright finish with minimum requirement for expensive mechanical finishing of the part. They were initially produced simply to reduce costs, but it was soon noticed that, because the undercoat of sulphur-free semi-bright nickel is electrochemically more noble than the final bright nickel above it, corrosive attack when it does occur is preferentially directed towards the bright nickel, and penetration to the basis metal is markedly delayed. Figure 13.7 shows how pits in a single-layer nickel deposit start at small pores or other imperfections in the chromium top coat2’. The pits are initially hemispherical; those shown here were produced by 6 months in an industrial atmosphere on a copper plus nickel plus chromium plated car bumper.
13 :82
NICKEL COATINGS
Fig. 13.7 Commencement of corrosion at discontinuities in chromium topcoat over nickel; x 1 OOO (after Reference 22)
In double-layer nickel coatings however, a flat-based pit is formed in the nickel coating, giving marked resistance to penetration to the basis metal. Figure 13.8 shows a pit in a double-layer nickel plus chromium coating after 58 months service.
Fig. 13.8 Flat-based pit in double-layer nickel plus chromium coating after 58 months service; x 300 (after Reference 22)
NICKEL COATINGS
13 :83
Triple-layer nickel coatings In order to minimise the effect of corrosive attack on the appearance of the deposit while still retaining the resistance to penetration to the substrate afforded by double-layer nickel, triple-layer nickel coatings have been developed in which the semi-bright and bright layers are separated by a thin nickel layer electrochemically less noble than both of them. This thin layer of nickel, highly activated by incorporated sulphur, is described by Brown23.Figure 13.9 shows a section through such a triple-layer coating. In service, corrosive attack is substantially confined to that part of the coating adjacent to the highly-activated layer.
Fig. 13.9 Triple-layer nickel deposit consisting of semi-bright and bright nickel layers with a thin, highly activated layer of nickel between them (after Reference 23)
Nickel coatings that induce microporosity in chromium topcoats In addition to the methods invoked in double- and triple-layer nickel coatings to ensure that the inevitable corrosion currents developed in a corrosive environment are directed away from the basis metal, another method of protecting the basis metal is to ensure that the conventional chromium top-coat (0.3pm) is made sufficiently porous for the corrosion current to be dissipated over a large number of exposed nickel sites. This is achieved conveniently by applying, between the nickel coating and the chromium, a further thin nickel layer containing incorporated solid particles which are inert and which induce in the chromium a large number of pores. The rate of attack at any one pore is then small. Such coatings are increasingly used under severely corrosive service conditions and are described by Oderkerken 24 and Williams2’, among others. Microcracked chromium topcoats Historically, microcracked chromium preceded the micro-porous chromium just described, but it is related to it in that the deposition conditions and thickness of the chromium topcoat are controlled to give porosity through a network of very fine cracks. A thickness of at least 0 - 8 p m is normally needed to ensure that the required crack pattern is formed all over a shaped part. Such microcracked chromium coatings have a slightly lower lustre than the thinner conventional chromium deposits and take longer to deposit. The improved resistance to
13 :84
NICKEL COATINGS
corrosion that they impart t o nickel has been chiefly of interest to the automotive industry. In an attempt to avoid the slightly diminished lustre of thick microcracked coatings, an alternative process has been developed whereby a thin, highly stressed nickel layer is deposited upon the normal bright nickel layer. A conventional chromium topcoat is then applied, causing the thin nickel layer to crack, thereby cracking the chromium layer itself so as t o give a microcrack pattern”. Supplemental films The Batelle Memorial Institute” has developed a post-treatment for nickel plus chromium coatings in which the plated part is made cathodic in a solution containing dichromate. A film thereby formed on the surface seals pores in the coating through which corrosion of the nickel might otherwise occur. Later work3’ suggests, however, that microcracked chromium gives superior results. Control of quality of decorative nickel coatings Increasing international effort has been spent during the past few years in drawing up agreed recommendations aimed at ensuring that incorrect plating procedures d o not diminish the high performance of nickel, or nickel plus chromium, coatings. During 1970, the International Standards Organisation issued Recommendation 1456 Electroplated Coatings of Nickel plus Chromium and Recommendation 1457 Electroplated Coatings of Copper plus Nickel plus Chromium on Steel (or Iron) which were used as guidelines by the British Standards Institution in drawing up BS 1224: 1970 Electroplated Coatings of nickel and Chromium and BS 460 1:1970 Electroplated Coatings of Nickel Plus Chromium on Plastic Substrates. These British standards specify the type and thickness of deposits required for various service conditions, appropriate accelerated corrosion test procedures, and methods of measuring other important properties. The quality of nickel salts and anodes for plating is specified in BS 558 and 564: 1970Nickel Anodes, Anode Nickeland Salts for Electroplating.
Engineering Electroplating Engineering nickel coatings are used to improve load bearing properties and provide resistance to corrosion, erosion, scaling and fretting. The coatings are applied to new parts such as rolls for glass making, laundry plates, wire and tube. They are also used for reclaiming worn gears, shafts and other parts of buses and ships, and as undercoats for engineering coatings of chromium. Deposits from Watts-type solutions Most coatings of nickel for engineering applications are electro deposited from a Watts-type bath Typical mechanical properties of deposits from Watts and sulphamate solutions are compared with those of wrought nickel in Table 13.15. The uncertain effects of impurities are avoided by periodic or continuous electrolysis of the solution at low current densities to remove metallic contaminants and by filtration through active carbon to remove organic substances. A concise review of the effects of impurities and their removal is given by Greenall and Whittington3’.
’.
13:85
NlCKEL COATINGS
Table 13.15 Typical mechanical properties of nickel deposits and wrought nickel
(Hd
Ductility I% elongarion)
Tensile strength (MN/m 2)
(MN/m 2)
-
90-140
47
460
-
Watts nickel
Dull, matt
130-200
25
420
150
Conventional sulphamate nickel
Dull, matt
160-200
18
420
14
Appearance as plated
Hardness
Tensile slesss
~
Hot rolled and annealed Nickel 200
The mechanical properties of Watts deposits from normal, purified solutions depend upon the solution formulation, pH, current density and solution temperature. These parameters are deliberately varied in industrial practice in order to select at will particular values of deposit hardness, strength, ductility and internal stress. Solution pH has little effect on deposit properties over the range pH 1-0-5.0, but with further increase to pH 5 - 5 , hardness, strength and internal stress increase sharply and ductility falls. With the pH held at 3-0, the production of soft, ductile deposits with minimum internal stress is favoured by solution temperatures of 50-60°C and a current density of 3-8A/dmZ in a solution with 25% of the nickel ions provided by nickel chloride. Such deposits have a coarse-grained structure, whereas the harder and stronger deposits produced under other conditions have a finer grain size. A comprehensive study of the relationships between plating variables and deposit properties was made by the American Electroplaters’ Society and the results for Watts and other solutions reported 34.
Hard nickel deposits When the plating variables are adjusted to give deposits with a hardness much above 200 H, with a Watts solution, internal stress is usually too high and ductility too low for the deposits to be fully satisfactory. Higher hardness coupled with reasonable ductility can be achieved by addition of ammonium salts and operation at higher solution pH. A solution used for this purpose35and some deposit properties are as follows:
Composition Nickel sulphate (NiS04.7HzO):180 g/l Nickel chloride (NiCI, .6H20):30 g/l Ammonium chloride (NH4C1):25 g/l Boric acid (H, BO,): 30 g/l Operating conditions Temperature: 60°C pH: 5-6 Cathodic current density: 5 A/dm2 Deposit properties Hardness: 400 H, Tensile strength: 1 . 1 GN/m* Elongation: 6%
13:86
NICKEL COATINGS
Values of hardness higher than 400 Hv (up to 600 Hv) can be obtained by addition of organic substances to a conventional Watts solution. Similarly, internal stress can be made less tensile, zero or compressive, by the use of organic addition agents of the type used in organic bright nickel solutions. in practice, such hard nickel deposits are seldom used in engineering applications unless the required coating is so thin that no machining will be required. Increased hardness and wear resistance may also be achieved by incorporating approximately 25-50070 by volume of small non-metallic particles. These may be carbides, oxides, borides or nitrides, and hardness values up to 560 H, have been reported36.
Deposits from sulphamate solutions The concentration of nickel ions in a conventional sulphamate plating solution is similar to that in a Watts solution, but nickel coatings deposited from the sulphamate bath have lower internal stress. Consequently higher plating rates than those employed in the Watts solution may often be used and this compensates for the higher initial cost of nickel sulphamate compared with nickel sulphate. Typical solution compositions and suitable operating conditions are given in Table 13.16 for the conventional solution and for a concentrated solution used for deposition at high rates:
Table 13.16 Typical solution compositions and suitable operating conditions for the conventional and the concentrated sulphamate baths
Compositions: Nickel sulphamate [Ni(NH2S03)24H20] Nickel chloride (NiCI, .6H20) Boric acid (H, BO,) Operating conditions: Temperature ("C) Agitation
PH Cathodic current density (A/dm Mean deposition rate (pm/h)
2,
Conventional solution
Concentrated solution
Wb
Wl)
300
600 10
30 30 25-50 air 3.5-4-5 2-15 25- 180
30 60-70
air 3-5-4.5
2-80 25-1 OOO
Replacement of nickel chloride by nickel bromide has been claimed3' in the USA to reduce deposit stress, but subsequent German work38was unable to substantiate this finding. Deposition of nickel at rates up to 1 mm/h in the concentrated solution is described by Kendri~k'~. If pure nickel anodes are operated at a current density between 0.5 and 1 -0A/dm2 in sulphamate solutions, a substance which behaves as a stress reducer is produced continuously in sufficient quantity that the stress in deposits can be varied at will from compressive to tensile by adjusting cathode current density and solution temperature. This finding is exploited with the concentrated sulphamate solution in the Ni-Speed process@,and in a further development4' cobalt is added to give deposits of
NICKEL COATINGS
13 :87
hardness up to NOH,. The nature of the stress reducer conveniently produced at the nickel anode is unknown but it differs42 from the azodisulphonate produced43at an insoluble anode such as platinum. A comprehensive and authoritative study of the sulphamate bath has been made by Hammond".
Deposits from all-chloride solution Nickel deposits from a solution of nickel chloride and boric acid are harder, stronger and have a finer grain size than deposits from Watts solution. Lower tank voltage is required for a given current density and the deposit is more uniformly distributed over a cathode of complex shape than in Watts solution, but the deposits are dark coloured and have such high, tensile, internal stress that spontaneous cracking may occur in thick deposits. There is therefore little industrial use of all-chloride solutions. Deposits from other solutions Nickel can be deposited from solutions based on salts other than the sulphate, chloride and sulphamate. Solutions based on nickel fluoborate, pyrophosphate, citrate, etc. have been extensively studied but none of them is used to any significant extent in Europe for engineering deposits. Resistance to corrosion O ~ w a l has d ~ ~surveyed the resistance of engineering coatings of nickel to corrosion by various chemical environments. Environments in which nickel has proved satisfactory include: (a) dry gases including ammonia, the atmosphere, carbon dioxide, coal gas, fluorine, hydrogen, nitrous oxide; (b) carbon tetrachloride, cider, creosote, hydrogen peroxide, mercury, oil, petrol, soaps, trichlorethylene, varnish; (c) alkalis (incl. fused), nitrates (incl. fused at SOO"C), cheese, cream of tartar, eggs, fish, gelatin, fused magnesium fluoride, synthetic resins. On heating in air, nickel forms a protective oxide and gives good service up to 70°C.Nickel is not recommended for exposure to chlorine, sulphur dioxide, nitric acid, sodium hypochlorite, mercuric or silver salts. Where nickel is provided as a corrosion-resistant finish, a thickness of 120-130 pm is usually applied, but for well-finished basis metals and in mild environments, a lower thickness may be adequate. For parts machined after plating however, up to 0.5 mm may be required. Effect of nickel coatings on fatigue strength In general, a coating of high fatigue strength raises the fatigue resistance of a basis metal having low fatigue strength, and vice versa. Thus nickel coatings applied to steels of tensile strength greater than about 420 MN/mm2 can lead to reduced fatigue strength. In practice, this reduction in fatigue resistance is often taken to be negligible for industrial Components because the safety factor used in design is high enough to accommodate the degree of The loss can also be minimised either by using high-strength nickel deposits with compressive internal stress obtained by using appropriate addition agents, or by shot peening the surface of the steel before plating. Effect on corrosion fatigue The combination of corrosion and fatigue can cause rapid failure, and a coating of nickel, by preventing corrosion, can increase the life of the parts. Figure 13.10 shows results obtained by the National Physical Laboratory on mild steel Wohler specimens sprayed with
13 :88
NICKEL COATWIGS
ton f / in2
MN / m m 2
170
\
g IO
1154
ig3 5
,
, ,
: a s s , ,
8
77
3% sodium chloride solution during testing at 2 200 cycle/min; the benefit given by the 75 pm nickel coating is clearly shown. Effect on galling and fretting corrosion (Section 8.7) Even when well lubricated, nickel tends to gall, i.e. stick, when rubbed against some metals, including other nickel surfaces. Nickel also tends to give galling in contact with steei and it is necessary to chromium plate the nickel. Nickel does not form a good combination rubbing against chromium or against phosphorbronze, owing to the action on the nickel of the hard particles contained in the phosphor-bronze. Good performance is given by well-lubricated nickel against normal white-metal bearings, brasses or bronzes. When two metals in intimate contact are subjected to vibration, a dark powder forms at the areas of contact. The effect is referred to as fretting corrosion though it is due to wear rather than true corrosive attack. The galling effect between nickel and steel ensures good resistance t o fretting corrosion and lubricated nickel against steel is a very satisfactory combination used widely in industry for components assembled by press-fitting. Heat treatment after plating Heat treatment may be necessary after plating t o improve the adhesion of coatings on aluminium and its alloys when certain processes, e.g. the Vogt process, are used, or to minimise hydrogen embrittlement of steel parts. Care is needed since heating may distort the part and impair the mechanical properties of the substrate. Heat treatment to improve adhesion on aluminium and its alloys is normally carried out at 120-140°C for 1 h. Heat treatment to minimise hydrogen embrittlement (Section 8.4) should be carried out immediately after plating and before any mechanical finishing operation. Delay is especially undesirable with steels having a tensile strength exceeding 1 .4GN/m2. Steels with tensile strengths below 1 GN/m2 are usually not heat treated. For the stronger steels, heat treatment is carried out
NICKEL COATINGS
13:89
at 190-230°C for not less than 6 h with steels of tensile strengths in the range 1-1 -85 GN/mZ, and for not less than 18 h in the case of even stronger steels.
Other aspects of engineering electrodeposited coatings A great deal of information has been published on important, but specialised, aspects of engineering nickel coatings. General guidance is provided by BS 4758: 1971 Electroplated Coatings of Nickel for Engineering PurposesM.Cleaning, stopping off, etching, plating and subsequent machining of the coatings are discussed by Oswald4’, and the special pretreatments for maraging steels are described by Di Bari4’. Detailed recommendations for turning, grinding, milling and boring nickel coatings are given by Greenwood4*.Treatments that promote strong adhesion of subsequent nickel deposits after intermediate machining operations are discussed by The physical and mechanical properties of nickel at elevated and sub-zero temperatures, determined with electroformed test pieces, have been described by Sample and Knapp”. Other details of the properties of electrodeposited nickel coatings are given in Reference 5 1.
Electroless Nickel In contrast to electrodeposited nickel, electroless nickel is deposited without application of electric current from an external supply. The metal is formed by the action of chemical reducing agents upon nickel ions in solution and, although several substances including h~drazine’~-’~ and its derivatives will give metallic nickel, commercial processes use either sodium hypophosphite which gives a nickel-phosphorus alloy, or, sodium borohydride or various alkyl aminoboranes which give a nickel-boron alloy. These reducing agents can be used in either batch or continuous deposition processes. The amount of boron (typically 3-770)or phosphorus (usually 5-12%) incorporated in the deposits depends upon solution composition and deposition conditions, and it. determines to a large extent the properties of the deposit. A major advantage of the electroless nickel process is that deposition takes place at an almost uniform rate over surfaces of complex shape. Thus, electroless nickel can readily be applied to internal plating of tubes, valves, containers and other parts having deeply undercut surfaces where nickel coating by electrodeposition would be very difficult and costly. The resistance to corrosion of the coatings and their special mechanical properties also offer advantages in many instances where electrodeposited nickel could be applied without difficulty.
Commercial processes Commercial electroless nickel plating stems from an accidental discovery by Brenner and Riddell made in 1944 during the electroplating of a tube, with sodium hypophosphite added to the solution to reduce anodic oxidation of other bath constituents. This led to a process available under licence from the National Bureau of Standards in the USA. Their solutions contain a nickel salt, sodium hypophosphite, a buffer and sometimes accelerators, inhibitors to limit random deposition and brighteners. The solutions are used as acid baths (pH 4-6) or, less commonly, as alkaline baths (pH 8-10). Some compositions and operating conditions are given in Table 13.1755.
13:90
NICKEL COATINGS
Table 13.17 Brenner and Riddell electroless nickel solutions” A lkaline
solution Composition (&I) Nickel chloride, NiCI, -6H,O Nickel sulphate, NiSO, .7H,O Sodium hypophosphite. NaH, PO, .H20 Sodium acetate, NaC, H 3 0 2 - 3 H 2 0 Sodium hydroxyacetate. NaCz H3O3 Sodium citrate, Na,C,H,O, -5fH 2 0 Ammonium chloride. NH,CI Operoting conditions PH Temperature (“C) Plating rate (pm/h) Appearance
Acid solutions 1
2
3
10 50
-
30
30
30 10
10
10
8-10 90
30
10
-
-
10
-
-
-
10
4-6
4-6
90
90
4-6 90 5 -
7.5
25
12.5
Bright
Rough, dull
Semi-bright
-
Further development was made by the General American Transportation Corporation, and their Kanigen p r o c e s ~has ~ ~been * ~ ~available since 1952. Other commercial processes based on the use of hypophosphite have since been developed. Work with reducing agents containing boron has given rise which has been available since 1965. to the Nibodur Plating on plastics Electroless nickel is used in thin deposits in order to provide an initial electrically-conductingsurface layer in the preparation of plastics parts for electroplating. A typical procedure has as its first step an etching treatment of the plastic moulding in a solution of chromic and sulphuric acids in order to give a surface into which subsequent metallic deposits can key. The surface is then made catalytically active for electroless nickel deposition, usually by successive treatments in solutions containing tin compounds and compounds of a platinum group metal. Electroless nickel deposition is then followed by electrodeposition of the required coating which is usually copper plus nickel plus chromium. Thorough rinsing between the pretreatment steps is essential to prevent carry-over of solutions. The commonest plastic plated is ABS (acrylonitrile butadiene styrene copolymer) but procedures are also available for polypropylene2 and other plastics. In some proprietary processes, electroless copper solutions are used to give the initial thin conducting layer. Engineering coatings In the field of engineering coatings of electroless nickel, use of boron compounds as reducing agents has up until now been confined largely to Germany. A comprehensiveaccount of electroless nickelphosphorus and nickel-boron plating in Germany was published by International Nickel Electroless nickel-boron deposits have broadly similar mechanical, physical and chemical properties to those of electroless nickelphosphorus deposits, and in the following discussion of deposit properties, data refer to nickel-phosphorus coatings unless otherwise stated.
@.
Preparation of basis metals for plating Preliminary cleaning of various basis metals follows the broad principles used for electrodeposited nickel.
NICKEL COATINGS
13:91
Electroless nickel deposition may then be carried out directly onto steel, aluminium, nickel or cobalt surfaces. Surfaces of copper, brass, bronze, chromium or titanium are not catalytic for deposition of nickel-phosphorus and the reaction must be initiated by one of the following operations: 1. Apply an external current briefly so as to electrodeposit nickel. 2. Touch the surface with a metal such as steel or aluminium while immersed. 3. Dip in palladium chloride solution (this gives only modest adhesion and carries the danger of contamination of the bath by solution carryover).
Antimony, arsenic, bismuth, cadmium, lead, tin and zinc cannot be directly plated by these techniques and should be copper plated. Resistance to corrosion Most authors who compare resistance to corrosion of electroless nickel with that of electrodeposited nickel conclude that the electroless deposit is the superior material when assessed by salt spray testing, seaside exposure or subjection to nitric acid. Also, resistance to corrosion of electroless nickel is said to increase with increasing phosphorus level. However, unpublished results from International Nickel's Birmingham research laboratory showed that electroless nickel-phosphorus and electrolytic nickel deposits were not significantly different on roof exposure or when compared by polarisation data. Resistance to corrosion of electroless nickel, both as-deposited and, in most cases, after heating to 750°C is listed by Metzger6' for about 80 chemicals and other products. Resistance was generally satisfactory, with attack at a rate below 13 pm/year. The only substances causing faster attack were acetic acid, ammonium hydroxide or phosphate, aerated ammonium sulphate, benzyl chloride, boric acid, fluorophosphoric acid, hydrochloric acid, aerated lactic acid, aerated lemon juice, sodium cyanide and sulphuric acid. Electroless nickel-phosphorus should not be used with either fused or hot, strong, aqueous caustic solutions because the coating offers lower resistance to attack than does electrodeposited nickel. As-deposited electroless nickelboron, however, offers good resistance to hot aqueous caustic solutionsm. It is also resistant to solutions of oxidising salts such as potassium dichromate, permanganate, chlorate and nitrate. Heat treatment, e.g. 2 h at 6OO"C,improves the resistance to corrosion of nickel-boron and nickel-phosphorus electroless nickel deposits, especially to acid media. This presumably results from formation of a nickel-iron alloy layer ". Mechanical properties
Ductility The ductility of electroless nickel deposits is low, but the brittleness of deposits containing less than 2% phosphorus can be reduced by heating to approx. 750°C for some hours followed by slow cooling. Hardness The hardness of electroless deposits is higher after heating to intermediate temperatures, the final value depending upon temperature and time of heating. Values of maximum hardness of nickel-phosphorus after heating to various temperatures6' are plotted in Fig. 13.11; the variation of
13:92
NICKEL COATINGS
500
~
200
Id0
'0°'
300
LOO
Temperature
500
660
'
)O
(OC)
Fig. 13.1 1 Heat-treatment curve for electroless nickel (after Reference 61) kp/rnm2
I 1,11111
,
I
1 I 1111
I I 1 11111
I I I I11111
I I I I11111
I I I IT
1200~
1000-
5:
-X
800-
-
600
500 O C
C
P
2
400-
200-
Time ( m i d
Fig. 13.12
Relationship between hardness and heat-treatment time for electroless nickel (after Reference 62)
hardness with heating time62 is shown in Fig. 13.12 for various heattreatment temperatures. These curves show that hardness can be made to exceed 1 OOO H, by appropriate heat treatment. Nickel-boron deposits can similarly be heat treated to values up to 1 200H,. Resistance to abrasion The resistance to abrasion of electroless nickelphosphorus hardened to 600 H,, assessed by Taber abrasion tests, has been found to be double that of electroplated However, electroless nickel coatings are not suitable for applications where two electroless nickel surfaces rub together without lubrication unless the values of hardness are made to differ by over 200Hv units. Galling of aluminium, titanium or stainless steel may be overcome by applying electroless nickel to one of the two mating surfaces.
NICKEL COATINGS
13 :93
Applications In 1970, according to best estimates, 60000t of nickel coatings were deposited in the western world. This figure corresponds to 13% of the nickel consumed for all purposes. Decorative coatings It is impossible to give a comprehensive list of the uses of nickel coatings but applications of decorative nickel coatings, usually with a chromium top-coat, are given below: 1. Automotive: bumpers, grills, handles, over-riders, hubcaps, exhaust trim, locks, aerials, ash-trays, knobs. 2. Bicycles: rims, handlebars, spokes, cranks, hubs, bells, brake levers. 3. Perambulators: wheels, handles, springs, wing-nuts, body trim. 4. Door furniture: numbers, letter boxes, handles, bells, locks, keys. 5 . Bathrooms: shower attachments, taps, chains, handles, locks, holders for soap and toothbrushes, mirror surrounds. 6. Kitchens: window fasteners, toasters, can-openers, trim for cooker, washing machine and dishwasher, clips. 7. General household: irons, needles, pins, press-studs, birdcages. 8. Tools: spanners, nuts, bolts, screws, screw-drivers, hacksaw bodies. Toys, office equipment, sports equipment and shop furniture also provide large markets for decorative coatings of nickel or nickel plus chromium. Engineering electrodeposits Engineering electrodeposits are used to give improved properties on new components, or to replace metal lost by wear, corrosion or mis-machining, or as an undercoat for thick chromium deposits. For new components, the nickel coating is usually 25-250pm thick. Normally, the deposits are not machined. Applications include pump bodies, laundry plates, heat exchanger plates, evaporator tubes, alkaline battery cases and food-handling equipment of various sorts. Machined deposits on new equipment, including undercoats for chromium, are usually 125-500 pm thick. Applications include cylinder liners (on the water side), cylinders used in the rubber, pulp and paper handling industries, compressor rods and armatures for electric motors. Machined deposits for salvaging worn parts, with or without a chromium topcoat, are limited in thickness only by the economic limitation when it becomes cheaper to manufacture a new part; thicknesses up to 5-6 mm are used. Applications include axles, swivel pins, hydraulic rams, shafts, bearings and gears. Among larger installations, the repair shop of the London Transport Executive at Chiswick houses many nickel plating tanks devoted to reclaiming worn engine parts from 8 OOO buses, and repairs are carried out on about 15 buses each week. Electroless nickel engineering deposits Electroless nickel is not usually deposited to thicknesses greater than about 125 prn. Where a greater total thickness is required, an electrolytic nickel undercoat should be used. The number of applications has been growing at a considerable rate in recent yearsw and amongst the most common are hydraulic cylinders, tools for handling plastics, machine parts, printing cylinders, internal plating of valves and tubes, cooling coils, compressor housings, parts for pumps, storage vessels for chemicals, braking equipment, industrial needles, reaction
13:94
NICKEL COATINGS
vessels, filters, moulds for glass, and precision gears. Electroless nickel is also used as a pretreatment stage in the preparation of some printed circuitsM. Electroforms Electroforming is electrodeposition onto a suitable mandrel which is subsequently removed so that the detached coating becomes the desired product. The process has the advantages that an object of intricate form can be produced in a single stage, a variety of desired surface textures can be reproduced simultaneously, a high order of accuracy is obtained in reproducing mandrel shape, and tools can be replicated exactly for massproduction work. Nickel has the particular advantages that its internal stress, hardness and ductility can be varied at will between wide limits and the final electroforms are strong, tough and highly resistant to abrasion, erosion and corrosive attack. The many applications of electroforming with nickel in Europe have recently been reviewed by Bailey, Watson and Winkler4*.
Recent Developments Decorative Plating
There has been continuing progress in recent years in the formulation of proprietary nickel electroplating solutions. Bright nickel processes are available with improved brightness and levelling on unpolished substrates, improved ductility, and with brightness obtained over a wider range of current densities. Wearmouth and Bishop6' have developed a process for applying a pattern to decorative nickel plus chromium coatings after plating, by laying a stencil over the surface and exposing the bare areas to the peening action of a slurry of glass beads in water to form a satin texture. Microcracking of the chromium occurs over the satin regions and resistance to corrosion is thereby improved. New pretreatments for aluminium to enable it to be nickel plated more easily have led to novel decorative applications including large mirrors. WyszynskiMhas described a proprietary process applicable to a wide range of aluminium alloys. Concern over the health hazards of the hexavalent chromium solutions used to form the top coat of conventional nickel plus chromium coatings have encouraged research into trivalent chromium plating solutions. A process with better throwing power and improved covering power than those of hexavalent chromium has been described by Smart etaL6'. A process for depositing a chromium-iron, or chromium-nickel-iron alloy, has been outlined by Lawa. Engineering Electroplating
W e a r m ~ u t hhas ~ ~ described the production of nickel-cobalt, nickelmanganese, and nickel-chromium alloy coatings for non-decorative uses. The nickel-cobalt and nickel-manganese are electrodeposited direct from sulphamate-based solutions, the nickel-cobalt alloys offering higher hardness than the nickel-manganese alloys, which are restricted to a relatively
NICKEL COATINGS
13 :95
low manganese content. However, the manganese prevents embrittlement on heating that would otherwise arise from sulphur incorporated in the plating from conventional sulphur-bearing organic additives in the plating solution. The nickel-chromium alloys are formed by incorporating chromium carbide in nickel electrodeposit, followed by heat treatment in hydrogen at 1OOO"C to decompose the carbide. Composite Coatings
A wide range of applications for hard, wear-resistant coatings of electroless nickel containing silicon carbide particles have been discussed by Weissenberger'O. The solution is basically for nickel-phosphorus coatings, but contains an addition of 5-15gA silicon carbide. Hubner and 0stermann7' have published a comparison between electroless nickelsilicon carbide, electrodeposited nickel-silicon carbide, and hard chromium engineering coatings. Electroless nickel coatings containing PTFE particles have been discussed by Tulsi 72, and non-stick coatings of electrodeposited nickel containing 30% by volume PTFE particles are described by Naito and Otaka73.They found that the addition of organic additives to increase the hardness of the nickel matrix to 500-600HV reduced the incorporation of the PTFE to 10-15070 by volume. E k t r ofonning
~ ~ described the continuous electroNickel Foil Jones and M ~ G r a t hhave deposition of nickel on to a rotating titanium drum, and detachment of the nickel coating to give a process for manufacturing foil up to 500mm wide. The foil is used directly as an intermediate layer in fire-resistant blankets on North Sea oil rigs and as the substrate for grafihite gaskets employed in high-temperature applications, where the nickel replaces asbestos. The main use, however, is as a foil 0.013 mm thick carrying a solar energy absorbing surface. For this purpose, the foil is coated with a thin black mixed nickel oxide layer which out-performs conventional nickelblack (an electrodeposited zinc-nickel sulphide complex) and chrome-black coatings. A significant advantage of the foil approach is that the foil can be fixed adherently on to a variety of collector surfaces and shapes that could not be electroplated and blackened directly. This solar foil is already used in 25 countries in Europe, North and South America, and Asia. Other uses for nickel foil include printed circuits where welded, instead of soldered, connections are specified, in heating elements for panel heaters, and in the manufacture of bursting discs and explosion release device^'^. Abrasive Sheets Abrasive sheets, polishing and lapping foils are electroformed in nickel using a photoresist te~hnique'~. The sheets bear tiny cutting edges all at the same level and have almost a planing effect when rubbed against the surface to be worked. The nickel is hardened to 600 H, and the spaces between the cutting edges have a mirror-like finish to minimise retention of abraded material which could otherwise clog the surface.
13:%
NICKEL COATINGS
Tubes and Perforated Tubes Electrodepositing nickel non-adherently all over the curved surface of a cylinder, and then sliding off the coating, produces a tubular nickel product. Some tubes manufactured in this way are plain, but most are perforated. They are used industrially for screen printing textiles, carpets and wall paper7’. Bands and perforated bands Detached nickel coatings in the form of bands are made by a similar technique to that used for tubes except that their diameter is usually greater and their width much less. The outside layer of nickel can itself be an integral coating comprising the nickel matrix and incorporated diamonds. Such bands are used as cutting tools75.Some practical aspects of the incorporation of the diamonds in nickel have been Electroformed nickel perforated bands are used in cigarette making machines to transport the shredded tobacco at a constant rate. Since the bands contain no joins, they resist fatigue and have long service life75. Bellows Nickel bellows can be made by electrodeposition onto a grooved cylinder. In this case, the nickel coating cannot be slid off, and so the substrate must be removed destructively. The grooved cylinders or mandrels are frequently of aluminium alloy which is dissolved away in caustic alkali when the nickel deposition is completed. Uses include pressure switches, flexible couplings, and pressure transducers 75.
Discs Discs of nickel electroformed on to mandrels bearing grooves modulated with recorded sound have been used for many years for stamping sound recording discs. This process has been adapted and refined for the manufacture of digital records, including video discs 78,79. Video disc stampers must be hard, stress-free, and flat to within 0.1 pm; results of a short investigation directed towards these requirements have been reported *O. Other applications of nickel electroforms are reviewed in Reference 75. S. A. WATSON
REFERENCES 1. Saubestre, E. B., Trans. Inst. Melal Finishing, 41 No. 5, 228-235 (1969) 2. Innes, W. P., Grunwald, J . J., D’Ottavio, E. D., Toller, W. H. and Carmichael, L., Plating, 56, 51-56, January (1969) 3. Watts, 0. P., Trans. Electrochem. SOC., 29, 395 (1916) 4. Raub, E. and Disam, A., Metalloberpuche, 13, 308-314, Oct. (1959) 5. Sellers, W. W. and Carlin. F. X., Plating, 52 No. 3. 215-224 (1965) 6. Saubestre. E. B.. Plating, 45 No. 9. 927 (1958) 7. Tucker, W. M. and Beuckman, F. 0.. h o c . 43rd Ann. Meeting Amer. Electroplater’s Society, 118-122 (1956) 8. Bouckley. D. and Watson, S. A., Electroplating and Metal Finishing. 20. 303-310 and 348-353 (1967) 9. Weisberg. L.. Trans. Electrochem. Soc.. 73. 435-444 (1938) IO. Hinrichsen, O., UK Pat. 4 6 1 126 (1937) 11. Brenner, A. and Senderoff, S., J. Res. Bur. Standards, 42, 89-104 (1949) 12. Hoar, T. P. and Arrowsmith, D. J., Trans. Inst. Metal Finishing, 36, 1-6 (1959) 13. Fry, H. and Morris, F. G., Electroplating, 12, 207-214 (1959) 14. Sykes, J. M., Ives, A. G. and Rothwell, G. P., JournalofPhysics E. ScientifcInstrumenls, 3, 941-942 (1970)
NICKEL COATINGS
13:97
15. Watson, S. A. and Edwards, J., Trans. Inst. Metal Finishing, 34, 167-198 (1957) 16. Foulke, D. G. and Kardos, 0.. Proc. Amer. Electroplater’s SOC.,43, 172-180 (1956) 17. Kruglikov, S. S., Kudryavtsev, N. T. and Semina, E. V . , Proc. ‘Interfinish’68’,66-71, May (1968) 18. Brown, H., Electroplating and Metal Finishing, 15 No. I , 14-17 (1962) 19. Edwards, J., Trans. Inst. Metal Finishing, 41. 169-181 (1964) 20. Brown, H., Trans. Inst. Metal Finishing, 47, 63-70 (1969) 21. Siegrist, F. L., Metal Progress, 85, 101-104, March (1964) 22. Flint, G. N. and Melbourne, S. H., Trans. Inst. Metal Finishing, 38, 35-44 (1961) 23. Brown, H., Electroplating and Metal Finishing, 15 No. 1 I , 398 (1962) 24. Oderkerken, J. M., Electroplating and Metal Finishing, 17 No. I , 2 (1964) 25. Williams, R. V . , ibid., 19 No. 3, 92-96 (1966) 26. Seyb, E. J., Proc. Amer. Electroplater’s SOC.,50, 175-180 (1963) 27. Millage, D., Romanowski, E. and Klein, R., 49th Ann. Tech. Proc. A.E.S., 43-52 (1962) 28. Hairsine, C., Longland, J. E. and Postins, C., Electroplating and Metal Finishing, 21, 41-43, Feb. (1968) 29. Carter, V . E., Trans. Inst. Metal Finishing, 48 No. 1, 19-25 (1970) 30. UK Pats. 1 122 795 and I 187 843 31. Safranek, W. H. and Miller, H. R., Plating, 52, 873-878, Sept. (1965) 32. Davies, G. R., Electroplating and Metal Finishing, 21, 393-398, Dec. (1968) 33. Greenall, G. J. and Whittington, C. M., Plating, 53, 217-224, Feb. (1966) 34. Brenner, A., Zentner, V. and Jennings, C. W., Plating, 39, 865-899, 933 (1952) 35. Wesley, W. A. and Roehl, E. J., Trans. Electrochem. SOC.,82, 37 (1942) 36. Kedward, E. C. and Kiernan, B., Metal Finishing Journal, 13, 116-120, April (1967) 37. Searles, H., Plating. 53. 204-208, Feb. (1966) 38. Brugger, R., Nickel Plating, Robert Draper (1970) 39. Kendrick, R. J., Trans. Inst. Metal Finishing International Conf.. 42, 235-245 (1964) 40. Kendrick, R. J. and Watson, S. A., Electrochimica Metallorum, 1, 320-334, July-Sept. (1%) 41. Belt, K., Crossley, J . A. and Watson, S. A., Trans. Inst. Metal Finishing, 48 No. 4, 132-138 (1970) 42. Bailey, G. L. J., Watson, S. A. and Winkler, L., Electroplating and Metal Finishing, 22, 21-34 and 38. Nov. (1969) 43. Greene, A. F., Plating, 55, 594-599, June (1968) 44. Hammond, R. A. F., Metal Finishing Journal, 16, Part I , June, 169-176; Part 2, July, 205-211; Part 3, August, 234-243 and Part 4, September, 276-285 (1970) 45. Oswald, J. W., Heavy Electrodeposition of Nickel, International Nickel Ltd., Publication No. 2 471 (1962) 46. BS 4758, Electroplated Coatings of Nickel for Engineering Purposes (1971) 47. Di Bari, G. A,, Plating, 52 No. 11, 1157-1161 (1965) 48. Greenwood, A., Metal Finishing Journal, 11, 484-490, Dec. (1965) 49. Carlin, F. X . , Plating, 55, 148-151, Feb. (1968) 50. Sample, C. H. and Knapp, B. B., A.S. T.M. Special Technical Publication No. 318, 32-42 ( 1962) 51. Watson, S. A., ‘Engineering Uses of Nickel Deposits’, Electroplating and Metal Finishing, May (1972) 52. Dini, J. W. and Coronado, P . R., Plating, 54, 385-390 (1967) 53. Levy, D. J., Electrochem. Technology, 1, 38-42 (1963) 54. Kozlova, N. I. and Korovin, N. V . , Zh. Prikl. Khim., 40, 902-904 (1967) 55. Kreig, A., A.S.T.M. Special Technical Publication No. 265, 21-37 (1959) 56. Colin, R., Calvanotechnik, 57 No. 3, 158-167 (1966) 57. Heinke, G., Metalloberflache, 21 No. 9, 273-275 (1967) 58. Lang, K., Galvanotechnik, 55, 728-729 (1964) 59. Lang, K., Metalloberflache, 19, 257-262 (1965) 60. Stromloses Dickvernickeln, International Nickel Deutschland GmbH, Publication No. 63, (1971) 61. Metzger, W. H., A.S.T.M. Special Technical Publication No. 265, 13-20 (1959) 62. Wiegand, H., Heinke, G. and Schwitzgebel, K., Metalloberfloche, 22 No. IO, 304-311 ( 1968) 63. Chinn, J. L., Materials and Methods, 41, 104-106, May (1965) 64. Lonhoff, N., Trans. Inst. Metal Finishing. 46, 194-198 (1%8)
13:98 65. Wearmouth, W. ( 1984)
NICKEL COATINGS
R. and Bishop, R. C. E. B., Trans. Inst. Metal Finishing, 62,
104-108
66. Wyszynski, A. E., Trans. Inst. Metal Finishing, 58, 34-40 (1980) Smart, D., Such, T. E. and Wake, S. J., Trans. Inst. Metal Finishing, 61, 105-110 (1983) Law, M., Finishing, 8, 30-31 (1984) Wearmouth, W. R., Trans. Inst. Metal Finishing, 60, 68-73 (1982) Weissenberger, M.,Metall, 30, 1134-1 137 (1976) Hiibner, H. and Ostermann, A. E.,Metalloberflache, 31, 456-463 (1979) Tulsi, S. S., Trans. Inst. Metal Finishing, 61, 147-149 (1983) Naitoh, K. and Otaka, T. New Materials and New Process, 1, 170-176 (1981) Jones, P. C. and McGrath, J. P., Proc. 36th Annual Conf. Australasian Inst. of Metals, pp. 139-145 (1983) 75. Watson, S. A., Plating and Surface Finishing, 62 No. 9, 851-861 (1975) 76. Lindenbeck, D. A. and McAlonan, C. G., Industrial Diamond Review, 34-38 (1974) 77. Zahavi, J. and Hazan, J.. Plating and Surface Finishing, 70, No. 2, 57-61 (1983) 78. Schneck, R. W., Plating and Surface Finishing, 71 No. I , 38-42 (1984) 79. Legierse, P. E. J., Schmitz, J. H. A., Van Hock, M. A. F., and Van Wijngaarden, S., Plating and Surface Finishing, 71 No. 12, 21-25 (1984) 80. Wearmouth, W. R. and Bishop, R. C. E., Trans. Inst. Metal Finishing, 62, 32 (1984)
67. 68. 69. 70. 71. 72. 73. 74.
BIBLIOGRAPHY Dennis, J. K. and Such, T. E., Nickel and Chromium Plating, Butterworths, London (1972)
13.8 Chromium Coatings It is not economically or technically feasible to use chromium in a fabricated form, but the high resistance of the metal to corrosion can be utilised by applying a thin coating of chromium to less resistant metals. Although the metal is base (EoCr3fIcr= -0.74 V (SHE)) it is protected by a thin, stable, tenacious, refractory, self-sealing film of Cr,O, . This is preserved by oxidising conditions, and the metal is very resistant to high-temperature oxidation and to atmospheric exposure in most natural environments. Unlike silver and copper, it is not tarnished by hydrogen sulphide, nor is it ‘fogged’ like nickel by atmospheres containing sulphur dioxide. The high reflectivity, pleasing blue-white colour , and the oxidation- and tarnish-resistance of the metal are the main reasons for its application in the form of thin coatings to cheaper and less resistant metals, for decorative purposes. In addition, the extreme hardness of the metal, its low coefficient of friction and its non-galling property, combined with its corrosion resistance, make it particularly valuable as a coating where resistance to wear and abrasion are important. Thick deposits applied for this purpose are referred to as hard chromium to distinguish them from the thin decorative deposits.
Methods of Applying Chromium Coatings The only methods of significance for producing chromium coatings are electrodeposition, chromising and vapour deposition. The last-mentioned is used only to a negligible extent for special high-temperature applications, as the coatings are less porous than electrodeposited chromium and are less liable to spall (see Section 12.5). The metal is deposited in vacuo from chromous or chromic iodide. Chromising produces coatings which are essentially alloys, and which are considered in Section 12.3. Electrodeposited chromium is one of the most widely used metallic coatings. Electdeposition (Sectrbn 12. 1)
Electrodeposited chromium, both decorative and ‘hard’, is produced with the use of a solution of chromic acid containing a small amount of catalyst which is usually sulphuric acid, although fluosilicic or fluoboric acid may be used. A typical electrolyte contains 250-400g/l of chromic acid and 13 :99
13: 100
CHROMIUM COATINGS
2-5-4*0g/l of sulphuric acid; the CrO,:SO:ratio is important and for satisfactory plating it should be maintained at about 100: 1- If the catalyst content is too low no metal will be deposited, and if it is too high throwing power will be considerably reduced. The cathode efficiency is usually only 10-12% although up to 20% can be achieved with a silicofluoride catalyst. The evolution of hydrogen at the cathode and oxygen at the anode (6% antimonial-lead, which becomes coated with lead peroxide) necessitates provision for removal of the toxic spray by extraction or for the suppression of bubble formation by the addition to the baths of a perfluoro-carbon type of surface-active agent, the only type known to be stable under the prevailing conditions’. The chromium content of the bath is replenished by addition of chromic acid, as a chromium anode is not technically feasible. The voltage used is 4-8 V, current density 9-22 A/dm’, and temperature 3843°C. Higher current densities, up to 55A/dm’, are used for thick deposits. A considerable amount of heat is generated during electrodeposition and provision must be made for cooling of the electrolyte during operation. The covering and throwing power of the electrolytes is low, and bright plating which is required for decorative purposes can be obtained for a given composition and temperature only within a relatively narrow range of current densities. Outside this range, the deposits are not bright and the hardness of chromium is such that polishing is very difficult and uneconomical. Hence special care must be taken in the racking of irregularly shaped articles to avoid unplated areas, or dull and burnt deposits. Rack design is important in obtaining uniform deposits and good coverage in view of the low efficiencyof the bath. The use of computer modelling has been examined for this purpose3’. Much attention has been given of late to the development of chromic acid baths with higher efficiencies, especially for hard chromium plating, since it is here that the greatest potential for savings in time and energy can be achieved (see below). Amongst the addition agents which have been found to be effective in increasing the efficiency of the conventional chromic acid solution are the bromates, iodates and dichloromalonic acid36. Several commercial high efficiency baths are currently in use. Self regulating chromium The self-regulating chromium solutions were introduced to eliminate the need for maintaining the correct catalyst concentration by periodic analysis; they depend on the addition of a sparingly soluble sulphate to the bath which supplies the correct amount of SO:automatically. Initially strontium sulphate (solubility approx. 1 75 g/l at 30°C and 21 g/l at 40°C) was employed for this purpose’. The strontium sulphate forms a layer on the bottom of the bath, which must be stirred from, time to time. A bath with a CrO, concentration of 250g/l would have a catalyst content of 1*52g/l SrSO, and 4*35g/l of K,SiF6. Potassium dichromate and strontium chromate have also found application as additives for the control of the saturation solubility of the catalyst. Succinic acid has also been proposed3 for the stabilisation of a selfregulating bath, a recommended bath composition consisting of 375g/l of CrO,, 8g/l of SrSO,, and 40gA of succinic anhydride; the bath is
-
CHROMIUM COATINGS
13: 101
operated at 35°C. Self-regulating solutions generally have a higher current efficiency (18-25%) than the conventional bright solutions4. Tetrachromate electrolytes The alkaline tetrachromate baths are used to a small extent chiefly for the direct chromium plating of zinc die-castings, brass or aluminium, since the solutions do not attack these metals5. The original bath was developed by Bornhauser (German Pat. 608 757) and contained 300 g/l of chromic acid, 60 g/l of sodium hydroxide, 0.6-0-8 g/l of sulphuric acid and 1 ml/l of alcohol. The essential constituent of the bath is sodium tetrachromate, Na,Cr,O,,, which is, however, only stable at temperatures below about 25°C. This temperature should therefore not be exceeded in the operation of the bath. Current densities of 75-150A/dm2 are used. The current efficiency of the bath is high (30-35%) so that the metal is deposited at the rate of about 1 pm/min. The deposits are normally matt in appearance, but are comparatively soft and readily polished. A proprietary tetrachromate bath has been used in Germany under the name of the D process6. By the use of additions of magnesium oxide and sodium tungstate it is claimed that the current efficiency of the bath can be raised to as high as 3540%. Other additives such as indium sulphate, sodium selenate or sodium hexavanadate enable bright deposits to be obtained. Trivalent chromium baths Considerable attention has been given recently to the possibility of depositing chromium from trivalent chromium solutions. One bath uses a dimethyl formamide-water solvent system having chromic chloride as an active salt with additions of ammonium chloride, sodium chloride and boric acid to improve current efficiency and conductivity. Plating efficiencies are of the order of 30-40% based on Cr(lI1) and bright deposit can be obtained over the normal plating range of 25-1 -25 A/dm2 at a plating speed of at least 0.3 pm/min. The deposits are micro-discontin~ous~-'~. A commercial trivalent chromium bath which is entirely aqueous and based on chromic sulphate (Cr203),with complexing agents, conductivity salts, a buffer (e.g. boric acid) and a wetting agent, has been introduced and has had some success" although it has the disadvantage of having to be operated in a diaphragm cell. The deposit has a more greyish colour than that obtained from a chromic acid bath, and has a lower hardness. The bath is mainly used for decorative purposes, being unsuited to producing thicker coatings. The attractions of trivalent baths are their lower toxicity, greater efficiency, and a considerably simplified procedure for efficient treatment. Properties of electrodeposited chromium
Structure Although massive chromium has a body-centred cubic structure, electrodeposited chromium can exist as two primary modifications, i.e. a(b.c.c.) and 6- (c.p.h.). The precise conditions under which these forms of chromium can be deposited are not known with certainty. Muro'' showed that at 40°C and 2-0-22 A/dmz the deposit was essentially a-chromium but small amounts of 0-and y- were present, while Koch and Hein" observed
13: 102
CHROMIUM COATINGS
the 8- form at 5OoC and 40A/m2. This form is unstable, however, and is converted rapidly by heating or more slowly by storage at room temperature to the CY- form. The crystal structure is exceedingly fine and cannot be revealed by the microscope; WoodI3 has shown by X-ray diffraction that the grain size is 1.4 x 10-9m.
Porosity and discontinuities Chromium plate of 0.5 pm or less in thickness is invariably porous. An increase in thickness above this value, however, when plating is carried out under conventional conditions (Le. 38-43OC, 11-16 A/dm2 and a CrO, :SO:- ratio of 100: 1 to 120:l) results in a cracked deposit which can be revealed by microscopical examination at about x 350 magnification. CohenI4 considers that the cracks are filled with a transparent film, probably of hydrated chromic oxide, which dehydrates on heating to form Cr,O,. According to Snavely”, cracks and included material in the cracks are caused by the formation of unstable chromium hydrides during plating. A hexagonal form of the hydride (CrH to CrH3 is formed initially, but decomposes spontaneously to a-chromium and free hydrogen. This involves a decrease in volume of over 15%, and since the plate is restrained by the basis metal, surface cracks form normal to the surface. The chemical constituents found in the electrodeposit are due to the drawing of electrolyte into the cracks, which are then covered over by subsequent layers of electrodeposit. Black chromium plating Black chromium deposits are frequently required for the optical and instrument industries. The deposits contain large amounts of chromium oxides and are not strictly speaking chromium deposits. Graham j 6 recommends a solution consisting of 250 g/l of chromic acid, 0-25 g/l of hydrofluosilicic acid and a CrO,:H,SiF, ratio of 1 OOO: 1. The bath is operated at about 32°C with a current density of about 30 A/dm2 and a bath voltage of 6 V. The electrolyte solution must be free from sulphuric acid, excess sulphate ions being removed by treatment with barium sulphate. Silvery deposits of chromium containing some nickel are obtained at 70-100A/dm2 from a bath consisting of 200g/l of chromic acid, 20g/l of nickel chloride and 5 ml/l of glacial acetic acid. By a short immersion (5-30s) in concentrated hydrochloric acid, the deposit becomes greyish black. Good black deposits are produced from a bath containing 200 g/l of chromic acid, 20g/l of ammonium vanadate and 6 - 5 ml/l of glacial acetic acid at a current density of 95A/dm2 and a temperature of 35-5OOC”. Some types of black chromium deposits are claimed to have very good corrosion resistance. Hard chromium plating The so-called ‘hard’ (or thick) chromium deposits are applied on carbon and alloy steels, cast iron and light alloys, to improve resistance to wear, abrasion and corrosion. The solutions employed generally contain 150-500g/l of chromic acid and employ a Cr0,:H2S0, ratio of 80-120. Deposit thicknesses of 12-150pm are applied, but the use of thicker deposits is limited to parts which are not subject to bending or stress. Plastic moulds are generally plated with coatings of 10-15 pm, which are considered adequate. Hard chromium deposits are normally ground or lapped before being put into service, and allowances must be made for this
CHROMIUM COATINGS
13: 103
operation to be carried out. Applications for hard chromium deposits include cylinder liners, crankshafts, pump shafts, plastic moulds, dies, cams, rockers, journals and bearings. Plating is carried out by suspension in the bath in the usual manner, areas which are not to be plated being protected by ‘stopping of€’materials, such as lacquers, waxes or plastics. Chromium can also be deposited locally without the use of a tank by the tampon method. The Dalic’* process makes use of an insoluble anode and an absorbent pad containing the electrolyte solution, which is a slightly alkaline organic complex amino-oxalate compound of chromium dissolved in an alcohol, with a wetting agent added. High current densities are used, and rates of deposition of up to 2 . 5 pm/min are practicable. The deposit is slightly softer than the conventional hard chromium deposits. A trivalent hard chromium bath has recently been described38. The bath contains potassium formate as a complexing agent, and thicknesses in excess of 20pm can be deposited. Hardnesses of up to 1 650HV can be obtained by heat treatment at 700°C. The deposits contain 1.6-4.8% carbon, and the bath is suitable for the deposition of composite deposits containing diamond or silicon carbide powder. Several high-efficiency hard chromium plating baths are now available commercially. A solution which does not contain fluoride, and does not therefore attack steel or aluminium, has been described by Schwartz3’. At 50A/dm” and 53°C the cathode efficiency is about 25%, enabling deposition to be carried out at the rate of 1 pm/min, with a consequent substantial saving in power and time. The deposit is bright, and has a hardness of about 1 050 H,. Hard chromium plating provides excellent resistance to atmospheric oxidation both at normal temperatures and at temperatures of up to 650°C. It is unattacked by many chemicals, owing to its passivity. When attack takes place, this usually commences at cracks in the chromium network; hence the most corrosion-resistant deposits must have a very fine structure, such as is obtained from relatively high solution temperatures using low current densities. Corrosion
Electrodeposited chromium, if it is required to protect an underlying metal against corrosion, has to be applied in considerable thicknesses, owing to its high porosity and tendency to crack. Such deposits are expensive to produce and are not fully bright, and, as the polishing of chromium is difficult, it is the general practice to use a protective undercoat, usually nickel, when ferrous or non-ferrous metals have to be protected. For wear resistance or engineering applications, however, ‘hard’ chromium coatings are usually plated directly on to steel and other metals at thicknesses of up to approximately 0.50 mm as against 0-000 25-0.002 mm for decorative chromium deposits on a nickel undercoat. When corrosion of a chromium-coated metal takes place, the corroding current concentrates its action on fissures in the deposit. There appears to be an incubation period, after which rapid attack occurs in the form of pits, and
13: 104
CHROMIUM COATINGS
sometimes a network of corrosion can be observed. The chromium becomes cathodic, the underlying metal (usually nickel) which is exposed at the pores or stress cracks of the chromium plate being anodic (see Fig. 13.13). CORROSION PRODUCTS C r PLATE -THICK NI PLATF
Fig. 13.13 Two stages in the corrosion commencing at a discontinuity in the chromium plating (after Electroplating ond Metol Finishing, 12 No. 1, 3 (1959))
Dettner'' claims that the degree of polish of the base metal has a definite influence on the corrosion resistance of chromium deposited directly on steel, a high degree of polish leading to improved protection; electrolytic polishing is said to lead to particularly good durability2'. Polishing of the chromium layer usually has little effect, but excessive heat generation can lead to reduced corrosion resistance. It is desirable, from a practical point of view, that the chromium be deposited within the bright plating range, and although this does not always coincide with the conditions necessary for maximum protection, a reasonable compromise can be reached.
Crack-free Chromium As has already been stated, attempts to reduce the porosity of chromium plating by increasing its thickness much above 0.OOO 5 mm result in cracked coatings when normal solutions and conditions are employed. It is, however, possible to obtain crack-free deposits in thicknesses up to 0.002 5 mm, with a consequent improvement in corrosion resistance as shown by accelerated tests, by the operation of the bright chromium plating bath at 49-54"C, and higher CrO, :SO:- ratios of 150:1 to 200: 1 21-23. Crack-free deposits can be obtained under conditions outside these ranges, but for practical operation these are the ones which it is most convenient to employ. The main drawback to plating chromium under these conditions is that the current requirements are greater owing to the need to work at twice or three times the conventional current density. There is also some tendency for the deposit to be rather more blue in colour, while frostiness can develop at highcurrent-density areas. Better results can be obtained by allowing the work to enter the plating tank at a lower voltage (2-3 V) before applying the full plating voltage, or by working at a temperature of around 49°C. Some confirmation of these findings has been reported by Safranek et a/.24in their work on the corrosion resistance of plated die-castings. They found that 0-OOO64 mm (minimum) of bright crack-free chromium
CHROMIUM COATINGS
13: 105
deposited in a high-ratio bath at 54°C will extend the ‘corrosion-free’ life of plated die-castings in highly corrosive environments to at least one year, as compared with less than six months for normal deposits. Bright, crack-free chromium deposited on 0.007 6 mm of copper and 0-020 mm minimum of bright nickel furnished good protection against accelerated corrosion. Deposits of more than around 0.002 0 mm in thickness cannot be applied in this way without the initiation of cracks visible to the naked eye particularly at high-current-density areas. This is a disadvantage, since owing to the poor ‘throw’ of chromium it is sometimes necessary to exceed this thickness at such areas on articles of complex shape in order to secure an adequate deposit in recesses. A method of overcoming this problem known as duplex chromium plating (see below) has been developed. Using pulse plating techniques with a duty cycle of 509’0,it is also possible to produce crack-free chromium deposits from a sulphate- or silicofluoridecatalysed solution with a hardness similar to deposits obtained by direct current A high frequency (2 000-3 OOO Hz)is required to give the hardest deposits at a current density of 40A/dm2 and a temperature of 54°C. It is important to avoid conditions that will co-deposit hydrides.
@.
Microcracked Chromium Duplex Chromium
It has been claimed that better corrosion protection than that afforded by the high-temperature chromium-plating method can be obtained by the use of a double chromium plate. In this system, bright crack-free chromium is deposited as described above, followed by an equal thickness of a bright, finely cracked chromium plate. The total chromium thickness should be not less than 0-OOO75 mm, and should preferably be greater. The initial crack-free deposit may be obtained by plating from a solution of chromic acid and sulphuric acid (250-400g/l of chromic acid, chromic acid to sulphate ratio 125 to 175: l), operating temperature 49-54’C. Baths containing fluoborates, which are self-regulating so far as the ratio of chromic acid to catalyst is concerned, can also be used successfully for producing crackfree deposits2’. Immediately after this deposit has been plated, a cracked chromium layer is applied from a dilute bath containing about 200g/l of chromic acid at 46-52°C. The articles t o be plated should enter the bath at a low voltage t o prevent streakiness. Fluoboric or fluosilicic acid in the bath, in addition to sulphuric acid, helps to produce the required fine stress cracking. Single-layer Chromium
A number of proprietary solutions are now available for producing the same result from a single bath. The plating time tends to be rather longer, but this can be reduced either by increasing the current density (which may upset the crack pattern), by decreasing the chromium thickness at which cracking occurs, or by increasing the cathode efficiency.
13: 106
CHROMIUM COATINGS
The effect of the finely cracked chromium layer is to equalise the anode and cathode areas more nearly, so that corrosion of the nickel under the chromium takes place more slowly than it would at larger, isolated cracks. Moreover, the corrosion proceeds laterally along the nickel surface and not in depth as is the case with conventional chromium; hence’failure of the coating under adverse conditions is less likely to occur. A crack count of 30-80 crackslmm is desirable to maintain good corrosion resistance. Crack counts of less than 30 crackslmm should be avoided, since they can penetrate into the nickel layer as a result of mechanical stress, whilst large cracks may also have a notch effectz6. Measurements made on chromium deposits from baths which produce microcracked coatings indicate that the stress decreases with time from the appearance of the first cracks”. It is more difficult to produce the required microcracked pattern on matt or semi-bright nickel than on fully bright deposits2*. The crack network does not form very well in low-current-density areas, so that the auxiliary anodes may be necessary. Corrosion tests have shown that a system based on copper, double nickel and microcracked chromium gives good corrosion resistance, although automobile parts plated with microcracked chromium are not as easy to clean as those plated with crack-free chromium deposit.
Microporous Chromium One of the best methods of improving the corrosion resistance of nickelchromium deposits is to apply a uniformly porous layer, rather than a microcracked chromium layer, this having the advantage that the microporosity is not greatly dependent on the current density at which the chromium plating is carried out. Hence the chromium can be deposited in microporous form on quite complex-shaped articles from a single, conventional chromium solution. The method of achieving this is to suspend inert particles in the underlying nickel coating; the presence of these, being non-conducting, results in the formation of a highly microporous chromium deposit. Severe electrochemical attack of the underlying nickel at large cracks or pores in the chromium is thus prevented, and a substantial improvement in the corrosion resistance of the combined coating is obtained. A relatively thick copper deposit (75 pm) underneath the nickel layer has been found to add considerably to the protective value of the coating. Thereafter very little improvement occurs. The large number of microscopic anode nickel sites which develop when about 0.25 pm of chromium is applied results in very weak corrosion currents with extremely low corrosion penetration. The number of pores in the chromium can be varied from about 3 OOO per square centimetre to several million per square centimetreZ9.The variation in the porosity of microporous chromium with thickness is shown in Fig. 13.14. In practice a special nickel solution containing the suspended particles is applied over the normal bright nickel deposit. The plating time in this solution is from 20 s to 5 min; the most suitable ratio of the two deposits has to be determined in each particular case. The use of a chromium deposit with a fine porosity pattern of 15 OOO to 45 O00 pores per square centimetre in the usual thickness results in a sharp
13 :107
CHROMIUM COATINGS
60 -
50 -
.
"E LOU
z
0
--30X
c .-
In
2 a
20-
10 -
I
0
0.1
02
I
I
03
OL
I
0.5
06
Chromium deposit thickness ( p m )
Fig. 13.14 Variation of porosity of microporous chromium with thickness of deposit
slowing down of the corrosion rate. Such corrosion as does occur develops laterally, thus very greatly delaying the downward penetration into the vulnerable base metal (Fig. 13.15). Since the theory of the mechanism of the microporous chromium system depends on the fact that the occlusions in the underlying nickel provide a
Fig. 13.15 Lateral corrosion in nickel deposit layer containing inert particles beneath microporous chromium
13: 108
CHROMIUM COATINGS
large number of sites where nickel can be corroded at discontinuities in the nickel deposit, it was at one time believed that increasing the chromium thickness excessively would be disadvantageous, as it would seal some of the active nickel sites. Carter3' has shown that this is not the case provided that the porosity in the chromium is not reduced below about 15 OOO pores per square centimetre. When copper was present under the nickel plus chromium coating, full protection of steel was obtained in industrial atmospheres for two years in all environments. This effect on copper is not found under conventional chromium coatings. The reason for this is ascribed to the fact that the copper remains cathodic to the nickel because of the large area of nickel involved in the corrosion reaction when the chromium layer is discontinuous3'. With conventional deposits, however, the smaller area of the corroding nickel allows high current densities to occur in the pits so that the copper becomes anodic and is readily penetrated. The effect is reduced when the nickel layer adjacent to the copper is of a less corrodible type (i.e. semi-bright, dull nickel) and hence the advantages of the copper undercoat are less in the systems employing double nickel deposits. There is some deterioration or dulling in the appearance of articles plated with microporous chromium (as is also the case with the microcracked deposits), but this is only significant on exposure in the severest environments. Dulling was progressively reduced by increasing thickness of the chromium deposit within the range studied, without adverse effect on the protection of the basis metal.
Chromium-nickel-chromium A further development is the use of a combined chromium-nickelchromium or nickel-chromium-nickel-chromium deposit on steel- or zincbase alloy articles3*.An advantage of this system is that the first chromium layer need not be plated within the bright range of the chromium bath, so that plating can be carried out under conditions giving deposits of maximum corrosion resistance; such conditions do not coincide with those under which fully bright chromium plate is obtained. K n a ~ reports p ~ ~ that a chromium deposit of O.OOO25 mm from the usual type of chromium bath, followed by 0.013mm nickel and a further 0-OOO25 mm of chromium gave protection equal to that of a nickel coating of double the thickness applied in the form of normal nickel and chromium plate.
Porous Chromium Porous chromium is largely used on cylinder liners for automobile engines, its advantage being that it retains lubricants better than normal chromium 34. The porosity is usually created by etching the metal. Appropriate etching methods include reversal of current to make the work anodic in the plating solution, and cathodic or chemical treatment in a separate bath. Hydrochloric, sulphuric or oxalic acid may be used as the etching electrolyte in a separate bath, the work being made cathodic; alternatively, chemical etching
13: 109 without current in a hot dilute sulphuric acid or hydrochloric acid bath, to which an inhibitor such as antimony oxide is added, can be employed. Whether pin-point or channel porosity is produced depends primarily on the conditions of deposition, solution temperature and composition being the principal factors. Generally, higher temperatures and higher sulphate ratios in the bath favour channel-type porosity. The degree of porosity must be carefully controlled in order to ensure that excessive roughness is not produced. The ideal condition is one where the chromium becomes adequately receptive for oil but remains smooth. It is usual to hone or lap the porous chrome; careful cleaning is then essential to remove the debris produced by honing. H. SILMAN CH ROMl UM COATINGS
REFERENCES 1. UK Pat. 758 025 (1953) 2. Stareck, J. E. Passal, F. and Mahlstedt. H., Proc. Am. Electroplatecs’ Soc.. 37, 31-49 (1950) 3. Dutch Pat. 6 513 035 (1966) 4. Griffin, J. L., Pfuting. 53. 1%-203 (1966) 5. Twist, R. D., Prod. Fin., 25 No. 20, 37 (1972) 6. Kutzelnigg, A., Metulloberpiiche, 5 No. 10, 156-160 (1953) 7. UK Pat. I 144 913, March (1969) 8. Bharncho, N. R. and Ward, J. J. B., Prod. Fin., 33 No. 4, 64-72, Jan. (1%9) 9. Ward, J . J. B., Christie, I. R. A. and Carter, V. E., Trans. Inst. Met. Fin., 49, 97 (1971) 10. Ward, J. J. B. and Christie, I. R. A., Trans. Inst. Met. Fin., 49, 148 (1971) 11. Munro, Z. and Batsuri. 0.. J. Appl. Phys.. 21. 321 (1952) 12. Koch, L. and Hein. G.. Metalloberflache. 7. A145 (1953) 13. Wood, W. A., Trans. Farad. Soc.. 31, I 248 (1935) 14. Cohen. J. B., Trans. Electroehem. SOC., 86. 441 (1944) 15. Snavely, C. A., Trans. Electrochem. Soc., 92, 537 (1947) 16. Graham, A. K.,Proc. Am. Electroplaters’Soc., 46, 61-63 (1959) 17. Queseley, M.F., Plating, 40,982 (1953) 18. Electroplating, 6 No. 4, 131 (1953) 19. Dettner, H. W., Metalloberflache, 4, A33 (1950) 20. Eilender. W., Arend, H. and Schmidtmann, E., Metafloberpiiche, 2. 141 (1948) 21. Dow. R. and Stareck. J. E., Proc. Amer. Electropl. Soc., 40, 53 (1953) 22. Brown, H., Weinberg, M.and Clauss, R. J., Plating. 2. 144 (1958) 23. Brown, H. and Millage. D. R., Trans. Inst. Met. Fin.. 37, 21 (1960) 24. Safranak, W. H., Miller, H. R. and Faust, C. L., 46th Ann. Proc. Amer. Electropf. SOC., 133 (1959) 25. Morriset, P., Oswald, J. W., Draper, C. R. and Pinner, R., Chromium Plating, Robert Draper, London, 123-129, 325-326 (1954) 26. Gabe, D. R. and West, J. M.,Trans. Insr. Mer. Fin., 40, 197-202 (1963) 27. Such, T. E. and Partington, M., Trans. Inst. Met. Fin., 42, 68-75 (1964) 28. Dennis, J. K., Trans. Inst. Met. Fin.. 43. 8 6 % (1965) 29. Brown, H.and Silman, H.. Proc. 6th International Conference on Electrdeposition and Mefal Finishing. 50-56 (1964) 30. Carter, V. E., Trans. Inst. Met. Fin., 48, 19 (1970) 3 1. Claus, R. J. and Klein, R. W., Proc. 7th International Conference on Metal Finishing, I24 ( 1968) 32. Weinberg, M. and Brown, H., US Pat. 2 871 550 (1959) 33. Knapp. B. B., Trans. Inst. Mer. Fin., 35, 139 (1958) 34. Gray, A. G., Modern Electroplating, Wiley, New York. 135-188 (1953) 35. McCormick. M., Parn. S. Y. S. and Howe, D.. Trans. Inst. MetalFinishing, 64,39 (1986) 36. McCormick, M.. Pate, M.A. and Howe, D., Trans. Insl. Metal Finishing. 63, 34 (1985) 37. Smart, D., Such, D. E. and Wake, S. J.. Trans. Inst. Metul Finishing. 6 . 105 (1983)
13: 110
CHROMIUM COATINGS
38. Takaya, M., Matsunaga, M. and Otaka, T., Proc 12th World Congress on Surface Finishing, p. 161 (1988) 39. Schartz, G . K., Proc 12th World Congress on Surface Finishing, p. 527 (1988) 40. Pearson, T. and Dennis, J . K., Proc 12th World Congress on Surface Finishing, p. 407 (1988)
BIBLIOGRAPHY Dennis, J . K. and Such, T. E., Nickel and Chromium Plating, Butterworths, London (1972)
13.9 Noble Metal Coatings The most widely used methods for the application of coatings of gold, silver and the platinum group metals (platinum, palladium, rhodium, iridium, ruthenium, osmium) to base metals are mechanical cladding and electroplating. In cladding, the coating is applied in the form of sheet, which may be bonded to the underlying metal by brazing or by elevated-temperature working processes such as swaging and drawing for the production, for example, of platinum-clad molybdenum or tungsten wire, or hot-rolling followed by spinning, cupping or drawing, for the production of coated dishes, tubes, etc. Silver coatings are extensively applied in this way in the lining of chemical reaction vessels, distillation and evaporation equipment, etc. particularly in fine chemical manufacture and food processing where product purity is vital and the protective coating must therefore be completely impervious. The main advantage of silver in this type of application, apart from its relative cheapness as compared with other metals of the group, is its good resistance to organic acids and other compounds and to chloride-containing media. Its high heat-transfer capacity is also a useful asset. Platinum and gold find application in a similar sense where particular conditions warrant their cost. Coating thicknesses may vary from less than 0.025mm to 0.640 mm, depending on service requirements. Palladium can be applied in the same way, but is not employed to a significant extent in this form since its corrosion resistance is inferior to that of platinum. Application of other metals of the platinum group, Le. rhodium, ruthenium and iridium, as protective claddings is hindered by limitations in working technology. In experimental work on the protection of soldering iron bits by ruthenium, the expedient has been adopted of fabricating small hollow cones by compacting and sintering ruthenium powder, and fixing these to the tips by brazing'. In the case of silver and gold, thick coatings of equivalent protective value to those produced by cladding can be obtained by electrodeposition; both metals have, in fact, been successfully employed for electroforming. In the more general case, however, electrodeposited coatings, particularly those of the platinum group metals and, to a lesser extent, gold plated in the bright condition, are subject to some degree of porosity and, with increasing thickness, to the possibility of spontaneous cracking due to internal stress in the as-deposited condition. Nevertheless, the bulk of precious metal coatings used for decorative and industrial purposes, including tonnage use in the electronics field, are applied by electroplating, since protective requirements, 13:111
13: 112
NOBLE METAL COATINGS
though arduous, are in most cases less critical than those demanded by long term exposure to liquid or high-temperature corrosive environments, and some degree of porosity can often be tolerated. Processes for electrodepositing silver, gold, platinum, palladium and rhodium have been long established. In this group the most striking development of recent times has been the emergence of bright gold plating solutions which utilise the stability of gold cyanide at relatively low pH values to plate under acid condition^'*^, and more recently still, of noncyanide electrolytes based on sulphite complexes4. Relatively new electrolytes have also been formulated for the deposition of ruthenium, iridium and even osmium, though these are subject to limitations with regard to the thickness of sound coatings. Platinum metal coatings may also be produced from fused cyanide electrolytes', a technique which is useful in those cases (e.g. ruthenium and iridium) where coatings of sufficient thickness cannot be produced from aqueous solutions. Iridium coatings of this type have been studied in connection with the high-temperature protection of molybdenum6, and thick coatings of rhodium have been produced in a similar way'. Since they are deposited at a temperature of the order of 600°C from a non-aqueous medium, such coatings tend to be less stressed, softer and less porous than coatings from aqueous solutions. For example, rhodium from a bath of this type shows a hardness of approximately 300H, compared with about 900 Hvfrom a conventional sulphate electrolyte. Limitations imposed upon the thickness of coating obtainable from aqueous solutions due to internal stress have been overcome in several directions. Atkinson' has reported the production of ductile, crack-free platinum coatings from a chloro-platinic acid plus hydrochloric acid electrolyte. Tripler, Beach and Faust' have achieved improvement in the protective value of platinum coatings from a diamminodinitritoplatinum(lr) electrolyte by the use of the periodic reverse (p.r.) current technique, which is also widely applied in gold plating. Patent claims have been made for the production of crack-free rhodium deposits from sulphate electrolytes modified by the addition of magnesium salts" or of selenic acid". Highly ductile palladium coatings have been produced by Stevens in thicknesses up to 5 mm from a tetramminepalladium(I1) bromide electrolyte I'. Non-electrolytic plating processes of the displacement type and of the auto-catalytic type have also been described. In the former, a thin film of the noble metal is formed on a base metal substrate by chemical replacement. The reaction may cease when the substrate is completely covered, or, as in the processes described by Johnson l3for the platinum group metals, attack of the substrate may continue through an essentially porous top-coat, which may exfoliate on prolonged treatment. Such processes are of main utility for short-term protection purposes, e.g. retention of solderability of electronic components during storage. In auto-catalytic processes, further deposition of metal is catalysed by the initial layer of the coating itself. Processes of this type have been described for both goldi4and
Electrodeposited Coatings Silver and gold (Chapter 6) Apart from their traditional decorative
NOBLE METAL COATINGS
13: 113
applications, both silver and gold find important industrial use in various types of chemical processing equipment. In the electrical and electronics industries they are employed as plated coatings on contacts and as finishes on wave guides, hollow conductors for high-frequency currents, etc. Silver plating is particularly used in the latter application where its high electrical and thermal conductivity are required in addition to its protective value. The thickness of coating necessary for adequate protection depends on the conditions of service and the nature and condition of the basis metal to which the coating is applied. For electrical purposes for the protection of aluminium, steel and copper, DTD 919A specifies a minimum thickness of 0.007 5 mm (O.OO0 3 in) of silver, with a total thickness of undercoat (copper or nickel) plus silver of 0.038 mm (0.001 5 in) on aluminium parts and 0.020mm (O.OOO8 in) on steel. BS 2816:1957 Efectropafed Coatings of Silver for Engineering Purposes is also relevant in this context. Laister and Benham ” have shown that under more arduous conditions (immersion for 6 months in sea-water) a minimum thickness of 0.025 mm of silver is required to protect steel, even when the silver is itself further protected by a thin rhodium coating. In similar circumstances brass was completely protected by 0-012 5 mm of silver. The use of an undercoating deposit of intermediate electrode potential is generally desirable when precious metal coatings are applied to more reactive base metals, e.g. steel, zinc alloys and aluminium, since otherwise corrosion at discontinuities in the coating will be accelerated by the high e.m.f. of the couple formed between the coating and the basis metal. The thickness of undercoat may have to be increased substantially above the values indicated if the basis metal is affected by special defects such as porosity. In view of its susceptibility to sulphide tarnishing, silver may itself require some measure of protection in many decorative and industrial applications. Chromate passivation processes are commonly employed, but as an alternative, thin coatings of gold, rhodium or palladium may be used. Although usage of gold plating in industrial applications has long outstripped that in traditional decorative fields, it was not until 1968 that an appropriate British standard was issued to cover both spheres of application’*. The high reflectivity of gold in the infra-red region accounts for its use on reflectors in infra-red drying equipment, for which purpose a coating of 0.005 mm gives excellent service on beryllium-copper. This order of thickness became general for electrical contacts in the electronics field, where the main area of industrial gold plating is to be found; but thinner coatings are now usual. The basis metals involved are most commonly copper or copper-base alloys, e.g. brass, nickel-silver, beryllium-copper and phosphor bronze, and coating thickness is dictated not only by environmental conditions but by the need for mechanical wear resistance in sliding and wiping contacts, in which context the softness of pure gold deposits from cyanide electrolytes is generally a disadvantage. Numerous proprietary electrolytes have been developed for the production of harder and brighter deposits. These include acid, neutral and alkaline solutions and cyanide-free formulations and the coatings produced may be essentially pure, where maximum electrical conductivity is required, or alloyed with various amounts of other precious or base metals, e.g. silver, copper, nickel, cobalt, indium, to develop special physical characteristics.
13: 114
NOBLE METAL COATINGS
The hardness of such coatings may reach a maximum of about 400Hv as compared with approximately 50 H, for a soft gold deposit. A series of corrosion studies in industrial and marine atmospheres by Baker has indicated that the protective value of hard gold coatings is comparable with that of the pure metal, and that a thickness of only 0-002 5 mm gives good protection to copper base alloys during exposure for six months. In view of the high cost of gold there is a continuing urge to reduce coating thickness in industrial applications to the bare minimum consistent with adequate service life. It is claimed for example that the thickness of gold on a wiping contact can be reduced by using an undercoating of silver, e.g. 0.0075 mm of silver plus 0-OOO25 mm of gold. In this case a special problem arises, particularly at elevated temperature, due to diffusion of silver outwards through the gold layer, with formation of a tarnish film at the surface. This can be prevented by interposing a thin deposit of palladium or rhodium between the gold and silver layers2'. At gold thicknesses below 0.005 mm significant porosity is likely to be present, and a great deal of work was directed to the study of factors affecting the degree of porosity of gold coatings2'-", and to possible means of reducing this, or at least minimising its practical effect. Reduction of porosity can be achieved by the use of copper or nickel under-coats and patents claimed that a coating of platinum only 0.38 pm thick would substantially reduce porosity and improve the high temperature stability of 0.002 5 mm gold coatings on copper25. The effect of corrosion through pores in thin gold coatings on copper- or silver-base substrates can be minimised by applying a thin coating of palladium or rhodium, since sulphide tarnish products do not spread on these metals26,whereas they readily spread over gold to form large areas of high contact resistance. Gold coatings on sliding contacts are often lubricated, and it is claimed that pores in the coatings may be effectively sealed, with marked increase in service life, by incorporating a suitable corrosion inhibitor in the lubricating system". The Platinum Metals (Chapter 61
Rhodium Rhodium is the most important of the platinum group of metals as an electrodeposited coating for protective purposes as shown by the fact that it is the only metal of the group for which a DTD Process Specification exists (No.931). Major fields of application are the protection of silver from tarnishing in both decorative and industrial spheres, and the finishing of metallic reflectors and electrical contacts (particularly sliding or wiping contacts subject to mechanical wear and concerned with the transmission of very small electrical signals, e.g. in radar, telecommunication, and allied equipment, where freedom of the contact surface from films is a critical requirement). The special properties of the electrodeposited coating on which these applications depend are its high reflectivity, virtual immunity from attack by corrosive environments, its consequently low and stable contact resistance, and its extremely high hardness (approximately !NOHv). A disadvantage of the deposit, as produced from conventional acid sulphate or phosphate plus sulphate electrolytes, is a high internal tensile stress, which may give rise to cracking in deposits thicker than
NOBLE METAL COATINGS
13:115
0.002 5 mm and which, as indicated earlier, places strict limitations on the usefulness of the coating for protection against severely corrosive liquid environments. The value of rhodium in resisting atmospheric corrosion in environments ranging from domestic to marine and tropical exposure has, however, been amply demonstrated by experience, and it appears probable that further developments in technology may lead to still wider application. In view of the high cost, when tarnish resistance of the surface is the only requirement it is customary to use the thinnest possible coatings of rhodium (0.000 25-0-000 5 mm). Since rhodium deposits in this thickness range, like thin electrodeposits of other metals, show significant porosity, readily corrodible metals, e.g. steel, zinc-base alloys, etc. must be provided with an undercoating deposit, usually of silver or nickel, which is sufficiently thick to provide a fairly high level of protection to the basis metal even before the final precious metal deposit is applied, and, in this way, to prevent accelerated electrochemical corrosion at pores in the rhodium deposit. It is not possible to plate rhodium directly on to reactive metals of the type mentioned above, in view of the acid nature of the electrolyte, but copper and its alloys, e.g. nickel-silver, brass, phosphor-bronze, berylliumcopper, which are of special importance in the electrical contact field, may be plated directly. Even in this case, however, an undercoat is generally desirable. Whether nickel or silver is selected for use as an undercoating is determined by a number of factors, relative resistance to particular corrosive environments being clearly of primary importance. Laister and Benham ” have discussed the respective merits of the two metals on the basis of corrosion tests in a number of environments. Generally speaking, silver is preferred when the composite coating is required to resist exposure to marine or other chloride-containing atmospheres, the potential difference between silver and rhodium in sea-water at 25°C being only 0.05 V2*. A nickel undercoat is better for sulphide atmospheres and for operation at elevated temperatures (up to 500°C). In this connection, it should be noted that rhodium itself will begin to oxidise at temperatures in the range 550-600°C.
Silver is often preferred as an undercoat for rhodium by reason of its high electrical conductivity. A further advantage of silver in the case of the thicker rhodium deposits (0-0025 mm) applied to electrical contacts for wear resistance is that the use of a relatively soft undercoat permits some stress relief of the rhodium deposit by plastic deformation of the under-layer, and hence reduces the tendency to cracking”, with a corresponding improvement in protective value. Nickel, on the other hand, may be employed to provide a measure of mechanical support, and hence enhanced wear resistance, for a thin rhodium deposit. A nickel undercoating is so used on copper printed connectors, where the thickness of rhodium that may be applied from conventional electrolytes is limited by the tendency of the plating solution to attack the copper/laminate adhesive, and by the lifting effect of internal stress in the rhodium deposit. A thickness of 0.000 38 mm may be regarded as a good quality finish for general decorative and industrial use for tarnish protection at normal temperatures. For optimum tarnish resistance at temperatures up to 500”C,
13:116
NOBLE METAL COATINGS
0-00125 mm of rhodium on a nickel undercoat is recommended. In sliding contact applications, where the ability of the coating to withstand some degree of mechanical wear is almost as important as tarnish resistance, the order of thickness employed is 0-0025-0.005 mm and, in a few special circumstances, this may be increased to 0.012 5 mm or more.
Palladium Although satisfactory palladium plating processes have existed for many years, the metal was slow to attain industrial significance as an electrodeposited coating, but then became of considerable interest as an alternative to rhodium or gold in the finishing of electrical contacts, especially in copper end connectors of printed circuits3’. Apart from its relatively low cost, palladium has special technical advantages in this type of application. It may be deposited from neutral or slightly alkaline non-cyanide electrolytes which virtually do not attack the copper-laminate adhesives, t h e deposit shows only a low tensile stress, and it may readily be soldered, whereas rhodium presents some difficulty in this respect. Palladium has good contact properties and, in the electrodeposited condition, has a hardness of 200-300Hv which, while considerably lower than that of rhodium, is higher than that of most gold deposits, and affords a useful degree of wear resistance. Thicknesses of 0,0025-0-005 mm are usual, and the comments made previously regarding the porosity of thin coatings and the importance of undercoatings are applicable here too. In sliding electrical contact applications, palladium plating has been criticised on the basis of a tendency due to its catalytic activity to cause polymerisation of organic vapours from adjacent equipment with the formation of insulating films on the surface3’. This effect is important in certain circumstances, but is not serious in many practical applications 32. Platinum Since the ready workability of platinum permits cladding of base metal with sound coatings which may be as thin as 0.002 5 mm uses of the metal in the electrodeposited condition for corrosion protection are relatively few. As in the case of palladium, electrolytes for platinum plating have been available for many years but interest in the process was greatly increased, chiefly in connection with the plating of titanium for the preparation of inert anodes for electrolytic processes”. Attempts to use bare titanium as an anode in aqueous solutions result in the formation of a resistive oxide coating on the metal which prevents the passage of useful currents below about 15 V applied potential. At this potential, complete breakdown of the firm occurs, with the onset of catastrophic corrosion. The presence of a thin layer of platinum on the titanium surface permits the passage of high currents at voltages well below the critical value, and in this application the presence of discontinuities in the electrodeposited platinum coating does not affect performance, since the exposed basis metal is sealed by a protective anodic film. This composite material, with a coating of platinum up to 0-002 5 mm thick, was adopted for many electrode applications in which platinum-clad base metals or graphite were previously used, e.g. in brine electrolysis, peroxide and per-salt production, electrodialysis, cathodic protection, etc. Studies suggested that under certain conditions platinum would become mechanically detached from titanium anodes owing to attack of the substrate through pores in the coating. Anodes became available with a
NOBLE METAL COATINGS
13: 117
mechanically-clad platinum coating, and alternative coatings, e.g. of platinum-iridium alloy or ruthenium oxide were developed.
Ruthenium, iridium and osmium The use of a fused cyanide electrolyte is the most effective means for the production of sound relatively thick coatings of ruthenium and iridium, but this type of process is unattractive and inconvenient for general purposes and does not therefore appear to have developed yet to a significant extent for industrial application. This is unfortunate, since these metals are the most refractory of the platinum group and in principle their properties might best be utilised in the form of coatings. However, several interesting improvements have been made in the development of aqueous electrolytes. For ruthenium, electrolytes based on ruthenium sulphamateU or nitrosylsulphamate 35 have been described, but the most useful solutions currently available are based on the anionic complex 36- 37 (H, 0 CI, Ru * N Ru Cl,. OH,)'-. The latter solutions operate with relatively high cathode efficiency t o furnish bright deposits up to a thickness of about 0.005 0 mm, which are similar in physical characteristics t o electrodeposited rhodium and have shown promise in applications for which the latter more costly metal is commonly employed. Particularly interesting is the potential application of ruthenium as an alternative to gold or rhodium plating on the contact members of sealed-reed relay switches. Iridium has been deposited from chloride-~ulphamate~~ and from bromide electrolytes3', but coating characteristics have not been fully evaluated. The bromide electrolytes were further developed by Tyrrella for the deposition of a range of binary and some ternary alloys of the platinum metals, but, other than the platinum-iridium system, no commercial exploitation of these processes has yet been made. Electrodeposition of osmium4' was reported from a strongly alkaline electrolyte based on an anionic complex formed by reaction between osmium tetroxide and sulphamic acid. Little is known concerning the general soundness of such coatings, but they appear to show excellent mechanical wearresistance, since in comparative abrasion tests an osmium coating lost only one-quarter the thickness of a hard chromium deposit. Both iridium and osmium have very high melting points and high work functions, which suggest application in the coating of tungsten valve grids to suppress secondary electron emission, but in both cases application is likely to be restricted by the high cost and limited availability of the metals.
Other Coating Techniques 'Brush' plating" is a variant of electrodeposition in which the electrolyte is held in a pad of cotton wool or other absorbent material and applied by wiping over the article to be plated. Though very old in principle, modern developments in equipment and applicational techniques render the method extremely useful in the case of precious metals in view of the possibility of localising the coating t o selected areas. It is also useful in repair and salvage operations in the plating of electronic components. Another method entails application of the coating by spraying, brushing
13: 118
NOBLE METAL COATINGS
or silk-screen printing onto the surface a liquid composition containing organic salts of the metal in a suitable vehicle, which, on firing, decomposes to produce a metal film. This process has been used for many years to apply very thin coatings of gold and other precious metals to non-conductors for decorative purposes, and has served as a basis for technological improvement designed to make it possible to apply thick coatings of platinum in a single a p p l i ~ a t i o n ~Though ~. developed initially for the coating of refractories for critical applications in the glass industry, the process is useful also in the coating of metals carrying refractory oxide films, e.g. titanium, zirconium. It has the merit that the properties of the coating are sometimes closer to those of the pure metal than is generally the case for electrodeposits.
Protection at High Temperatures Although the platinum metals have high melting points, covering a range from 1 552°C (palladium) to approximately 2 500°C (ruthenium and iridium) only platinum retains its freedom from oxide films at temperatures up to the melting point. Palladium and rhodium form stable protective oxide films over a temperature range of approximately 500-1 O0OoC,above which the oxides dissociate. The oxides formed by ruthenium and osmium are readily volatile, hence these two metals are quite unsuitable for high temperature application. The behaviour of iridium in this respect is intermediate between that of rhodium and ruthenium. At temperatures above the melting point of gold, which represents the chief range of interest, the life of a platinum coating on a base metal is limited by the extent to which inter-diffusion with the substrate metal, and gaseous diffusion through the outer coating (leading to formation of base metal oxide initially along grain boundaries of the coating and ultimately at the surface) is possible. The problems involved are exemplified in the application of platinum coatings for protecting molybdenum against oxidation at temperatures in the region of 1200°C in gas turbines, and in the preparation of clad-molybdenum stirrers for molten glass. Useful life of the composite material is obtained only with claddings 0.25-0.5 mm thick, and in this connection Rhys" has demonstrated the importance of an intermediate layer of gold or an inert refractory oxide as a barrier to outward diffusion of m ~ l y b d e n u m ~ ~ . Although electrodeposition permits the coating of relatively complex shapes, the permeability of coatings so applied to gases at temperatures of the order of 1200°C is, in the present stage of development, too great for them to have a protective value comparable to that of wrought metal coatings. For example, an electrodeposit of platinum 0.10 mm thick protected molybdenum for only 16 h in air at 1 20O0C, whereas a mechanical cladding of this thickness had a life of some 300 h under similar conditions. It is possible, however, that modified coatings might be more akin to the pure metal in this respect. Coatings produced by vapour-phase deposition may possibly have advantages in this type of application. An interesting approach to the inter-diffusion problem was made by Rhys& who protected ruthenium-rich ruthenium-gold alloys by palladium-
NOBLE METAL COATINGS
13:119
gold coatings of composition corresponding to the opposite ends of the ‘tie-lines’in the palladium-gold-ruthenium ternary system. Since substrate and coating compositions are in thermal equilibrium, diffusion between the two does not occur to an appreciable extent over long periods at high temperatures. Unfortunately, coating life is again limited by diffusion of oxygen through the coating.
Recent Developments Electrodeposited Coatings
Silver and gold Silver is nearly always deposited from cyanide baths, though other baths have been described. To limit oxidation and polymerisation in high-speed selectiveplating with insoluble anodes, low-cyanide baths have beem developed containing salts such as phosphate47. Silver coatings may blister above 200°C because of oxygen diffusion. A nickel undercoat stops interdiffusion with a copper substrate above 150°C. Alloying with antimony, selenium, sulphur or rhenium increases hardness-the coefficient of friction is also much reduced in the last case4*. Clarke’s study of the porosity of gold deposits lasted for a d e ~ a d e ’ ~ , ~ ~ . Reviewing the topic, Garte concluded that, to reduce porosity: (1) the substrate surface should be smoothed chemically or electrochemically; (2) certain undercoats are beneficial; and (3) plating conditions must be tightly contr~lled’~.Better procedures led to a reduction of thickness requirements-e.g. from 5 to 2.5 km on connectors, provided other requirements are satisfied. Abbott studied the corrosion of contacts, and proposed quality tests in dilute mixtures of hydrogen sulphide, nitrogen dioxide and chlorine in air at controlled temperature and humidity”. These gave good results in a project seeking improved procedures for British and IEC standards”. Hundreds of baths exist for electrodeposition of gold and its alloys53. The latter are more wear resistant, so better for contactss4.Polymers incorporated in cyanide-bath deposits affect wear and contact resistance”.
The Platinum Metals Rhodium Patents have been filed on low-stress deposits, better undercoats and use of soluble anodes; there have been several reviewss6,but no major recent developments. Palladium Advantages have been claimed for new baths (e.g. using chelated complexess7).Antler summarised the use of palladium as coatings, inlays and weldments in electronic connector^'^. Crosby noted that palladium deposits are of two kinds: (1) soft but continuous or (2) hard but porous or cracked. To resist wear and substrate corrosion on contacts, he proposed the application of type 1 (from a bath with tetranitropalladium(I1) anion) over type 2 (from solution containing tetramminepalladium(I1) cation) ’9.
13: 120
NOBLE METAL COATINGS
Industry, however, favours electrodeposited palladium-nickel alloy since it is cheaper than palladium, harder and less prone t o cracking, fingerprinting and formation of polymer filmsw. Its wear resistance is poor, so it is usually given a thin topcoat of hard (sometimes, soft) gold6’. Palladium-silver alloy has greater resistance to fretting. Inlays are common, and coatings will be adopted as deposition processes improve6’. Platinum Platinum-coated titanium is the most important anode material for impressed-current cathodic protection in seawater. In electrolysis cells, platinum is attacked if the current waveform varies, if oxygen and chlorine are evolved simultaneously, or if some organic substances are present 63. Nevertheless, platinised titanium is employed in tinplate production in Japan”. Although ruthenium dioxide is the most usual coating for dimensionally stable anodes, platinum/iridium, also deposited by thermal decomposition of a metallo-organic paint, is used in sodium chlorate manufacture6’. Platinum/ruthenium, applied by an immersion process, is recommended for the cathodes of membrane electrolysis cells&. Characteristics of established platinum plating baths have recently been reviewed67.Advantages have been claimed for new baths based on the complex tetrammineplatinum(l1) cation6*. Ruthenium, iridium and osmium Baths based on the complex anion (NRu,C~,(H,O),)~- are best for ruthenium electrodeposition6’. Being strongly acid, however, they attack the Ni-Fe or Co-Fe-V alloys used in reed switches. Reacting the complex with oxalic acid gives a solution from which ruthenium can be deposited at neutral pH. To maintain stability, it is necessary to operate the bath with an ion-selective membrane between the electrodesw. Iridium and osmium are rarely deposited. A new osmium bath is based on the hexachloroosmate ion 70. Procedures were outlined for depositing osmium on targets for nuclear reactions”.
Other Coating Techniques
The largest uses of platinum group metals in electronics are: ruthenium for resistors and palladium for multilayer capacitors, both applied by thick film techniques”. Most anodes for brine electrolysis are coated with mixed ruthenium and titanium oxide by thermal d e c o m p o ~ i t i o n ~Chemical ~. vapour deposition of ruthenium was patented for use on cutting Protection et H&h Tempef8tufes
The life of gas turbine blades is improved by platinum and/or rhodium, applied below or above, or co-deposited with, aluminised, thermal-barrier or MCrAlY-type layers75. The performance of modified aluminides was demonstrated in long-term engine trials 76. J. EDWARDS F. H. REID
NOBLE METAL COATINGS
13: 121
REFERENCES 1. Angus, H. C., Berry, R. D. and Jones, B., Engineering Materials and Design, 11, 1965,
Dec. (1968) 2. Rinker, E. C., and Duva, R., US Pat. 2 905 601 (1959) 3. Erhardt, R. A., Proc. Amer. Electropl. SOC.,47, 78 (1960) 4. US Pat. 3 057 789 (1962) 5. Rhoda, R. N., Plating, 49 No. I , 69 (1962) 6. Withers, J. C. and Ritt, P. E., Proc. Amer. Electropl. Soc., 44, 124 (1957) 7. Smith, G. R. et al., Plating, 56 No.7, 805 (1969) 8. Atkinson, R. H.,Trans. Inst. Met. Finishing, 36, 7 (1958) 9. Tripler, A. B., Beach, J. G. and Faust, C. L., J. Electrochem. SOC.,195, 1 610 (1958) 10. US Pat. 2 895 889 and 2 895 890 (1959) 11. UK Pat. 808 958 (1959) 12. Stevens, J. M., Trans. Inst. Met. Finishing, 46 No. I , 26 (1968) 13. Johnson, R. W., J. Electrochem. SOC., 108, 632 (1961) 14. Okinaka, Y., Pluting, 57 No. 9, 914 (1970) 15. Rhoda, R. N., Trans. Inst. Met. Finishing, 36, 82 (1959) 16. Pearlstein, F. and Weightman, R. F., Plating, 56 No. 10, I 158 (1969) 17. Laister, E. H.and Benham, R. R., Trans. Insr. Met. Finishing, 29, 181 (1953) 18. Electroplated Coatings of Gold and Gold Alloy, BS 4292 ( 1968) 19. Baker, R. G.,Proceedings of 3rd E.I.A. Conference on Reliable Electrical Connections, Dallas, Texas (1958) 20. US Pat. 2 897 584 (1959) 21. Carte, S., Plating, 53 No. 11, 1 335 (1966) 22. Carte, S., Plating, 55 No.9, 946 (1%8) 22% Antler, M., Plating, 56 No. 10, 1 139 (1969) 23. Clarke, M. and Leeds, J. M., Trans. Inst. Met. Finishing, 43, 50 (1965)and 47, 163 ( 1%9) 24. Leeds, J. M., Trans. Inst. Met. Finishing, 47, 222 (1969) 25. UK Pats. I 003 848 and 1 003 849 (1%5) 26. Egan, T.F. and Mendizza, J., J. Electrochem. SOC., 107, 253 (1960) 27. Krumbein, S. J. and Antler, M., Proceedings ofNEP/COW67 WestMeeting, Long Beach, Calif., Feb. (1%7) 28. Corrosion and its Prevention at Bi-Metallic Contacts, Admiralty and Ministry of Supply Inter-Service Metallurgical Research Council, HMSO, London (1956) 29. Reid, F. H., Trans. Inst. Met. Finishing, 33, 195 (1956) 30. Philpott, J. E.,Platinum Met. Review, 4, 12 (1960) 31. Hermance, H. W. and Egan, T. F., Bell System Tech. Journal, 37, 739 (1958) 32. Reid, F. H., Plating, 52 No. 6, 531 (1%5) 33. Cotton, J. B., Chem. and Ind. (Rev.), 17, 492 (1958) 34. US Pat. 2 600 175 (1952) 35. Reid, F. H. and Blake, J. C., Truns. Inst. Met. Finishing, 38, 45 (1961) 36. Reddy, G. S. and Taimsalu, P., Trans. Inst. Met. Finishing, 47,187 (1969) 37. Bradford, C. W., Cleare, M. J. and Middleton, H., Plat. Mer. Rev., 13, 80 (1969) 38. Conn, G. A., Plating, 52 No. 12, I258 (1%5) 39. Tyrrell, C. J., Trans. Inst. Mer. Finishing, 43, 161 (1965) 40. Tyrrell, C . J., Paper presented at International Metal Finishing Conference, Hanover, May ( 1968) 41. Greenspan, K. L., Engelhard Industries Tech. Bull., 10 No. 2, 48-49,Sept. (1969) 42. Hughes, H. D., Trans. Inst. Met. Finishing, 33, 424 (1956) 43. UK Pat. 878 821 (1957) 44. Rhys, D. W., Proceedings of Symposium sur la Fusion du Verre, Brussels, 6-10 Oct. (1958). Union Scientifique du Verre, 677 45. Safranek, W. H. and Schaer, C. R., Proc. Amer. Electropl. SOC.,43, 165 (1956) 46. UK Pat. I 150 356 (I%5) 47. Blair, A., 54th Metal Finishing Guidebook and Directory, Metal and Plastics Publications, Hackensack, N.J., p. 282 (1986) 49. Leeds, J . M. and Clarke, M., Trans. Inst. Mer. Finishing, 46, I (1968);Clarke, M. and Leeds, J . M., ibid., 46, 81 (1968);Clarke, M. and Chakrabarty. A . M . , ibid., 50, 1 1 ,
13: 122
NOBLE METAL COATlNGS
(1972); Clarke, M. and Sansum, A. J., ibid.. 50, 211. (1972); Clarke, M.. ibid., 51, 150 (1973); Clarke, M. and Subramanian. R., ibid.. 52. 48 (1974) 50. Garte, S. M., in Cold Plating Technology, ed. Reid, F. H. and Goldie, W., Electrochemical Publications. Ayr, p. 295 (1986) 51. Abbott. W. H.. Proc. 12th Int. Conf. on Elecfric Contact Phenomena, Chicago, p. 47 (1984);MaferialsPerformance, 24 No. 8.46 (1985); Proc. 13fhInt. Conf. on Electric Contacts, Lausanne, 343 (1986); Proc. 33rd Int. IEEE/Holm Conf. on Electrical Contacts, Chicago (1987);Proc. 14fh Inf. Cong. on Electric Confucts, Paris (1988);Koch, G. H., Abbott. W.H. and Davis, G. 0.. MaferialsPerformance, 27 No. 3, 35 (1988) 52. Gwynne, J. W. G., EnvironmenfalEngineering, 2 No. 1, 26 (1989) 53. Weisberg, A. M. W., 54fh Metal Finishing Guidabook and Direcfory, Metal and Plastics Publications, Hackensack, N.J.. p. 224 (1986) 54. Antler, M.. Thin Solid Film, 84,245 (1981) 55. Munier, G. B., Plating, 56, 1 151 (1969) 56. Foster, A. J.. Electroplafing Met. Finishing, 28, 8 (1975); Branik. M. and Kummer F., Galvanotechnik,72, I 175 (1981); Kubota. N.. Mer. Finishing, 85 No. 5, 55 (1987) 57. Morrissey. R. J., Platinum Mer. Review, 27, 10 (1983) 58. Antler, M., Platinum Met. Review, 26, 106 (1982) 59. Crosby. J. N.. Proc. Symposium on Economic Useof andSubsfifutionfor Precious Metah in fheElectronics Industry. Amer. Electroplaters' Soc.. Danvers. Mass., September (1980) 60. Pike-Biegunski. M. J. and Bazzone, R., Symposium on Economic Use of and Substitufion for Precious Mefals in fhe Elecfronics Indusfry, Amer. Electroplaters' Soc., Danvers, Mass., September (1980); Schulze-Berge, K.. Galvanotechnik. 72, 932 (1981); Whitlaw, K. J.. Trans. Inst. Met. Finishing, 60, 141 (1982) 61. Sato. T.,Matsui, Y.,Okada, M., Murakawa, K. and Henmi, Z., 26th Holm Conf. on ElectricalContacfs, Chicago, p. 41 (1980);Graham, A. H..30lhHolm Conf on ElectricalConfacfs. Chicago, p. 61 (1984); Whitlaw. K. J.. Trans. Insf. Met. Finishing. 64,62 (1986) 62. Sturzenegger, B. and Puippe, J. C., Platinum Mer. Review, 26, 117 (1984);Nobel, F. I., Martin, J. L. and Toben, M. P., 13fhSymp. on Plating in the Electronics Industry. Amer. Electroplaters' and Surface Finishers' Soc.. Kissimmee. FA (1986) 63. Hafield, P. C. S., Plafinum Met. Review. 27. 2 (1983) 64. Saito, H., Int. Tinplate Conf., London (1984) 65. Modern Chlor-alkali Technology, Ellis Horwood, Chap. 9 (1980) 66. Grove, D. E., Platinum Met. Review, 29, 98 (1985) 67. Baumgartner, M. E. and Raub, C. J., Plufinum Met. Review, 32, 188 (1988) 68. Skinner, P.E.,Inf. Surf. Finishing '89, Brighton, 12 April (1989) 69. Crosby, J. N.. Symposium on Economic Use of and Substitufionfor Precious Metals in the Electronics Industry, Amer. Electroplaters' Soc., Danvers. Mass, September (1980) 70. Notley, J. M., Trans. Insf. Mef. Finishing, 50, 58 (1972) 71. Stuchbery, A. E. , Nucl. Insfr. Methods Phys. Res., 211, 293 (1983) 72. Davey. N. M. and Seymour, R. J.. Platinum Met. Review, 29. 2 (1985) 73. Modern Chlor-alkali Technology. Ellis Horwood, Vol. 2, Chap. 13 (1983) 74. U.K. Patent 1 499 549 75. U.K. Patents I 495 626 and 2 041 246; U.S. Patents 4 070 507,4 346 137,4 399 199 and 4 530 805; European Patent Appl. 107, 509 76. Cocking, S.L., Richards, P. G. and Johnston, G. R.. Surf. Coat. Technol., 36,37 (1988)
14
14.1 14.2 14.3 14.4 14.5 14.6
PROTECTION BY PAINT COATINGS
14:3 Paint Application Methods 14:7 Paint Formulation The Mechanism of the Protective Action of Paints 14:22 Paint Failure Paint Finishes for Industrial Applications
14:39 14:53
Paint Finishes for Structural Steel for Atmospheric Exposure Paint Finishes for Marine Application
14.7 14.8 Protective Coatings for Underground Use 14.9 Synthetic Resins 14.10 Glossary of Paint Terms
14: 1
14:69 14:76 14:89 14:105 14:114
14.1 Paint Application Methods
Methods of applying paint today are numerous and it is impossible to list and describe them in detail in a section of this size. Where corrosion resistance of the finished article is a major consideration, it is possible to apply controls to ensure that maximum corrosion protection is obtained from the selected application process.
Application Methods In any method of application, either an excess of paint is applied and the surplus is removed, or the desired thickness of paint is put on directly. For simplicity, methods can be divided into cycling and non-cycling processes, Le. procedures which return surplus paint to the plant and those which d o not. Cycling Processes
For these processes, the paint must not only meet the specification and end requirements of the finished product, but it must also be stable over long periods under operating conditions. The simplest form of the cycling paint process is the standard dip tank which can vary from a simple hand dip in a container of paint, to a sophisticated mechanised system. In the conveyorised process, articles pass into the dip tank, are withdrawn at a controlled rate, and, after draining, are allowed to air dry or are cured in a stoving oven. In such a system a large volume of paint is involved with a large surface area exposed t o atmosphere. The bulk paint is continuously contaminated by the excess paint draining from the articles and any extraneous substances introduced with the articles. These contaminants can be controlled by filtering and circulating the paint. Constant control of viscosity, and paint composition are necessary if uniform results are to be obtained and, in modern sophisticated plants, these factors are regulated continuously by automatic equipment. FIow coating In this process, paint is directed on to the workpiece from a series of strategically placed jets in an enclosed area and the excess paint 14:3
14:4
PAINT APPLICATION METHODS
drains back to the main supply tank. The workpieces are then allowed to drain in a solvent-saturated zone (to delay evaporation and permit paint flow) before passing on to a flash-off zone and final stoving. Articles are usually hung from a monorail and accurate jigging is essential. Curtain coating With this process, paint falls in a continuous curtain from a closely machined gap in a header tank on to the flat article passing below on a horizontal conveyor; the excess paint is collected in the main tank and then passed up to the header tank. It is an ideal method of applying thicker coatings (60 pm and above) to sheet metalwork.
Electrodeposition This method of paint application is basically a dipping process’. The paint is water-based and is either an emulsion or a stabilised dispersion. The solids of the paint are usually very low and the viscosity lower than that used in conventional dipping. The workpiece is made one electrode, usually the cathode, in a d.c. circuit and the anode can be either the tank itself or suitably sized electrodes sited to give optimum coating conditions. The current is applied for a few minutes and after withdrawal and draining the article is rinsed with de-ionised water to remove the thin layer of dipped paint. The deposited film is firmly adherent and contains a minimum of water and can be stoved without any flash-off period. This process is used for metal fabrications, notably car bodies. Complete coverage of inaccessibleareas can be achieved and the corrosion resistance of the coating is excellent (Fig. 14.1). 1
2
3
6
5
6
’
8
7
9
10
Fig. 14.1 Typical plant layout for electropainting (courtesy Stein, Atkinson Stordy Ltd.)
LEGEND 1. 2. 3.
Alkali degrease Cold water rinse Hot water rinse Zinc phosphate and Cold water rinses Demineralised water rinse
::} 6. 7.
8. 9. 10. 11. 12. 13. 14.
Paint dip First rinse (town water) Second rinse (demineralised water) Stoving oven Cooling leg Jig strip Jig rinse
Fluidised bed This process is used for powder coating’. Basically, the equipment consists of a dip tank with a perforated shelf near the bottom. The powder is placed on this shelf and low pressure air is fed under the perforated shelf, resulting in a cloud of fine powder in the body of the dip tank. The article is heated to a little above the melting point of the powder and is then dipped into the fluidised bed for a short period. It is then withdrawn
PAINT APPLICATION METHODS
14:5
and the coating cured in an oven. Thick films are formed, but it is difficult to obtain uniform film thickness with varying gauge metals.
Non-cycling Processes
These are processes in which the paint is used once only and the excess material is not returned to the main bulk. A typical example is the normal spray system in which the paint is fed to the spray gun, atomised by air jets and applied to the article as a stream of small droplets. The excess paint and overspray are deposited on the walls of the booth and are collected by various methods depending on the type of spray booth used3. There are many modifications of the conventional spray system which include the following.
Hot spray The paint is heated to 60-80°C. The hot paint flows better and gives better coverage. Transfer efficiency is increased and drying time is shortened. Airless spray In this process, a high pressure (12-35 MN/m2) is applied to the paint to force it through a fine orifice in the spray gun. This process allows rapid transfer with reduced overspray. Air assisted airless spray This concept is a combination of air spray and airless methods. Paint can be atomised with full spray patterns at low pressures. Turbulence is reduced significantly and overspray is minimised. Electrostatic spraying This process takes advantage of electrostatic attraction. It is suitable for applying either liquid or powder coatings. Paint droplets or powder particles are passed through a powerful electrostatic field and become charged. They are attracted to the earthed workpiece and coat not only the front surface but also the back surface to a large extent if the object is not too deep or too wide. Automatic plant is available with spray guns on reciprocators. Electrostatic hand guns The same principle has been adopted on portable hand guns. These modifications facilitate the coating of electrically shaded areas. Brush Application of paint by brushing is still a commonly used method for maintenance painting. Coverage of large areas is slow and the quality of finish achieved relies heavily on the skill and motivation of the painter. Paint Processes
Although the results obtained from a particular process depend almost entirely on the nature and design of the article, the plant layout and other local conditions play an important part. Table 14.1 gives some indications of the limitations of the processes mentioned. Pressures to reduce atmospheric pollution, increase safety in the workplace, and save energy have all influenced. paint application methods in
14:6
PAINT APPLICATION METHODS
Table 14.1 Summary of process limitations Process
Uses
Lirnilalions and defecfs
Dipping
All types of articles of suitable shape and size
Flow coating
Suitable for use on most articles. Gives good penetration into pores of castings Only suitable for flat sheets of uniform dimensions Suitable for most articles
Curtain coating
Electrodeposition
Fluidised bed Conventional spray
Airless spray Electrostatic spraying (automatic)
Brushing
Most suitable for small articles Suitable for most components Suitable for most components With suitably designed plant, most articles can be painted with this process Suitable for use on most articles, but use limited for economic reasons. Results rely almost entirely on skill of the operator
Requires large throughput Gives uneven film thickness on large flat sheets (from top to bottom) Does not cover sharp edges or interior of channel sections, etc. Possibility of solvent wash, i.e. solvent vapour from the hotter area condensing on cooler areas during flash-off and stoving Similar to dipping, but the defects not so marked. Tendency for greater solvent loss. On suitable articles this method will produce a uniform paint film providing the surface of the article is perfectly clean. Gives uniform film even on rough surfaces. With suitable plant will completely coat interior surfaces, sharp edges, etc. but generally only economic on mass production Produces thick films. Varying metal gauge could produce uneven films and weak spots Very difficult to obtain adequate cover on inside corners, etc. and results depend on the skill of the operator Superior penetration in awkward areas to normal spraying. Only economical in long runs. Very difficult to obtain adequate cover in electrically shaded areas and interiors of hollow articles Labour cost is high
recent years. Emerging trends can be expected to continue and automation wherever possible, with increased use of water-thinned coatings and powder coatings, can be expected. W. H . TATTON N.R. WHITEHOUSE REFERENCES I . Lambourne, R. (Ed.), Paint and Surface Coafings, Ellis Horwood (1987) 2. Harris, S. T., The Technology of Powder Coatings, Portcullis Press (1976) 3. Chandler, K . A. and Bayliss, D. A., Corrosion Profection of Steel Structures, Elsevier (1985)
14.2 Paint Formulation Constituents of Paint Paint consists essentially of a pigment dispersed in a solution of a binding medium. The binding medium or binder, which in most instances is organic, will decide the basic physical and chemical properties of the paint, but these will be modified by the nature and proportion of pigments present. In a decorative finish, for example the primary function of the pigment is to provide colour, but in a primer it should contribute to the durability of the whole system in a variety of ways depending on the substrate to which it is applied. The sole function of the volatile component or solvent is to control the viscosity of the paint for ease of manufacture and for subsequent application. Thereafter the solvent evaporates and is lost. A further class of paint is based on a binder emulsified in water. This type of paint has increased in importance in recent years and there is considerable evidence that good anticorrosive properties can be built into paints which themselves are thinned with water. The development of these paints is attracting considerable attention because of the absence of fire hazards, a low level of harmful vapours, and performance comparable with products carried in stronger solvents. Since the possible variations in binder alone are limitless, it is possible to produce an infinite number of paints. As the range of raw materials available to the formulator becomes wider, their chemical purity is continually being improved. Mathematical models of binders can be constructed using computers and it is usually possible to predict fairly accurately the properties of a particular formulation before it is made. Nevertheless, the formulation of paints for specific purposes is still considered to be very much a technological art. Although formulation is an art, science finds its place in the characterisation of the raw materials, in the design and testing of the series of experimental formulae and in the interpretation of the results. In addition to possessing an intimate knowledge of pigments, binders and solvents, the paint formulator must also be well acquainted with raw material costs and availability, paint making machinery, and the market’s performance requirements.
Basic Principles of Formulation Before any attempt is made to formulate a paint it is necessary to know a great deal about the conditions under which it will be used and subsequently 14: 7
14:8
PAINT FORMULATION
exposed. The more comprehensive the information relating to requirements, the greater the probability of achieving complete success with the first practical trial. The conditions under which the paint will be dried, e.g. air drying or stoving, and the properties demanded in service will dictate the choice of binder. This, in turn, will limit the choice of solvents, and further constraints may be imposed by the presence of potential fire hazards at the user’s works, or the problem of toxic fumes in enclosed working spaces. The quantity of solvent in the paint will depend on the intrinsic viscosity of the binder and the paint viscosity appropriate to the method of application, e.g. brushing, spraying, dipping, electrostatic spray, electrocoating, flow coating, etc. A single paint will rarely possess all the required properties and it therefore becomes necessary to formulate a system comprising a primer, a finish, and possibly one or more intermediate coats. A primer, as its name implies, is the first coat of a system. Its principal functions are to provide adhesion and good protection to the substrate. The manner in which these properties are obtained will vary with the substrate, but frequently involves the use of a large proportion of a specific pigment. This may impose a restriction on the colour and gloss, and probably on other desirable properties such as durability. The finish or final coat must make up the deficiencies of the primer by affording it protection and providing the required colour and degree of gloss. These last two requirements will dictate the quality and quantity of the pigments to be used. In the majority of cases maximum durability will be produced by a multicoat system comprising priming and finishing paints only. Other considerations, however, such as uniformity of colour and smoothness of surface, may make it desirable to introduce intermediate coats, e.g. putty, filler and undercoat. The appearance of the final film or of the final painted structure is of some importance, and a final colour coat, which may contribute very little resistance to corrosion, may be necessary. Putties are heavy-bodied pastes of high pigment content that are applied by knife for rough filling of deep indentations, more especially in rough castings. Fillers are used to level-out shallower imperfections. Ease of flatting is an important consideration and to a large extent influences the composition and proportion of the pigment mixture. Undercoats are invariably of high pigment content and low gloss. Their function is to provide a foundation that is uniform in both colour and texture for the finishing coats, thereby enhancing the final appearance of the completed system. On occasions, it is possible to achieve the same result by substituting an additional coat of finish for the undercoat, and this may improve the durability of the system. When considering the number of coats of paint to achieve adequate protection it is worth noting that the cost of applying the paint usually far outweighs the cost of the paint. This is leading to a class of relatively more expensive paints which can be applied in very thick coats. The increasing mechanisation of painting methods, such as the airless spraying of structural areas, influences the paint formulator in the selection of the most suitable formulations.
PAINT FORMULATION
14:9
However comprehensive the information relating to requirements, the paint technologist cannot proceed with the problem of formulating a suitable paint unless he is in possession of considerable data on the properties of the raw materials at his disposal, but within the scope of the present work it is impossible to do more than indicate the important properties of the more commonly used ingredients*.
Binding Media The most important component in the majority of paints is the binding medium, which determines the physical and chemical properties of the paint. Blends of binding media are often used to impart specific properties to the dry paint film or to suit a particular application method. The compatibility of chemically different types of binders is an important factor to be taken into account by the paint formulator. These properties will be modified, however, to a greater or lesser extent by the nature and quantity of the other components, more especially the pigment. The general characteristics of various binding media are given in Table 14.2.
Drying Oils (treated and untreated)
Apart from being basic ingredients of oil varnishes and alkyd resins, drying oils are occasionally used as the binder in paint. Linseed oil is an important drying oil and is the only one used to any extent in its natural state. Its main use is in corrosion-inhibiting primers. Disadvantages of paints based on raw linseed oil are their very slow drying, lack of gloss, and inability to flow sufficiently for brushmarks to level out. The mechanism of the protective action of these primers is considered in Section 14.3. Heat-treated oils fall into three categories: boiled oils, stand oils and blown oils. Boiled oils are prepared by heating linseed oil in the presence of catalysts. They have somewhat higher viscosities and better drying properties due to their higher molecular weights and more complex molecular structure than raw linseed oil. They are commonly used in oil-based primers and, in conjunction with oil varnishes, in undercoats. Stand oils range in viscosity up to about 20 N s/m2 and are prepared by heat-polymerising linseed oil either alone or in admixture with tung oil. They are used mostly in combination with oil varnishes and alkyd resins to improve application properties and, when desirable, to increase the total oil: resin ratio. Blown oils differ from stand oils in that they are partially oxidised in addition to being polymerised. The oxidation is achieved by blowing air through the heated oil. This treatment results in a product having poor drying properties, and blown oils are therefore effective plasticisers and are used as such in nitrocellulose finishes. * A list of relevant standard texts is given at the end of this section, but the principal sources of detailed information are in the form of technical data sheets issued by raw-material suppliers.
Table 14.2
General characteristics of binding media
L
?! Type of binder
Mode of drying
Solvents
L
Alkali resistance
Water resistance
Solvent resistance
Fair
Bad
Fair
Poor
Poorlfair
Binder for anticorrosive primers for wire-bushed steel Slow drying
Bad
Fairlgood
Poor
Fairlgood
Pale-coloured finishes that yellow on exposure
resistance
Raw linseed oil Boiled linseed oil Stand oils
Air drying Oxidative polymerisation
Oleoresinous varnishes
Air drying and/or Aliphatic and/or stoving aromatic Condensation hydrocarbons and/or oxidative polymerisation
Fair
Air drying Oxidative polymerisation
Aliphatic hydrocarbons
Fair
Medium oil length alkyds
Air drying and/or stoving Oxidative and/or condensation polymerisation
Aliphatic and aromatic hydrocarbons
Fair
Poor
Fairly good
Fair
Very good
Short oil length alkyds
Stoving Condensation polymerisation
Aromatic hydrocarbons
Fair
Fair
Good
Fairly good
Very good
Modified alkyds
Air drying and/or stoving Oxidative and/or
A wide range of solvents depending on
Fair
Fair
Usually good
Fair/
Usually good very good
Long oil length alkyds
Aliphatic hydrocarbons
Exterior
Acid resistance
Special features
0
-0
2:
3 Bad
Fair
Poor
Very good
2
: r
5
8 z
Mode of drying
Type of binder
Solvents
Table 14.2
(continued)
Acid resistance
Alkali resistance
Waier resisiance
Solvent resistance
Exrerior weathering resisiance
Special features
Urea formaldehyde Stoving /alkyd blends Condensation polymerisation
Aromatic hydrocarbons and alcohols
Fairly good Fairly good
Very good
Good
Fair
Water white Gives white finishes of excellent colour
Stoving Condensation polymerisation
Aromatic hydrocarbons
Fairly good Fairly good
Very good
Good
Very good
Water white Gives white finishes of excellent colour
Air drying Epoxide/aliphatic Addition amine or polymerisation polyamide blends
Blends rich in higher ketones
Fairly good
Poor
Very good
Epoxide/amino or phenolic resin blends
Stoving Addition and condensation polymerisation
Blends rich in higher ketones and alcohols
Good
Epoxide/fatty acid esters
Air drying and/or stoving Oxidative polymerisation
Aliphatic and/or aromatic hydrocarbons
Fair
Polyester/ polyisocyanate blends
Air drying or stoving Addition polymerisation
Blends rich in Fairly good ketones and esters Alcohols excluded
Melamine formaldehyde/ alkyd blends ~
~~
Very good
Good
Very good
Very good
Fairly good/ Finishes need to be good supplied in two separate containers and mixed just prior to use
->z -0
+
n
%
I
5
Good
5
z
~
Fair
Fairly good
Poor
Poor/fairly good
Good
Fairly good
Very good
Very good
Finishes need to be supplied in two separate containers and mixed just prior to use
L
P
li: ..
c
t 4
Type of binder
Mode of drying
Solvents
Table 14.2
(continued)
Acid resistonce
Alkali resistance
Water resistance
Solvent resistance
Exterior resistance
Special features
Vinyl resins
Air drying Solvent evaporation
Blends usually rich in ketones
Very good
Very good
Very good
Poor
Good
Fire hazard Flash point usually below 23°C
Chlorinated rubber
Air drying Solvent evaporation
Aromatic hydrocarbons
Good
Good
Verygood
Poor
Good
Very poor heat resistance
Cellulose nitrate
Air drying Solvent evaporation
Blends of esters, alcohols and aromatic hydrocarbons
Fairly good
Bad
Good
Poor
Very good
Fire hazard Statutory regulations governing use
-z 2 -3
5
5a z
PAINT FORMULATION
14: 13
Oil Varnishes
The current practice is to classify as ‘oil varnishes’ all varnishes and paint media prepared from drying oils and natural or preformed oil-free synthetic resins. Examples of such resins are rosin, rosin-modified phenolics and oilsoluble 100% phenolics. The introduction of the resin results in improved drying and film properties. Oil varnishes are capable of producing primers for ferrous metals which perform excellently on clean o r pretreated surfaces, but they have not the same tolerance as their oil-based counterparts for wirebrushed rusted surfaces. The undercoats that follow are frequently also based on oil varnishes. Since the individual members of this group of media differ considerably in properties, so also do the finishes that can be made from them. As a class, however, they are generally inferior to the better alkyds for durability under normal conditions. A particular exception is the tung-oil 100% phenolic type of medium, which produces finishes with very good resistance to water and mildly acidic or alkaline conditions; pale colours, however, discolour by ‘yellowing’ on exposure.
Alkyd Resins*
Introduced some 50 to 60 years ago, alkyd resins quickly established themselves and are still widely used. They are essentially polyesters of moderate molecular weight prepared by the reaction of polyhydric alcohols with the mixtures of monobasic fatty acids and dibasic acids. Ethylene glycol (dihydric), glycerol (trihydric) and pentaerythritol (tetrahydric) are the more commonly used alcohols. Phthalic anhydride is the most commonly used dibasic acid. Isophthalic acid and adipic acid are also used for special purposes. An unsaturated dibasic acid called maleic anhydride is widely used and can give polymers of high molecular weight. There is a very wide range of fatty acids available, the ultimate choice being dependent upon the properties required. The fatty acid is frequently added in the form of a vegetable oil which is a tri-ester of fatty acid and glycerol. Individual alkyds are usually described in terms of the proportion and type of fatty acid and of the alcohol that they contain. Thus a 70% linseed-oil pentaerythritol alkyd would be expected to comprise linseed oil fatty acids, pentaerythritol and phthalic anhydride, with an equivalent of 70% linseed oil calculated on the weight of the non-volatile resin. The members of this family are so diverse that only the fundamental properties can be considered here. For convenience they will be subdivided according to their use, i.e. (a) air-drying, (b) stoving, (e) plasticising and ( d ) modified alkyds. Air-drying alkyd resins Alkyds capable of air drying do so through the oxidation of the drying oils that they contain. Such alkyds are consequently *The synthesis of various types of resins is given in Section 14.9.
14: 14
PAINT FORMULATION
usually of long oil length*, i.e. 65 to 75% and based on the tetrahydric alcohol pentaerythritol. The oils most commonly used are linseed and soya bean. The latter imparts more freedom from yellowing to white- and palecoloured finishes, especially where there is little natural light. Tung oil is less frequently used because it promotes yellowing. Sunflower oil, cottonseed oil, safflower oil and tall-oil fatty acids are being used more and more frequently for high quality white gloss paints. In some cases, the fatty acid is partly replaced by a synthetic organic monobasic acid which modifies the polymer solubility and film properties. Among the outstanding properties of airdrying alkyds are (a) their convenience in use and (b) their ability to give finishes of unrivalled durability in all but heavily polluted atmospheres. Where premature failure does occur, it probably results from poor surface preparation or an inadequate priming system. Air-drying alkyds may also be used for the production of primers and undercoats. In the case of primers, the shorter the oil length of the binder the faster the drying, but the lower the tolerance for wire-brushed rustedsteel surfaces. Alkyd-based undercoats are not significantly different in performance from those based on oil varnishes; the choice is frequently dictated by economic considerations. Stoving alkyd resins When drying is to be effected by stoving, the oxidative properties of drying oils are of less importance, and advantage can be taken of the tougher properties of the phthalic ester component of the resin. Hence stoving alkyds may be based on drying or semi-drying oils, and the oil length is invariably shorter than for air-drying finishes, usually in the range 50 to 65 Yo. For high-quality stoving finishes the alkyd is frequently blended with a lesser quantity of an amino resin. This reduces the stoving schedule and enhances most of the physical properties of the finish. The inclusion of a small proportion of rosin during the manufacture of the alkyd will also improve application and initial film properties. Such binders are commonly used for stoving primers and for cheap stoving finishes that will not be subjected to exterior exposure.
Plasticising alkyd resins The term plasticking alkyd is a loosely used one, embracing those alkyds that are employed in conjunction with a larger proportion of another, and usually harder, stoving resin, e.g. an amino resin. In certain compositions the shorter-oil-length stoving alkyds referred to before may function as plasticisers, but in general plasticising alkyds are of even shorter oil length, usually 40 to 50%, and consist of fatty acids of nondrying oils, e.g. coconut oil. *Oil lengfh is the relative proportion of oil to resin in a binding medium. It is expressed in a variety of ways, including simple ratios and, as in the present text, the percentage oil calculated on the weight of the non-volatile binder. In the case of traditional varnishes it is a precise value calculated directly from the relative quantities of oil and resin used. With more complex binders, including alkyds, such simple calculations are not possible; various assumptions must be made and the values then obtained are essentially theoretical. The terms long, medium, and shorr oil length are used loosely to indicate respectively, high, medium and low proportions of oil. There are no generally agreed limits but in the present context long oil length is applied to binders containing more than 65% oil, medium oil length to binders having between 65 and 50% oil and short oil length to those containing less than 50% oil.
PAINT FORMULATION
14: 15
Modified alkyd resins In this group one finds styrenated alkyds, vinyl toluenated alkyds, oil-modified vinyl resins, acrylic alkyds, silicone alkyds and polyurethane alkyds. The modifying component usually has a number of effects. It always increases the molecular weight of the alkyd polymer, and may impart hardness, durability, or chemical resistance. It also affects the solubility of the polymer in solvents. Amino Resins
The two amino resins in common use are urea formaldehyde and melamine formaldehyde, and most stoving finishes contain one or the other. They have many properties in common; urea formaldehyde, however, while substantially cheaper, has poor exterior durability, whereas melamine formaldehyde imparts excellent exterior durability. As they are both water white they give white finishes of excellent colour, with the additional advantage of retaining their colour on over-stoving. Urea formaldehyde is commonly used in conjunction with a lesser quantity of an alkyd to give finishes with excellent resistance to water and mild chemicals, which are therefore well suited to use on domestic equipment, e.g. washing machines. Melamine formaldehyde is also used in conjunction with an alkyd, but the ratio varies considerably according to the ultimate use of the finish. Epoxide Resins
Epoxide resins are essentially long-chain polyhydric alcohols with epoxide groups at either end. They make useful building blocks because both the hydroxyl and the epoxide groups are available for reaction with other compounds. Aliphatic polyamines, amine adducts and polyamides react with epoxide resins at normal temperatures to give complexes with outstanding chemical resistance. Paints based on this type of reaction must be supplied in two separate containers, one containing the epoxide resin and the other the ‘curing agent’, the two being mixed in prescribed proportions immediately before use. Amino resins and certain phenolics react with epoxide resins at elevated temperatures to give somewhat similar results. As the combination is nonreactive at normal temperatures this type can be supplied in the form of ready-for-use stoving finishes. Epoxide resins can be esterified with fatty acids to give media ranging from air-drying to stoving types. The presence of fatty acid reduces the chemical resistance to the same order as that of the alkyds. It is nevertheless sometimes found advantageous to use an epoxy ester for certain specialised purposes. Polyurethanes
Polyurethanes are essentially the reaction products of polyisocyanates and polyesters containing free hydroxyl groups. They are comparable with the
14: 16
PAINT FORMULATION
epoxide types in that they possess excellent chemical resistance but, by contrast, have very good colour and gloss retention. It is necessary to supply the air-curing types in two-pack containers. One-pack stoving types are formulated by using less reactive 'masked' isocyanates. Another important group of products is the polyurethane oils and polyurethane alkyds. In these binders, the chemical linkages are a mixture of the highly chemically resistant urethane links and the less resistant ester links. It is very misleading and difficult to classify their properties because the ratio of urethane to ester linkages varies widely from one product to another. In many properties, they are very similar to alkyds, but usually possess more rapid drying, even at low temperatures, and give a slightly harder film initially. They are, however, less flexible than comparable alkyds and often slightly worse for exterior durability. The name polyurethane on a product cannot be taken as an indication of chemical resistance unless it is a two-pack polyurethane or a moisture-curing polyurethane. Moisture-curable urethane systems (one-pack) can be considered as twocomponent systems which use atmospheric moisture as the second component. One-pack urethane coatings can be produced that are similar in physical properties to the two-pack systems for almost all applications. These highly complex systems can have a great deal of flexibility. Claimed advantages are: a one-pack system, rapid cure, even at low temperatures, excellent chemical and abrasion resistance and good flexibility. Although these systems have been available for some time in other countries of Europe, they are only recently beginning to be of interest in the UK. Vinyl Resins
A wide range of resins prepared by polymerisation of compounds containing vinyl groups is available. Those most commonly used in paint manufacture are of the following types:
(a) Essentially copolymers of vinyl chloride and vinyl acetate or vinyl ether. (b) Emulsified vinyl acetate copolymers. (c) Acrylic modified alkyds, etc. A characteristic of the group (a) of resins is that they air-dry solely by solvent evaporation and remain permanently solvent soluble. This fact, combined with the need to use strong solvents, makes brush application very difficult, but sprayed coats can be applied at intervals of one hour. A full vinyl system such as (a) possesses excellent chemical and water resistance. Many members of group (a)have very poor adhesion to metal, and have therefore been exploited as strip lacquers for temporary protection. Excellent adhesion is, however, obtained by initial application of an etching primer; the best known of such primers comprises polyvinyl butyral, zinc tetroxy-chromate and phosphoric acid. The chemical resistance of group (b), frequently used in emulsion or latex paints, is often upset by the presence of water-soluble emulsion stabilisers and thickeners, which remain water soluble in the dried paint film. Group (e) has already been discussed under the heading modified olkyds.
PAlNT FORMULATION
14: 17
Chlorinated Rubber
Chlorinated rubber is soluble in aromatic solvents, and paints made from it dry by solvent evaporation alone. In contrast to the vinyls, there is less difficulty in formulating systems that are suitable for brush application. It has excellent resistance to a wide range of chemicals and to water, but as it is extremely brittle it needs to be plasticised. T o preserve chemical resistance it is necessary to use inert plasticisers such as chlorinated paraffin wax. Due t o the presence of ozone depleting solvents, chlorinated rubber coatings are being phased out and largely replaced by vinyl acrylic coatings which have very similar performance and can be formulated from lower aromatic or aliphatic solvents.
Paints containing nitrocellulose are of importance in relation to the protection of metals because of their excellent durability combined with very fast drying. They may, on this account, be used for mass-production work where stoving facilities are not available, and it is interesting to recall that had such paints not been available the mass production of motorcars would inevitably have been delayed. Their rapid drying makes them unsuitable for brush application to large areas, but a more serious disadvantage is the fire hazard associated with nitrocellulose, and users of such paints must comply with stringent statutory regulations. Nitrocellulose alone will not give a continuous coating. It must, therefore, be blended with other components comprising a plasticiser and a hardening resin. An extensive range of such products is available, the ultimate choice depending on the properties required.
Miscellaneous Binders
These consist of the following: (a) Silicone polymers having high heat stability and excellent chemical resistance are available. They are very expensive and hence are not commonly found in paint coatings. (b) Silicate binders are used in conjunction with zinc powder to give paints of excellent corrosion resistance. The organo-silicates, e.g. ethyl orthosilicate, are most commonly used. The full potential of this type of binder has probably not yet been exploited. (c) Thixotropic binders. ( d ) Fluorinated polymers such as polytetrafluorethylene are available for specialised applications. Titanium polymers with excellent heat stability are available. New polymers are being developed all the time, especially by the plastics industry, and the aforementioned groups of binding media are merely those commonly used, and do not constitute a complete list. It may, however, illustrate the range of products
14: 18
PAINT FORMULATION
and properties available to the paint formulator for the selection of the most appropriate binders in the paint.
Pigments In a finish, the function of the pigment is to provide colour, but in a primer it should contribute to the protection of the metal substrate and enhance the adhesion of the finishing system. Pigments are essentially dry powders which are insoluble in the paint medium and which consequently need to be mixed in it by a dispersion technique. They range from naturally occurring minerals to man-made organic compounds and may be subdivided broadly into priming pigments, colour pigments, extenders and metal powders. Extenders are chemically inert, naturally occurring or synthetic, inorganic compounds which are included to confer specific properties to the paint. Such properties include suspending the pigment to prevent the formation of hard settlement, improvement of ‘build’, and the provision of ‘tooth’ or ‘key’ to improve intercoat adhesion. Red lead, zinc chromate, calcium plumbate and zinc dust were for many years of special importance as pigments for metal primers. When dispersed in raw or lightly-treated linseed oil, the first three possess the ability to inhibit the corrosion of mild steel and will function very well on wire-brushed rusted surfaces. In other media the tolerance towards rusted surfaces decreases with decreasing quantities of available oil, but performance on clean steel will usually be maintained and often improved. Zinc phosphate is now probably the most important pigment in anticorrosive paints. The selection of the correct binder for use with these pigments is very important and can dramatically affect their performance. Red lead is likely to accelerate the corrosion of non-ferrous metals, but calcium plumbate is unique in providing adhesion to newly galvanised surfaces in the absence of pretreatment, and is claimed to behave similarly on other metals in this group. Primers containing 93-95% zinc dust by weight in non-saponifiable media provide sacrificial protection to clean steel (see Section 14.3). Pigments for finishes are selected on the basis of their colour, but special attention must be paid to inertness in the chosen binder and stability and light fastness under the conditions of application .and exposure. Flake pigments such as aluminium and micaceous iron oxide give finishes of lower moisture-vapour permeability than conventional pigments, and consequently contribute to better protection.
Paint Additives A paint rarely consists solely of pigment dispersed in a solution of a binder. For one reason or another, small quantities of ancillary materials called additives are included. The oldest and still the most important are the ‘driers’ which are used in all air-drying and many stoving paints containing drying oils. They are organic salts of certain metals, notably cobalt, calcium, barium, zirconium and manganese, with lead very much in decline.
PAINT FORMULATION
14: 19
Anti-oxidants are of value in preventing skinning in containers, but care must be taken to ensure that they d o not adversely affect the drying properties of the paint. They are also used to reduce the oxidation of the excess paint that drains from dip-coated articles back into the dip tank. Surface-active agents are used to facilitate the dispersion of pigments, to keep the pigment in suspension during storage of the paint, and to preserve the homogeneity of pigment mixtures while a paint is drying. Another group of additives are used as thickeners and antisettle agents. They affect flow and reduce sagging of thick films.
Solvents The term solvent is loosely applied to the volatile component of a paint, though this component may in fact consist of a true solvent for the medium plus a non-solvent or diluent. When such a mixture is used, usually with the aim of reducing cost or obtaining a higher solids content at a given viscosity, care must be taken to ensure that the diluent is more volatile than the true solvent in order that the medium shall remain in solution during the drying process. A small amount of a particular solvent may be needed to aid application, t o enable the release of small air bubbles in sprayed films, or to activate thickeners. Classification of solvents is normally by chemical composition, e.g. aliphatic or aromatic hydrocarbons, alcohols, esters, ketones, etc. In addition to knowing which are appropriate for use with particular media, the paint formulator must also be acquainted with the fire hazards associated with the individual solvents and mixtures thereof, and the toxicity of various mixtures. Regulations governing Occupational Exposure Limits and ventilation requirements play an important role in the choice of solvents in a coating composition. There are both statutory and transport regulations relating to the use and carriage of paints, according to their ‘flash point’ and the composition of their solvent.
Paint-making Machinery For the purpose of paint formulation the most important units of equipment are the laboratory ball mill, bead mills and high speed dispersers. The most common, the ball mill, consists of a cylindrical porcelain vessel a little more than half filled with steel, porcelain balls or pebbles. Pigment, together with sufficient binder and solvent to make a free-flowing mix, is loaded into the mill until it is approximately two-thirds full. The mill is then closed and fixed into a device whereby it is made to rotate about its major axis. Normally, a period of about 16hours is required for thorough dispersion of the pigment, whereupon the mill-base is emptied out and blended with the remainder of the ingredients. The selection of the appropriate type of machinery and the determination of the optimum conditions for bulk manufacture is usually the subject of discussion between the paint formulator and a senior member of the
14 :20
PAINT FORMULATION
production department. For most pigmentlresin bases there will usually be more than one milling machine suitable for producing the required degree of dispersion. Different types of dispersion equipment can be classified on the basis of milling action, and by considering how pigment agglomerates are broken up. All mills operate by crushing or shearing or both together, and each one will work best within fairly close limits of mill-base viscosity. Machines working mainly by crushing require a low mill-base viscosity and those relying on shearing need a high one. In principle, the selection of dispersion equipment for a given purpose is very simple. The obvious choice is the one that will give the required degree of dispersion most economically. In practice it is not so easy. Availability of equipment, the nature of the raw materials, mill-base formulation, batch size, product type and the time available all influence the decision of which machine to use.
Formulating a Paint The paint technologist entrusted with the task of formulating a paint to meet a specified set of conditions must first decide what type of binders he should use and the type of solvent blend that this will require. In the particular case of a finish, he must then select the pigments most likely to give the required colour, bearing in mind any limitations imposed by his choice of binder system or by the conditions to which the paint will be subjected. With the aid of a palette knife, weighed quantities of the several pigments are ground by hand into a binder such as linseed oil until an approximate match to the colour pattern is obtained. The consistency of this paste can be adjusted instrumentally to obtain the maximum work from the particular dispersion unit to be used. On the basis of this rough estimate, a premix is prepared with the appropriate quantities of pigment, binder and solvent, and a high-pigment-content mill-base is produced. From this and subsequent mill-bases, ordered series of paint samples are prepared and tested to establish the following data: (a) The most appropriate pigment: binder ratio. (b) In the case of a composite binder system, the optimum proportion of each. (c) The optimum addition of additives, e.g. driers, that may be necessary. ( d ) The appropriate viscosity and solvent composition.
If, as is possible, the first mill-base gives a poor colour match, the relative proportions of the several pigments are suitably adjusted in subsequent experiments. The ultimate aim should be to obtain a colour that is slightly deficient in the stronger tinting strength pigments, since it is more convenient to produce an exact match to the pattern by making small additions of high tinting strength mill-bases than by making larger additions of weaker bases. Assuming that a paint with satisfactory properties has now been produced, there remains the possibility that it may deteriorate on storage. This must be investigated, and any faults that develop must be corrected.
PAINT FORMULATION
14:21
The ability to apply knowledge gained by practical experience is the hallmark of a good paint formulator, for it frequently enables him to proceed to an acceptable basic formulation without delay. The greater part of the limited time that he has been allowed can then be devoted to perfecting his product. It is worthy of note, however, that the development of new products for exterior exposure is inevitably a slow process because there is no accelerated weathering cycle that can be relied upon to reproduce faithfully the effects of natural weathering. M. W. O’REILLY J. T. PRINGLE BlBLlOGRAPHY Paint Technology Manuals, Oil and Colour Chemists’ Association, Chapman and Hall, London Banor, A. Paints and Coatings Handbook for Contractors, Architects, Builders and Engineers, Structures Publishing Co. Turner, G . P. A., lntroducrion to Paint Chemistry and Principles of Paint Technology, Chapman and Hall, London Weismantel. G. E., Puinr Handbook. McGraw-Hill Book Company Payne, H. F., Organic Coating Technology, Vols. I and 11, Chapman and Hall, London
14.3 The Mechanism of the Protective Action of Paints From time to time astronomical estimates are made of the annual destruction of metals, particularly iron and steel, by corrosion (Section 1.1). Paint is one of the oldest methods used for delaying this process and consequently it is somewhat surprising that its protective action has only recently been systematically examined. Since iron is the commonest structural material, the following discussion will be limited to the behaviour of this metal. The general principles can readily be extended to non-ferrous metals.
The Corrosion of Iron and Steel (Sections 1 . 4 and 3.1) Corrosion is essentially the conversion of iron into a hydrated form of iron oxide, i.e. rust. The driving force of the reaction is the tendency of iron to combine with oxygen. It has long been known that iron is not visibly corroded in the absence of either water or oxygen. The overall reaction in their presence may be written: 4Fe
+ 30, + 2H,O
-+
2Fe20,.H20
When the supply of oxygen is restricted the corrosion product may contain ferrous ions. The overall reaction can be broken down into two reactions, one producing electrons and the other consuming them: 4Fe -+ 4Fe2+ + 8e (anodic reaction) 202
+ 4H,O + 8e
+
8 0 H - (cathodic reaction)
or 4Fe
+ 20, + 4 H 2 0
--t
4Fe(OH),
In the presence of oxygen the ferrous hydroxide will be converted into rust, Fe,O, .H,O. Ferrous hydroxide is soluble @To) in pure water, but slight oxidation renders it appreciably less soluble. Thus in the presence of water and oxygen alone the corrosion product may be formed in close contact with the metal and attack will consequently be stifled. In the presence of an electrolyte such 14 :22
THE MECHANISM OF THE PROTECTIVEACTION OF PAINTS
14:23
as sodium chloride, however, the anodic and cathodic reactions are modified, ferrous chloride being formed at the anode and sodium hydroxide at the cathode. These two compounds are very soluble and not easily oxidised, so that they diffuse away from the sites of formation and react at a distance from the metal surface to form ferrous hydroxide, or a basic salt, which then combines with oxygen to form rust, with the regeneration of sodium chloride:
+ 2NaOH 4Fe(OH), + 0, FeCl,
-+
+
Fe(OH),
+ 2NaCl
2Fe,O,.H,O
+ 2H,O
Consequently rust is formed at a distance from the metal and stifling cannot occur. It follows that when iron rusts, the conversion is accompanied by a flow of electrons in the metal from the anodic to the cathodic regions, and by the movement of ions in solution. This conclusion has been firmly established by Evans’ and his co-workers, who have shown that, in the case of a number of metals under laboratory conditions, the spatial separation of the anodic and cathodic zones on the surface of the metal was so complete that the current flowing was equivalent to the corrosion rate (see Section 1 .ti). In order to inhibit corrosion, it is necessary to stop the flow of current. This can be achieved by suppressing either the cathodic or the anodic reaction, or by inserting a high resistance in the electrolytic path of the corrosion current. These three methods of suppression are called cathodic, anodic and resistance inhibition respectively (Section 1.4). The effect of paint films on the cathodic and anodic reactions will now be considered and the factors which influence the electroIytic resistance of paint films will be discussed.
The Cathodic Reaction The cathodic reaction in neutral solutions usually involves oxygen, water and electrons :
0,
+ 2H,O + 4e
--t
40H-
If a paint film is to prevent this reaction, it must be impervious to electrons, otherwise the cathodic reaction is merely transferred from the surface of the metal to the surface of the film. Organic polymer films do not contain free electrons, except in the special case of pigmentation with metallic pigments; consequently it will be assumed that the conductivity of paint films is entirely ionic. In addition, the films must be impervious to either water or oxygen, so that they prevent either from reaching the surface of the metal. The rate of corrosion of unpainted mild steel immersed in sea-water was found by Hudson and Banfield* to be O.O89mm/y. Hudson’ obtained a similar average value for steel exposed in the open air under industrial conditions (0.051 mm/y at Motherwell and 0.109mm/y at Sheffield). This rate of corrosion corresponds to the destruction of 0.07 g/cm2 per year of iron. Assuming that the corrosion product was Fe,O,.H,O, this rate of
14:24
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
attack represents the consumption of 0.011 g/cm2 per year of water and 0.03 g/cm2 per year of oxygen. Diffusion of Water
The diffusion of water through paint films has been measured by various workers. The weight of water which could diffuse through three clear vehicles and eight paint films, each 0.1 mm thick, at 85-100% r.h. has been calculated on the assumption that the water would be consumed as soon as it reached the metal surface, i.e. that the rate-controlling step was the rate of diffusion of water through the film, and is shown in Table 14.34.5. Table 14.3 Diffusion of water through paint films of thickness 0.1 m m Rate of water consumed (g cm-2y-’)
Vehicle
Pigment
Glycerol phthalate varnish Phenolformaldehyde varnish Epoxy coal tar Glycerol phthalate varnish Phenol formaldehyde varnish Linseed oil Ester gum varnish
None None None Flake aluminium Flake aluminium Lithopone White lead/ zinc oxide Iron oxide 15% P.V.C. Iron oxide 35% P.V.C. Iron oxide 35% P.V.C. Iron oxide 35% P.V.C.
Linseed penta-alkyd Linseed penta-alkyd Epoxypol yamide Chlorinated rubber
Reference
0.825 0.718
0.391 0.200 0.191 1.125 1 ’ 122 0.840
5
0.752
5
1.810
5
1.272
5
~
Nore. Unpainted steel consumes water at a rate of 0.008-0.023
g cm-’y-’
By means of an ingenious instrument which measured the ‘wetness’ of a painted surface, Gay6 found that although the relative humidity of the atmosphere varies appreciably, this is not reflected in the behaviour of paint films. He found that under normal conditions paint films are saturated with water for about half their life, and for the remainder the water content corresponded with an atmosphere of high humidity; furthermore, the relative humidity of sea-water is about 98%. It follows from Table 14.3 that the rate at which water passes through paint and varnish films is many times greater than the water consumed by an unpainted specimen exposed under industrial conditions or immersed in the sea. Diffusion of Oxygen
The diffusion of oxygen through polymer films has been examined by a number of workers. Guruviah5 measured the permeability to oxygen of films cast from five paints (Table 14.4) and compared the results with the
14~25
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
corrosion rates of painted steel panels, when exposed t o salt spray and humidity for 1 OOO h. He concluded that ‘the low corrosion rate could be explained by the low permeability to oxygen of the films’; however, when his values for the permeability are plotted against the corrosion it is clear that this conclusion is without foundation. The weight of oxygen which could diffuse through unit area of a 0.1 mm thick film under a pressure gradient of 2 kN/m2 of oxygen has been calculated, and is shown in Table 14.4’*’,*. B a ~ m a n nhas ’ ~ claimed that these figures are about 100 times too high, but this is because he compared the amount which could pass through in a day with that passing in a year. Haagen and Funke5’ concluded that the permeation of water was too great and that of ions too small to be the controlling factor and suggested that the rate controlling step was the rate of the diffusion of oxygen. However, if this were the case then painted steel upon exposure should corrode at a rate varying from that of unpainted steel to about a tenth of that value. Since painted steel upon exposure does not corrode immediately at this rate, it is concluded that the rate of the diffusion of oxygen is not the controlling factor. Table 14.4 Diffusion of oxygen through paint films of thickness 0.I mm Rate of Vehicle
Pigment
oxygen consumption (g c m - * y - ’ )
Asphalt Epoxy coal tar Polystyrene Polyvinyl butyral Asphalt Linseed penta-alkyd
None None None None Talc Iron oxide 15% P . V . C . Iron oxide 35% P . V . C . Iron oxide 35% P.V.C. iron oxide 35% D.V.C.
0.053 0.002
Linseed penta-alkyd Epoxypol yamide Chlorinated rubber
Reference
0.013
0.027 0.039 0.003 0.003 0.002 0.006
Note. Unpainted steel consumes oxygen at a rate of 0.020-0.030g crn-’y-’
The general conclusion drawn from these considerations is that paint films are so permeable to water and oxygen that they cannot inhibit corrosion by preventing water and oxygen from reaching the surface of the metal, that is to say they cannot inhibit the cathodic reaction.
The Anodic Reaction The anodic reaction consists of the passage of iron ions from the metallic lattice into solution, with the liberation of electrons, which are consumed at the cathode by reaction with water and oxygen. There are two ways in which the anodic reaction can be suppressed:
14:26
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
(a) If the electrode potential of iron is made sufficiently negative, positively
charged iron ions will not be able to leave the metallic lattice, i.e. cathodic protection. (b) If the surface of the iron becomes covered with a film impervious to iron ions, then the passage of iron ions into solution will be prevented, i.e. anodic passivation. Cathodic Protection (Chapter 7 01
In order to make the potential of iron more negative, the iron must receive a continuous supply of electrons. As has already been pointed out, polymer films do not contain free electrons; there remains the possibility of obtaining these from a pigment. The only pigments which contain free electrons are metallic ones, and such pigments will protect iron cathodically if the following conditions are fulfilled: (a) The metallic pigment must be of a metal less noble than iron, otherwise
the iron will supply electrons to the pigment, which will be protected at the expense of the iron. (b) The pigment particles must be in metallic, Le. electronic, contact with each other and with the coated iron; if they are not the movement of electrons cannot occur. It has been shown’ that zinc dust is the only commercially available pigment which fulfils both conditions. Paints capable of protecting steel cathodically can be prepared with zinc dust, provided that the pigment content of the dried film is of the order of 95% by weight; both organic and inorganic binders have been used, the latter being very useful when resistance to oil or organic solvents is required. These paints are quite porous and function satisfactorily only in the presence of an electrolyte-e.g. water containing a trace of salt, or acid -which completes the circuit formed by the two metals. It might be thought that the useful life of these paints is limited to the life of the electronic contact between the zinc particles, but this is not correct. Under normal conditions of exposure the electrons supplied by the zinc to the steel are consumed at the surface of the steel by reaction with water and oxygen (cathodic reaction), with the formation of hydroxyl ions. Consequently the surface becomes coated with a deposit of the hydroxides, or carbonates, of zinc, calcium, or magnesium, which blocks the pores in the film and renders it very compact, adherent and impervious. Thus, although metallic contact between the steel and the zinc dust particles is essential in the early stages of exposure, the paints provide good protection after that contact has been lost. Paints containing less zinc dust have been known for a long time, but as the zinc dust concentration is decreased, protection at scratch lines or at gaps in the coating, decreases; however, such paints frequently afford good general protection owing to the formation of deposits (consisting of oxides and carbonates) on the metal at the base of the coating. Recently it has been pointed out that manganese satisfies both conditions, since the oxide film around the particles contains ions in two states of oxidation, and it has been claimed that cathodically protective paints can be
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
14127
prepared with this pigment Io. Exposure trials in this country have indicated that at an inland site their behaviour is comparable with the zinc dust controls, but that they were inferior t o zinc-rich paints under severe marine conditions. It has been suggested that they might be of interest where zinc was unsuitable owing to toxicity”.
Anodic Passivation (Section 10.8)
When a piece of iron is exposed to the air, it becomes covered with an oxide film. Upon immersion in water or solutions of certain electrolytes, the airformed film breaks down and corrosion ensues. In order to prevent corrosion the air-formed film must be reinforced with similar material, or a ferric compound, and there are two ways in which this may be achieved: (a) The pigment may be sufficiently basic to form soaps when ground in
linseed oil; in the presence of water and oxygen these soaps may autoxidise to form soluble inhibitive degradation products. (b) The pigment itself may be an inhibitor of limited solubility. Basic pigments Typical pigments in this class are basic lead carbonate, basic lead sulphate, red lead and zinc oxide. It has been established that water becomes non-corrosive after contact with paints prepared by grinding basic pigments in linseed oil”; it was also shown that lead and zinc linoleates, prepared by heating the oxide with linseed oil fatty acids in xylene, behave in a similar way. Later this observation was extended to the linoleates of calcium, barium and strontiumI3. Determinations have been made of the solubility of lead linoleate prepared in the absence of oxygen and extracted with air-free watert4.Under these conditions, lead linoleate had a solubility of 0-002Vo at 25°C and the extract was corrosive when exposed to the air. When, however, the extraction was carried out in the presence of air, the resulting extract contained 0.07% solid material and was non-corrosive. It was concluded that in the presence of water and oxygen lead linoleate yielded soluble inhibitive degradation products. In order to obtain information regarding the composition of these degradation products, aqueous extracts of the lead soaps of the linseed oil fatty acids were analysed, mainly by chromatography. The extracts contained formic acid 46070, azelaic acid 9% and pelargonic acid and its derivatives 27%, the remaining 18% consisting of a mixture of acetic, propionic, butyric, suberic, pimelic and adipic acids. It was shown that whereas the salts of formic acid were corrosive, those of azelaic and pelargonic acid were very efficient inhibitors. Ramshaw I s has obtained information regarding the origin of these various acids by examining the degradation products of the lead soaps of the individual acids present in linseed oil. He found that it was only the unsaturated acids which degraded to give inhibitive materials, and that the lead soaps of linoleic and linolenic acid yielded in addition short-chain acids which were corrosive. He also examined the relative inhibiting powers of the lead, calcium and sodium salts of a range of mono- and di-basic acids
14: 28
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
in the pH range 4-6 at concentrations of to I O - ’ N ’ ~ . Under these conditions the lead salts were always more efficient than the sodium and calcium salts, and the optimum efficiency occurred when both the mono- and di-basic acids had a chain length of 8-9 carbon atoms. The mechanism of inhibition by the salts of the long chain fatty acids has been examinedI7. It was concluded that, in the case of the lead salts, metallic lead was first deposited at certain points and that at these points oxygen reduction proceeded more easily, consequently the current density was kept sufficiently high to maintain ferric film formation; in addition, any hydrogen peroxide present may assist in keeping the iron ions in the oxide film in the ferric condition, consequently the air-formed film is thickened until it becomes impervious to iron ions. The zinc, calcium and sodium salts are not as efficient inhibitors as the lead salts and recent work has indicated that inhibition is due to the formation of ferric azelate, which repairs weak spots in the air-formed film. This conclusion has been confirmed by the use of I4C labelled azelaic acid, which was found to be distributed over the surface of the mild steel in a very heterogeneous mannerI8. Zinc phosphate was introduced as an inhibitive pigment by Barraclough and and in the early tests vehicles based on drying oils were used. Later it was claimed57that it was an effective inhibitive pigment when used with all paint media in current use. Variable results have been reported with this pigment and an examination of its inhibitive action5’ has led to the conclusion that under rural and marine conditions, where the pH of the rain-water is above 5 , it behaves as an inert pigment owing to its limited solubility. However, in industrial and urban areas, where the pH of the rain-water may be in the region of 4 or lower, it is converted into the more soluble monohydrogen phosphate. This reacts in the presence of oxygen, with the steel surface to form a mixture of tribasic zinc and ferric phosphates, which being insoluble protects the steel from further attack. Soluble pigments The most important pigments in this class are the metallic chromates, which range in solubilities from 17.0 to O-ooOOSg/l CrO$”’. An examination has recently been carried out of the mechanism of inhibition by chromate ions and it has been shown by chemical analysis of the stripped film, Mossbauer spectroscopy and electron microprobe analysis that the airformed film is reinforced with a more protective material in the form of a chromium-containing spinel” (Chapter 17). The situation is, however, complicated by the possibility that some chromates, particularly the basic ones, may inhibit through the formation of soaps. There is evidence that lead chromate can function in this way. It has been found that red lead, litharge and certain grades of metallic lead powder render water alkaline and inhibitive 12; this observation has been confirmed by Pryor”. The effect is probably due to a lead compound, e.g. lead hydroxide, in solution. Since, however, atmospheric carbon dioxide converts these lead compounds into insoluble basic lead carbonate, thereby removing the inhibitive materials from solution, these pigments may have only limited inhibitive properties in the absence of soap formation. Work by BeckmannZ3 indicated that lead hydroxide was only very slightly better as an inhibitor than sodium hydroxide, and the mechanism
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
14: 29
of inhibition is probably similar to that suggested for alkaline Owing to the low dielectric constant of organic vehicles, these pigments can ionise only after water has permeated the film, consequently their efficiency is associated with the nature of the vehicle in which they are dispersed, a point which is sometimes overlooked when comparing the relative merits of chromate pigments.
Resistance Inhibition It has been shown that paint films are so permeable to water and oxygen that they cannot affect the cathodic reaction, and that the anodic reaction may be modified by certain pigments. There are, however, many types of protective paint which do not contain inhibitive pigments. It is concluded that this class of paint prevents corrosion by virtue of its high ionic resistance, which impedes the movement of ions and thereby reduces the corrosion current to a very small value. It is assumed that conduction in polymer films is ionic - it is difficult to see how it could be otherwise-and the factors which break down this resistance, or render it ineffective, will now be considered. The effective resistance of paint films may be influenced by ions derived from three sources: (a) Electrolytes underneath the film. (6) Ionogenic groups in the film substance. (c) Water and electrolytes outside the film, i.e. arising from the conditions of exposure.
Electrolytes Underneath the Film
Atmospheric exposure trials, carried out in Cambridge, established the fact that when rusty specimens were painted in the summer, their condition, after some years’ exposure, was very much better than that of similar specimens painted in the winter’’. It was found that steel weathered in Cambridge carried spots of ferrous sulphate, deeply imbedded in the rust, and that the quantity of ferrous sulphate/unit area was very much greater in the winter than in the summer26;this seasonal variation was attributed to the increased sulphur dioxide pollution of the atmosphere in the winter, caused by the combustion of coal in open grates. It was concluded that there was a causal relationship between the quantity of ferrous sulphate and the effective life of the paint. It was suggested that these soluble deposits of ferrous sulphate short-circuit the resistance of the paint film and, since paint films are very permeable to water and oxygen, the ferrous sulphate will become oxidised and hydrolysed with the production of voluminous rust, which will rupture the film at numerous points, thus giving rise to the characteristic type of failure seen on painted rusty surfaces. It can be claimed that the problem of painting rusty surfaces is now understood. A method for estimating the ferrous sulphate content of any rusty surface has been put forwardz6,but the amount of ferrous sulphate
14:30
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
which can be tolerated by various paints has not yet been established. Thus it is bad practice to apply paints to surfaces carrying electrolytes.
lonogenic Groups in the Film Substance
Ionogenic (ion-producing) materials may be present, in the form of electrolytes, in both the pigments and the vehicle. Their presence in the pigments may be eliminated by the selection of suitable raw materials by the paint manufacturer, consequently it does not concern us here, but it is of importance to consider the possibility of the existence of ionogenic groups, such as carboxyl groups, in the polymer itself. When paint films are immersed in water or solutions of electrolytes they acquire a charge. The existence of this charge is based on the following evidence. In a junction between two solutions of potassium chloride, 0.1 N and 0.01 N , there will be no diffusion potential, because the transport numbers of both the K + and the C1- ions are almost 0 - 5 . If the solutions are separated by a membrane equally permeable to both ions, there will still be no diffusion potential, but if the membrane is more permeable to one ion than to the other a diffusion potential will arise; it can be calculated from the Nernst equation that when the membrane is permeable to only one ion, the potential will have the value of 56 mV. It is easy t o measure the potential of this system and it has been found” that membranes of polystyrene, linseed oil and a tung oil varnish yielded diffusion potentials of 43-53 mV, the dilute solution being always positive to the concentrated. Similar results have been obtained with films of nitrocellulose**, cellulose acetate”, alkyd resin and polyvinyl chloride3’. This selective permeability is ascribed to the presence on the membrane of a negative charge, which is attributed to carboxyl groups attached to the polymer chains. Paint films can, therefore, be regarded as very large anions. It has been shown3’ that the charge influences the distribution of the primary corrosion products, and recent work has indicated that the existence of carboxyl groups in the polymer film has an important influence on its behaviour when immersed in potassium chloride solutions.
Water and Electrolytes outside the Film
Here we are concerned with the effect of ions in the environment on the resistance of polymer films. Kittelberger and Elm3*measured the rate of diffusion of sodium chloride through a number of paint films. Calculations based on their results2’ showed clearly that the rate of diffusion of ions was very much smaller than the rate of diffusion of either water or oxygen. Furthermore, they found that there was a linear relationship between the rate of diffusion and the reciprocal of the resistance of the film. This relationship suggests that the sodium chloride diffused through the membrane as ions and not as ion pairs, since the diffusion through the film of un-ionised material would not affect the resistance, because if a current is to flow, either ions of similar charge
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
14: 3 1
must move in one direction, or ions of opposite charge must move in opposite directions. An examination has, therefore, been made of the effect of solutions of potassium chloride on the electrolytic resistance of films cast from a pentaerythritol alkyd, a phenolformaldehyde tung oil and an epoxypolyamide v a r n i ~ h ~Potassium ~ ’ ~ ~ . chloride was chosen because its conductivity is well known and unpigmented films were first examined in order to eliminate the complexities of polymer/pigment interaction. The experimental procedure consisted of casting the varnish on glass plates by means of a spreader bar having an 0.102 mm (0.004 in) gap; this produced a wet film 0.051 mm (0.002 in) thick that yielded a dried film of 0.025 mm (0.001 in). This standard thickness was used throughout and resistances are quoted in cm’. The cast films were dried for 48 h in a glove box followed by a further 48 h in an oven at 65°C. The films were then soaked in water and removed from the plates. Portions were mounted in glass cells which were filled with potassium chloride solution; two Ag/AgCl electrodes were inserted into the limbs of the cells and the unit was placed in a thermostat. The resistance of the films was determined, from time to time, by connecting the cells in series with a known resistance and applying a potential of 1 V to the combination; the potential drop across the standard resistance was measured by means of a valve potentiometer. When samples of about 1 cmz were taken from a single cast film of 100 x 200mm of a number of paint and varnish films, their resistances varied with the concentration of potassium chloride solution in one of two ways (Fig. 14.2). Either the resistance increased with increasing concentration of the electrolyte (inverse or I conduction) or the resistance of the film followed that of the solution in which it was immersed (direct or D conduction). The percentage of I and D samples taken from different castings varied, but average values for a number of castings were 50% D for the pentaerythritol alkyd and the tung oil phenol formaldehyde varnishes, 57% for urethane alkyd, 76% for epoxypolyamide and 78% for polyurethane varnishes 50. The effect of iron oxide, zinc oxide and red lead on the percentage of D areas has been determined. Three vehicles were used, a pentaerythritol alkyd, a tung oil phenolic and an e p ~ x y p o l y a m i d e ~In~ .the case of iron oxide, the D areas increased with all three vehicles; in contrast zinc oxide had very little effect on the percentage D areas. However, red lead when dispersed in the alkyd and tung oil vehicles behaved in a similar way to iron oxide, whereas red lead when dispersed in the epoxypolyamide vehicle had very little effect. A careful examination has been made of the properties of I films when immersed in solutions of electrolytes. It was found that when a film of a pentaerythritol alkyd varnish was transferred from 0.001 N KCl to 3 - 5N KCl its resistance rose, fell upon returning it to the 0-001N KCl, rose again to the same high value when immersed in a sucrose solution isotonic with 3.5 N KCI and fell to the original value when returned to the dilute KCI sohtion (Fig. 14.3). It was concluded that the changes in resistance were dependent only upon the available water in the solution and were associated, therefore, with the entry of only water into the varnish fiw.
14: 32
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS 10.51 I X
r
g
I
8.0 -
\
7.5
-
7.0
-
I
I
X I
I
l
l
13I
I
I
-5
mrn __
In contrast, D films followed the resistance of the solution in which they were immersed, and this behaviour was originally explained by assuming that D films contained holes, or pores, filled with solution that controlled the resistance of the film. Thus a typical value for the resistance of a D film in 3.5 N potassium chloride is 1O8Qcm2and if this resistance was due to a pore, then it would have a radius of about 500A. In order to test this
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
14: 33
explanation the distribution of I and D areas in a given piece of film has been determined by means of a series of gaskets fitted into a dismountable cell36.It was found that I films were free from D areas, but that in the case of the three vehicles examined, samples of D films always contained a mixture of I and D areas in an interlocking mosaic structure. It was concluded that those portions of the film having D properties were distributed over an appreciable area of the sample and not confined to a single area, as would have been the case had the sample contained a single pore. It was concluded that D conduction cannot be attributed to the presence of pores, unless they were of molecular dimensions. In general, the water uptake of D films tended to be higher than that of I films, but a more significant difference was shown by microhardness measurements. The results obtained with a11 three vehicles showed that the D areas were significantly softer than the I areas and that the distribution of the hardness values corresponded to that of the resistances. It was concluded that these films have a very heterogeneous structure and that I and D areas are brought about by differences in crossIinking density within the film. An investigation has been made of the factors which control I and D conduction and it has been found that the difference is only one of degree and not of kind3’. Thus, if the varnish films are exposed to solutions of decreasing water activity, then the resistance falls with increasing concentration of electrolyte, but a point is eventually reached when the type of conduction changes and the films exhibit I-type behaviour. It appears that D films can be converted into I films, the controlling factor being the uptake of water. The discussion so far has been limited to the behaviour of polymer films after immersion in potassium chloride solutions for only a short time. When varnish films were immersed in potassium chloride solutions for a month or more a steady fall in resistance took place. Further experiments indicated that the effect was reversible and dependent on both the pH of the solution and the concentration of potassium chloride. It was concluded that an ion exchange process was operative3’. In view of this, the properties of Ifilms were examined after they had been subjected to increasing amounts of ion exchange34. In order to do this, detached films were exposed at 65°C for 7 h to a universal buffer adjusted to a suitable pH and the resistance of the film measured at 25°C in 3 N and 0.001 N potassium chloride. The results obtained with a pentaerythritol alkyd are shown in Fig. 14.4 from which it can be seen that as the pH of the conditioning solution increased, the resistance of the film fell, until at a pH of about 7.5 it suddenly dropped. The resistance of the film then followed that of the solution in which it was immersed, Le. it became a D-type film. Similar results were obtained with films of a tung oil phenolic varnish, although in this case the change-over point occurred at a higher pH, i.e. about 9. In the case of the epoxypolyamide varnish, however, as the pH increased the resistance of the film at first rose, then at about pH 8.8 it started to fall until at pH 11 the change-over in the type of conduction occurred. This suggests that the resin was acting as a zwitterion with an isoelectric point at about pH 8.8. Thus before the isoelectric point the membrane would be positively charged and an increasing concentration of hydroxyl ions would
14: 34
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
x
in 3~
0
in 0.001~KCI
KCL
Fig. 14.4 Variation of the resistance of I films (log scale) with the pH of the conditioning soht ions
depress the ionisation of the ionogenic groups; above the isoelectric point the membrane would be negatively charged and ion exchange with potassium ions would take place. This conclusion was confirmed by diffusion potential measurements. In the case of all three varnishes after ion exchange had taken place, a point was reached when the type of conduction changed from I to D. The change-over in the type of conduction was found to occur at the same pH as a fall in the temperature coefficient of resistance, and the lower value corresponded to that of the aqueous solution. The phenomenon of ion exchange has been confirmed by chemical analysis3’. Films were exposed to potassium chloride solutions of increasing pH, ashed and their potassium content determined by flame photometry. It was found that the potassium content of the films increased as the pH of the solutions rose until saturation was reached at a value which corresponded to that of the change-over in the mechanism of conduction. It was concluded that the change-over in the mechanism of conduction corresponded to the point at which the exchange capacity of the film had reached its limit. Rothwell 38 found by resistance measurements that ion exchange occurred in films of eight unpigmented varnishes, and he confirmed this for penta-
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
14:35
erythritol alkyd films by the determination of the uptake of radioactive potassium in the form of 42KCI; however, his films, with one exception, were all D type. Fialkiewicz and Szandorowski 39 examined the penetration of wSr and 36CIions through air-dried films of a styrenated alkyd pigmented with iron oxide over the range 10-60% P.V.C. (pigment volume concentration) and found intense penetration of the strontium cations, but negligible penetration of the chloride ions. Ulfvarson et ai." examined the ion exchange properties of free films of a soya alkyd, and later Khullar and Ulfvarson4' extended the examination to clear films of 20 vehicles. They concluded that those binders with low ion exchange capacities provided the best protection. In a later study4*they examined the relationship between ion exchange capacity and corrosion protection of 22 paints based on three alkyd binders and concluded that ion exchange was not the dominating factor, but a secondary one. This conclusion was confirmed by van der H e ~ d e nwho ~ ~ ,suggested that a process of ion exchange combined with the diffusion of cations into the film was operative.
Physical Factors Affecting Resistance
The influence of temperature, the concentration of the electrolyte, film thickness and solvent on the resistance of paint and varnish films is discussed below. Temperature An examination has been made of the effect of temperature on the structural changes in polymer films produced from the three vehicles described earlier". Three methods were used: dilatometry, water absorption and ionic resistance. It was concluded that dilatometry was the most reliable method and water absorption is difficult to determine. Both methods use appreciable quantities of film, which contain both D and I areas. Resistance measurements, however, can be carried out on small areas of film and the relative properties of D and I areas studied. It was established that significant changes in resistance took place at the transition temperature and consequently sharp changes in protective properties. The resistance always fell with an increase in temperature and this may provide an explanation for the fact that accelerated tests using the same corrosion cycle, may not produce the same results if carried out at different temperatures. Concentration of Electrolyte Myer and applied the Donnan equilibrium to charged membranes and developed a quantitative theory of membrane selectivity. They expressed this selectivity in terms of a selectivity constant, which they defined as the concentration of fixed ions attached to the polymer network. They determined the selectivity constant of a number of membranes by the measurement of diffusion potentials. Nasini et and Kumins4' extended the measurements to paint and varnish films.
14:36
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
When the Donnan equilibrium is operative the entry of ions into the membrane is restricted. Consequently as the concentration of ions in the solution increases the resistance of the membrane remains constant until the concentration of ions in the solution reaches that of the fixed ions attached to the polymer network. At this point their effect will be swamped and the movement of ions will be controlled by the concentration gradient. Films of a pentaerythritol alkyd, a tung oil phenolic and an epoxypolyamide pigmented with iron oxide in the range 5-7% P.V.C. were exposed to solutions of potassium chloride in the range 0.0001-2.0 M4*.It was found that in all cases the resistance of the films steadily decreased as the concentration of the electrolyte increased. Since the resistances of the films were at no time independent of the concentration of the electrolyte, it was concluded that the Donnan equilibrium was not operative and that the resistance of the films were controlled by the penetration of electrolyte moving under a concentration gradient.
Film Thickness Varnishes prepared from the three standard polymers were cast at two thicknesses and the percentage of D areas compared with that obtained from films produced by casting one thin coat, allowing it to dry and then casting a second coat on top5'. Similar results were obtained from all three varnishes and the results obtained with the epoxylyamide varnish are given below. Thickness of coating (pm)
Single coat Single coat Double coat
35-40
070
D Type
80
75-80
50
70-75
0-5
Earlier it was shown that D type areas are small; consequently the chance of Dareas overlapping each other is low. It follows that two coats of all three
varnishes, which are based on crosslinking polymers, are more effective in improving the resistance of the films than single coats of equal thickness.
Solvents All the films discussed so far have been cast from paints or varnishes containing solvents. In order to examine the effect of solvents, films of a solvent-free epoxypolyamine were cast, mounted in cells and their resistances measured in dilute and concentrated potassium chloride solution5'. All the films had I properties with resistances in the range 10ko-10L2 $2cm2. It appears that during the drying of paint or varnish films the presence of solvent molecules interferes with the process of cross-linking; consequently the films have a heterogeneous structure and films of improved protective quality arise when solvents are eliminated. It is suggested that future work should be directed towards the pigmentation of solvent-free systems, either with inert pigments, when they would form coatings of high electrolytic resistance which would protect by the exclusion of ions, or as sealing coats applied over primers containing inhibitive pigments.
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
14: 37
Prediction of Performance
If protection by paints or varnish films is due to their ability to restrict the penetration of corrosive ions, then it follows that resistance measurements should form the basis of the prediction of their behaviour. In 1948 Bacon et al.52measured the resistance of over 300 paint systems immersed in seawater using a d.c. technique, and concluded that for good performance coatings should have a resistance in excess of 1O8Qcm2.Coatings having resistances in the range 106-10*Qcm2 were found to be unreliable, and those of lower resistance behaved poorly. It has frequently been suggested that during d.c. measurements the specimens become polarised, consequently a.c. current should be used. A comparison of d.c. and a.c. methods has been made and it was found that at a frequency of 1 592 Hz, and over the range 0-2-20 kHz the values of the resistances were always lower when a.c. was useds3. The situation has now been clarified54,and it has been shown that, with a.c., the values obtained are controlled by the capacitance until the frequency has fallen to about 1 Hz. It was shown that under these circumstances, in the absence of corrosion, the resistances of paint films measured by d.c. or a.c. were the same; furthermore, no polarisation resistance was detected. The conclusions are that when coatings have resistances greater than 108Qcm2(i.e. when corrosion is absent) then their resistances may be measured by either d.c. or a.c. However d.c. measurements can be made more quickly, they are easier to make and the apparatus is less costly. It has also been suggested that such measurements provide a basis for the prediction of performance. On the other hand, when corrosion has started, then a.c. should be used, since the values obtained can be resolved into two components, which provide a means of detecting and following the corrosion beneath the coating. J. E. 0. MAYNE
REFERENCES 1 . Evans, U . R . , The Corrosion and Oxidation of Metals, Chapt. 21, Arnold, London 1960) 2. Hudson, J. C. and Banfield, T. A., J . Iron St. Inst., 154, 229 (1946) 3. Hudson, J. C., The Corrosion of Iron and Steel, Chapman and Hall, London, 66 1940) 4. Edwards, J. D.and Wray, R. I . , Industr. Engng. Chem., 28, 549 (1936) 5. Guruviah, S., J . Oil Col. Chem. Ass., 53, 669 (1970) 6. Gay, P. J., J. Oil Col. Chem. Ass., 31, 481 (1948) 7. Anderson, A. P. and Wright, K . A , , Industr. Engng. Chem., 33, 991 (1941) 8. Davis, D.W., Mod. Packag., 145, May (1946) 9. Mayne, J. E. 0, J . SOC. Chem. Ind., 66, 93 (1947) 10. Kelkar, V. M. and Putambekar. S. V.. Chemy. Ind., 1315 (1964) 11. Wild, G.L. E., Paint Technology, 30,9 (1966) 12. Mayne, J. E. O . , J . SOC. Chem. Ind., 65, 196 (1946);68, 272 (1949) 13. Mayne, J. E. 0.. Oil Col. Chem. Ass., 34 473 (1951) 14. Mayne, J. E. 0. and van Rooyen, D., J . Appl. Chem., 4, 384 (1954) 15. Mayne, J. E. 0. and Ramshaw, E. H . , J . Appl. Chem., 13, 553 (1963) 16. Mayne, J. E. 0. and Ramshaw, E. H., J . Appl. Chem., 10, 419 (1960) 17. Appleby, A . J. and Mayne, J . E. 0.. J . Oil Col. Chem. Ass., 50, 897 (1967) 18. Mayne, J. E. 0. and Page, C. L., Er Corros, J., 5, 94 (1970)and 7, 1 11, I I5 (1972)
14: 38
THE MECHANISM OF THE PROTECTIVE ACTION OF PAINTS
19. Bauman, K., Plaste and Kautschuk, 19, 694 (1972) 20. Sherman, L. R., OBcial Digest, 28. 645 (1956) 21. Bancroft, G. M., Mayne, J. E. 0.and Ridgway, P., Br Corros. J , , 6 , 119 (1971) 22. Pryor, M. J., J. Electrochem. SOC., 101, 141 (1954) 23. Beckman, P. and Mayne, J. E. O., J . Appl. Chem., 10, 417 (1971) 24. Gilroy, D. and Mayne, J. E. 0.. Br. Corros. J., 1, 161 (1966) 25. Mayne, J. E. 0.. J . Iron St. Inst., 176, 143 (1954) 26. Mayne, J. E. O., J. Appl. Chem., 9, 673 (1959) 27. Mayne, J. E. O., Research, Lond., 6 , 278 (1952) 28. Sollner, K., J . Phys. Chem., 49, 47, 147 and 265 (1945) 29. Meyer, K. H. and Sievers, J. F., Helu. Chim. Acta., 19, 665 (1936) 30. Nasini, A. G., Poli, G. and Rava, V., Premier Congrh Technique International de Industrie des Peintures, Paris, 299 (1947) 31. Mayne, J. E. 0.. J. Oil Col. Chem. Ass., 40, 183 (1957) 32. Kittleberger, W. W. and Elm, A. C., Industr. Engng. Chem., 44, 326 (1952) 33. Maitland, C. C. and Mayne, J. E. O., OBcial Digest. 34, 972 (1962) 34. Cherry, B. W. and Mayne, J. E. O., First International Congress on Metallic Corrosion, Butterworths, London, 539 (1962) 35. Kinsella, E. M. and Mayne, J. E. O., Br. Polym. J., 1, 173 (1969) 36. Mayne, J. E. 0. and Scantlebury, J. D., Br. Polym. J., 2, 240 (1970) 37. Cherry, B. W. andMayne, J. E. O., Second Internationalcongress onMetallic Corrosion, National Association of Corrosion Engineers, Houston 2, Texas, 680 (1966) 38. Rothwell, G. W., J. Oil Col. Chem. A s . , 52. 219 (1969) 39. Fialkiewicz, A. and Szandorowski, M., J. Oil. Col. Chem. Ass., 57, 258 (1974) 40. Ulfvarson, U., Khullar, M. L. and WAhlin, E., J. Oil. Col. Chem. Ass., 50, 254 (1967) 41. Khullar, M. L. and Ulfvarson, U., IX Congres Fatipec, Section 3, p. 165 (1968) 42. Ulfvarson, U. and Khullar, M. L., J. Oil. Col. Chem. Ass., 54, 604 (1971) 43. van der Heyden, L. A., XI Congres Fatipec, p. 475 (1972) 44. Mayne, J. E. 0. and Mills, D. J., J. Oil. Col. Chem. Ass., 65, 138 (1982) 45. Meyer, K. H. and Sievers, J. F., Helvetica Chemica Acta, 19, 649 (1936) 46. Nasini, A. G., Poli, G. and Rava, V., Premier Congres Technique International de I’lndustrie des Peintures et des Industries Associees, p. 299 (1947) 47. Kumins, C. A., Oflcial Digest, 34, 843 (1%2) 48. Maitland, C. C., Mayne, J. E, 0. and Scantlebury, J. D., Proocedings 8th International Congress on Metallic Corrosion, p. 1032 (198 1) 49. Mills, D. J. and Mayne, J. E. O., J . Oil.Col. Chem. Ass., 66, 88 (1983) 50. Mills, D. J. and Mayne, J. E. O., Corrosion Control by Organic Coatings, NACE, p. 12 (1981) 51. Mayne, J. E. 0.. ‘Advances in Corrosion Protection by Organic Coatings’, Electrochem. SOC. PV, 89-13 (1989) 52. Bacon, R.C., Smith, J . J. and Rugg, F. M., Ind. Eng. Chem., 40, 162 (1948) 53. Buller, M., Mayne, J. E. 0. and Mills, D. J., J . Oil.Col. Chem. Ass., 59, 351 (1976) 54. Burstein, G. T., Gao, G. and Mayne, J. E. O., J . Oil. Col. Chem. Ass., 72, 407 (1989) 55. Haagen, H. and Funke, W., J. Oil. Col. Chem. Ass., 58, 359 (1975) 56. Barraclough, J. and Harrison, J. B., J. Oil. Col. Chem. Ass., 48, 341 (1965) 57. Harrison, J., Br. Corros. J . , 4, 55 (1969) 58. Burkill, J. A. and Mayne, J. E. O., J . Oil. Col. Chem. Ass., 71, 273 (1988)
14.4 Paint Failure*
In view of the wide scope of the subject, paint failure can be treated here only in general terms; detailed accounts will be found in the literature’-6.
Forms of Paint Failure A frequent defect of paintwork is cracking in all its forms, including checking, crazing and alligatoring, followed by such effects as flaking, scaling and peeling. These defects expose the underlying metal surface to the environment so that corrosion is not prevented. It should be observed that checking and crazing begin in the upper coat and extend gradually down towards the substrate, the fissures being wider on top and narrower towards the base. If such signs of breakdown become noticeable after a coating system has had a reasonable length of life in relation to the given conditions of exposure, it is not proper to consider them as film defects. Paint films start their gradual decomposition due to oxidation, erosion, weathering, etc. from the moment of exposure onwards at a rate dependent on their constituents, the environmental conditions and circumstances of application. With increasing age the elasticity of the film usually decreases (in the case of an oil-based or modified paint this results, in the main, from continued oxidation). Expansion and contraction of the metal base caused by severe temperature changes will result in the formation of discontinuities in a relatively inelastic paint film unless the paint has been formulated to withstand these conditions. Excessively high temperatures cause unsuitable paint films to become brittle, crack and loose adhesion. Loss of adhesion can also be caused by swelling. Penetration of rust through an otherwise intact paint film is usually a result of inadequate surface preparation before painting, especially over weathered and hand-cleaned steel ’. However, superficial rust staining may be traceable to dissolved iron salts, e.g. in bilge water from a ship’s deck.
Causes of Paint Failure A consideration of the most important causes of paint failure must include the following: inadequate surface preparation, application of the paint A glossary o f the terms frequently used in this field will be found in Section 14.10.
14: 39
14 :40
PAINT FAILURE
under unfavourable conditions or by inappropriate methods, use of unsuitable paints, adhesion difficulties, the nature of the corrosive environment, etc't. Premature failure can also occur as a result of lack of attention to design. Facilities should, therefore, be provided for ventilatory drainage of water (rain, condensation, etc.), and all structures should be designed so as to permit ready access for repainting. Due consideration by architects and structural engineers at the design stage can indeed help to obviate certain of the causes of paint failure mentioned in this section (see also Sections 9.3 and 11.5).
Pretreatment and Paint Failure The majority of failures of paint applied to metal surfaces are undoubtedly due to insufficient or unsatisfactory preparation of the metal surface, and it is essential that residues of dirt, grease, oil, silicone compounds, etc. be removed from the metal and that all loose paint be removed from surfaces which have been painted previously. The most common cause of premature failure is omitting to remove (as far as is practicable) corrosion products, e.g. rust and millscale, before painting. If a metal has been adequately pretreated, it is then desirable to apply the primer immediately at the factory, if this is possible, in order to ensure that the metal surface remains free from contamination and corrosion products. This is particularly important after grit blasting or any mechanical operations where a protective air-formed film may have been disrupted so that the metal is sensitive to corrosion. The condition of shop-applied primers should be examined before further painting, e.g. on site, and defective areas made good. Edges, welds and rivets need special attention. Prepainted structural steel should not be left exposed unduly long to the weather, particularly under damp conditions or in a marine or industrial atmosphere, and should be handled carefully.
Priming The first coat of paint applied to a surface has the major responsibility for establishing adhesion and for preventing corrosion. Hard-drying primers applied over loose millscale can result in wholesale stripping of scale and paint. Traditionally, oil-based red-lead primers were used over weathered and wire-brushed steel but it is safest to remove all the scale before priming. Etch primers, incorporating phosphoric acid t o etch and clean the surface, substantially increase adhesion to metals that have not had a chemical pretreatment. Pretreatment with organo-functional silanes or incorporation of these materials in epoxy and polyurethane primers has been shown to improve adhesion to simply degreased steel and aluminium to levels approaching those on grit-blasted surfaces " * I * . Corrosion-inhibiting pigt Photographic standards are in use for the identification of the state of rusting at steel surfaces and of the quality of preparation before painting9; they distinguish between manual scraping and blast cleaning. Other photographic standards are used to classify the degree of rusting of painted steel lo.
PAINT FAILURE
14:41
ments and their effectiveness in preventing underfilm corrosion are discussed in Section 14.3.
Effects of Climatic Conditions on Paint Films Paint failure is related to climatic conditions, and the weather prevailing during application of the paint and during subsequent exposure will determine the life of the paint system. This applies, of course, particularly to outdoor work. In unfavourable weather conditions, cracking and blistering can be promoted as a consequence of the expansion of the products of corrosion, and in the case of iron and steel this can lead to under-rusting. Low Temperatures and Wet Weather
When severe drops in temperature occur, outdoor work should if possible be halted, as hail, hoar frost and freezing conditions at the time of painting or shortly afterwards will greatly reduce the life of any paint film, as well as being detrimental to its appearance. Temperatures below 7"C, particularly in still, damp conditions during or immediately after application will prolong the drying time and may leave the film tacky for a long time. During this period dirt will adhere to the tacky film and rain or condensed moisture will tend to reduce the gloss by displacement of some of the paint. Moisture-curing polyurethane paints and bituminous paints, specially formulated for the purpose, are also suited for application to damp substrates; other polyurethane paints should not even be applied to dry surfaces if the relative humidity is high. Suitable paints for use underwater include vinyl resin systems, coal tar paints over inorganic zinc-rich primers, and some coal-tar epoxy primers have also proved themselves 1 3 . Special paints are available for application under water, e.g. epoxy modifications with polyamides. Loss of matter by weathering induces hazing and loss of gloss, which are followed by chalking, usually a white film due to increased light scattering by loose pigment particles, but black on tar or bitumen. The chalking caused by the presence of titanium dioxide (especially anatase) in the top coat initiates rapid erosion. As the pigments in such defective films become more exposed to rain and wind and wash away, the films become increasingly permeable to moisture, with consequent corrosion of the underlying metal although they have the advantage of looking clean. In special cases (see below) controlled chalking may be desirable. Painted metal exposed at coastal areas, ports and docks often suffers most from such hazards, which may be aggravated by high levels of U.V.radiation and the erosive action of blowing sand. Such conditions can prevail up to approximately 3 krn inland. Stripping of top coats soaked by rain or sea-water has occurred with alkyd-resin-based paint systems, mainly on ships. The risk of such intercoat failure is reduced if the time interval between application of coats is reduced, but is best controlled by modification of the alkyd resin with a proportion of a different material.
14 :42
PAINT FAILURE
Factors which can contribute to unsatisfactory drying, in particular of paints containing drying oils, include application of too-heavy paint films (especially during cold weather when viscosities and relative humidities are higher), overdoses of driers, gaseous pollution, use of unsuitable heavy thinners and residues of tar oil, wax or grease present on metal surfaces. Curing reactions of paints based on epoxy and polyurethane resins are markedly temperature dependent becoming extremely slow at temperatures below 5°C unless special hardeners are used; such paints should never be used below the minimum temperature specified by the supplier.
Tropical Conditions
Tropical conditions in general contribute to faster paint breakdown, owing to high temperatures, moisture-laden atmosphere, high ultra-violet content in the solar radiation, or to a combination of some or all of these effects. Thus in the tropical regions, fading and discoloration, matting, chalking and cracking, followed by peeling and general embrittlement, can take place rapidly. Chlorinated rubber paints fail especially in dry tropical environments, probably due to autocatalysed dehydrochlorination, whilst alkyd resin paints chalk more rapidly in wet tropical environments than in temperature conditions because of the greater amount of short wavelength U.V. radiation.
Effect of Industrial Atmospheres In towns which are not heavily industrialised, the life of a paint film may be about equal to that in rural areas, but, because of traffic disturbance, dirt collection from soot and dust will be noticeable earlier. The use of selfcleaning (chalking) paints can overcome the premature loss of some of the decorative effect without noticeably reducing protection. Industrial towns, especially those having heavy or chemical industries, have an acid atmosphere and the pH of rain water is sometimes as low as 3, owing t o the presence of sulphuric acid. This can cause gradual attack on certain pigments and extenders, resulting in discoloration (e.g. red lead can be transformed into white lead sulphate, and atmospheric hydrogen sulphide results in the blackening of lead pigments) and decomposition of the paint film, and can lead t o premature failure. Aluminium finishing paints applied over, for example, a red lead primer, are liable to be attacked in industrial atmospheres, owing t o the formation of water-soluble aluminium salts, and the aluminium colour may disappear quickly. Similarly, top coats of zincrich paints may lose their metallic colour by formation of zinc salts (e.g. on iron chimney stacks), and contamination in the atmosphere may also endanger intercoat adhesion.
PAINT FAILURE
14 :43
Effects of Moisture Hot steam and severe condensation acting on a film surface exert a very destructive effect, comparable with that of a paint remover; they are particularly liable to cause swelling. Dry steam, in contrast to condensed steam, does not cause corrosion'4. Less severe attack by water vapour can cause blistering, which can be of two types: intercoat blisters between paint films, and blisters through the complete film system. Only the latter leads to corrosion of the underlying material. Paint films exposed to condensation often fail unexpectedly by very early blistering between primer and finishing coat, usually associated with soluble salts trapped under the relatively impermeable finishing coat. Relatively more permeable latex-based paints are less prone to this failure. Dampness often accounts for the promotion of mould growth on painted surfaces, e.g. in breweries, laundries and dairies; fungi develop faster under tropical conditions. There are special media which are resistant to mould growth, in particular those which are based on the chlorinated compounds, such as chlorinated rubber, polyvinyl chloride, its various copolymers and other halogen-containing polymers. By addition of suitable fungicides and careful selection of the pigments, traditional hard-drying paints and varnishes can also be made to resist mould growth. Infected surfaces and films should be washed with fungicidal solutions before painting, but unless the source of infection is removed the trouble is likely to recur.
Factors Which Cause Paint Failure in Industrial Applications Anti-oxidising Environments
Where fumes or deposits which act as anti-oxidants are present, no orthodox paint which dries by oxidation can give satisfactory service. Instead, a coating which dries either by evaporation (e.g. a selected chlorinated rubber paint), or by a cross-linking reaction (e.g. a catalysed epoxy or twocomponent polyurethane paint) must be used. Oxidising and Acid Environments
Atmospheres polluted by oxidising agents, e.g. ozone, chlorine, peroxide, etc. whose great destructive power is in direct proportion to the temperature, are also encountered. Sulphuric acid, formed by sulphur dioxide pollution, will accelerate the breakdown of paint, particularly oil-based films. Paint media resistant both to acids, depending on concentration and temperature, and oxidation include those containing bitumen, acrylic resins, chlorinated or cyclised rubber, epoxy and polyurethane/coal tar combinations, phenolic resins and P.V.C. Acid conditions occur in the vicinity of, for example, coke ovens, gas works, oil-fired plant, galvanising plant and paper pulp mills, and in these
14 :44
PAINT FAILURE
conditions, cracking is a frequent form of failure; the cracking and peeling in acid environments is usually much more severe and occurs much earlier in the life of the paint film than is the case in other environments. Failure in acid environments results from the specific properties of pigment, medium, or drier used in the paint, e.g. in sulphuric acid environment zinc pigments form zinc sulphate, which appears on the paint surface. Non-oxidising and weak acids, in contrast to oxidising acids, can penetrate paint films without destroying them; they then react with the metal base to form salts with resultant stresses which cause cracks. Magnesium-rich alloys are particularly prone to attack by acids; their salts, having considerable volume, in severe cases effloresce through the broken paint films. For resistance t o acid conditions alone, traditional filled and unfilled bituminous solutions (which have economic advantages), chlorinated rubber and shellac have been used. Crosslinking coatings, e.g. amine-cured epoxy resins, often blended with coal-tar which develops resistance to oils and solvents, have obvious advantages on chemical plant.
Alkaline Environments
Oil-base (including oil-modified alkyd resin) paint films should not be used in alkaline environments as the paint will deteriorate owing to saponification; alkali-resistant coatings are provided by some cellulose ethers, e.g. ethyl cellulose, certain polyurethane, chlorinated rubber, epoxy, p.v.c.1 p.v.a. copolymer, or acrylic-resin-based paints. In particular, aluminium and its alloys should be protected by alkali-resistant coatings owing to the detrimental effects of alkali on these metals.
Salt Solutions
Corrosive solutions, e.g. salt solutions as present in saltems, refrigeration plant and sea-water, are particularly active at the water-line (cathodic zone), where alkali may accumulate and creep up between paint and metal’’ and cause softening and loosening of the paint. This process may also occur where the metal is completely immersed, particularly below paint films pigmented with zinc or aluminium”. Caustic soda is formed at the steel surface (which is made cathodic by the zinc) resulting in the softening of oilbase paints and consequent loss of adhesion. In sea-water, at the local cathodes the total concentration of ions will exceed that in the surrounding sea-water, and water may be drawn in by osmosis, with resultant alkaline blistering 15. This is usually the first sign of electrochemical corrosion; alkaline peeling and corrosion of the metal become apparent only later. Good results in the salt-rich Mediterranean have been reported 16* with anticorrosive primers containing a proportion of chromium fluoride, including those for ships’ bottoms.
’’
PAINT FAILURE
14:45
Marine A tmosphere
Iron girders, etc. are frequently supplied to a site in the grit-blasted and primed condition, but occasionally this work is carried out on site. If the structures lie about afterwards for some time in a salt-laden environment, e.g. a marine atmosphere, and are not thoroughly washed with fresh water and dried before further painting in order to remove all traces of sodium chloride, the latter will soon play havoc with the steel and anticorrosive film system. This will occur after erection and possibly even inside buildings owing to under-rusting accompanied by severe blistering and followed by flaking with rustbacking. The rust can be in various states. Analogies can be drawn in connection with the repainting of ships in dry docks. High relative humidity has an aggravating effect. Corrosion-promoting Hgments
Some pigments promote corrosion owing to their content of soluble salts, their reactivity, or their electrochemical action, and thus should be avoided. Rust of the spotted type can be the consequence of their presence in a paint, especially the first coat, e.g. of graphite (noble to steel), some red oxides of iron, gypsum, ochre or lamp black. Paint containing potentially soluble copper, such as antifouling compositions, if applied directly to steel, may stimulate corrosion by plating out of copper anodes. Antifoulings are always separated from the steel by an effective anticorrosive primer, but interaction between the two must be avoided by suitable formulation to avoid corrosive and excessive leaching, i.e. making the antifouling ineffective ”. Mercury compounds, used as fungicides or for antifouling can promote rapid attack of aluminium and its alloys under wet or humid conditions.
Effects of Stoving and Storage Conditions During stoving in convection-type box ovens, drying can be delayed (as it can on air drying when the ventilation is insufficient, e.g. in a ship’s hold) if the vents are closed too far, or if the coated articles are too closely packed. In the latter case there may even be trouble caused by solvent wash, i.e. redissolution of the uncured film by stagnant solvent vapours, which occurs mostly on surfaces near the top of the oven. This can lead to the establishment of practically unprotected areas. Damp conditions contribute to ‘gas-checking’* of some synthetic stoving lacquers, quite apart from the effects of foul oven gases, or the presence of detrimental solvent vapours, e.g. from a trichlorethylene degreasing plant. Overstoving, too, can result in embrittling due to overpolymerising or oxidising, followed by cracking or crazing. Stoving enamels, etc. which are based mainly on cross-linking epoxy resin combinations, behave for all * A fine or coarse wrinkling due to irreversible swelling of a surface-dried film.
14 :46
PAINT FAILURE
practical purposes in exactly the contrary manner to this, forming almost the only exception to the general rule. They are brittle when undercured and become tougher after complete cure, and even remain so when they have been somewhat overcured. Certain members of this class of coatings, therefore, do not perform too satisfactorily on air drying. It has been observed, for instance, that air-dried amine-cured anticorrosive epoxide paints over new steel were not able to hold down millscale, which appeared still adherent at the time of painting, for any practicable length of time in contrast to the performance of traditional anticorrosive oil paints. Very premature flaking occurred, the brittle paint flakes being backed with millscale. Epoxide resin esters, however, perform quite well, apart from a tendency to chalking. If infra-red heating or any other radiation curing method is employed, areas which are shaded from the rays or are outside the area of greatest flux density, cannot dry as hard as the fully irradiated surfaces, and may form weak spots susceptible to mechanical damage and consequent corrosion. After long storage in their packages, certain oxidising, i.e. drying-oil or drying-oil-modified alkyd-resin-based paints containing certain pigments, of which iron oxides, iron blues, toluidine red and carbon blacks are the most important, lose some of their drying properties, probably owing to inactivation of driers by adsorption on the pigment surface, followed by slow deactivation of the adsorbed catalysts'*. Such paints, often used as primers, dry and harden satisfactorily when freshly made, but storage may make them increasingly sensitive to the application of a second coating. Discoloration due to mixing of the films, drag of the brush, and in severe cases even lifting, may result. Lifting may also occur if a paint containing strong solvents (xylol or solvent naphtha, not to mention such active solvents as esters and ketones) is applied (not necessarily by brushing) over a paint which is not resistant to them. The older an oxidising paint film becomes, the more solvent-resistant it will be. Short-oil media and pigment-rich paints are not so prone to lifting. This type of failure is not restricted to oil-base materials; it can, for example, also occur with chlorinated rubber paints.
Effects of Application Methods Excessive thinning of a paint of good quality is often the cause of the application of films which are too thin. The temptation for operators to do this is great as it often increases the ease of application and their bonuses, especially in the case of paints for brushing. Overthinning is particularly common when surface coatings based on e.g. medium to short oil-modified alkyd resins, or coatings which dry by evaporation are being used. It is particularly difficult to check with highly opaque aluminium paints. Overthinning is also frequently responsible for running and sagging, which in turn promotes excessive pigment flotation. If application is by spraying, this can be countered by the use of thinners which evaporate quickly. In brushing, however, such thinners would cause dragging of the brush. If the evaporation rate of thinners is too fast, they may promote
PAINT FAILURE
14 :47
cobwebbing when highly polymerised resins such as the vinyls or chlorinated rubber are being sprayed. Again, if heavy thinners containing strong solvents are used in the second coating, lifting trouble may be experienced in addition to sagging. Some specially formulated paints can be applied wet on wet by spraying, without the aforementioned disadvantages. Some water-thinned industrial paints exhibit anomalous viscosity changes during drying and therefore need careful control of air flow and humidity to ensure satisfactory film formation. If paint is insufficiently stirred before use, over-pigmented paint from the bottom of the container will, when it comes to be used, act as a short-oil nonelastic coating of poor binding power, while under-pigmented mixture from the upper strata will perform more as a longer-oil, more elastic coating, and will possibly run. Two- or three-pack materials mixed immediately before use present special hazards. The supplier’s recommendations on mixing ratios and pot life must be followed carefully. Pot life is highly temperature dependent and may be reduced greatly if materials mixed in bulk are heated by exothermic reaction. Thinning of material that has partially cured in the pot results in unsatisfactory films. If an elastic or insufficiently hard primer or paint has been applied under a less elastic top coat, or if the first coat (or set of coats) of oil-base paint has been second-coated before it is completely dry, not only will the paintwork remain soft for an unduly rong period, but cracking will also follow, as the upper layer cannot follow the movement. If the last coat is very thick this fault will frequently manifest itself in the form of alligatoring, Le. the formation of cracks which do not penetrate all the films down to the substrate, and which may be present in the top layer only. Repainting
For painted structures it is essential that an additional paint coating be applied as soon as there is evidence of paint breakdown. The Protective Coating Sub-committee of BISRA4 recommend painting of steel surfaces when 0.2-0-5Voof the surface area shows evidence of rust. Delay in repainting may be a false economy, as if rusting is extensive it may be necessary to clean down to bare metal before paint can be applied. Should an old bituminous paint layer have t o be recoated, this should be done only with another bituminous paint, unless the surface is first insulated with one of the special primers which are available for the purpose. Bleeding and premature checking may otherwise occur. Damage to prefabrication primers or even the whole film system can be caused on transport or on erection, leaving for example, bare edges. Good supervision is necessary to ensure that defective areas are conscientiously touched-up before applying further paint films. It has been recommended to disregard the coat of prefabrication primer when deciding the number of coats to be specified’.
14 :48
PAINT FAILURE
Adhesion Difficulties It is important to realise that various factors contribute to good adhesion of paint films. These include: 1. Cleanness of the base, Le. freedom from grease, which improves the wettability of the metal surface, and the removal of oxides, dust or loose paint, etc. already described. The closer the surfaces of paint film and metal, the more secondary valencies originating in the polar constituents of the medium are brought into play. 2. Mechanical pretreatment of the metal by weathering, sanding, shotblastingI9, etc. for the removal of corrosion products and loose millscale, and chemical pretreatment by phosphating, pickling, etc. to create a mechanical, or, in the case of etch primers, a chemical key. Wet abrasive blasting is particularly effective in removing contaminants from rough surfaces. Degrees of cleanliness of steel surfaces can be compared with BS 4232, etc2' 3. Selection of suitable coatings possessing good wetting properties, which are elastic enough to expand and contract with the metal base over a reasonably long period and which, as far as priming is concerned, have an affinity with the metal t o be painted. It is often not appreciated that the adhesion properties of a given coating material may vary according to the type of metal to which it is applied, although it is suggested that the degree of retention of contaminants is the real cause2'.
So far as iron and steel are concerned, the adhesion problem is simple, and the oleoresinous coatings which are generally applied to them form a good bond with them. Mechanical pretreatments are always extremely useful. Cracking, flaking, scaling or blistering due to under-rusting (the latter often being accompanied by brown discoloration of the film) is, as has already been explained, due to mechanical action by the products of corrosion. This may at times pose the problem of whether the paint or the painting system was responsible for the corrosion, or whether, on the other hand, it was the corrosion (possibly residual) which was responsible for the unsatisfactory performance of the paintwork. The better the adhesion of the paint to the metal, the less damage there will be to the paint film, and the less premature corrosion will ensue. This is similarly the case with nonferrous metals. Rough (especially blasted) steel surfaces which have received too thin a paint coverage will be indicated by the presence of pinpoint rust spots in the film surface, wherever the metal peaks have not been sufficiently protected. A patchy form of rust that attacks paint films from underneath, can be caused by sweaty hands, residues from fluxes, etc. Examples of the latter include residues from phosphating and soluble salts (including those from unsuitable rinsing water) and they can manifest themselves on steel in the form of a creeping filiform corrosion, i.e. as progressing threads of rust which loosen the coating. This can be followed visually through transparent films. It occurs, however, only when the relative humidity of the surround-
PAINT FAILURE
14 :49
ings is above 82%, and if oxygen diffuses through the film. Diffusion of carbon dioxide seems, however, to suppress filiform c o r r ~ s i o n ~A~some*~~. what similar type of corrosion that causes the destruction of paint films as a result of the presence of salts beneath, is termedfiligrun corrosion, and has been observed on painted shipsz4. Filiform corrosion is considered in Section 1.6. Aluminium and magnesium alloys, copper and its alloys, and zinc and zinc-base discastings, including galvanised iron, to name the most important groups of non-ferrous metals, can offer serious adhesion problems. These are aggravated if the surfaces are very smooth, as, for example, on diecastings or hard rolled sheets. For light metals, p.v.b.-based* etch primers are ideal; long-oil alkyd-resin-based zinc chromate primers may also be satisfactory. Etch primers and alkyd-resin-based coatings are very suitable for zinc and its alloys and alkyd resin-based coatings for copper and its alloys. If the first coat has been selected for good adhesion, the subsequent ones may be chosen from a wider range of products to satisfy other requirements involved in the particular application. A number of cold-rolled alloys based on aluminium, copper and zinc are susceptible in varying degrees to recrystallisation on exposure to heat. This can have a detrimental effect on the adhesion of paint films. While there may, at first, be no sign of trouble, the defect will become obvious by brittleness of the film after some storage time has elapsed. To avoid peeling of oleoresinous top coats from zinc-rich primers, a sufficient interval should be allowed between coats to permit the zinc-rich primer to weather first; in sheltered conditions soluble products should be removed before recoating. Lacquers drying by evaporation to rather rigid films, e.g. some nitrocellulose lacquers, may not be able to follow the movements of metals caused by changes in temperature, and rapid cracking, followed by flaking of the paint film, can result. In all such cases the smoother the metal surface, and the less affinity the coating has for the grease- and oxide-free metal surface, the more likely breakdown is. The presence of various proportions of minor constituents in alloys, including those of iron, can have a profound effect on the behaviour of the main metal in this respect. Reference has already been made to the detrimental consequences which bad weather conditions occurring during or shortly after application usually have on the life and protective value of a film”. To protect buried metals from premature breakdown it must suffice to say that protective coatings and other methods must be applied against factors such as the effects of galvanic currents, composition of the moisture in the groundz6,humus acids, bacteria, etc. (See Section 14.8.) In conclusion, it should be emphasised that surface coatings which are of the highest quality, and which where necessary have the special protective properties required, should always be used. Good supervision, careful working, and common sense can contribute a great deal to reduce paint failures and the wasteful work which is necessary to put a job right. *p.v.b. is an abbreviation for poIyvinyI butyral.
14:50
PAINT FAILURE
Paint Adhesion and Corrosion When corrosion develops on painted steel the question is often raised as to whether corrosion was a result of paint failure or the paint failure was caused by corrosion. Several studies have shown that adhesion forces are reduced greatly after water soaking or even at very high h ~ m i d i t y and ~ ~ ~it*has ~ been argued that film detachment by water usually precedes underfilm corrosion2’. Against this view others have claimed that those paints known to have reduced wet adhesion, e.g. those based on alkyd resins, are not uniquely, or even especially, subject to underfilm corrosion 30. Several factors should be considered in this discussion: 1. A continuous intact film of water-resistant paint forms an effective elec-
2. 3.
4.
5.
6.
trical resistance to the flow of a corrosion current (a resistance of over lo9Q cm2through the film is easily achieved). Underfilm corrosion can then only occur if a channel of electrolyte connecting anode and cathode can be established by local adhesion failure between the coating and the metal substrate. Localised adhesion failure occurs most easily where broken scale or rust, or deposits of salts, have impeded wetting of the metal substrate by the film-forming constituents of the paint. After major surface contaminants have been removed, e.g. by wet abrasive blasting of hot-rolled structural steel, application of a thin coat of an etch primer greatly reduces the incidence of underfilm corrosion, presumably by eleminating localised areas of poor adhesion. Phosphate pretreatments followed by effective rinsing have a similar effect over cold reduced sheets. Even small traces of certain corrosion stimulants, notably soluble chlorides and sulphates, can maintain a continuing corrosion process under a paint film because the salts accelerate the initial dissolution of ferrous iron (and other metal ions) but are not immobilised in the hydrated oxide corrosion products. Filiform corrosion is the most spectacular example of this phenomenon, but progressive spread, preceded by blistering, is also observed from scratches or other breaks in a coating, for example during salt spray tests. Soluble salts in or under a coating, even if not active corrosion stimulants, can induce osmotic blistering and thus expose underlying metal to possible corrosion-”. Such salts may be present in a pigment (even some soluble chromates are suspect), may be formed by reaction with basic pigments (e.g. barium carbonate), or by reaction of organic acids from drying oil oxidation with metal oxide substrates (zinc or magnesium formate are especially likely to be found at interfaces with the appropriate metals). Residual salts from rinse water have been shown to cause ‘snail trail’ blistering and subsequent corrosion under motor car finishes. Fears have been expressed that soluble flash-rusting inhibitors used in wet abrasive blasting could have similar effects, but no problem has been found with the concentrations normally used. Paint stripping by water is most likely to occur from cathodic areas, the phenomenon of cathodic disbonding sometimes observed on steel protected by external anodes or impressed current being a particularly
PAINT FAILURE
14:51
spectacular case. This failure may involve direct attack on the paint binder by cathodic alkali32,but some workers have claimed that the attack is more often on the metal/paint interface, possibly having more in common with alkali degreasing processes33. Overall there is good evidence for the presumption that the best way to avoid corrosion under paint films is to prepare the substrate in such a way as to maximise adhesion and then to apply an insulating film of paint. Provided that the substrate is free from coarse sharp-edged profile the insulating coating need be no more than 100pm, thick; indeed a very thick film may be more likely to crack or be damaged by external mechanical action. The value of some chemical pretreatments and self-etching primers has already been mentioned; the possible advantages of incorporating specific adhesion promoters in primers have yet to be fully explored.
Long-life Coatings All organic and some inorganic, coatings are subject to a continuous process of erosion by chemical breakdown to volatile or water-extractable products. The processes involve oxidation, depolymerisation and other bond-splitting reactions. Many of the breakdown reactions are stimulated by U.V. radiation, particularly the high quantum radiation at the short wavelength limit of the sun’s spectrum. Some pigments, notably certain titanium dioxide pigments, accelerate breakdown under U.V. radiation; others, such as red iron oxide or metallic aluminium protect by absorbing the radiation, as do specific U.V. absorbing additives. Typical erosion rates for coatings fully exposed to full weathering, facing south in temperate areas are 1-2 pm/year for white alkyd paints as against 5 pm/year for the earlier oil-based paints. In tropical areas with more short wavelength u.v., rates may be two or three times higher. Modification of alkyd resins with high proportions of silicones considerably reduces rates of attack, but the most spectacular extension of life is shown by fluorinated polymers such as polyvinylidene fluoride where erosion rates can be reduced to 0.1 pm/year. If this level of durability can be achieved an initial coating, if firmly adherent and free from any breaks, may often be expected to maintain protection over a metal substrate for the likely life of the structure. The considerably increased first cost, as compared with more conventional coatings, has to be balanced against the probable saving in maintenance costs or consequences of failure. Acknowledgment A number of suggestions by W. A. Edwards have been incorporated in this Chapter, and these are gratefully acknowledged. M. HESS T. R. BULLETT REFERENCES 1. Hess, M., et al., Prrint Film Defects, Their Causes and Cure, 2nd edn, Chapman and Hall,
London (1965) 2. Hudson, J . C., The Corrosion of Iron and Steel, Chapman and Hall, London (1940) 3. Third Report of the Corrosion Committee of the Iron and Steel Institute, London (1959)
14:52
PAINT FAILURE
4. Fancut, F. and Hudson, J. C., for the Protective Coatings (Corrosion) Sub-committee of B.I.S.R.A., protective Painting of Structural Steel, Chapman and Hall, London (1957) 5. Evans, U. R.. The Corrosion and Oxidation of Metals, Arnold, London (1960) 6. Mayne, J. E. O., J. OilCol. Chem. Ass., 3, 183 (1957); Mayne, J. E. 0. et at., ibid., 7,649 ( 1967) 7. BS 5493:1977. Code of Practice for Protective Coating of Iron and Steel Structuresagainst
Corrosion 8. Breakdown ofPaint FilmsandSteel, the B.I.S.R.A. Scale, Degrees of Rusting, British Iron and Steel Research Association (1949). 9. Swedish Standard SIS 05 59 00;see also IS0 8501, Visual Assessment of Rust Grades and 10. 11. 12. 13. 14. 15. 16. 17.
18. 19.
20.
21. 22. 23.
of Preparation Grades BS 3900:Part 1-13:1983, Designation of Degree of Rusting (equivalent to I S 0 4628/3) Walker, P., J. Oil Col. Chem. Ass., 65, 415 (1982) Walker, P., 1. Oil Col. Chem. Ass., 65, 436 (1982) Suggitt, J. W. and Graft, C. M.,J. Paint Techn., 38, 150 (1966) Waeser, B., Rostschuden und Rosrschutz, Wilh. Pansegrau Verlag, 115, W. Berlin (1956) Evans, U . R., Metallic Corrosion, Passivity and Protection, Arnold, London (1946) Communicated by Paint Research Association, Haifa Munk, F. and Rothschild, W., ‘Interaction between Anticorrosive and Antifouling Coatings in Shipsbottom Painting’, J. of Paint Techn., 43, 557 (1970) Bell, S. H., J. Oil Col. Chem. Ass., 38, 5 9 9 (1955) Comparisonsof Pretreatments:A Background to the Corrosionof Steel and its Prevention, No. 3, ‘Effect of Surface Preparation and Paint Performance’, 20, The Corrosion Advice Bureau of B.I.S.R.A. In BS 4232, Surface Finish of Blast Cleaned Steel for Painting, first quality corresponds to SA3,2nd to SA2.5 and 3rd to SA2 of the much more extensive Swedish Standards Commission’s SIS 055 900-1962, Rust Gradesfor Steel Surfaces and Preparation Grades Prior to Protective Coating, Stockholm (1%2); see also BS 7079 Part A1 (1989) Bullett, T. R. and Prosser. J. L., Adhesion Problems with Paint and Powder Coatings. Chap. 7, Industrial Adhesion Problems, Orbital Press (1985) Slabaugh, W. H. and Chan, E. J., Ofl Dig., 38 No. 499, 417 (1966) Slabaugh, W. H. and Kennedy, G. H., Amer. Chem. Soc., Org. Coatings Diu., 26 No. 1,
1-9 (1966) 24. Rathsack, H.A., Schiffsanstriche. Korrosions und Bewuchsschaden am Schixsboden, Berlin, 18 (1967) 25. Comparison of Pretreatments: A Background to the Corrosion Control of Steel and its Prevention, No. 3, ‘Rates of Rusting in Different Environments’, 16, Corrosion Advice
Bureau of B. I .S.R. A. 26. Comparison of Pretreatments: A Background to the Corrosion Control of Steel and its Prevention, No. 3, ‘Water Composition’, 11, Corrosion Advice Bureau of B.I.S.R.A. 27. Walker, P., Oficial Digest, 31, 1561 (1%5) 28. Prosser, J. L., Paint R. A., Research Memo No. 332 (1963) 29. Funke, W.. J. Oil Col. Chem. Ass., 68. 229 (1985) 30. Walker, P., J. Oil Col. Chem. Ass., 68, 319 (1985) 31. Bullett, T.R. and Rudram, A. T. S., J. Oil Col. Chem. Ass., 44, 787 (1961) 32. Hammond, J. S., Holubka, J. W. and Dickie, R. A,, J. Corrosion Tech., 51, 45 (1979) 33. Koehler, E. L., Corrosion, NACE, 40, 5 (1984)
14.5 Paint Finishes for Industrial Applications
Introduction Industrial finishing systems are those paint systems that are applied in factories, not in homes, construction sites or shipyards. In factories it is possible to obtain considerable control over all stages of the painting process. The application process may be selected to give accurate control over film thickness. The temperature and time of drying may be chosen to obtain a given throughput of finished articles. Furthermore, a wide range of polymer types are available to give particular combinations of properties to the dry films. Industrial finishing systems are applied to a wide variety of substrates, the majority of which are metallic, but they are also applied to paper, wood, wood composites, cement products and plastics. Often a high quality of decoration is required, as well as protection from a number of hazards, such as knocks, abrasions, bending or forming and contact with noncorrosive liquids. Resistance to the weather may be required. Outdoor finishing systems, and many others, are also required to protect metal against corrosion.
Finishing Systems: Factors Governing the Choice It may be possible to decorate or even to protect some surfaces with a single coat or finish, but protection of metal against corrosion always requires a finishing system. A full finishing system will require some or all of the following coatings. 1. A metal pretreatment or conversion coating This is a specially formulated mixture of inorganic chemicals which react with the metal to produce a strongly adherent, corrosion-inhibiting conversion coating, such as a phosphate or chromate, on the metal surface (see Sections 15.2 and 15.3). This coating often provides a better surface than the original metal oxide layer for obtaining good adhesion of the paint layers. 14:53
14:54
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
2. A primer On metal, the purposes of a primer are to enhance corrosion protection and to give excellent adhesion. The primer will contain anticorrosive pigments, such as strontium chromate or zinc phosphate, which will slowly release ions that can repair damage or faults in the underlying conversion coating. 3. An undercoat This coat is required to provide bulk cheaply and to be capable of being sanded easily to give a smooth surface for finishing. This coat is required to provide all decorative properties (colour and gloss) and the main resistance to external damage (e.g. U.V. degradation).
4. The finish
The selection of each of these coatings (or the decision to omit one or more of them) is dependent on a number of factors, which will now be considered. Painting systems are selected by the manufacturers of industrial articles, advised by their paint and their equipment suppliers, taking into account the following factors: -the -the -the -the
size and shape of the article; physical and chemical nature of its surface; appearance and protection required from the paint; required output rate.
The selection is made in the light of various constraints, such as: -existing equipment and space; -money and space available for new equipment; -acceptable running costs (including paint, energy and labour); -the maintenance of safe working conditions; -conformity with regulations on environmental pollution. Selection is therefore a compromise. The variety of choices available to the manufacturer will now be illustrated by considering how these factors can operate in the selection of finishing systems for metal articles to be protected from corrosion. Size and shape can have a dramatic effect. The immense size of a jumbo jet immediately rules out any possibility of putting the aeroplane in an oven; the coatings must all dry at ambient temperature. The size and shape also rule out all automated methods of application. On the other hand, flat sheet can be processed on an automated painting line using economical methods of painting, such as roller coating or curtain coating, followed by cure by stoving of by infra-red or electron beam radiation. If the surface is smooth, then a high quality appearance may be obtained with low film thickness and only one or two coats. On the other hand, a rough casting can only be given a good appearance if the film thickness is built up with surfacer and sanding is carried out before finishing. It may be possible to use an automated application technique, like electropainting, on a casting, but stoving of the paint will be very inefficient with the large amount of metal acting as a heat sink. A relatively inert surface like tinplate may not need a pretreatment. Zinc, on the other hand, may be pretreated to improve adhesion of the paint
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
14:55
coat. Steel invariably needs thorough cleaning and an iron or zinc phosphate pretreatment for passivation and subsequent optimum corrosion resistance. Assuming maximum corrosion resistance is required, then an anticorrosive primer will be needed, with best protection coming from a crosslinked epoxy stoving primer. Most other properties are dominated by the finish, which will be based on a high molecular weight-polymer, either linear or (more usually) crosslinked. The precise selection of the polymer depends on the balance of properties required, but will be constrained by the type and rate of curing necessary. With infinite space, drying rate does not determine output rate, but usually space is at a premium and drying must be hastened, if possible, with heat (or other forms of energy). If heat cannot be used, fast air movement aids solvent removal, and lacquers based on linear polymers or emulsions dry fastest. They d o not, however, confer more than limited resistance to solvents. Increasingly, industrial painters, especially in the USA, are turning to paints of low solvent content to minimise air pollution. These include powder coatings, 100% polymerisable coatings, high solids coatings and water-based materials. These coatings can demand more energy to obtain good throughput, though radiation-curing finishes are both fast and economical with energy.
Methods of Application and Drying Since these methods are selected by the industrial finisher at an early stage and can, as discussed above, have a major effect on the polymer options available to the paint supplier, they will be discussed next.
Apprication
The range of application methods available is extremely wide. A number of these are described in Section 14.1 and include: brushing; a wide range of spraying techniques; techniques involving total immersion, such as dipping, electrodeposition and fluidised beds; methods such as flow coating and curtain coating, in which paint is made to flow over the article. Additionally, the techniques of centrifuging and tumbling or barreliing are especially suitable for very small articles. In the latter method, the articles are tumbled in a rotating barrel with just enough paint to coat them to the required thickness. In the former, excess paint is used and the excess removed by centrifugation after coating. Extrusion coating is ideal for rods, tube and wire. The article is passed through a paint reservoir and then out via a die, which leaves only the correct thickness of paint in place. There are further techniques suitable for flat articles in sheet or web form. Knife coating is ideal for very thin coats, especially on continuous paper or plastic webs. The knife is either a metal doctor blade or a curtain of high velocity air (an air knife) directed onto the surface and it removes surplus material applied previously.
14:56
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
Even more widely used are a variety of roller-coating techniques. In forward roller coating a controlled amount of paint is metered onto the surface of a rubber or gelatine roller rotating such that, at its point of contact with the sheet or web, roller and sheet are moving in the same (forward) direction. Even finer control of thin coatings is obtained if the paint is transferred from gravure cells onto the application roller. In the coating of continuous metal coils, reverse roller coating is often used. In this technique the web is moving counter to the application roller direction, so that the paint is partly wiped off by the moving coil. Shear leads to better flowout. Another type of reverse roller coating is used for the application of stiff paste fillers to chipboard. Application is by forward roller, but this is immediately followed by a reverse roller, which presses the filler into the board and doctors it smooth. Tumbling and centrifuging are batch processes, but all the others can be included in a continuous line process and, for suitable articles, the process can be fully automated. If the shape of the articles is unsuitable, some kind of hand spraying is usually selected. A matter of considerable importance in the selection of an application method is its efficiency. Spray techniques are usually inefficient, since many droplets drift past the target and are lost. Even electrostatic spraying can waste as much as 35% of the paint. There is some loss of paint in most methods, but roller coating, curtain coating and electrodeposition are very efficient. Electrodeposition is also a very useful technique where corrosion resistance is important, since it applies a uniform coating over nearly all surfaces of even the most complex-shaped article. Drying
Lacquers dry simply by the evaporation of the solvent, leaving behind high molecular weight linear polymers which provide the properties of the films. Only air movement is necessary, but heat speeds up the process. For some emulsion paints the process is similar, though heat may be necessary to soften the polymer particles, allowing them to integrate to form a film. For all other types of paint, low molecular weight polymers must be converted into high molecular weight crosslinked polymers by chemical reaction. Many of these reactions are extremely slow below certain threshold temperatures; these temperatures must be exceeded in drying. Other reactions, which proceed slowly at room temperature, are accelerated considerably b y heat. There is a third group of reactions which depend mainly on the creation of free radicals, and there are ways of creating these without heat. In industrial painting throughput rate is critical and drying equipment will usually be needed. This equipment will control the rate of air movement, to remove solvents and/or volatile reaction products, and is also likely to include devices for raising the temperature of the paint film, or creating free radicals within it. The simplest and most widely used method of increasing the film temperature is to pass the coating through a convected hot-air oven. This is relatively inefficient, but effective with articles varying widely in shape and size. If
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
14: 57
the article is flat, speed can be increased by directing jets of very hot air at high velocity onto the surface. The next most frequently used technique is to raise the film temperature by infra-red irradiation. Emitters vary from low-energy long-wavelength (3.6-8 pm) black emitters (90-50O0C surface temperature), through medium wave length (2.0-3.6 pm) red-hot emitters (500-1 200°C) to high-energy short-wave length (1 .O-2.0pm) white-hot (1 200-2 200°C) emitters I . The radiant energy must be directed to reach all parts of the film; shadowing on complex shapes can cause difficulties. Infra-red heating is often combined with convected hot air. For specific end uses (e.g. exteriors of small containers) flame drying is a means of very rapidly increasing temperature (0.02-0.04 s). An air curtain surrounding the flame prevents solvent ignition. An alternative, fast method, suitable for simply-shaped metal articles, is induction heating of the metal with conduction to the coating. For removal of water from flat films on nonconducting substrates, radiofrequency heating can be used. If the film-former is designed to be polymerised by a free radical mechanism, free radicals can be created in the film by decomposing a photoinitiator within the film using ultra-violet radiation2: 0
II
OCH,
I
0
II
Ph - C - C - P h S P h - C *
I
OCH,
OCH,
I
+ * C- P h I
OCH,
The free radicals then initiate curing by attacking residual double bonds in acrylic oligomers and monomers, or in styrene and unsaturated polyester resins. Since most pigments absorb U.V. radiation and can prevent it reaching sufficient photoinitiator molecules, this technique is best suited to transparent coatings or thin pigmented layers (e.g. inks). Alternatively, the same coatings can be cured by electrons from an electron accelerator without the use of photoinitiators. Electrons from a 150600 kV accelerator are energetic enough to create free radicals on impact with the polymer molecules and curing ensues. Clear and pigmented coatings can be cured. Electron accelerators are extremely expensive, but are cheap to run. Both U.V. radiation and electron beam curing are best suited to flat or nearly flat objects, because the beams are directional and shielding must be avoided. Electron beam curing also requires the coating to be in an oxygenfree gaseous atmosphere. Both techniques cure in a fraction of a second and are suitable for fast, high-volume production lines.
Materials and Methods for Various Industrial Finishing Tasks It is not possible, in a section of this size, to deal adequately with the painting systems used by all industrial finishers. Instead a selection will be covered,
14:58
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
to illustrate the range of problems, finishing materials and methods of application and drying encountered. In the sub-sections that follow, there will be frequent references to polymers and resins. Where the detailed chemistry is not shown, it will be found in Section 14.9.
Motor cars: the original finish
The modern motor car is made from steel, zinc or zinc alloy-coated steel and some plastic parts, all of which require painting. The main component is the body shell, made from the above metals, and this is coated in a continuous production process. A full finishing system with all four coatings is usually applied for maximum protection and a high quality appearance. First comes the pretreatment stage. After rust removal and alkaline degreasing, a zinc phosphate formulated pretreatment (see Section 15.2) is applied by dip or spray-dip. Crystalline iron-rich zinc phosphate forms on the metal surface at a coating weight of 0.5-4.5 g/m2. After rinsing and dry-off, the primer is applied. In most modern plants this means electrodeposition of the primer (Section 14.1). The most widely used primers are cathodic. The body shell is made the cathode and current flows between it and inert anodes in the electropaint bath. The paint is formulated so that the resin is basic and, when neutralised with an acid such as lactic acid, becomes positively charged. The most widely used resins are epoxy-amine adducts:
+ 2R2NH+R2N-CHz-CH
v * * - T 7
0 0 epoxy resin
*.*CH-CHz-NR2-
1 OH
I OH
+
+
RzN-CH-CH*..CH-CH2-NR2+
I H
I
I
OH
OH
2HfA-
2A-
I H
The primer contains fine particles of paint in water, each particle being pigmented resin and therefore carrying a positive charge. At the cathode, hydrogen is discharged by electrolysis of water, leaving an excess of hydroxide ions. This pushes the polymer ionisation equilibrium to the right:
+
- N -Rz
+ OH- S
-N-Rz
+ HzO
I H The particles therefore lose their charge. Since the charge provides the colloidal stability, the colloidal paint destabilises and deposits on the nearest surface, the car body. Primer coatings 12-35 p m thick are applied according to primer type. Each particle also contains a crosslinker for the resin, usually a blocked isocyanate. After rinsing, the primed article is passed into a hot
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
14: 59
air oven at 180°C for about 20-30 min, during which time the isocyanate unblocks and reacts with the epoxy. After a de-nib, spray surfacer is applied to build up the film thickness before top-coating. The surfacer contains a high level of pigment and extender (at least 35% by volume) and frequently a saturated polyester resin with a melamine - or urea-formaldehyde crosslinker. The coating is applied at thicknesses up to 35 pm and stoved for 20 min at 150-165°C. Sanding is carried out at this stage and, after clean-up, the final colour or top-coat is applied. There is some variation in the resin chemistry used. Alkyds crosslinked with melamine-formaldehyde are widely used for nonmetallic pigmentation. Metallics are usually based on acrylics for better durability. The acrylic may be thermoset with melamine-formaldehyde or a thermoplastic lacquer (plasticised copolymer of methyl methacrylate). A thickness of about 50 pm is applied and stoved for 20 min at 130°C (lacquers receive a bake-sand-bake process for a smoother appearance).
Motor cars: repair finishing (or refinish)
If a motor car has to be refinished after repair, commonsense suggests that the original finishing system would be ideal for maintenance of protection and durability. However, with tyres, upholstery, fabric and plastic trim fitted and petrol in the tank, the use of such high stoving temperatures is not practical. The practical upper temperature limit is 80°C. This means that none of the original materials is suitable, not even the acrylic lacquer, since this is designed to be sanded and the scratches 'reflowed' at 155°C. A range of lacquer and low-bake thermosetting materials is available and, since many refinishers are small operators with no oven facilities, all of these materials have to be capable of drying at room temperature. For a complete panel replacement, the refinisher starts with a panel preprimed in the appropriate stoving primer. For spot repairs or larger repairs without replacement of metal, there will be areas which have to be rubbed through to clean metal. Any indentations then have to be filled with a stopper or spray filler, probably based on unsaturated polyester resins and styrene, with cure initiated by mixing in an organic peroxide. After sanding, remaining bare metal areas are sprayed with a two-pack etch primer. Etch primers partially fulfil the roles of both pretreatment and primer. They contain phosphoric acid for surface passivation and are based on polyvinyl butyral:
[ - ~ - c H ~ - c HI 0
This provides excellent adhesion to the metal. The PVB will crosslink in the presence of the acid with phenolic resin, and epoxy or epoxy ester resin
14 :60
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
may also be included. Zinc tetroxychromate anti-corrosive pigment is an essential part of the pigmentation, since it contributes to a zinc phosphate conversion layer by reaction with phosphoric acid and additionally provides chromate passivation. After the bare metal is primed, the whole area is built up with primersurfacer. After light sanding where necessary, the repair is completed with topcoat. The materials used in the primer-surfacer are matched to the selection of topcoat. Topcoat is chosen from four main types. Two of these types are lacquers, giving quick drying to the dust-free state at ambient temperature, but at the expense of lower film build. Nitrocellulose-based lacquers are preferred in some European countries and acrylic lacquers in North America. Nitrocellulose is plasticised with nondrying alkyds, polyester and liquid plasticiser. Acrylics are plasticised internally by use of plasticising monomers with methyl methacrylate and by solvent plasticiser. Acrylics give better durability and nitrocellulose gives easier application. With these lacquers, nitrocellulose-based primer-surfacers are used. As well as liquid plasticisers, a wide range of materials are used as plasticising resins: short oil alkyds, maleinised oils, ester gum, rosin and bodied castor oils. Pigmentation is usually inert. Thermoplastic acrylics are often preferred under acrylic lacquers; these are based on acrylic resins and cellulose acetate butyrate. The other two main finish types are thermosetting enamels. The older enamels are based on quick drying short oil alkyds which dry by oxidative drying. Alternatively, a second component containing either melamine formaldehyde or polyisocyanate may be added to give cure with heat. Higher film thicknesses can be obtained, but drying to the dust-free stage is slower, polishing properties are poor and the enamel may be sensitive to solvent attack if recoated. Nitrocellulose or alkyd primer-surfacers are used. In recent years the two-pack acrylic/polyisocyanate finishes have gained ground widely, giving a good balance of properties, including excellent durability. Heat is preferred for drying if available. These finishes are widely specified by motor manufacturers for repair of damaged cars which are still under corrosion warranty. Primer-surfacers may also be acrylic/polyisocyanate-based, or alternatively the acrylic resin may be replaced with alkyd or polyester. Whereas aliphatic polyisocyanates must be used in the topcoats for good colour and durability, aromatic polyisocyanates can be used in the primer-surfacer for fast cure and economy. Coil Metal for Exterior Cladding
This is steel or aluminium sheet made in a continuous ribbon and wound tightly onto a bobbin to form a coil of metal. On a coil finishing line, the coil can be fitted at one end, and wound up pretreated, primed and finished on both sides at the other end. Sheets of painted metal can be cut from the coil and formed for use as the exterior cladding for, for example, industrial buildings and caravans. There are some similarities between coil finishing and original motor car finishes: both are required to give good exterior durability and both can be
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
14:61
dried at high temperature. There the similarities end. Because coil is a continuous web, the finishing process can be completely automated and carried out at high speed with extreme efficiency. Lines run at 30-200 m/min and this means that stoving temperatures must be very high and times very short (15-60s) if ovens are not to be excessively long. Temperatures peak at 180-250°C just as the coil leaves the oven and the paint is then crash-cooled by water spray. The first stage is a cleaning and spray-applied or immersion pretreatment process. If the metal is hot-dipped galvanised steel, a complex metal oxide pretreatment may be applied, followed by a passivating chromate rinse, to improve paint adhesion and inhibit ‘white rust’. Afrer drying, primer is applied by roller coater at a thickness of about 5pm. Epoxy resin crosslinked with amino resin is often preferred and chromate pigmentation is used. Application is followed by stoving, quenching and topcoat application at a thickness of 20 pm, again by roller coater. The coil is then passed through another high-velocity hot-air oven, followed by quenching and cooling, and is then wound up. For industrialised buildings long life is required and coating systems are expected to be more durable than those on motor cars, even though paint thicknesses are lower. For this reason, the lowest durability type offered is the thermosetting acrylic (7 years). Longer life can be obtained from polyester resin crosslinked with hexamethoxymethyl melamine (10 years), siliconised polyester with the same crosslinker (12-15 years) or polyvinylidene fluoride/acrylic (20 years). Alternatively, cheaper PVC plastisol can be applied at a thickness of 100-250 pm to give a very damage-resistant coating with a life of 10-15 years. The back of the coil is simultaneously roller-coated at each station (if necessary) with a 10 pm coat of polyestermelamine backer or a 3-5 pm coat of primer and 8-10 pm of backer. The very high durability of PVF, comes from the polymer structure: F H F H F H 1 l 1 1 1 1
-c-c-c-c-c-cI F
l H
I F
l H
l F
l H
This material does not absorb U.V. radiation at all and so is not degraded by sunlight. The structure of polyvinyl chloride is quite similar: H
H
H
H
H
H
I
I
I
I
I
I
I H
I I C1 H
I C1
I H
I C1
-c-c-c-c-c-c-... However, this structure does not give the same properties, and the polymer degrades slowly, eliminating HCI. Plastisols (PVC + plasticiser) lose gloss rapidly and gradually chalk even in temperate climates, but the high film thicknesses that lower cost permits lead to long life.
14 :62
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
The increased durability obtained by siliconising a polyster resin comes from reacting a high hydroxyl value polyester with 20-30% of appropriate silicone resin. R
R
R
R- .I k - O f ~ i - O ~ i i - R R where R = -0-CH,
R
R
or -0-Ph
Agricuitural equipment
Tractors, combined harvesters, ploughs, harrows, etc. are large and complex machines with many parts. Some of these are sheet metal and others are castings, and all are mainly steel. The assembled product is finished in a uniform single ‘house colour’ of the manufacturer, even though the parts may be painted with different systems in different finishing shops. Like coil and motor cars, agricultural equipment must have exterior durability, though the main emphasis is placed on showroom appearance. However, because of the variety of components and systems, some are air dried, some force dried and some stoved at temperatures varying from 15°C to 150°C. Short-medium oil alkyds are used for these coatings, with driers at ambient temperature or force-dry temperatures (60-80°C) and with amino resin crosslinkers at stoving temperatures (120-150°C). Relatively high solids can be obtained, leading to the full-bodied glossy appearance required at lowest possible cost. Parts are normally degreased, but not pretreated. Primers are applied to critical areas, but much of the metal receives only topcoat. Primers and one-coat finishes are applied by dipping, electrodeposition or flow coating. Waterborne alkyds are increasingly used, for reduced fire hazard and lower environmental pollution. Water solubility or dispersibility is achieved by making alkyd molecules with higher concentrations of acid end-groups; these are neutralised with ammonia or amines to a pH value of about 8. In such alkaline media, hydrolysis of the polymer’s ester linkages can occur rapidly, and storage life has to be extended by the use of more expensive hydrolysis-resistant acids and alcohols (e.g. 5 or 6 carbon diols, shielded hydroxyls, as in neopentyl glycol, and isophthalic rather than o-phthalic acid ’). Some water-miscible solvent is also necessary. Topcoats over primer are often applied by airless spray. Trends to higher standards of exterior durability have encouraged the use of rnethacrylated alkyds and two-component urethane finishes. Aircraft
This is the last of the end usages in this section for which exterior durability is required from the painting system. The substrate here is mainly aluminium
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
14:63
alloy in various forms. The aircraft is constructed from many components and where possible these will be coated at least as far as primer before the aircraft is assembled. Protection from corrosion is a major requirement. Chromic acid anodising and chromate conversion coatings (Section 15.3) are used at the pretreatment stage and these are followed by two-pack epoxypolyamide (or polyamine) primers, which will cure at ambient temperature. The primers contain leachable chromate pigments for maximum corrosion protection. From the topcoat a number of properties are required. First, a high-gloss quality appearance at least as good as that obtained from motor car finishes. Next, U.V. resistance, including resistance to the more destructive shorter wavelengths emitted by the sun which are usually screened out by moisture in the atmosphere, and resistance to extremes of temperature, varying from -50°C in flight to over 70°C on a tropical airstrip (due to absorption of energy by the paint film, especially in darker colours). A special requirement is resistance to the aggressive phosphate ester hydraulic fluids used in aircraft. These requirements are usually met with two-pack paints based on hydroxyl-rich polyester or acrylic resins in the pigmented pack and aliphatic polyisocyanates in the activator pack. Cure with this type of finish is relatively fast and complete even at low ambient temperatures. An alternative finish is an acrylic lacquer, similar to the lacquer used for refinishing motor cars. These finishes are applied t o the assembled aircraft by operators protected by air-fed hoods and using airless or conventional spray guns. High durability pigments are included.
Domestic Appliances
The key properties here are hard :ss and wear resistance. ability t stand minor knocks and dents without cracking and resistance to various domestic chemicals. These vary with type of appliance, e.g. detergent solutions are important for washing machines, while a fridge will be required to withstand fruit juices, ketchup and polishes. Good colour and appearance in white and mainly pastel shades will be expected. Corrosion resistance is required, especially for washing machines, and domestic appliances frequently have to withstand humid conditions in kitchens. Good quality steel is used and electrozinc is preferred for washing machines. Steel is pretreated with iron phosphate for economy; electrozinc with a fine crystal zinc phosphate. N o primer is normally used: 2540 pm of finish is applied direct to metal. The required properties are best obtained with a thermosetting acrylic or polyester/melamine-formaldehyde finish. Self-reactive acrylics are usually preferred; these resins contain about 15% N-butoxymethyl acrylamide (CH,=CH -CO- N H -CH2-OC,H,) monomer and cure in a manner similar to butylated melamine-formaldehyde resins. Resistance or anti-corrosive properties may be upgraded by the inclusion of small amounts of epoxy resin. Application is usually by electrostatic spray application from disc or bell. Shapes are complex enough to require convected hot-air curing. Schedules of 20min at 150-175°C are
14:H
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
obtainable with the use of p-toluene sulphonic acid (or blocked PTSA) catalyst. A very high quality finish can be obtained with little or no organic emission if the liquid coatings are replaced by a powder coating. Powder coatings are paints in powder form, with a particle size range from 10 t o 80 pm. Each particle contains all the pigments necessary to give the colour, the filmforming ingredients and the additives. Milling of pigment is done in an extruder under polymer melt conditions, with all other ingredients present. Extruded paint is then rolled into sheet, broken up into flakes and then ground to powder in a pin mill. Classification is necessary to reject ultra-fine and coarse particles. For this use, the preferred powders are based on acrylic, epoxy or polyester and epoxy resins. For best colour, epoxy resins are crosslinked with anhydrides of dicarboxylic acids in the straight epoxy coatings, or with saturated polyesters of high acid content in the epoxy-polyester type. Acrylics contain epoxide rings via, for example, glycidyl methacrylate (CH2=C(CH3)- CO-0- CH2-CH- CH2), and these groups crosslink \ /
0 by reaction with carboxyls in diacids or other acrylic molecules. The powder for this use is applied using electrostatic guns and, since the transfer is not very efficient, unused powder is recovered in a cyclone. Curing times are around 15 min at 170-190°C. Yet another option for domestic appliances is to make the appliance from precoated coil. The appliance has to be designed to minimise the problem of unprotected cut edges. Electroplated zinc-coated steel, pretreated, primed and finished with special polyester-melamine, is used. The finish is designed to be hard at room temperature, yet accept bending and forming, probably, but not necessarily, at somewhat higher temperatures (ca. 60°C). Heating and Ventilating Equipment
Ducted hot air heaters or airconditioners are made largely from sheet metal and finishing systems are similar to those for domestic appliances. Alkydamino resin finishes will usually give sufficiently good performance, since resistance to household chemicals is not important in the specification. However, European panel radiators are made largely from cast metal, though corrugated sheet metal is often welded to the back, or between panels, to create a larger, ‘extended’ surface from which convection can occur. For these radiators, a finish able to withstand knocks and to accept repainting by decorative house paints is required. A painting method that gives good coverage of the complex shapes of extended radiators is also required. After degreasing and pretreatment with iron phosphate, the finish is applied by electrodeposition or by dipping in a waterborne coating. Acrylic or polyester finishes are applied, usually anodically if by electrodeposition. For even better appearance, the dip layer is a primer and this is followed by an electrostatically applied liquid polyester-melamine or by a powder coating.
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
14:65
Cans
Cans are used for packaging in a wide range of industries. The market divides into three main sectors: beer and beverage, food and general line (covering a multitude of usages outside food and drink). Cans are made from several metals, principally aluminium (especially in the USA), tinplate (especially in Europe) and tin-free steel (steel with a chromium/chromium-oxide coating). Part or all of the can may be made from flat sheet or coil. The body is often drawn from metal discs and this may be done before or after painting. Interior coatings for food and drink have to conform to stringent food regulations and have excellent resistance to the can’s contents. Exterior coatings are more concerned with appearance, ink acceptance, resistance to machine handling and to processing. This is a very wide range of requirements, so this section concentrates on the interior usages, with their main requirement for corrosion resistance, and mainly on beer and beverage containers. These containers are commonly of a two-piece design: body plus end. The body is made from aluminium or tinplate by drawing from a disc and then wall-ironing to stretch and smooth the metal further. The coating is applied by airless spray into the revolving body and must protect the metal from attack by contents which are often acidic. However, once the end is sealed in place, the pack is under carbon dioxide pressure and virtually anaerobic. Under these conditions it has been found that satisfactory protection is obtained from 3-4 pm of a waterborne acrylic-modified epoxy resin clear coating on aluminium. On tinplate, the wall ironing exposes a high proportion of steel and higher coat weights are needed: 5 pm for beer and up to 11 pm for soft drinks. Coatings must be completely continuous and lacquers are tested for pinholes in an electrical conductivity test. Drying is by convected hot air: 3 min in the oven, with one minute at the peak temperature around 200°C. The epoxy-acrylic resin referred to above is a graft copolymer prepared by the polymerisation of acrylic monomers in the presence of the epoxy resin in such a way that grafting of the acrylic onto the epoxy takes place. Water dispersibility is achieved by neutralising carboxyl groups in the acrylic polymer chain with ammonia or amine. Amino or phenolic resins are used as crosslinkers. Alternatively, solvent-borne epoxy-amino or epoxy-phenolic lacquers can be used. Two-piece food cans may be made by a draw-redraw process, in which lacquer is first applied to and cured on sheet. Blanks are then cut from the sheet and the can is drawn from the blank in two or three stages. The lacquer deforms with the drawing process and lubricates the draw. It then becomes the interior protective coating. Although epoxy-phenolic solvent-borne lacquers are used, even better drawing properties are obtained from organosols. These are dispersions of colloidal polyvinyl chloride powder in solutions of other mixed resins in solvent, e.g. chosen from epoxy, polyester, vinyl and phenolic.
14 :66
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
Wood and Paper
This is the one example in which metal is not the substrate. Corrosion takes on a new meaning; the coating here is required to protect the substrate from direct attack by ‘corrosive’ substances, from water to more powerful household or industrial chemicals, such as grease, alcohols and bleach. We are concerned with the industrial application of thin protective layers to paper (e.g. labels), card (e.g. playing cards) and many wooden articles, including industrially finished doors, window frames and, particularly, furniture. Most paper and card finishing operations require only a finish (though, two coats may be needed), while wood may require a primer for sealing the porous surface and then fillers and undercoat to level grain and build up thickness before the topcoat. These operations have the common need to dry the coating without damaging the sensitive substrate. This may be done with cool conditions (room temperature to 60°C), fast air movement and relatively long times, or by short bursts of heat from high velocity hot air or infra-red heaters. Alternatively, curing may be brought about by ultraviolet radiation or electron beams. Coating materials may be based on short or medium-oil alkyds (e.g. primers for door and window frames); nitrocellulose or thermoplastic acrylics (e.g. lacquers for paper or furniture finishes); amino resin-alkyd coatings, with or without nitrocellulose inclusions, but with a strong acid catalyst to promote low temperature cure (furniture finishes); two-pack polyurethanes (furniture, flat boards); unsaturated polyester resins in styrene with free-radical cure initiated by peroxides (furniture); or unsaturated acrylic oligomers and monomers cured by U.V. radiation or electron beams (coatings for record sleeves; paperback covers, knock-down furniture or flush interior doors). These coatings are applied by spray on more complex shapes, but on flat sheet or board roller coating is the preferred method, with curtain-coating used for thicker layers. Nitrocellulose, of the resins used in these end uses and in car refinishing, is the nitrate ester of cellulose. The structure is linear and a wide range of (high) molecular weights is available as well as various degrees of nitration: H
0-NO,
I
H
CH~O-NOZ
0-NO2
-
n A cellulose nitrate
Unsaturated polyesters are similar to the saturated polyesters shown in Section 14.9, but include maleic anhydride or fumaric acid to introduce unsaturation:
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
HC-CO
I1 HC-CO
14 :67
HOOC-C-H
>o
Maleic anhydride
II
H-C-COOH Fumaric acid
Unsaturated acrylic oligomers are made from unsaturated acrylic monomers. For example, an epoxy acrylate may be made by reaction of acrylic acid with epoxy resin.
Newer Developments The pressures leading to new developments in industrial painting derive from the drive for better quality, the need for economy, and the demand for increased safety in the workplace and in the environment.
Better quality Nowhere is this more evident than in the motor industry, where warranty times for corrosion protection have steadily lengthened. There is a target towards which the industry is moving of a ‘10-5-2’, warranty, i.e. a 10 year guarantee against perforation, a 5 year guarantee against cosmetic corrosion damage on the outer face of the metal and a 2 year guarantee against corrosion at edges. T o achieve this, more and more of the car body steel is coated with zinc or with a range of zinc alloys. In Japan, some of these alloys are delivered to the car manufacturer already coated with a conversion coating and a 1 pm organic coating, for greater protection for those parts which cannot at present receive paint. These changes have created many new difficulties and challenges for pretreatment process suppliers and paint suppliers alike. New multi-metal pretreatments are becoming available, and more versatile electropaints are required. Economy Economies can be achieved in various ways: lower cost paint, fewer painting operations, less paint, faster throughput, more automation and less energy for cure. Pressures continue on paint suppliers on all these fronts. Attempts are being made to extend the etch primer principle to uses other than refinishing by developing primers that also have a pretreatment action. The most widespread pressure is to bring stoving temperatures down, and decreases of 10-30°C have proved possible in many end uses. Alternatively, much greater throughputs are being required without temperature reduction to increase line capacities. An interesting process, which eliminates heat altogether, is the vapour curing process. In this, isocyanate-containing coatings are cured rapidly by exposure to catalytic amine vapour at ambient temperature. Safety No year goes by without some widely used chemical being declared suspect on toxicity grounds. The paint industry has responded rapidly to eliminate toxic chemicals from coatings or to show how they can be used safely in an industrial environment. Examples are the elimination of specific ether-alcohol solvents and the introduction of air-fed hoods for spraying isocyanates. Of particular interest in corrosion prevention is the current pressure to eliminate chromate pigments. Currently there are no equally effective alternatives and the emphasis has had to be on safe usage. The search for replacements continues.
14:68
PAINT FINISHES FOR INDUSTRIAL APPLICATIONS
Pollution of the environment is increasingly regarded as undesirable and the trend over the last 15 years to change to less polluting coatings (water-based, higher solids, 100% polymerisable and powder coatings) will continue. Where lower solids solvent-borne coatings are still necessary, after-burners are often installed to burn the solvent fumes and recycle the energy released. Waste disposal problems with chromate pretreatments are being minimised by the introduction of ‘no-rinse’ or ‘dried-in-place’ pretreatments, which are roller coated onto flat metal surfaces, with virtually no waste. G . P.A. TURNER
REFERENCES 1. Pray, R . W . , Radiation Curing, 5 No. 3 , 19-25 (August 1978) 2. Sandner, M. R. and Osborn, C. L., Tetrahedron Letters, 415 (1974) 3 . Turpin, E. T., J Paint Techno/, 47 No. 602, 40-46
BIBLIOGRAPHY Turner. G. P. A.. Introduction to Paint Chemistry. 3rd edn, Chauman and Hall, London (1988)
Lambourne. R., Paint and Surface Coarings: Theory and Practice, Ellis Horwood, Chichester (1987) O.C.C.A. Australia, Surface Coatings, Vol. 1 and 2, Chapman and Hall, London (1983) Solomon, D. H . , TheChemistryof Organic Film Formers, Robert E. Krieger, Malabar, Florida ( 1977)
14.6 Paint Finishes for Structural Steel for Atmospheric Exposure
Paint for structural steelwork is required mainly to prevent corrosion in the presence of moisture. In an industrial atmosphere this moisture may carry acids and in a marine atmosphere this moisture may carry chlorides. Paint is therefore required to prevent contact between steel and corrosive electrolytes, and to stifle corrosion, should it arise as a result of mechanical damage or breakdown of the coating through age and exposure. For an adequate barrier against moisture, sufficient thickness of paint must be applied. The modern trend is to apply high-build coatings based on media having high intrinsic water resistance. Such paints may be pigmented with corrosion inhibitors or minerals which impede the flow of moisture through the film. Correct surface preparation is of paramount importance. High performance paints will almost certainly fail if applied over badly prepared surfaces whilst simple, low performance coatings may perform surprisingly well over correctly prepared surfaces. Good adhesion is essential and the biggest single factor in good adhesion is good surface preparation.
Methods of Preparing Structural Steel Degreasing
The first stage in any method of surface preparation is to ensure that any oil or grease is removed, otherwise the preparation method is likely to spread the contamination over a wider surface. Large quantities of oil or grease should be physically removed by scraping, and then the rest is best removed by emulsion cleaners, followed by thorough water rinsing. Under site conditions, degreasing by wiping the surface with solvent is not recommended because this invariably leads to the spreading of a thin film of oil over a wider area. In a factory, however, solvent vapour degreasing can be a very effective process.
14:69
14 :70
PAINT FINISHES FOR STRUCTURAL STEEL
Manual, Wire-brush and Mechanical Methods
Cleaning with mechanical or hand wire brushes, grinders, chippers or scrapers rarely removes millscale, paint or other tightly adhering contaminants, or traces of rust or deposits in pits and crevices. Results can be very variable and the process must generally be a relatively slow one in order to be effective. On the other hand, for very heavily rusted surfaces, initial chipping and scraping can save time by removing loose, heavy deposits before more thorough surface preparation methods are employed. Photographs of different levels of hand cleaning are included in the British Standard 7079:Part A1:1989, St Series’. Since wire brushing as a method of surface preparation is unlikely to remove much contamination, the old practice of ‘weathering’ beforehand should be avoided if possible. It can only result in the transformation of new steelwork, with its admittedly undesirable millscale, into corroded and pitted steelwork, with corrosion products which are even more undesirable and difficult to remove. Dry Abrasive Cleaning
This is the most important and most widely used mechanical method of surface preparation. Originally, sand was used as an abrasive but now, because of the hazard to health, it has already been replaced in the UK by metal or non-silicon materials. There are two main types of process. In the first, the abrasive (generally a non-reusable, nonmetallic type) is carried by a jet of compressed air through a hand-held nozzle. In the second, the abrasive (generally round iron or steel shot) is thrown centrifugally from rotating impellors in a fixed plant. Both types are suitable for factory work but compressed-air blast-cleaning systems are more versatile and are most commonly used for on-site cleaning. Smaller blast-cleaning equipment incorporating a vacuum at the head to collect the abrasive is also available. This is slower in use than the conventional system but it can sometimes be used in situations where open blasting is not possible. It is particularly useful for small-scale repair work. Photographic standards and written descriptions of various stages of visual cleanness of steel surfaces after surface preparation by blasting are available in British Standard 7079:Part A1 :1989, Series Sal. Wet Abrasive Cleaning
High-pressure water jetting can be a dangerous process. Also, it is not a very efficient method of cleaning a surface for painting. The addition of an abrasive, generally sand, to the water gives a considerable improvement in cleaning. There are now even more effective wet processes using low-pressure water added to a high-pressure air stream containing sand. Since all wet processes leave wet surfaces these will soon form a powdery film of rust which, although generally iron oxide rather than iron sulphate or chloride, would be an undesirable surface to paint over because of its powdery nature. Some wet processes use inhibitors in the water to prevent such rusting, but it is
PAINT FINISHES FOR STRUCTURAL STEEL
14:71
important to establish that any traces of such inhibitors will be compatible with the subsequent paint finish. The inhibitors themselves are obviously water soluble and if left in quantities on a clean surface would be another cause of subsequent breakdown. Flame Cleaning
In this method an oxyacetylene or oxypropane flame is passed across the steel. The sudden heating causes millscale and other rust scales to flake off as a result of the differential expansion between the scale and the metal. In addition, any rust present is dehydrated. Immediately after the passage of the flame, any loose millscale and rust that remains is removed by wire brushing. This generally leaves a powdery layer which must also be removed by dusting down. Acid Pickling (Section 11.21
Pickling as a method of surface preparation is generally carried out by immersing the steel in an acid bath and then rinsing with clean water. It is essentially a works process because it must be carefully controlled. Site application of acid washes, etc, is not recommended.
Types of Paint (Section 14.2) Protective coatings are usually applied as systems. The simplest system would be: (i) A primer in contact with the metal. This usually contains a corrosioninhibiting pigment, capable of stifling either the anodic or the cathodic reactions in electrolytic corrosion. (ii) Finishing coats capable of adhering to the priming coat, resisting the ambient exposure conditions and providing the necessary decoration, light reflection, etc. where necessary. It is usual to define primers in terms of the principal inhibiting pigment e.g. zinc phosphate, zinc dust or zinc chromate, and the topcoats in terms of the binder, e.g. alkyd, chlorinated rubber, etc. This practice can be confusing, however, and lead to the selection of incompatible coatings. The paint system needs t o be chosen carefully for demanding environments, particularly marine situations. In general, interior steelwork is exposed to less severe conditions than exterior, but in some chemical factories the reverse is true and here special types of paint are needed. Much structural steel is encased in concrete; it is therefore hidden from view and is given some protection while the concrete remains alkaline. Where the concrete is thick, corrosion may be delayed, but as the concrete becomes carbonated and particularly if it is penetrated by acidic rain water, the metal will corrode. In general it is advisable that steel which is to be encased in concrete, especially for industrial plants, should
14: 72
PAINT FINISHES FOR STRUCTURAL STEEL
be prepared by one of the procedures outlined above and coated with an anticorrosive alkali-resisting composition.
Air-drying Paints
The selection of paint is a matter for the expert, but some knowledge of composition is of help to the user. Paints based on the drying oils, usually linseed and tung oil, are still used for decoration and protection, though the traditional oil paints have been superseded by those based on synthetic resins. Of these the alkyd resin and phenolic resin paints are the most widely used because they have excellent durability. The normal decorators’ paints, however, do not have the necessary resistance to chemical attack required for protecting steelwork in industrial conditions. Alkyd paints, for instance, are sensitive to alkali and are frequently softened and degraded by prolonged exposure to hot steamy conditions. Alkali formed locally at the cathodic area of a steel surface may destroy the adhesion between such paints and the metal. A high degree of resistance to water and chemical attack is provided by some oil-based paints, notably those based on tung oil and pure phenolic resin, but for the greatest resistance to these forms of attack, oil-free paints are recommended. Of these, bitumen is widely used, because it is cheap. Bituminous coatings fulfil an important rale in protecting hidden steelwork, where appearance is of little account. In recent years there have been considerable advances in the technology of bituminous compositions, and heavy-duty compositions now available give hard, tough coatings which can withstand rough handling without damage and virtually exclude all water from the steel. Chlorinated rubber-based paints have the advantage of combining acid and alkali resistance with weather resistance and decorative qualities. Highly impermeable anticorrosive systems can be built up and these paints have been used with great success to protect industrial plants where low maintenance costs are needed. The alkali resistance of chlorinated rubber paints makes them suitable for protecting concrete where it is desirable to safeguard embedded steel from corrosion. Chlorinated rubber finishes are now also available as high-build coatings and the combination of high intrinsic resistance with thickness provides excellent protection.
Chemically Cured Paints
These are supplied as separate components which are mixed together and then applied. The paints cure by chemical reaction-a process which also occurs in the can and so limits the time available for application after mixing. The films are tough and have good chemical resistance. There are three main types of these coatings: (i) Epoxy resin-based materials, which are cured with amino compounds or their derivatives.
PAINT FINISHES FOR STRUCTURAL STEEL
14 :73
(ii) Polyurethane coatings which cure by the interaction of polyisocyanates with hydroxylated resins. (iii) Polyester resin finishes which cure by peroxide-stimulated polymerisation. All these materials are capable of giving durable coatings. The epoxide resin finishes are highly resistant to alkali and acid and, like the other chemically cured finishes, are resistant to a wide range of oils, greases and solvents. They are used for protecting steelwork. The adhesion of paint to steel is good if proper attention is paid to preparation of the surface and if due attention is given during formulation to the ultimate structure of the cured film. In this respect both curing agents and solvents play a significant part. Thick Coatings
Chemically cured coatings differ from air-oxidised coatings in that they dry throughout the film regardless of thickness. In thick films, oil paints may not cure satisfactorily. The chemically cured materials lend themselves to protective coatings of considerable thickness with the consequent advantages of good performance and long life, and they have contributed significantly to the protection of steel in corrosive conditions. It is possible to apply high build systems which equal in thickness and performance many coats of orthodox paints, with consequent savings in labour costs. The extra cost of materials is more than compensated for by savings in time and application costs, and where scaffolding and shut-down time are involved this may be a matter of great importance. Quite apart from the economic advantage of thick films, the lower the solvent content the lower the intrinsic permeability to moisture and aggressive ions. Solvents, particularly polar solvents as used in many polymer resin-based paints, influence the structure of films over the early weeks of their life. Small quantities of many solvents are retained in the cured films for a long time, and water and aqueous solutions are able to penetrate the solvated films more easily.
The Paint System The priming coat provides the bond between the metal and subsequent coats. It gives electrochemical control of corrosion. Adhesion is dependent largely on the nature of the binder and the cleanliness of the metal surface. The pigment is the principal agent in the electrochemical control of corrosion by primers (see Section 14.3). Probably the best known anticorrosive pigment is red lead. When used in conjunction with linseed oil as the binder it gives very good primers which will perform well over relatively poorly prepared (manually abraded) steel surfaces. Present-day use of red lead (and lead pigments, generally) in paints has been drastically curtailed as a result of understandable pressure from the environmentalists. Zinc chromate and zinc tetroxychromate have also been used successfully in anticorrosive paints. Both pigments function by releasing chromate ions which passivate the steel surface. In common with lead pigments, those
14 :74
PAINT FINISHES FOR STRUCTURAL STEEL
based on chromates are now also under toxicological suspicion and their use in paints has declined significantly. In the United Kingdom, zinc phosphate has been the mainstay of many anticorrosive primers in recent years. It can be incorporated into most binders and primers can be manufactured in a range of colours because of its transparent nature. The mechanism of protection is still uncertain. Metallic zinc is also used widely in anticorrosive primers and zinc-rich paints are considered by many to afford best protection. Initially, the zinc protects the steel by galvanic action but, with time, zinc salts form an impermeable barrier and provide a second, reinforcing mode of corrosion protection. For effective galvanic protection, high concentrations of zinc are required (more than 90% by weight of zinc in the dry paint film) and the steel must be cleaned to a high degree in order that the zinc may be in intimate contact with the substrate. The search for new, effective anticorrosive pigments with low toxicity to replace red lead and chromates in paints has occupied the attention of many paint-making companies recently. Barium metaborate, calcium molybdate and zinc molybdate have been identified as possible compounds but they have not found general acceptance in the United Kingdom and western Europe, most probably because of their lower cost effectiveness. Welds on steelwork need special attention because of the different composition of weld metal and adjacent steelwork, the rough surface and spatter caused by welding and the presence of welding flux. The latter is often alkaline and destructive to many paints. It is necessary to clean thoroughly, preferably by reblasting for 25-50 mm each side of the weld, fare the rough metal and wash off residual flux. The cleaned surface should then be stripe coated with the primer used on the remainder of the surface.
Methods of Application (Section 14. I 1 Paint is applied to structural steelwork most commonly by airless spraying. This method of application is particularly well suited to high build coatings where the combination of rapid working and great film thickness allows work to be completed quickly and cost effectively. Application of paints by brushing is still often used for maintenance painting programmes involving small areas. The weather has an important effect on the drying of paint and on subsequent performance. Paint applied in bad weather may be slow to dry and remain susceptible to damage by rain and fog for a long time. Heavy steelwork has a large heat capacity and follows temperature changes of the ambient air only slowly. Careful consideration of weather conditions and planning of work is frequently repaid by improved results. With new construction there is much to be said for applying the primer and intermediate coats of a paint system at works and applying only the finishing coats on site.
PAINT FINISHES FOR STRUCTURAL STEEL
14: 75
Economic Considerations The costing of painting structural steelwork is a complex subject. The main items of costing are: 1. Scaffolding. 2. Labour, which may be further subdivided into surface preparation and application labour charges. 3. Materials. 4. Supervision and transport.
The proportion of the whole contributed by each of these items will vary with each job, but it will be immediately apparent that the cost of scaffolding and labour far outweighs the cost of materials and supervision. Therefore even a large increase in the cost of the last two items will produce only a fractional increase in total cost. On the other hand, first-quality materials and rigid supervision will give greatly increased protection and the best value from the expensive items of scaffolding and labour. It is economically sound to consider not only the initial cost of protection, but also the annual cost over the life of the structure, taking into account initial work, maintenance charges and the cost of shutdown. It is now widely recognised that highquality initial preparation and protection leads to reduction of total costs on an annual basis.
Maintenance Painting All the foregoing has been concerned with the initial protection of steelwork, but there is far more maintenance painting than new work. The same principles apply to maintenance painting, with the exception that it is often only in isolated patches and in complicated situations, such as around flanges, etc. that the steelwork is bare of paint, and then it is frequently heavily contaminated with corrosion products. The first necessity, therefore, is to clean down these areas to bare steel, but often it is not possible to use blasting methods. Often hand cleaning is all that can be done. Careful supervision is needed, and the cleaned areas must be primed without delay and then brought forward with a suitable anti-corrosive system.
P. J. GAY N. R. WHITEHOUSE REFERENCE 1. BS 7019:1989,Preparation of Steel Substrates before Application of Paints and Related
Products, Part A1 . Specification for Rust Grades and Preparation Grades of Uncoated Steel Substrates and of Steel Substrates after Overall Removal of Previous Coatings.
14.7 Paint Finishes for Marine Application In considering the requirements of paints for marine use it is necessary to distinguish between the parts of ships that are subject to different conditions of service. The exterior area of ships may be divided broadly into three parts: (a) the bottom, which is continuously immersed in the sea; (b) the boot-topping or waterline area, which is immersed when the ship is loaded and exposed to the atmosphere when cargo has been discharged; and (c) the topsides and superstructure areas, which are exposed to the atmosphere but subject to spray. In addition to these weather factors, the outsides of ships are also subjected to attack arising from the conditions of use, e.g. the boottopping is subject to abrasion by rubbing from quays, wharves and barges, while the topsides, superstructures and decks may receive mechanical damage during cargo handling. The interior surfaces, too, present varying requirements according to the conditions of use; cabins and accommodation spaces for crew and passengers call for treatment other than that demanded by cargo holds. A particular problem of ship interiors, to which special attention has been devoted in recent years, is the protection of the cargo tanks of oil and chemical tankers, and in particular those carrying acids and elemental sulphur. Although light alloys and non-metallic materials such as reinforced plastics are finding increasing applications in shipbuilding, the principal construction material is generally mild steel. Hence the protective painting of ships is basically a special aspect of the painting of steel. In relation to atmospheric exposure, the main principles of the subject are: (i) Proper surface preparation. (ii) Appropriate composition of the paint, in particular the use of an inhibitive priming paint. (iii) Adequate film thickness. (iv) Good conditions of application. These apply also to marine painting, but here additional factors must be taken into account. The present section refers specially to differences between ships’ painting and structural steel painting.
Surface Preparation and Pretreatment This is the most important factor determining the life of a protective paint system on steel. The best surface is one free from rust, scale, grease, dirt 14:76
PAINT FINISHES FOR MARINE APPLICATION
14:77
and moisture, Le. it is completely clean and dry. The removal of millscale is particularly important under marine conditions I , especially for ships’ bottoms, because the environment has a high conductivity which enables corrosion currents to pass easily between cathodic scale-covered and anodic scale-free areas. This results in pitting when the ratio of scale-covered to scale-free areas is high. A small scale-covered area with a large scale-free area is not so serious because the corrosion is spread over the larger area. Millscale and rust can be completely removed from steel by acid pickling or by blast cleaning. Pickling was formerly used in some shipyards, but during the years 1960-65 nearly all shipbuilders installed automatic airless blast-cleaning machines for the treatment of steel plates and sections prior to fabrication. In these machines the abrasive*, generally steel shot, is thrown against the steel by impeller wheels. A series of wheels directs the shot against each side of the plates as they pass through the machine at about 2m/min, this speed being adjusted in relation to the quantity, size and velocity of the shot so that the millscale and rust are properly removed. The finish produced by these machines is normally Second Quality of BS 4232~1967or SA2.5 of Swedish Standard S.I.S. 05 59 00-1967, and with a surface profile not exceeding 100 pm. The process is rapid and dry, and the machines are totally enclosed to prevent particles of abrasive and millscale getting into the atmosphere - accordingly they can be installed in the steel fabrication shops of modern shipyards. (Acid pickling, on the other hand, is a wet process requiring the steel to be immersed for some hours in a bath of acid and then rinsed thoroughly in water - it tended to be messy and was often banished to a corner of the shipyard.) Automatic blast cleaning of plates in these machines is much cheaper than blast cleaning after erection because labour charges are low and the abrasive is recovered, graded and re-used, fresh abrasive being added to make up for the fine particles rejected with the millscale. The cleanliness of the surface may be checked (a) visually using a hand lens, with which residual millscale and rust can be seen, (b) by the copper sulphate test3, or (c) by a reflectance method4. The surface profile may be checked (a) by examining the surface, or a replica, using a stylus type of surface profile instrument (6) by a simple probe type instrument6, or (c) by using a roughness gauge4 depending on the rate of leakage of gas from a cup held against the surface. The clean, dry, slightly rough steel surface produced by blast cleaning is ideal for the application of paint, but will not remain in this state for more than a few hours under average shipyard conditions. General practice’ is to apply a thin coat of prefabrication primer (also known as a blasf or shop primer) to the steel as it emerges from the blast-cleaning machine. The primary function of this primer is to protect the surface of the steel for the six to nine months during the fabrication and erection of the ship, but it must also meet other requirements to permit its use under practical conditions in shipyards, e.g. it must dry rapidly to permit the steel to be handled in 2-3 min, must withstand abrasion, must not affect the speed of flame cutting or welding, must not affect weld quality, must not cause any health hazards from fumes when coated steel is welded or flame-cut, and must be compatible with any type of paint system likely to be used on the different parts of ships. The principal types of prefabrication primer in commercial use are
’,
14:78
PAINT FINISHES FOR MARINE APPLICATION
(a) cold-cured epoxies pigmented with zinc dust, (b) zinc silicates, (c) phenolic-reinforced wash primers pigmented with red iron oxide and ( d )cold-cured epoxies pigmented with red iron oxide and inhibitive pigment. In many shipyards there are objections to the zinc types because zinc oxide fumes are evolved during welding and flame cutting, and for this reason the red oxide types are more widely accepted. The wash primer types are not universally compatible with marine paint systems, and the epoxy types are there fore recommended.
Selection of Paint Systems for Use on Ships Exterior Surfaces above the Waterline
As indicated earlier in this section, the choice of paints for marine use depends upon the conditions of service to which the part in question will be subjected. Thus the paints used on the exteriors above the waterline and on most of the interiors do not differ fundamentally from those used on structures ashore. Inhibitive priming paints are used on steel, including those based on red lead, calcium plumbate, zinc phosphate or zinc chromate. The best known structural steel primer, i.e. red lead in linseed oil, is still used on ships, although it requires a long drying time. Slow drying is a disadvantage for marine paints, particularly on ships in service which have to be painted between voyages, since when out of commission ships are not earning any revenue. Zinc chromate primers, usually based on alkyd or phenolic media, dry more quickly than red lead in linseed oil; they are frequently used on the interiors of ships because they may be sprayed without any risk of lead poisoning and may be applied either to steel or to aluminium alloys. Leadbased priming paints should not be used on aluminium. Finishing paints are also similar to those used ashore. Good-quality alkyds are used in accommodation spaces, and the standard of workmanship is high. Colour and decorative schemes receive careful attention, and the finish is kept up to standard by frequent cleaning and regular repainting. For exterior use on topsides and superstructures, finishing paints based on alkyd media are generally used; good water resistance is essential here. White is used extensively on the superstructures of ships; owing to the pollution of many estuaries and docks with sewage and the consequent evolution of hydrogen sulphide in warm weather, it is necessary to make marine white paints ‘leadfree’ in order to avoid discoloration by sulphide staining. Another feature of modern marine white paints is that they are usually made from alkyds based on a ‘non-yellowing’ oil such as soya-bean oil in order to prevent the yellowing which occurs on exposure of linseed-oil-based white paints. The British Navy’s topsides grey paint consists of rutile-type titanium dioxide in an alkyd medium based on non-yellowing oil. Black topsides paint which is used on many merchant ships may be based on phenolic media or alkyds reinforced with phenolics. Newer types of high-performance paints’ used on ship exteriors include those based on epoxy resins, polyurethane resins, vinyl resins (also vinyl/ alkyd or vinyl/acrylic blends) or chlorinated rubber. Epoxies and polyurethanes are chemically-curing types and present curing problems at low temperatures, whilst the overcoating intervals are critical for best adhesion
PAINT FINISHES FOR MARINE APPLICATION
14 :79
between coats. Chlorinated rubber’ does not suffer from these practical difficulties and is becoming widely used. A complete system based on one of these special coatings must normally be applied, and first class surface preparation is essential if the optimum performance is to be obtained from them. Simpler types of oil-based paints are generally less sensitive to the standard of surface preparation and may give better results than these special paints when imperfect surface preparation must be tolerated. Interior Surfaces
Aluminium finishing paints are frequently used for the interior of dry-cargo holds because they help to improve lighting. Aluminium paint is also used in engine rooms; the general requirement here is for hard-drying paints resistant to oils and to heat. Cargo and Ballast Tanks
Severe corrosion may occur in unprotected cargo and ballast tanks of oil tankers loas a result of the combined corrosive effects of the cargoes, fresh or salt-water ballast, and tank washing by cold or hot sea-water. Ships which carry cargoes of refined oil products (‘white oils’) suffer general corrosion, since these cargoes d o not leave any oily film on the interior surfaces of the tanks. Corrosion rates vary widely according to the conditions of service, rates of up to about 0.4 mm/y being reported. Cargoes of crude oil (‘black oil’) leave an oily or waxy film on tank interiors, and this has some protective action. As this film is not continuous over the whole surface, severe local corrosion may occur at areas of bare steel exposed to the action of sea-water ballast. The mechanism of the attack at these bare areas may be likened to that on small bare areas on steel which is a h o s t completely covered with millscale; the oil or wax-covered areas function as cathodes in the same way as millscale, and corrosion is concentrated on the anodic bare areas. Some crude oils contain appreciable quantities of sulphur compounds, and residues may react with water and oxygen to produce sulphuric acid. The attack in black-oil tanks therefore takes the form of pitting; rates vary widely, up to as much as 5 mm/y being known, depending upon the conditions of service. Corrosion in oil tankers is therefore a serious problem entailing costly steel renewals in unprotected tanks. Protective measures include (a) the use of cathodic protection, (b) oxygen elimination by the injection of inert gases, (c) dehumidification of the air above oil cargoes or in tanks when empty, (d) the addition of inhibitors to the oil cargoes or to the ballast water, or the spraying of inhibitors on to the interiors of tanks, or (e) protective coatings. Methods (a)-@) reduce the corrosion, but only (e) offers the prospect of complete protection. The coatings must have good resistance to many types of petroleum or other liquid-chemical cargoes, to ballast water and to normal tank cleaning, must not contaminate cargoes, and must be capable of being applied under shipyard conditions. Two main types of paint coating have been developed for this service, viz. epoxies and zinc silicates. Exoxy resin paints are supplied as two components, a base and hardener, to be mixed at the time of application. Curing of the film to a tough, oil-,
14:80
PAINT FINISHES FOR MARINE APPLICATION
chemical- and water-resistant state occurs by chemical reaction between the epoxy resin of the base component and a curing agent (amine or polyamide) forming the hardener. This reaction does not require the access of oxygen, so that the film cures right through, irrespective of thickness, It is, however, dependent on temperature, 10°C being the usual minimum practical recommendation. To ensure good intercoat adhesion, successive coats must be applied before the previous coat has fully cured, so that in practice there are maximum as well as minimum over-coating intervals, both varying with temperature. The early epoxy tank systems required application of four or even five coats to give a total dry film thickness of 200-250pm, but common practice now is to apply two high-build coats to achieve the same film thickness. Solventless types are also available which may be applied as single coats of 200-300 pm. Coatings based on epoxy resins modified with coal tar pitch may be used in tanks for the carriage of crude oils, but are not suitable for refined oils because the pitch would contaminate the cargoes. Zinc silicate tank coatings show good resistance to petroleum cargoes and many organic solvents, although their resistance to acids and alkalis is inferior to that of epoxies. The paints are supplied as two components, zinc dust being stirred into a silicate solution at the time of use; reactions take place during drying, the dry film consisting essentially of metallic zinc and silicic acid, together with zincates. Single coats with a thickness of 80-100 pm are normally applied. The choice of tank coating" depends upon the cargoes to be carried, and must be determined by the ship operator with the advice of paint manufacturers. The application of epoxy or zinc silicate tank coatings demands special techniques to ensure control of surface preparation, ventilation, over-coating intervals, curing times and temperatures if satisfactory service is to be obtained, and much of the work is undertaken by contractors with the necessary knowledge and equipment. When properly applied, tank coatings not only prevent corrosion of the tanks for up to 8-10 years, but also render tank cleaning easier and quicker since cargo residues are not retained by corrosion products on the interior steel surfaces. Ships' Bottoms
Paints used for protecting the bottoms of ships encounter conditions not met by structural steelwork. The corrosion of steel immersed in sea-water with an ample supply of dissolved oxygen proceeds by an electrochemical mechanism whereby excess hydroxyl ions are formed at the cathodic areas. Consequently, paints for use on steel immersed in sea-water (pH = 8.0-8-2) must resist alkaline conditions, Le. media such as linseed oil which are readily saponified must not be used. In addition, the paint films should have a high electrical resistance I 2 to impede the flow of corrosion currents between the metal and the water. Paints used on structural steelwork ashore do not meet these requirements. I t should be particularly noted that the well-known structural steel priming paint, i.e. red lead in linseed oil, is not suitable for use on ships' bottoms13. Conventional protective paints are based on
phenolic media, pitches and bitumens, but in recent years high performance paints based on the newer types of non-saponifiable resins such as epoxies,
14:81
PAINT FlNlSHES FOR MARlNE APPLlCATlON
coal tar epoxies, chlorinated rubber and vinyls have become widely used. With conventional paint systems the usual interval between drydockings is about 9 to 12 months, but with a high performance system used in conjunction with impressed-current cathodic protection, Lloyds Register and other Classification Societies permit this interval to be extended to 2f years. Antifouling compositions The finishing paints on ships' bottoms are required to prevent attachment of marine growths. These paints, known as antifouling cornpo~itions'~~'~, contain chemicals poisonous to the settling stages of marine plants and animals. The poisons are slowly released into the sea-water, maintaining a thin layer of water next to the surface of the paint in which the spores and larvae cannot survive; settlement and further growth are thereby prevented. The most widely used poison is cuprous oxide but its action, particularly against some types of plant growths, may be reinforced by other poisons, e.g. compounds of mercury, arsenic, tin, lead or zinc, The arsenic, tin and lead poisons are organometallic compounds. In addition, many hundreds of purely organic compounds have been examined as possible antifouling poisons, but none has yet proved so non-selectively effective against a wide range of organisms as the metallic poisons mentioned. It will be realised that antifouling compositions must have a limited effective life, because when the bulk of the poison in the film has been released, the poison release rate falls below that necessary to prevent attachment of marine organisms. On merchant ships the compositions are generally effective for about 9 to 15 months, but special long life types are effective for 2+-3 years. Details of typical marine painting systems are set out in Table 14.5. Table 14.5 Typical marine painting systems Type of paint 1. SHIP'S BOTTOM
Method of application
coats
Dry film thickness (pm)
SYSTEMS
(a) Conventional bituminous system Bitumen or pitch solution pigmented with aluminium flake Antifouling composition
Airless spray, brush or roller Airless spray, brush or roller
( b ) Conventional non-bituminous system Tung oil/phenolic medium Airless spray, pigmented with basic lead brush or roller sulphate, aluminium flake and extenders Antifouling composition Airless spray brush or roller ( c ) High performance epoxy system Coal tar epoxy (2-pack) Airless spray Antifouling composition Airless spray brush or roller ( d ) High performance chlorinated rubber system Chlorinated rubber primer Airless spray, brush or roller High build chlorinated rubber Airless spray Antifouling composition Airless spray or chlorinated rubber based brush
2-3 1
2-3
150-200 50-80
150-200
1
50-80
2 1
200-300 80- loo
1
50
2
175-225 80- 100
1
14:82
PAINT FINISHES FOR MARINE APPLICATION Table 14.5
Type of paint
(continued)
Method of application
2. TOPSIDES AND SUPERSTRUCTURE SYSTEMS (a) Conventional system Red lead primer in quick-drying Airless spray, alkyd or phenolic medium brush or roller Airless spray, Gloss finish, alkyd medium pigmented with rutile titanium brush or roller dioxide (white) and tinting pigments as required ( b ) High performance epoxy system High build epoxy (2-pack) Airless spray Gloss finish, epoxy or Airless spray or polyurethane (2-pack) brush (c) High performance chlorinated rubber system Chlorinated rubber primer Airless spray, brush o r roller High build chlorinated rubber Airless spray Gloss finish, chlorinated rubber Airless spray, or brush
3. INTERIOR ACCOMMODATION SYSTEMS (a) Conventionai system Zinc phosphate primer in quickAirless spray, brush or roller drying alkyd or phenolic medium Semi-gloss undercoat, alkyd Airless spray, medium pigmented with titanium brush or roller dioxide and tinting pigments Gloss finish, alkyd medium Airless spray, pigmented with titanium brush or roller dioxide and tinting pigments ( b ) High performance system Epoxy primer (2-pack) Airless spray, brush or roller Gloss finish, epoxy or Airless spray or polyurethane (2-pack) brush
4. DRY CARGO HOLD SYSTEM Zinc chromate primer in quickdrying alkyd or phenolic medium Bright aluminium finish, leafing aluminium flake in oleoresinous medium
Airless spray, brush or roller Airless spray, brush or roller
5. SYSTEMS FOR CARGO/BALLAST TANKS (a) Crude oil carriers Airless spray Coal tar epoxy (2-pack) ( b ) Refined oil and chemical carriers Airless spray High build epoxy (2-pack)
coals
Dry film thickness (pm)
2
100-125
2
50-80
2 1
200-250 40-60
1
50
1 1
80- 120 50
2
80- IO0
1
40-60
1
40-60
2
100-120
1
40-60
2
80- IO0
2
50-80
2
250-300
2
250-300
Nole: The above systems are lor application to steed blast-cleaned to a ‘near-white’ finish (Second Quality of BS 42321967) and immediately shop-primed before fabrication. The shop primer must be thoroughly cleaned and degreased at the time o f painting.
PAINT FINISHES FOR MARINE APPLICATION
14:83
Methods of Application (Section 14.11 The paints used on ships may be applied by brush, roller or spray-airless spraying in particular being widely used when large areas are to be coated. High performance coatings are formulated to permit application of the full system in only a few coats, i.e. the paints must be capable of airless spray application at wet film thicknesses of 200-500pm without sagging or running on vertical surfaces, to give dry film thicknesses of 100-300 pm per coat. Time in drydock is generally restricted owing to high costs - figures of E20 OOO-€40 OOO per day being quoted for a 20 oo00 t tanker - so ships’ paints must dry rapidly and must tolerate application under non-ideal weather conditions since owners are unwilling to incur extra costs from delays in painting. Possible health hazards, particularly when spraying some types of antifouling compositions, must be guarded against by wearing protective masks and equipment.
Economics In the painting of the general interior spaces and the exterior surfaces of ships above the waterline, protective and decorative aspects cannot be separated. Thus, on passenger liners the frequency of repainting the accommodation, superstructure and topsides is determined primarily by the decorative appearance, while on cargo ships this is usually less important than protection. For ships’ bottoms the maintenance of a smooth surface free from marine fouling growths is important because a rough or fouled bottom leads to reduced speed and/or increased fuel consumption. Fouling may easily cause a 50% increase in fuel consumption, involving an appreciable increase in running costs. For this reason the intervals at which ships’ bottoms are repainted depend on the efficiency of the antifouling compositions and on the degree of fouling encountered in service, marine growth being more vigorous in warm tropical seas than in temperate or polar waters. The cargo tanks of oil tankers present a special case, because of the high cost of steel renewals in unprotected tanks. For a 30 OOO t tanker, costs in the region of f5OOOOO for the initial painting of the tanks have been quoted; if the life of the paint system is 6-8 years, the tota1 cost over the normal 20-year life of a tanker is expected to be appreciably less than the sum otherwise spent on steel renewals, which may amount to several hundred thousand pounds.
Types of Failure (Section 14.41 Paints correctly applied to well-prepared surfaces on the above-water part of ships will normally fail first by chalking, with checking and crazing of the finishing paint following. Of the high performance systems, polyurethanes have better gloss retention than epoxies or chlorinated rubbers. In spite of a general improvement in conditions of application during recent years, however, ships’ paints are still liable to be applied to damp or otherwise imperfectly prepared surfaces, and this leads to failure by adhesion
14 :84
PAINT FINISHES FOR MARINE APPLICATION
breakdown and rust formation beneath the paint film. Intercoat adhesion failure is also likely with epoxy systems if recommended intervals between coats are exceeded. On ships’ bottoms the antifouling coat fails when its poison release rate (or leaching rate) falls below the value needed to prevent attachment and growth of marine fouling organisms. At this stage it becomes necessary to drydock the ship, clean the bottom and re-apply antifouling composition; the underlying protective paint system should normally only need renewal after about four or more years, depending on whether a conventional or a high performance system is used. For economic reasons (docking charges, interest, insurance, loss of earnings, etc.) no delay can be accepted in the repainting of ships’ bottoms, so painting sometimes proceeds under adverse weather conditions to a poorly prepared surface - in consequence failure may occur from loss of adhesion. Paints capable of application to damp surfaces are being developed to overcome this difficulty. It may also be mentioned that promising results have been obtained by cleaning and recoating ships’ bottoms under water, and this could eventually eliminate drydocking of ships for repaintingI6*’’.
Recent Developments During the years since the publication of the second edition there have not been any really fundamental changes in the materials and methods of painting ships, although there have been changes to meet differing application and health requirements and to take advantage of technical developments. Improved quality control has also led to better corrosion protection. The British Ship Research Association’* and the Dutch Paint Research Institute TNO19 have published ship painting manuals; reviews of marine paint technology have been published by De la Court and de Vries”, Phillip” and 23. BanfieldzZs In this section changes are described under the original headings, but some of the figures in Table 14.5 and in the subsections ‘Methods of Application’ and ‘Economics’ have also been updated. Surface Preparation and Pretreatment
Blast-cleaning in impeller-type machines is now almost universally used for the initial surface preparation of ships’ platez4, earlier methods by weathering, scraping and wirebrushing or by acid pickling being practically unknown in modern shipyards. The design and performance of the machines have been improved. More attention is given to the selection of suitable grades of abrasive, its recovery and grading before reuse to ensure that the most suitable balance of coarse, medium and fine particles is actually used. In addition to surface cleanliness the surfaces profile of the blast-cleaned surface is now frequently specified- this has a considerable bearing on the adhesion and performance of priming paints. The prefabrication primers previously described are still current, the phenolic-reinforced wash primers being most widely used for general ship
PAINT FINISHES FOR MARINE APPLICATION
14:85
construction. For cargo tanks designed to carry chemicals or solvents it is preferable to apply the epoxy tank coating direct to a freshly blast-cleaned surface because small amounts of some cargoes can become absorbed into the coating and soften a polyvinyl butyral primer, leading to adhesion failure.
Exterior Surfaces above the Watedine
The use of oleoresinous paints has declined, being confined to smaller ships -practically all large ships use high performance coatings. Priming paints containing lead pigments are hardly ever used because of a greater awareness of possible health hazards. Similarly, the use of zinc chromate primers is declining because soluble chromates are believed to be carcinogenic; this has led to the increased use of zinc phosphate primers. As stated above, high performance coatings based on epoxies, vinyls or chlorinated rubbers are used almost exclusively on all large ships. A general development in these materials has been the introduction of highly thixotropic” types that can be airless sprayed at wet film thicknesses of 300pm or more, that do not run or sag on vertical surfaces. This enables the requisite film thickness to be applied in fewer coats, saving time and reducing application costs.
Cargo and Ballast Tanks
The zinc silicate, epoxy and coal tar/epoxy coatings are still used. Coal tar epoxies are used for crude oil tanks, sometimes on all the interior surfaces but more often for (a) the bottom of the tank and about 2 m up the sides, (b) the top of the tank and about 2 m down the sides, and (c) other horizontal surfaces where seawater ballast may lie. These partly coated tanks are frequently also fitted with cathodic protection to prevent corrosion of the uncoated areas when seawater ballast is carried. The pure epoxy or coal tar epoxy coatings applied in bulk cargo tanks used for the carriage of grain must be approved by the North of England Industrial Health Service, or by similar independent authorities in other countries. In the case of some tanks used to carry wine or chlorinated solvents the final coat applied over an epoxy coating is sometimes an oil-free polyurethane enamel because this paint resists chlorinated solvents better than do epoxies, does not taint wines and is not stained by red wines.
Ships ’ Bottoms
The conventional bituminous or oleoresinous paints previously described are still used on the bottoms of smaller ships, the chief difference being that they are applied mainly by airless spraying. The formulations may be adjusted to permit application of thicker coats than by brush or roller, although the coats must not be too thick because oleoresinous paints require
14: 86
PAINT FINISHES FOR MARINE APPLICATION
access of atmospheric oxygen to permit drying - very thick coats would take an impractically long time to become dry. The outer hulls of large ships are protected by one or other of the high performance systems previously described -epoxies, vinyls or chlorinated rubbers, often blended with coal tar. Here, too, as already described for surfaces above the waterline, highly thixotropic types have been introduced permitting the required film thickness to be applied in fewer coats, saving time and reducing application costs. These large vessels are almost all fitted with cathodic productionz6 using an impressed current system in which inert anodes (e.g. platinised titanium, lead alloy) fitted on the hull are energised by a low voltage d.c. generator. This causes the entire surface of the hull to become a cathode at which electrons are discharged; in the presence of an ample supply of oxygen the reaction is: 4e - + 2H,O
+ 0,
-
40H-
The high performance coatings mentioned are all non-saponifiable types, so resist the alkaline conditions on the hull. In the vicinity of the anodes the current density is inevitably higher than elsewhere on the hull and the rate of production of hydroxyl ions is correspondingly higher, Le. conditions become highly alkaline. This leads to the deposition of calcium and magnesium carbonates (‘cathodic chalk’) near the anodes. Another effect of the high current density is that dissolved oxygen in pores in the coating becomes exhausted and the cathodic reaction then becomes: 4e - + 4H,O
-
40H-
+ 2H,
with evolution of gaseous hydrogen. These two effects both tend to disrupt the coatings. They are minimised by (a) electronic control of the cathodic protection installation to ensure that the hull potential is no more than required for protection, and (b) surrounding the anodes with rubber mats or glass reinforced plastic shields. Antifouling Compositions
Until recent years these paints could be classifiedz7broadly into two groups. In soluble matrix antifouling paints the particles of poisonous pigments (chiefly cuprous oxide) are distributed throughout the film of a resin-based binder which dissolves slowly in seawater. Dissolution of the binder exposes the particles to the action of the seawater, thus maintaining a thin layer of water next to the hull which is poisonous to the spores and larvae of marine plants and animals. In contact antifouling paints the poison content is high enough to ensure that particles of poisonous pigment (chiefly cuprous oxide) are in contact throughout the film. As the particles near the surface dissolve other particles deeper in the film become exposed to the action of the seawater, thus maintaining a toxic layer of water next to the hull. In more recent years two new types of antifouling composition have been developed, using organometallic compounds as poisons. In one type2’, based chiefly on vinyl resin and organotin compounds (e.g. tributyltin fluoride), the poison and resin form a solid solution. As the poison dissolves from the surface of the film, more poison diffuses from deeper in the film to
PAINT FINISHES FOR MARINE APPLICATION
14 :87
maintain a uniform concentration throughout the film, i.e. the poison released to the seawater is replenished by diffusion from within the film. This mechanism hardly disturbs the surface of the paint which therefore retains its original smoothness. The other new type29is based on a toxic component combined with a binder resin, e.g. tributyltin acrylate may be copolymerised with an acrylic resin, producing a film-forming copolymer resin with a high content of tributyltin groups. When applied as a paint to a ship’s bottom the polymer is slowly hydrolysed and toxic tributyltin groups released into the seawater. The residue of the polymer is water soluble. In this way the surface of the film is slowly eroded and the action is claimed to maintain a smooth finish on ships’ bottoms. Since 1986, however, an account of ecological and pollution problems associated with organotin compounds, and allied Health and Safety Regulations, the use of these compounds in antifouling compositions has markedly declined 30. Reference is made in the foregoing paragraph to the smoothness of ships’ bottoms, and the importance of this factor has become increasingly realised in recent years. A rough surface, whether caused by attachment of fouling organisms, by corrosion or by poor paint application techniques, leads to an appreciable increase in the resistance to movement of a ship and hence to increased fuel consumption to maintain the service speed. The British Ship Research Association ’* has developed a gauge to measure hull roughness, and this is used to check that the surface of the underwater hull of new ships is satisfactorily smooth - similar measurements are made after cleaning and repainting in service. Methods of Application
Reference has already been made to the greatly increased use of airless spraying for applying paints to ships. On the largest vessels the use of brushes or rollers is impracticable: the area of the outer hull of a 300 OOO t tanker exceeds 30 OOO m ’. Thus, high-build coatings cannot satisfactorily be applied by brush or roller - eight or ten coats would be needed, requiring many painters and a long time. One airless spray gun, however, is capable, under practical conditions, of applying thick coats at up to 400 m2/h. Four or six guns, therefore, will apply one coat to the entire area in a few days and the complete paint system in under 2 weeks. Airless spraying produces less spray mist than conventional air-assisted spraying, but there is some risk of inhalation of spray droplets by painters or by others working in the vicinity. The danger may be avoided by wearing a filter type face mask. When applying3’ antifouling compositions suitable protective equipment must be worn because of the poisonous compounds they contain- this applies particularly to some of the newer types containing organometallic compounds but also to the older types containing cuprous oxide. T. A. BANFIELD
14:88
PAINT FINISHES FOR MARINE APPLICATION
REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
11. 12. 13. 14. 15.
16. 17. 18. 19.
20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.
31.
Ffield, P., Trans. Soc. Nav. Archit., N.Y.. 50, 608 (1950) Singleton, D. W. and Wilson, R. W., Br. Corros. J., Supplementary Issue, 12 (1968) Singleton, D. W., Iron and Sfeel, 41. 17 (1968) Bullett, T. R., Br. Corros. J., Supplementary Issue, 5 (1%8) Wilson, R. W. and Zonsveld, J. J., Trans. N.E. Csf. Insfn. Engrs. Shipb., 78, 277 (1962) Chandler, K. A. and Shak, B. J., Br. Corros. J., 1, 307 (1966) Banfield, T. A., Proc. Conf. Profn. Mer., London, 95 (1970) Banfield, T. A,, Fairplay International Shipping J., Anti-corrosion Survey, 233, 37 (1969) Banfield, T. A., Shipping. 59 No. 1, 29 (1970) Logan, A,, Trans. Insf. Mar. Engrs., 60, 153 (1958) Rogers, J., Trans. Insf. Mar. Engrs., 83, 139 (1971) Mayne, J. E. O., J . Oil Col. Chem. Ass., 40, 183 (1957) Dechaux, G., Peinf. Pigm. Vern., 17, 758 (1942) Banfield, T. A., Ind. Fin. and Surface Coafings, 22 No. 266, 4 (1970) Banfield, T. A., Oceanology Infernafional72, Conference, Brighton (1972) Rudman, J. A., ibid. Jones, D. F., ibid. Recommended Pracfice for fhe Profecfion and Painting of Ships. British Ship Research Association and Chamber of Shipping of the UK, Wallsend, Tyne and Wear (1973) Berendsen, A. M.,Ship Painting Manual, Paint Research Institute TNO, The Netherlands, (1975). Translated from the Dutch version Verffechnisch Handboek uoor de Scheepsbouw en de Scheepsvaarf (1974) De la Court, F. H. and De Vries, H. J., f r o g . Org. Cfgs., 1, 375 (1973) Phillip, A. T., f r o g . Org. Cfgs., 2, 159 (1974) Banfield, T. A., f r o g . Org. Cfgs., 7, 253 (1979) Banfield, T. A., J . Oil Col. Chem. Ass., 63, 53, 93 (1980) McKelvie, A. N., .I. Oil Col. Chem. Ass., 60, 227 (1977) Pila, S.,J. OilCol. Chem. Ass., 56, 195 (1973); Birkenhead, T. F., J . OilCol. Chem. Ass., 52, 383 ( I 969) Bingham, M. H., and M u m , P. W., J . Cfgs. Tech., 50, 47 (1978) Partington, A., Paint Tech., 28 No. 3, 23 (1964) (also ref. 23) Mearns, R. D., J. Oil Col. Chem. Ass., 56, 353 (1973) Christie, A. 0.. J. OilCol. Ass., 60,348 (1977); Atherton, D., Verborgt, J. and Winkeler, M.A., J . Cfgs. Tech., 51. 8s (1979) J . Oil Col. Chem. Ass., 73, 39 (1990) Research Organisation of Ships Compositions Manufacturers Ltd, London (1970) see ref. IS, p. 822
14.8 Protective Coatings for Underground Use Introduction The general conception of a paint is of a cold-applied material containing thinners which evaporate to leave a higher molecular-weight base protective, of 25-50pm thickness per coat. For buried or submerged structures, where maintenance is difficult or even impossible and a degree of physical protection is also necessary, such thin protective paint barriers between metal and the corrosive electrolyte environments of soil or water are usually quite inadequate. In relatively non-corrosive soil, thin bituminous coatings on thick cast iron may be satisfactory, but this is the exception rather than the rule. In dealing with underground structures, therefore, the thicker protectives needed are regarded as coatings rather than as paint finishes. The most usual forms of buried metal structures are pipelines, piles, tanks and power and telephone cables. Power cables must usually have some metal protection, covered by expensive continuous factory-applied sheathings of considerable thickness. Since water, gas and petroleum pipelines provide the greatest area of metal surfaces to be protected below ground, a detailed discussion of the protection given to them would appear to be the best means of dealing with coatings for underground use. Improvements are continually being made in the quality of coating materials and their application, but it is still difficult to produce at economic cost a permanent coating for a buried pipeline. The disruptive effects of handling, construction, penetration by rocks, soil stress, material ageing, etc. inevitably result in areas of bare metal being exposed t o corrosive soil electrolyte at isolated locations, with ultimate pitting or holing of the metal. The aim is to supply the best possible coating at economic cost and to provide for any initial or later failures by application of cathodic protection. The combination of coating with cathodic protection shows the greatest economic advantage. In pipelining, the trend is towards all-welded steel for long lines, and since the wall thickness is less than that of cast iron, protection is the more important. Many types of coating are used, from thick concrete to thin paint films, and each has its own particular suitability, but the majority of pipelines throughout the world today are coated with hot-applied coal tar or petroleum asphalt-base-filled pipeline enamels, into which reinforcing wraps, such as glass fibre are applied. 14:89
14:W
PROTECTIVE COATINGS FOR UNDERGROUND USE
The use of coatings applied in the form of tape is also increasing. Polyethylene and polyvinyl chloride films, either self adhesive or else supporting films of butyl adhesive, petrolatum or butyl mastic are in use as materials applied ‘cold’ at ambient temperatures. Woven glass fibre or nylon bandage is also used to support films of filled asphalt or coal tar and these are softened by propane gas torches and applied to the steel surface hot, cooling to form a thick conforming adherent layer. Recently, sheets of high density polyethylene extruded on to the pipe surface over an adhesive have become available and the use of polyethylene or epoxy powders sintered on to the steel surface is becoming more frequent. Some use has been made in the water industry of loose envelopes of polyethylene sheeting and with the increasing lengths of submarine pipeline requiring heavy concrete coatings for reducing buoyancy, the use of a heavily filled bituminous coating is projected. In the special case of pipelines operating at relatively high temperatures such as for the transmission of heavy fuel oil at up to 85”C, heat insulation and electrical insulation are provided by up to 50mm of foam-expanded polyurethane. As a further insurance against penetration of water, and to prevent mechanical damage, outer coatings of polyethylene (5 mm), butyl laminate tape (0.8 mm) or coal-tar enamel reinforced with glass fibre (2.5 mm) have been used.
Properties Required of Buried Coatings The aim in applying a coating to a buried metal such as a pipeline is to prevent electrical contact with an electrolyte such as soil and/or water. The characteristics required are as follows: 1. Ease of application. It must be possible to apply the coating in the factory or in the field at a reasonable rate and to handle the pipe reasonably quickly after the coating has been applied without damaging the coating. 2. Good adhesion to the metal. The coating must have an excellent bond to steel. Priming systems are frequently used to assist adhesion. 3. Resistance to impact. The coating must be able to resist impacts without cracking. 4. Flexibility. The coating must be flexible enough to withstand such deformation as occurs in bending, testing or laying, as well as any expansion or contraction due to changes in temperature. It must not develop cracks during cooling after application or curing. 5. Resistance to soil stress. The coatings are often subject to very high stresses, due, for instance, to the contraction of clay soil in dry weather, and they must be able to resist such stresses without damage. 6. Resistance toflow. The coating should show no tendency to flow from the pipe under prevailing climatic conditions. It must not melt or sag in the sun and it must have sufficient resistance not to be displaced from the underside of large-diameter pipes. 7. Water resistance. Coatings must show a negligible absorption of water and must be highly impermeable to water or water-vapour transmission.
PROTECTIVE COATINGS FOR UNDERGROUND USE
14: 91
8. High electrical resistance. The coating must be an electrical insulator and must not contain any conducting material. 9. Chemical and physical stability. The coating must not develop ageing effects, e.g. denaturing due to absorption of the lower-molecularweight constituents, or hardening with resultant cracking from any cause including oxidation. It should be stable at operating temperatures. 10. Resistance to bacteria. The coating must be resistant to the action of soil bacteria. 11. Resistance to marine organisms. In the case of submarine lines, the coating should not be easily penetrated by marine life, e.g. mussels, borers, barnacles, etc. These characteristics cover the general ideal for a pipeline coating, but obviously modified conditions may impose requirements which are more, or less stringent; this of course also applies to other types of buried structures.
Preparation of Metal Surface Before applying a protective coating it is essential to ensure that the surface is free from rust, millscale, moisture, loose dust, or any other incompatible material which might prevent the electrically non-conducting coating from bonding properly with the metal surface or which might produce defects in the continuous film. The following cleaning methods are available and each may have a particular advantage in given circumstances:
(a) Mechanical cleaning. Hand or mechanical wire brushing, impacting or abrading are methods suitable for hot applied coatings, for repairs to damaged areas or for relatively small or inaccessible areas. Visual standards to assess the degree of cleanliness are available but are not commonly used. ( b ) Blast cleaning. Air-blast or centrifugally-impacted sand, shot or grit are appropriate for thin-film multicoat systems or for continuous factory production. Several visual standards are available. The cost of attaining a very high standard of cleanliness is considerable, and careful consideration should always be given to specifying the correct level of blasting for the particular application. (c) Pickling. Dipping in inhibited hydrochloric or sulphuric acid is commonly used in factory production, particularly in conjunction with hot phosphoric acid dipping (Footner process). The considerable facilities necessary for this method limit its use to the larger steel producers. Published standards are available for the phosphate surface conversion coating process. ( d ) Flame cleaning. This is appropriate only for field repair work where a dry or warm surface can be obtained only by flame application and must be preceded usually by mechanical cleaning. (e) Pipeline travelling machine. For long runs on continuously-welded pipelines, a machine with rotary wire brushes and/or impact tools and cutting knives may be used to prepare the surface. These machines,
14:92
PROTECTIVE COATINGS FOR UNDERGROUND USE
which are self-propelled along the pipe itself, are commonly combined with drip or spray apparatus to apply the primer which is spread over the surface by rugs or brushes so that the prepared surface is immediately primed.
No matter which method of cleaning is adopted, it is desirable to apply the primer or coating immediately after the cleaning operation. The preparation of the metal surface to receive the protective coating is of prime importance since a coating which is not bonded to the metal surface can allow electrolytes to contact the metal, with resultant corrosion. If water films develop between the metal and the electrically non-conductive coating, cathodic protection becomes ineffective.
Coating Techniques (Section 14.11 Dbping, Spraying and Brushing
These methods are generally appropriate for either thin-film solvent-based paints or for coatings up to about 150 pm thickness. The techniques are more usually used for the priming layer of the coating systems. Factory
or Yard Application
Protective coatings applied at a factory have the advantage that the work can be carried out under strictly controlled conditions but suffer from the disadvantage that they may be damaged during transport to the site. Pipes are frequently shot-blasted or descaled by acid pickling, then phosphated, either sprayed with primer or dipped into a bath of hot asphalt to provide a thin prime coat. The dry primed pipes are then slowly rotated by a lathe head, while hot enamel, mastic or asphalt/micro-asbestos paste is applied from a hopper travelling alongside the pipe. A pipe coating approximately 5 mm thick is produced by use of a heated pallet attached to the hopper feed. Reinforced-glass wrapping materials may also be spirally wound on to the coating according to requirements. ‘Rolling Rig, ‘Fixed-head and ‘Rotating-head‘ Coating Machines
The coating equipment under this heading may be used in permanent factories, but is often set up at temporary coating yards close to the location where the pipes are to be laid. The coating produced is usually 2-3 mm per pass. Rolling rig machines The rolling rig machine rotates the cleaned and primed pipe on mechanically driven ‘dollies’, while a tank travelling alongside the pipe floods it with hot asphalt or coal-tar-base enamel. At the same time internal and external reinforcing wraps may be spirally wound into or on to the hot enamel.
PROTECTIVE COATINGS FOR UNDERGROUND USE
14:93
Fixed-head machines Fixed-head machines are fed with the cleaned and primed pipe, which mechanically rotates as it passes through the fixed coating head which floods the hot enamel on to the pipe. At the same time reinforcing wraps are pulled on to the rotating pipe. Rotating-head machines In rotating-head machines the coating head and wrapping spindles rotate as the pipe is fed through the machine. Pipeline Travelling Machines
In the case of long continuously-welded steel pipelines the above pipecoating methods present the disadvantage that the joints have to be coated in the field after welding. To overcome this difficulty equipment which travels along the welded pipeline has been developed. A mechanically propelled cleaning machine travels along welded lengths of the pipeline. The machine has counter-rotating cutting knives or brushes, and also applies by rotating swabs, a thin coating (cold application) of primer to the clean metal surface. When the primer is dry, a coating and wrapping machine travels along the pipeline. The wrapping materials usually consist of staple glass tissue, pulled halfway into the hot enamel, and an outer wrap of glass impregnated with coaltar or asphalt enamel to produce a coating of approximately 2-5mm as shown in Fig. 14.5. OUTER GLASS, ASBESTOS OR KRAFT PAPER WRAP
GLASS FlSRE TISSUE
d
Fig. 14.5 Type of coating produced by mechanical flood coat and wrap machine
These machines can coat and wrap up to 5 km of pipeline per day. After the coating has been checked for pin holes by a high-voltage rolling-spring electrode, the pipe may be lowered directly into the trench, so that undue handling is avoided. The line travelling machine is usually used with coal tar or asphalt-base pipeline enamels. Similar line travelling machines are in use for the cold application of tape coatings.
Types of Coating Materials (Section 14.2) Plasticised Coal Tar and Petroleum Asphalt Enamels
The majority of pipelines today are coated with hot-applied plasticised coal tar or petroleum asphalt enamels. Both coal-tar pitch and petroleum asphalt have been used as protectives with and without filling materials. When filled
14 :94
PROTECTIVE COATINGS FOR UNDERGROUND USE
they are termed enamels or mastics. The term bitumen or bituminous has always been loosely applied and it is preferable to specify petroleum asphalt base or coal-tar pitch base.
Straight and filled enamels Fillers are normally added up to a maximum of about 30% weight (calculated on the mixture) which is equivalent to about 15 to 20% by volume. A filled coal-tar pitch has a higher softening temperature (as shown by the ‘ring and ball’ test) than the unfilled material, which results in a reduced tendency to flow. This fact is important in tropical countries or if a pipe is to operate at a somewhat elevated’temperature. Resistance to impact and abrasion of a coating is improved by the filler. The viscosity of the pipe coating is also increased; this entails a higher application temperature (193-249°C). A satisfactory filler must have the following characteristics: 1. Low water absorption. In this respect certain fine clays are unsuitable. 2. Ability to be readily wetted by the enamel. 3. Finely-ground composition, particles preferably of laminar shape to prevent settling when the enamel is molten. 4. Relatively low specific gravity, so that there is the minimum tendency for the filler to settle-out in the melting kettle. In present-day practice the materials which are commonly used and which satisfy most closely these requirements are talc, pumice powder, microasbestos and slate powders. It must be appreciated that there is an optimum percentage of filler which imparts to a coating the required melting point and toughness; beyond this point application. becomes more difficult and watertightness may be impaired.
Petroleum asphalt or coal-tar pitch as coatings The question of whether coal-tar pitch or petroleum asphalt is the more suitable for the coating of underground pipelines has raised a good deal of controversy. Asphalt and pitch are both waterproof materials, and they resemble one another in physical type. In the right circumstances both can be very effective in preventing the access of water to buried or submerged steel surfaces. Petroleum asphalts are manufactured in two general types: (a)a straight residue from distillation, which can be of the hard, high-melting type, and (b) so-called ‘blown’ grades which are prepared by partially oxidising the asphalt base by blowing in air. The general difference between the two grades is that ‘blown’ asphalt has a higher softening point than straight asphalt of the same penetration (Le. hardness). In assessing a pipeline coating the softening point is of considerable importance, since it determines the tendency to flow, and a certain minimum softening point is therefore necessary. A ‘blown’ asphalt has the advantage over straight material of the same softening point in that it has a better resistance to impact, since it is of a more rubbery nature. For this reason most petroleum asphalt coatings are based on the ‘blown’ variety. So far as coal tar is concerned, it was formerly the custom to use the straight residual pitch, but nowadays shock resistance is improved by a SOcalled plasticising process.
PROTECTIVE COATINGS FOR UNDERGROUND USE
14 :95
The differencesbetween asphalt and coal tar in relation to their application as pipeline coatings require comment. 1. It is often claimed that a coal-tar-base coating absorbs less water than
an asphalt coating and there is evidence in practice to support this claim, but some asphalt enamels in practice have been as good as the best coal-tar enamels. 2. Coal-tar enamels are claimed to have better adherence than the asphaltic enamels to clean metal, probably because of the presence of polar compounds, but little difference can be noted in practice under proper pipelining conditions. 3. The asphaltic enamels are easier to apply since they do not produce so much obnoxious fume and are usually applied at slightly lower temperatures. The field performance of the asphalt-base pipeline enamels was, at one time, erratic, probably because the material had been drawn from varying sources, without a close specification being used. The plasticised coal-tarbase enamel to the American Water Works Association Specification C-203 thus gained some favour in major pipelining organisations. The AWWA C-203 Standard remains a widely used specification suitable not only for the materials, but also for their associated reinforcing wraps and application procedures. The standard has been regularly updated. Hotapplied asphaltic and coal-tar coatings with their priming systems are now well classified, described and specified in BS 4164:1967 (coal tar) and BS 4147:1967 (asphalt), but no guidance is given in these specifications to application procedures.
Reinforcing materials
Internal At one time open-weave hessian cloth was very largely used as an internal reinforcement material, but experience showed that this is subject to rotting in the soil. Even when the material appears to be covered with enamel, some of the fibres must protrude, and thus moisture is absorbed so that after a period of years the hessian is generally found to be in a waterlogged condition and forming food for bacteria. The type of material to be used depends very largely on whether coating is carried out mechanically or by hand. For hand application it is not possible to use comparatively fragile staple tissues made of glass or asbestos and it is necessary to use a strong open-mesh fabric, such as woven asbestos or woven glass. The woven wraps are a great deal more expensive than the staple tissues, which are mechanically applied. It is not economical to use expensive woven material for long lines, which can be, and normally are, coated by mechanical means. For such lines the most commonly used material nowadays is a glass-fibre tissue of a nominal 0 . 5 mm thickness, consisting of glass fibres bonded together with a phenolic resin or starch. The improvement in coating quality achieved by using the internal glass wrap is illustrated by the following results. The tensile strength of a 3 . 2 mm thickness of 104°C softening-point enamel, 300 mm x 300 mm is virtually nil. A piece of 300 mm x 300 mm glass tissue 0.5 mm in thickness will break at about 50 kg under steadily increasing tensile load, but if it is embedded
14 :96
PROTECTIVE COATINGS FOR UNDERGROUND USE
in 3 - 2 m m of the enamel a tensile strength of the order of 150 kg is obtainable. These wraps are now longitudinally reinforced to prevent tearing on line-travelling or other coating machines. Where the pipeline is expected to have to withstand unusual dimensional variation due, perhaps, to temperature changes or near yield point pressure testing, the use of a woven glass or nylon reinforcement in place of the glass tissue is said to increase the flexibility of the coating system considerably.
External wrap The purpose of an external overlapping wrap is to provide a shield against the penetration of the enamel by stones and to prevent the pulling of the enamel away from the pipe by soil stress. It also reduces flow of the enamel owing to the weight of the pipe, and damage to the coating caused by handling can be more easily observed. The properties required of an external wrap are as follows: (a) Compatibility of impregnant to bond with the enamel used. (b) Tensile strength to prevent breaking while wrapping. (c) Hardness to resist penetration. ( d ) Flexibility to allow wrapping without cracking. (e) Free rolling from the reel while wrapping. (f)Resistance to soil conditions and bacterial attack. (g) Non-absorption or low absorption of water. These properties apply to a reinforcing outer wrap such as coal tar or asphalt-impregnated glass or asbestos bonded lightly to the outside of the hot-applied enamel. For some conditions kraft paper is adequate to facilitate handling and reduce soil stress. Were it not for its screening effect on cathodic protection with a consequent decrease in the effectiveness of the latter, the external wrap could be loose around the coating. It has become conventional to have the external wrapping lightly bonded to the coating to prevent lamination and water entry.
Armour wrapping In rocky ground it has always been considered good practice to pad the trench for a buried pipeline with clean sand. This procedure can be very expensive if the sand has to be hauled long distances, and an armour wrap has been developed to supplement the normal outer wrap to meet such conditions. A typical wrap is supplied in sheets about 6 mm thick, consisting of a sandwich of mastic enamel between sheets of asbestos about 1 . 5 mm thick. It may be longitudinally indented to allow the material to be wrapped around the pipe and secured by steel ribbon straps. An objection to this form of wrap is that its mode of application renders it extremely difficult to obtain a good uniform bond between the wrap and the enamel. In view of this, water could become trapped under the armour wrap, and because of the non-conducting nature of the wrap itself the effective application of cathodic protection would be difficult. Cold-applied Tapes
Hot-applied coatings require special melting and handling equipment to be available at the construction site. Clearly, considerable economies are possible if this equipment can be dispensed with, particularly in remote areas
PROTECTIVE COATINGS FOR UNDERGROUND USE
14 :97
with difficult access. Thus, the availability of cold-applied tapes for use either at the joints between factory-coated pipes or continuously over the pipeline has led to the increased usage of this type of wrapping. The tapes are usually relatively thin (0.5 mm) and easily damaged. It is, therefore, essential to take elaborate precautions to provide physical protection to the tape once it has been applied both during construction and after burial. Good results have been obtained when the tape is applied by line travelling machine and without further handling, immediately lowered into a sand padded trench and covered over with fine sand before the trench is back filled. Initial effective electrical resistance of tapes, as evidenced by the cathodic protection current demand, has been outstanding. There have been reports of increasing current demand with time which indicate a need for investigation. The current demand increase has been found, on occasion, to be due to poor construction practice, but not all tapes are affected in this way. On large diameter pipes having a raised seam weld, difficulty is encountered in covering the weld ‘shadow’ effectively.
Petrolatum-type tapes Petrolatum has, like lanolin, long been recognised as a means of preventing corrosion. It is easily cold-applied and has a definite place in corrosion engineering, but it is not suitable for buried structures, unless it is screened from soil and water by a woven glass or nylon cloth or an impervious membrane such as P.V.C. The polythenes normally tend to swell in contact with it. Earlier petrolatum coatings were frequently applied with cellulosic backing material; there were several objections to this type of protection, e.g. attack by sulphate-reducing bacteria on the cellulose, absorption of the grease by dry bentonite-type clays, lack of physical strength against stones, and water absorption. Petrolatum-type tape coatings now incorporate inhibitors against bacteria and with their backing film have high electrical and water resistance and therefore find extensive applications in the UK. A great advantage of the petrolatum-type coatings is ease of application and conformability to irregular surfaces. Pressure-sensitive tapes Unlike the more recently developed petrolatum tapes which rely on both the petroleum and backing films, the pressuresensitive tapes offer protection which depends almost entirely upon the prevention of ingress of moisture to the metal surface by the tape itself. The tapes are cold-applied, either by hand or by mechanically-operated equipment moving along the cleaned pipeline. The tapes are usually produced from polythene or polyvinylchloride films of 25 pm to 0 . 5 mm in thickness and the inner surface is coated with an adhesive, frequently rubber-based. The adhesive is usually between 25 and 100 pm thick. Earlier tapes frequently suffered from the migration of plasticiser from the tape to the adhesive with the result that the tape became detached from the metal, to which the adhesive remained attached. This has now been overcome by using a barrier between the tape and adhesive which itself may contain inhibitors against soil bacteria. Spiral corrosion due to inadequate overlap has been detected with selfadhesive tapes, and a 25 mm (or preferably half-tape-width) overlap is to be advocated. Within normal limits, the thicker the adhesive the better.
14 :98
PROTECTIVE COATINGS FOR UNDERGROUND USE
The self-adhesive tape coatings are thin and the adhesive itself does not necessarily come into contact with the valleys in the cleaned metal surface. Under these circumstances, the transmission of water vapour through the film to the metal may be possible. Moisture-transmission characteristics and other properties of P.V.C. and polyethylene tapes, as given by major manufacturers, are provided in Table 14.6. Table 14.6
P.V.C. and polyethylene tapes
Physical property Material
Thickness of film plus adhesive (mm)
P.V.C. 0.229 Polyethylene 0.203
Tensile strength (kg/crn width)
Elongaiion at break
(a)
(%o)
10 IO
175 70
0.19 0.02
+ 0.025 + 0.100
Moisrure absorption
Moisturevapour transmission rate (g,m2 per 24 h 24.0 3-1
Dielectric srrength
(V) 10 OOO
I4 OOO
Table 14.6 is only indicative of general properties, and the latest developments of specific manufacturers of self-adhesive tapes may show advances on these. P.V.C. tends to be more conformable to irregularities than polyethylene. Both types have their right and proper application for buried structures.
Laminated tapes In more general use now than pressure sensitive tapes are tapes consisting of polyvinyl chloride or polyethylene films in conjunction with butyl rubber. These tapes are applied with an adhesive butyl rubber primer. Thicknesses of up to 0-75 mm are in use and loose protective outer wraps of P.V.C. or polyethylene sheet are commonly applied. Tape quality control is exercised with reference to ASTM standard test methods and may include water vapour transmission rate and elongation. Conventional holiday-detection is of little value in the field but great attention should be given to preventing damage to the applied tapes. Coal-tar Epoxy Coatings
The epoxy resins when mixed with the correct amine produce tough films which adhere closely to metal. The chemistry of these resins is considered in Sections 14.5 and 14.9. The thickness and water resistance of the normal air-cured film can now be much improved by the incorporation of suitable coal-tar pitch material. A typical coal tar/epoxy coating material would be constituted as follows: Epoxy resin Coal-tar pitch Filler Solvent
30 25
25 20
and to the above would be added the amine curing mix.
PROTECTIVE COATINGS FOR UNDERGROUND USE
14:99
The coating is of the two-pack type, consisting of resin plus curing hardener. In practice the resin and amine may be mixed together and used for application by brush or spray, or by mechanical means at ambient temperature. Sometimes the clean metal is heated, as are the coating components which are then sprayed separately on to the metal to reduce curing time. Little reaction occurs below 4°C. For pipeline coating the pipes can usually only be handled after a few hours, depending on the mix and temperature, but it takes anything from two to seven days before the best characteristics of the coating develop. Information to date indicates that the total thickness of the coating should not be less than 0.3 mm and this requires several applications. These coatings are very tough and closely adherent (one pipeline company states that they handle coal tar/epoxy-coated pipe like bare pipe, including bending in the field). The first coal tar/epoxy coatings came into use only in 1953, and although they seemed most promising they have been little used to date compared to other materials. This is undoubtedly due to their relatively slow setting and curing time. Polyethylene Sheet
The practice has been developed amongst some water undertakings to envelop uncoated spun iron pipes in 0.5 mm thick polyethylene sheet, the ends of which are tied down to the pipe with a substantial overlap by means of adhesive tape. This method has great advantages in cost and simplicity. No long term performance figures have been published but many have grave doubts about the effectiveness of this method since the possibility of aggressive soil water entering at perforations or through overlaps, appears to be very high. Foam Polyurethane
These materials have been finding extensive use on transmission pipelines supplying heated heavy fuel oils to power stations. To prevent damage to the 50 mm thick coating, a mechanically stronger outer wrap which can also prevent water ingress is usually necessary. In one method of production, the foam is manufactured inside a polythene tube over the steel tube. In other methods where the foam is produced by spraying on to the steel surface, conventional tape or enamel coatings have been used. Weight Coatings
For pipelines to be placed under water, it is necessary to provide negative buoyancy. This is commonly achieved by placing lightly reinforced concrete up to 150 mm thick over the 3-5 mm hot enamal coating on the steel. Joints at the welded tube ends have to be coated with a minimum of delay due to the high production rate required on the laying barge, and tapes have therefore found application at this point. Where submarine pipelines are ‘pulled’
14: 100
PROTECTIVE COATINGS FOR UNDERGROUND USE
into position off the land, joint repair is more commonly carried out by means of the same hot enamel used as the pipeline coating. For the final joint between towed ‘strings’ of up to 300 m, fast setting epoxies have been used. A composite asphaltic mastic filled with high-density aggregate is now available as a combined insulation and weight coating, and this could be the development area in this field.
Internal Pipeline Coatings In some instances it is necessary to coat pipelines internally, and materials widely used are red lead, hot-applied enamels, concrete and epoxy resins. Internal coatings are usually applied at the factory and no difficulty exists in field construction if flanges, screwed, or spigot and socket joints are used, nor is there any difficulty with welded pipes above, say, 750 mm diameter, where patching can be carried out on the joints from the inside. Repair of internal coating on smaller-bore welded pipes presents many problems, which have not yet been satisfactorily overcome for all conditions. Pipelines in the ground can be mortar lined in situ by the use of travelling devices. Epoxy resin paints for long welded pipelines already laid have been applied in situ by placing two plugs in the pipeline with the paint between them, and then forcing them to travel through the pipeline by the use of compressed air.
Recent Developments Recent trends in protective coatings used on buried pipelines have been away from reinforced hot applied coal tar and asphalt enamels and butyl rubber laminate tapes, particularly where applied ‘over-the-ditch’. The more recently developed coatings based on fusion bonded epoxies, extruded polyethylenes, liquid-applied epoxies and polyurethanes, require factory application where superior levels of pipe preparation and quality control of the application process can be achieved. The longest most successful track record is still claimed by reinforced hot enamels, with their performance beneath concrete weight-coating making them first choice for the majority of North Sea offshore pipelines installed. However, reduced use of coal-tar enamel coating particularly in continental Europe, has been brought about mainly by an increasing awareness of the health hazards involved in the application of the material. The application procedures, properties and uses of buried pipeline coating materials are compared in Table 14.7. Fusion Bonded Epoxy Powders
After their initial development in the USA, fusion bonded epoxy coatings (FBE) are now factory-applied worldwide. Their specification as the first choice alternative to enamel coatings is still contested, although important
Table 14.7 Comparison of buried pipeline coating materials
Coating type
Typical system thickness (mm)
Applicable slandards
Application procedures
Glass fibre reinforced enamels
BS 4147 BS 4164 BS 514 AWWA/ANSI C203
Hot applied in factory and in field by line travel
2.5-6
Asphalts prone to water absorption and root damage. Coal tar resistant to oil products and root damage. Long successful service record, particularly coal tars. Proven under concrete weight coatings.
Cold applied tapes
AWWA C209 ASTM D-1000
By hand o r machine, in factory or field
62 (single wrap)
Various tapes, allowing suitable choice for individual projects. Particularly useful for coating weld joints, bends, specials in the field. Compatible with all factory coatings.
Polyethylene loose sleeving
AWWA/ANSI ClOS BS 6076
By hand in the field
0.2-0.25
Very economical and lightweight. Most commonly utilised over zinc sprayed ductile iron pipes. Will not allow application of effective cathodic protection. May not arrest all corrosion.
Fusion bonded epoxy
AWWA/ANSI C213 BS 3900
Electrostatic spray in factory and for joints in field
0.3-0.65
Higher temperature limitations and superior soil stress resistance compared with enamels. Requires careful handling in the field. Quality of pipe steel important.
Extruded/sintered polyethylene
DIN 30670 DIN 30674
By extrusion or sintering in factory
1 8-3 ' 5
Rugged, heavy coating. Limited track record.
Various thermosetting and thermoplastic resins
BS 3900
Powder/liquid system in factory. Airless spray/trowel in field
G5
Superior chemical and abrasion resistance compared with enamels. Comparatively expensive. Simultaneous coating internally/externally possible. Various resins available to suit particular requirements.
Heat-shrink crosslinked polyethylene
DIN 30672
Flame or heat gun in field
1'25-2.25
Utilised for coating of field weld joints and repairs on extruded polyethylene coated pipes. Careful application required to achieve consistent bond.
Characteristics and uses
14: 102
PROTECTIVE COATINGS FOR UNDERGROUND USE
improvements have been made in present powder systems over those first developed. The thermosetting powders are applied to a white metal blast-cleaned surface by electrostatic spray. Pipe is preheated to approximately 230°C, the quantity of residual heat being directly correlated to the maximum thickness of coating which may be achieved. On application, the powder melts, flows and cures to produce thicknesses in the range 250-650 pm and is then forced cooled by water quenching. Specifications normally place restrictions on pipe bending with thicknesses greater than 450 pm, but nearer maximum thicknesses are required where concrete weight coating is to be applied by impact methods. Strict control of the fusion process is imperative. In addition to thickness, hardness, continuity and adhesion checks, correct cure may be assessed by differential scanning calorimetry techniques, which are designed to measure any difference in the glass transition temperature of a laboratory-cured powder and the cured coating taken from the factory-coated pipe. Although in the UK, FBE powders have been chosen in preference to coaltar enamel coatings where stability at higher temperatures or resistance to soil stress situations has been required, doubts still exist over the powders long-term water absorption characteristics and resistance t o cathodic disbondment under high negative cathodic protection potentials. Many of these doubts are being overcome by the inclusion of a precoating chromate conversion treatment provided to the pipe immediately after normal surface preparation. This process has brought about significant improvement to FBE-coated pipe under test for cathodic disbondment and hot water immersion resistance. When applying epoxy powders special consideration must be given to the quality of the pipe steel. This factor has not posed problems to the heavier enamel coatings. However, due to the comparative thinness of the FBE coating, it is necessary to inspect the metal surface after blast cleaning and vigorously remove all slivers, scabs, gouges and similar defects by grinding to avoid consequential defects in the finished coating. FBE-coated pipe requires careful handling from factory to the pipe trench to avoid mechanical damage. Repairs are undertaken with either trowel or brush-applied, liquid two-pack epoxy resin-based paints or by melt sticks of compressed powder. Weld joints may be coated in the field with FBE powder, utilising a portable blast cleaning/induction heating and powder application system. Alternatively joints may be provided with self-adhesive laminate tapes or heat-shrink crosslinked polyethylene sleeves.
Polyethylene Resins
Polyethylene coating on ferrous pipes may be applied by means of one of the following processes: circular or ring-type head extrusion, side extrusion and wrapping or powder sintering. The commercially available coating systems also differ further in that the extruded polyethylene may be applied in conjunction with various primer/adhesive systems.
PROTECTIVE COATINGS FOR UNDERGROUND USE
14: 103
Generally, systems developed in the USA favour a combination of polyethylene with either butyl-rubber or hot-applied mastic adhesives, the latter consisting of a blend of rubber, asphalt and high molecular weight resins. In European and Far East coating plants, epoxy type primers and ‘hard’ ethylene copolymer adhesives have been successfully employed. The specification of these later coatings is covered by the German DIN 30670 standard for steel tubes and DIN 30674 for ductile iron pipes. These standards note that some 1 mm thickness of polyethylene is required for corrosion protection alone, but to improve the mechanical load-bearing capacity of the coating, total thicknesses of 1.8-3.0 mm, depending on pipe diameter, are to be specified. Repairs to the coating are made with either hot-melt polyethylene sticks or polyethylene sheet patches with mastic profiling compounds for small damaged areas. Large repair areas are best treated as for field weld joint coating, where either heat-shrink crosslinked polyethyiene sleeves or coldapplied self-adhesive laminate tapes are employed. Cold Applied Tapes
In addition to the petrolatum tapes and those based on a laminate of p.e. or P.V.C. with an elastomeric sealant or pressure-sensitive adhesive layer, recent developments have centred around self-adhesive bituminous Iaminates. These tapes are commonly constructed with a P.V.C. backing, whose thickness ranges from 0.08t o o . 75 mm and a bituminous adhesive compound layer to provide a total tape thickness of up to 2 mm. In order to maintain conformability without compromising impact values, tapes may also be manufactured with a fabric reinforcement within the bituminous layer. Being self-adhesive these tapes are produced on the roll with a protective paper interleaf designed to be removed as the tape is applied. Application may be by hand or by specially designed hand operated pipe wrapping machines which will accommodate the interleaf. Application is normally undertaken at either 25 mm or 55% overlap depending on the total coating thickness required. Most importantly, bonding at the overlaps will be achieved, as compared with tapes employing elastomeric sealant layers, where contact with fresh primer is required to activate adhesion. These bituminous tapes are compatible with all factory-applied coatings and thus are particularly employed for weld joint wrapping in the field. Tapes are produced in both temperate and tropical grades and heavy duty versions can be supplied for application under hot mastic asphalts at field joints of concrete weight-coated pipelines. Wrapping of complex shapes may be achieved by first profiling with a bituminous filler compound. Other Systems
Thermosetting epoxy and polyurethane chemically-cured liquid resins can provide, among other characteristics, superior abrasion resistance coatings. Solvent-free formulation applied by ‘hot’ spray techniques can achieve film thicknesses of up to 5 rnrn.
14: 104
PROTECTlVE COATINGS FOR UNDERGROUND USE
A typical application of these coatings is the use on carrier pipes installed by thrust boring techniques at major road, rail and river crossings. Sprayed polyurethane coatings of 900 pm thickness, are also commercially available on ductile iron pipes. Thermoplastic resins, such as vinyl chlorides, vinyl acetates and polyamides are employed, particularly in the water industry, on buried pipes and fittings. To provide both internal and external coating, application may be by one of these principle techniques: dipping in a plastisol, fluidised beds or electrostatic spray.
M. D. ALLEN D. A. LEWIS
BIBLIOGRAPHY Bigos, J., Steel Structures Painting Manual, Steel Structures Painting Council (1954) Coal Tar Based Hot Applied Coatings, BS 4164:1%7 Hall, R. E., Scott, F. S. and Weir, C. J., Materials Protection, 6 No. 8, 35 (1967) Hot Applied Bitumen Based Coatings, BS 4147:1%7 Peabody, A. W. and Woody, C. L., Corrosion, 5, 369 (1949) Romanoff, M., NBS Circular 579 (1957) Spencer, K. A. and Footner, H. B., Chem. Ind., Lond., 19 (1953) Sparrow, L. R., Petroleum, Lond., 21, 351 (1958) Shideler, N. T. and Whittier, F. C., Pipeline Ind., 6, May (1958) Shideler. N. T., Corrosion Technol., 17, 52 (1960) Hoiberg, A. J., (ed.), Asphalts, Tars and Pitches, Interscience Publishers (John Wiley) (1965) ANSI/AWWA C213-19, Standard f o r Fusion Bonded Epoxy Coatings for the Interior and Exterior of Steel Water Pipelines Omori, K., Watanabe, U. and Takeda, T., Improvement of Fusion Bonded Epoxy Coating, 5th International Conference on the Internal and External Protection of Pipes, Innsbruck Austria, pp. 61-19, BHRA, London (1983) NACE Recommended Practice RP-02-75, Application of Organic Coatings to the External Surface of Steel Pipe f o r Underground Service NACE Recommended Practice RP-08-85, Extruded Polyolefin Resin Coating Systems /or Underground or Submerged Pipe Schmitz-Pranghe. N. and von Baeckmann, W., Polyethylene-Exrrusion-Coating of Buried Steel Pipe: Properties, Experiences, Valuation, Corrosion 1977, NACE, San Francisco DIN 30674, Coating of Ductile Cast Iron Pipes-Polyethylene Coating DIN 30670, Polyethylene Sheathing of Steel Tubes and of Steel Shapes and Fittings ANSVAWWA C209-76, Cold Applied Tape Coatingsf o r Special Sections, Connections and Fittings f o r Steel Water Pipelines
14.9 Synthetic Resins
The term ‘synthetic resin’ was coined originally to distinguish these resins from natural resins such as rosin, shellac and the copals. Nowadays nearly all resins used in paint are synthetic, so the first term is often dropped. There is not enough space here to give a detailed classification, but only to delineate the major families from which resins for industrial coatings may be selected. Resins may be divided into two groups according to their modes of film formation which may or may not involve a chemical reaction. In the first, the components must react together to form a crosslinked structure which may require heat, radiation or catalysis to effect the reaction. The bulk of resins used in industrial finishes are of this type. They are commonly referred to as chemically convertible or, simply, convertible. In the second, the components are already of a large size and will form a film by a felting process. Here film formation depends on some physical change such as the loss of solvent by evaporation or heating, or the fusion of a dispersion. Cellulose nitrate is the classic example of a non-convertible resin and still is used extensively because of its unparalleled speed of drying. However, it has a number of disadvantages, being very highly flammable and prone to yellowing. Where better film properties are required, the thermoplastic acrylic resins will give excellent heat and light resistance.
CH3 -CH,-C-CH,-
i
I
COOR
CH3
I
C-CHZ-C-
I
COOR
CH3
I I COOR
Section of a thermoplastic acrylic resin. R represents an alkyl group. Note that the backbone is a chain of carbon atoms which is very resistant t o all forms of attack. The side groups determine properties such as solubllity, transparency and chemical resistance.
14: 105
14: 106
SYNTHETIC RESINS
Halogenated resins such as PVC and especially fluorinated resins such as polyvinylidene fluoride show a greater chemical resistance than any other type of resin.
Section of polyvinylchloride resin
Common to all non-convertible, resins is their very low solid content in solution, typically 10-20%, necessitating the application of a number of coats to give an acceptable film thickness unless one is able to use a dispersion rather than a solution of the resin. The rest of this section will be devoted to the chemically convertible resins. The variety of chemical types exploited in these resins are legion, so only the most widely used will be mentioned here. In some cases a single resin may be employed to produce a coating, but generally blends are used so enabling the film properties to be controlled by ratios of components as well as by choice of the components themselves.
Alkyd Resins and Polyesters These comprise a large group because almost any acid can be reacted with almost any alcohol to produce an ester which might be suitable as a coating resin. The distinction between an alkyd and a polyester is that the former contains monobasic acids usually derived from vegetable oils such as linseed, soyabean or coconut while the latter do not.
I C17H3Z
I C17H32
I C17H32
Section of an alkyd resin
The typical alkyd resin (see above) is comprised of three basic components: an aromatic diacid such as phthalic anhydride which together with a polyol such as glycerol, forms the backbone of the resin molecule and along which are distributed the fatty acids derived from vegetable oils. The solubility, film hardness and colour of alkyd resins depend on the nature of the modifying fatty acid which in most cases contributes some colour to the film. Today the user industries demand absolute colour stability which has been obtained by developing the so-called oil-free alkyds, also called polyesters
14: 107
SYNTHETIC RESINS
which have excellent colour stability. These are based on mixtures of diacids such as phthalic anhydride and aliphatic diacids such as adipic acid (which promotes extensibility) and a heat-stable polyhydric alcohol such as trimethylolpropane. Structures of these components are shown below. CH3
I
CHZ
I I
HOCH2-C-CH20H
co
CH2OH Phthalic an hydride
Trimethylolpropane
Adipic acid
There are basically two types of polyesters depending on the ratio of acids to polyols used in their preparation, as they may have a predominance of hydroxyl groups or of acid groups. These groups are the sites for crosslinking reactions, for example with formaldehyde resins or reactive isocyanates in the case of the hydroxyl groups or with solid epoxy resins in the case of the acid groups. The latter reaction is exploited in one type of powder coating.
Hydroxyl type polyester resin
CH3
CH
I
0
I
0
II
I1
HOOC (CH2),-C-O-CH2-C-CH-OC
CH
I
I
CH,
I
I
OH
0
Acid type polyester resin
0
II
14: 108
SYNTHETIC RESINS
The resin structures so far depicted represent the basic features of the alkyd and polyester molecules, but other components can be incorporated to enhance one or more film properties as required. One of the most widely used modification is that of vinylation. This is the free radical copolymerisation of unsaturated monomers during the manufacturing stage of the alkyd which must contain a proportion at least of unsaturated fatty acids preferably conjugated as in dehydrated castor oil. The two monomers most used are styrene and methyl methacrylate and the final product may contain up to 35% of combined monomer. This gives alkyds that are faster drying and paler having greater chemical resistance, but having less solvent resistance and outdoor durability than the unmodified alkyds. Saturated polyesters and saturated alkyds cannot undergo such modification with vinyl monomers but can be modified with other polymers such as silicone resins by alcoholysis. Here outdoor durability is considerably improved. A further type of ester resin is the unsaturated polyester where the unsaturation is built into the backbone by the use of maleic anhydride:
II C-OCH,-CH-OCHC=CHCO-CH,-CH-OCHC=CHCO-CH~-CH-OH II iH3i II i II
0 HO-c
0
0
7H3
0
CH3
I
Unsaturated polyester resin
This is a linear polyester containing phthalic anhydride to ensure hydrocarbon solubility and maleic anhydride to enable copolymerisation to take place, esterified with 2-propanediol. The ester is dissolved in styrene which initially acts as the solvent and subsequently as film former when it is copolymerised with the double bond in the ester by free radical induced polymerisation. Unsaturated polyester finishes of this type do not need to be stoved to effect crosslinking, but will cure at room temperature once a suitable peroxide initiator cobalt salt activator are added. The system then has a finite pot life and needs to be applied soon after mixing. Such a system is an example of a two-pack system. That is the finish is supplied in two packages to be mixed shortly before use, with obvious limitations. However, polymerisation can also be induced by ultra violet radiation or electron beam exposure when polymerisation occurs almost instantaneously, These techniques are used widely in packaging, particularly cans, for which many other unsaturated polymers, such as unsaturated acrylic resins have been devised.
Formaldehyde Resins These resins are prepared by an addition reaction of formaldehyde with either phenols, urea or melamine to prepare an intermediate such as the following:
14: 109
SYNTHETIC RESINS
OH
OH
OH
CH,OH Phenol formaldehyde intermediate
These intermediates are too small to be used alone, but need to be enlarged and modified to obtain compatibility with other resins. In the case of the phenol formaldehyde resins this is achieved by either using pura-substituted phenols where the substituent contains at least four carbon atoms or by reacting the intermediate with the natural resin, rosin, and then esterifying with glycerol or pentaerythritol. These resins have a limited use in stoved epoxy finishes where colour is not an important factor. In industrial finishes colour is very important and so a formaldehyde resin based on urea or melamine is usually chosen as both are virtually colourless. Here the intermediates are polymerised in the presence of an alcohol such as I-butanol which butylates some of the methylol groups. Few of these resins are capable of being used as such in surface coatings and are best considered as crosslinking agents for other resins such as stoving alkyds or thermosetting acrylics. Crosslinking occurs on stoving at about 120°C as follows: -__-- - - - - - - -Falkyd-,OH
C4H9 jOCH2
L
_ _ _ _ _ _ _ _ _J I
N-CH2-N
-_____ kay[ldO -H
I
c=o
I
HbCHz-NH
CH,OL
I I c=o I
H L_----
O
G alkyd F ]
NH2
I-_---A
ldealised crosslinking reaction of UF resin with alkyd resin
The melamine resins have many more reactable groups and so less are needed for crosslinking (25% of total compared with 50% of total with UF resins), and have greater heat resistance than the urea resins because of the pseudo-aromatic nature of the six membered ring.
Possible structure of part of an MF resin
14: 110
SYNTHETIC RESINS
In recent years the realisation of the danger to health from the presence of unreacted formaldehyde monomer in the working environment has led to the development of resins having very low free formaldehyde content, less than 0.5% instead of the usual 2-3%. This produces resins that are safer and less unpleasant to work with, though the solvent blend itself, xylene and butanol, has a very pronounced odour.
Epoxy Resins
13
The bulk of epoxy resins are still those based on epichlorhydrin and dihydroxy-diphenylpropane and may be represented by the following structure: 0 CH2CHCH20 /\
IH
0 ~ o c H ~ " c H 2 0 ~ c ~ O C H 2 CA H C H 2
C
I
I CH3
CH3
n
Generalised structure of an epoxy resin
When n has a value of 0 or 1, the resins are viscous liquids and have a high epoxy group content, while when n is 2 or greater, the resins are pale solids. These provide a range of highly chemically resistant coatings according to whether they are stoved, as when crosslinked with a formaldehyde resin in solution coatings or with an acid-terminated polyester in one type of powder coating, or cured at room temperature, as with the two-pack amine types using polyfunctional amines, such as diaminoethane or reactive polyaminopolyamides. 0
R NH-CH,OH
+
/\
CHz-CHCH-R
180°C
R' NH-CHz-OCHZ-CHCHZ-R
I
R' NH-CHz-OCHzCH-CHzR
+
R NH-CHzOH
I
OH
180°C
R'NHCHzOCHzCH-CH2-
I
OH
OCHZNHR
Curing reactions of epoxy resins with formaldehyde resins
0
R-COOH
/\
+ CHz-CHCHZ-OR
150°C R-COOCHZ-CHCHZ-OR
I
R-COOCHZ-CHCHZ-OR
I OH
+ R-COOH
180°C
OH
R-COOCHZCHCHzOR
I
OCO-R Curing an epoxy resin by esterification
14: 111
SYNTHETIC RESINS
The esterification reaction is also used to prepare epoxy esters from epoxy resins having an n value of 4 and vegetable oil fatty acids. They may be used in the same way as alkyds where better chemical resistance and adhesion are required. Unlike the alkyds, theepoxy esters contain virtually no acid groups. -0C H2-C HCH2 0 -
-0CHz-CHCHZ-0-
I
I
OH
0
+
co
I
I
NH
I
Room temperature .D
NCO CH3
+
co
OH
0
I
I
I
OCHzCHCH20-
OCH2CHCHZO-
Curing an epoxy resin by reaction with an isocyanate
This reaction is an example of a two-pack epoxy finish where the n value of the epoxy resin is 8 to 12. Although giving a high degree of chemical resistance the reaction is sluggish so the common two-pack finishes are usually based on polyamines with epoxy resins having n values of 0 to 2. 0 /\ OCHzCHCH2
Room temperature
+
4
H,N-CH,-CH,-NH,
OH
OH
I
I
a O C H 2 C > ;
C H z C H C H z O ~ N-CH2-C H2-N
OCH2CHCH2 dH
*
CH2CHCHz0 !IH
a
Curing an epoxy resin with a poiyfunctionai amine
14: 112
SYNTHETIC RESINS
The simplest polyamines are the aliphatic types such as diaminoethane, but these readily carbonate when exposed to the atmosphere as a thin film, so adducts (pre-reacted epoxy polyamines) are preferred. An alternative system is the polyaminoamides which are made by reacting dimerised fatty acids with an excess of polyamine. These themselves act as corrosion inhibitors and are noted for excellent adhesion. The curing mechanism shown below demonstrates the behaviour of one small polyamine molecule with four epoxy resin molecules. Similar reactions will occur at the other end of the epoxy resin molecules.
Isocyanate Resins A variety of types are available, each having different mechanisms of crosslinking but all dependent on the presence of the isocyanate (-NCO) group, either combined or free. Unlike the epoxy resins where the members differ only in their size, the isocyanate resins differ markedly according to the choice of components, but all have the common feature of a diisocyanate as one of the components. Two of the most widely used diisocyanates are tolylene diisocyanate and hexamethylene diisocyanate which have the following structures:
Nco Tolylene diisocyanate
Hexamethylene diisocyanate
From these, prepolymers are prepared where the diisocyanates may be completely reacted as in the case of the urethane oils which resemble the oilmodified alkyds but have urethane (-NHCOO-) links in place of the ester (-COO-) links of the alkyds, or where one only of the isocyanate groups is combined, leaving the other to participate in crosslinking reactions. Such a reactive prepolymer is the biuret that may be prepared from hexamethylene diisocyanate, has the following structure:
Biuret derived from hexamethylene diisocyanate
Such reactive isocyanates always contain about 1070 by weight of free diisocyanate monomer which is highly toxic, therefore when in use ventilation
14: 113
SYNTHETIC RESINS
must be excellent to maintain the occupational exposure limit below 0.02 PPm. The isocyanate group is more reactive than the epoxy group in that it will react at room temperature with water and hydroxyl groups as well as with amine groups. However, the latter reaction is too fast to be practicable so the standard two-pack coatings are based on isocyanate and polyhydroxy1 prepolymers such as hydroxyl terminated polyesters or polyethers as in the last example given in the section on epoxy resins. The moisture curing types are one-pack coatings, which, like the two-pack types have excellent chemical resistance and gloss but have a thickness limitation owing to the evolution of carbon dioxide during curing.
Formation of a crosslink by reaction of water and isocyanate group
P. J. BARNES BIBLiOGRAPHY General Miranda, T. J., J. Coat. Tech., 55, 696, 81-88 (1983) Boxall, J., Pol. Paint. Col. J., 175, 4154, 770-775 (1985) Seymour, R. B., J. Coat. Tech., 59, 745, 49-55 (1987) Boxall, J., Pol. Paint. Col. J . , 178, 4211, 240-244 (1988) Lambourne, R., Paint and Surface Coatings., Chap. 2, pp. 41-106, Ellis Horwood Ltd, Chichester (1987) Formaldehyde Resins Durr, H. and Schon, M., Pol. Paint. Col. J . , 177, 4205, 878-888 (1987) Sreeves, J., J. Oil. Col. Chem. Ass., 65, 2, 54-59 (1982) Alkyak and Polyesters Martin, J. C., Pol. Paint. Col. J., 175, 4134, 12-14 (1985), E p o q Resins and Powder Coatings
'Thermoset Powder Coatings', Fuel and Metallurgical J . , Surrey (1982) Jotischky, H., Pol. Paint. Col. J . , 175, 4136, 90-92 (1985) Kapilow, L. and Sammel, R., J. Coat. Tech., 59, 750, 39-47 (1987)
Isocyanate Resins Mirgel, V. and Nachtkamp., Pol. Paint. Col. J . , 176, 4163, 200-205 (1986) Stievater, P. C., J. Oil. Col. Chem. Ass., 70, 9, 262-267 (1987) Schonfelder, M., Pol. Paint. Col. J . , 177, 4203, 774-784 (1987)
14.10 Glossary of Paint Terms*
Adhesion: the degree of attachment between a paint or varnish film and the underlying material with which it is in contact. The latter may be another film of paint (adhesion between one coat and another) or any other material such as wood, metal, plaster, etc. (adhesion between a coat of paint and its substrate). Adhesion should not be confused with ‘cohesion’ (q.v.). Airless Spraying: the process of atomisation of paint by forcing it through an orifice at high pressure. This effect is often aided by the vaporisation of the solvents especially if the paint has been previously heated. The term is not generally applied to those electrostatic spraying processes which do not use air for atomisation. Bamer Coat: a coating used to isolate a paint system from the surface to which it is applied in order to prevent chemical or physical interaction between them, e.g. to prevent the paint solvent attacking the underlying paint or to prevent bleeding from underlying paint or material. Binder: the non-volatile portion of the vehicle of a paint; it binds or cements the pigment particles together, and the paint film as a whole to the material to which it is applied. Blast Cleaning: the cleaning and roughening of a surface by the use of natural grit or artificial ‘grit’or fine metal shot (usually steel), which is projected on to a surface by compressed air or mechanical means. Blistering: the formation of dome-shaped projections or blisters in paints or varnish films by local loss of adhesion and lifting of the film from the underlying surface. Such blisters may contain liquid, vapour, gas or crystals. Bubbling: a film defect, temporary or permanent, in which bubbles of air or solvent vapour, or both, are present in the applied film. Chalking: the formation of a friable, powdery coating on the surface of a paint film caused by disintegration of the binding medium due to disruptive factors during weathering. The chalking of a paint film can be considerably affected by the choice and concentration of the pigment. Cissing: a defect in which a wet paint or varnish film recedes from small areas of the surface leaving either no coating or an attenuated one. Cohesion: the forces which bind the particles of a paint or varnish film *For a full range of definitions see BS 2015:1992.
14: 114
GLOSSARY OF PAINT TERMS
14: 115
together into a coherent whole. It is distinct from ‘adhesion’ (q.v.), the forces binding the film to its substrate. Cracking: generally, the splitting of a dry paint or varnish film, usually as a result of ageing. The following terms are used to denote the nature and extent of this defect: Hair-cracking. Fine cracks which d o not penetrate the top coat; they occur erratically and at random. Checking. Fine cracks which d o not penetrate the top coat and are distributed over the surface giving the semblance of a small pattern. Cracking. Specifically, a breakdown in which the cracks penetrate at least one coat and which may be expected to result ultimately in complete failure. Crazing. Resembles checking but the cracks are deeper and broader. Crocodiling or alligatoring. A drastic type of crazing producing a pattern resembling the hide of a crocodile. Cratering: the formation of small bowl-shaped depressions in a paint or varnish film. Extender: an inorganic material in powder form which has a low refractive index and consequently little obliterating power, but is used as a constituent of paints to adjust the properties of the paint, notably its working and film-forming properties and to avoid settlement on storage. Filiform Corrosion: a form of corrosion under paint coatings on metals characterised by a thread-like form advancing by means of a growing head or point. Flaking: lifting of the paint from the underlying surface in the form of flakes or scales. Grinning Through: the showing through of the underlying surface due to the inadequate opacity of a paint film which has been applied to it. Holidays: skipped or missed areas, left uncoated with paint. Inhibitive Pigment: a pigment which retards or prevents the corrosion of metals by chemical and/or electrochemical means, as opposed to a purely barrier action. Red lead and zinc chromate are examples of inhibitive pigments as opposed to red iron oxide which has little or no inhibitive action. Medium: in paints or enamels. The continuous phase in which the pigment is dispersed; thus in the liquid paint in the can it is synonymous with ‘vehicle’ and in the dry film it is synonymous with ‘binder’ (q.v.). Opacity (Hiding Power): (a) Qualitatively. The ability of a coat of paint (or a paint system) to obliterate the colour of a surface t o which it is applied. ( b ) Quantitatively. The extent to which a paint obliterates the colour of an underlying surface of a different colour when a film of it is applied by some standard method. Orange Peel: the pock-marked appearance, in particular of a sprayed film, resembling the skin of an orange due to the failure of the film to flow out to a level surface. (See also spray mottle.) Pigment/Binder Ratio: the ratio of total pigment (white and/or coloured pigment plus extender) to binder (q.v.) in a paint; preferably expressed as a ratio by volume. Pinholing: the formation of minute holes in a film during application and drying. Sometimes due to air or gas bubbles in the wet film which burst,
14:116
GLOSSARY OF PAINT TERMS
forming small craters that fail to flow out before the film has set. Pitting: the formation of holes or pits in a metal surface, by corrosion. Plasticiser: a non-volatile substance, incorporated with film-forming materials in a paint, varnish or lacquer, to improve the flexibility of the dried film. Pot Life: the period after mixing the two packs of a two-puck (q.v.) paint during which the paint remains usable. Prefabrication Primer: a quick-drying material applied as a thin film to a metal surface after cleaning, e.g. by a blast cleaning process to give protection during the period before and during fabrication. Prefabrication primers should not interfere seriously with conventional welding operations or give off toxic fumes during such operations. Sagging: a downward movement of a paint film between the times of application and setting, resulting in an uneven coating having a thick lower edge. The resulting sag is usually restricted to a local area of a vertical surface and may have the characteristic appearance of a draped curtain, hence the synonymous term curtaining. Solids (Total Solids): the non-volatile matter in a coating composition, i.e. the ingredients of a coating composition which, after drying, are left behind and constitute the dry film. Solvent: a liquid, usually volatile, which is used in the manufacture of paint to dissolve or disperse the film-forming constituents, and which evaporates during drying and therefore does not become a part of the dried film. Solvents are used to control the consistency and character of the finish and to regulate application properties. Spray Mottle: the irregular surface of a sprayed film resembling the skin of an orange. The defect is due to the failure of the film to flow out to a level surface. (See also orange peel.) Tack: slight stickiness of the surface of a film of paint, varnish or lacquer, apparent when the film is pressed with the finger. Thinning Ratio: the recommended proportion of thinners to be added to a paint or varnish to render it suitable for a particular method of application. Thixotropic Paint: a paint which while free-flowing and easy to manipulate under a brush, sets to a gel within a short time when it is allowed to remain at rest. Because of these qualities a thixotropic paint is less likely to drip from a brush than other types and can be applied in rather thicker films without running or sagging. Two-Pack: a paint or lacquer the materials for which are supplied in two parts which must be mixed in the correct proportions before use. The mixture will then remain in a usable condition for a limited time only. The two parts of a two-pack paint are often (though not necessarily) supplied in the correct relative proportion either in entirely separate containers of appropriate sizes or in a single container divided into two compartments; the term 'dual-pack' is often used to describe the latter type of container. Zinc-Rich Primer: an anticorrosive primer for iron and steel incorporating zinc dust in a concentration sufficientto give electrical conductivity in the dried films, thus enabling the zinc metal to corrode preferentially to the substrate, i.e. to give cathodic protection. E. F. REDKNAP N.R. WHITEHOUSE
15
CHEMICAL CONVERSION COATINGS
15.1 Coatings Produced by Anodic Oxidation 15.2 Phosphate Coatings 15.3 Chromate Treatments
15: 1
15:3 15:22 15:38
15.1 Coatings Produced by Anodic Oxidation Practice of Anodising Anodic oxidation or anodising, as applied to metallic surfaces, is the production of a coating, generally of oxide, on the surface by electrolytic treatment in a suitable solution, the metal being the anode. Although a number of metals including aluminium, magnesium, tantalum, titanium, vanadium and zirconium, can form such anodic films, only aluminium and its alloys, and to a lesser extent magnesium, are anodised on a commercial scale for corrosion protection. The anodic oxidation of magnesium does not normally produce a film that has sufficient corrosion resistance to withstand exposure without further protection by painting, and the solutions used are complex mixtures containing phosphates, fluorides and chromates. In the case of aluminium, a relatively simple treatment produces a hard, compact, strongly adherent film of oxide, which affords considerably increased protection against corrosive attack2*-’. A further advantage of this process lies in the decorative possibilities of the oxide film, which may be almost completely transparent on very high purity aluminium (99.99% Al) and certain alloys based on this purity, and thus protects the surface without obscuring its polish or texture. On metal of lower purity, and other alloys, the oxide layer may become slightly milky, or coloured grey or yellowish, although the deterioration is hardly apparent with purities down to 99-7-99-870 Al. The appearance and character of the film may also be influenced by the type of anodising treatment, and the oxide film may be dyed to produce a wide range of coloured finishes. Anodising characteristics of a number of aluminium alloys are listed by Wernick and Pinner The anodising procedures in general use are shown in Table 15.1, sulphuric acid being the most commonly used electrolyte. Treatment time is 15min to 1 h. The articles to be anodised should be free from crevices where the acid electrolyte can be trapped*. They may be given a variety of mechanical and chemical pretreatments, including polishing, satin-finishing, etching, etc. but before anodising, the surface must be clean and free from grease and polishing compound.
’,
’.
T h e chromic acid process is preferred where the electrolyte is likely to be entrapped in crevices as it is an inhibitor for aluminium whereas sulphuric acid is corrosive.
15:3
15:4
COATINGS PRODUCED BY ANODIC OXIDATION
Table 15.1 Traditional anodising processes
Electrolyte
Temp ("C)
E.M.F.
(V)
Current density
Film thickness
(Am-*)
Gm)
17-22
12-24
110-160
3-25
3-10% chromic acid
30-45
30-45
32
2-8
2-5% oxalic acid
20-35
30-60
110-215
10-60
5-10% (v/v)
sulphuric acid
Appearance Transparent, colourless to milky Opaque, light to dark grey Transparent, light yellow to brown
After the anodic treatment, the work is removed from the tank and carefully swilled with cold water to remove all traces of acid. At this stage, the anodic film is absorptive, and care should be taken to avoid contamination with oil or grease, particularly if the work is to be dyed. Dyeing may be carried out by immersion for about 20min in an aqueous solution of the dyestuff at a temperature of 50-60°C. Inorganic pigments may also be incorporated in the oxide layer by a process involving double decomposition. Finally, both dyed and undyed work are sealed by treatment in boiling water (distilled or deionised) or steam, which enhances the corrosion resistance and prevents further staining or leaching of dye. Solutions of metal salts, usually nickel or cobalt acetates, are often used to seal work after dyeing, and sealing in 5-10Vo dichromate solution, which gives the coating a yellow colour, is sometimes employed where the highest degree of corrosion resistance is desired '. In the architectural field, increasing use is being made of integral colour anodising which is capable of producing self-coloured films in a number of fade-resistant tints ranging from grey, through bronze and brown, to a warm black. The electrolytesare developments of the oxalic acid solution and consist of various dibasic organic acids, such as oxalic, malonic or maleic, or sulphonated organic acids such as sulphosalicylicacid, together with a small proportion of sulphuric acid. For constant and reproducible results, a close analytical control of the electrolyte must be maintained, particularly with respect to aluminium which dissolves as treatment proceeds, and ionexchange resins are frequently used to regenerate the relatively expensive electrolyte and keep the aluminium in solution between controlled limits. Some typical colour anodising treatments are summarised in Table 15.2*. Alloys are generally of the Al-Mg-Si type with additions of copper and chromium or manganese. Colour varies with the particular alloy and the film thickness. For optimum control of coiour, the alloy must be carefully produced with strict attention to composition, homogenisation and heattreatment, where appropriate, and the anodising conditions must be maintained within narrow limits. It is usual to arrange matters, preferably with automatic control, such that current density is held constant with rising *Coloured metal compounds may also be introduced into the film by a x . treatment in a suitable electrolyte [Fuji process. UK Pat. 1 022 927 (26.2.63)l.
15:5
COATINGS PRODUCED BY ANODIC OXIDATION
Table 15.2 Integral colour anodising processes
Process Kalcolor
Electrolyte Sulphosalicylic acid,
Voltage
("C)
Current density (Am-')
(V)
Time (rnin)
22-25
215-320
25-10
20-45
15-30
130-370
>IO
30
15-25
130-160
34-61
50-90
Temp.
100 g/1
Duranodic6
Alcanadox'
Sulphuric acid, 50 g/l 4- or 5-sulphophthalic acid, 75-100 g/l Sulphuric acid 8-10 g/l Oxalic acid, 80 g/l to saturation
voltage up to a selected maximum, after which voltage is held steady; the whole cycle being for a fixed time. Refrigeration of the electrolyte may be necessary to maintain the temperature at the working level, owing to the relatively high wattage dissipation. Hard anodic films, 50-100 pm thick, for resistance to abrasion and wear under conditions of slow-speed sliding, can be produced in sulphuric acid electrolytes at high current density and low temperature'. Current densities range from 250 to 1 OOO Am-', with or without superposed alternating current in 20-100 g/l sulphuric acid at - 4 - 10°C. Under these conditions, special attention must be paid to the contact points to the article under treatment, in order to avoid local overheating. The films are generally dark in colour and often show a fine network of cracks due to differential expansion of oxide and metal on warming to ambient temperature. They are generally left unsealed, since sealing markedly reduces abrasion resistance, but may be impregnated with silicone oils' to improve the frictional properties. Applications include movable instrument parts, pump bodies and plungers, and textile bobbins. Decorative self-coloured films lo can also be produced in sulphuric acid under conditions intermediate between normal and hard anodising . Continuously anodised strip and wire, which may be given a dyed finish, are produced by special methods, and are now available commercially with a film thickness up to about 6pm. Uses include electrical windings for transformers and motors, where the light weight of aluminium and the insulating and heat-resistant properties of the film are of value, and production of small or light-section articles by stamping or roll-forming.
+
Mechanism of Formation of Porous Oxide Coatings The irreversible behaviour of an aluminium electrode, which readily passes a current when cathodically polarised, but almost ceases to conduct when made the anode in certain aqueous solutions, has been known for over a century. It has been established that in the case of electrolytes, such as boric acid or ammonium phosphate solutions, in which aluminium oxide is insoluble,
15:6
COATINGS PRODUCED BY ANODIC OXIDATION
this anodic passivity is due to the formation of a thin compact layer of aluminium oxide whose thickness is proportional to the applied voltage. In neutral phosphate solutions, for example, film growth practically ceases when the thickness corresponds t o about 1 - 4 nm/V*, and a similar value has been found for many other electrolytes of this type. These thin films have a high electrical resistance, and can withstand several hundred volts under favourable conditions. In electrolytes in which the film has a moderate solubility, film growth is possible at lower voltages, e.g. in the range 12-60 V, since the rate of formation of the oxide exceeds its rate of solution and current flow continues owing to the different structure of the oxide layer. Electron microscopy has revealed the characteristic porous structure of these films". The pore diameter appears to be a function of the nature and concentration of the electrolyte and of its temperature, being greatest in a solution of high solvent activity, while the number of pores per unit area varies inversely with the formation voltage. In any given electrolyte, the lower the temperature and concentration, and the higher the voltage, the more dense will be the coating, as both the pore diameter and the number of pores per unit area are reduced under these conditions. Table 15.3, taken from a paper by Keller, Hunter and Robinson1*,illustrates these points. Table 15.3 Number of pores in anodic oxide coatings
Temp.
E.M.F.
Electrolvte
("C)
(V)
15% sulphuric acid
10
3% chromic acid
2% oxalic acid
Pores/cmZ x
15
77
20
51 28 22 8
49
30 20
24
40 60 20 40 60
4
36 12 6
Nofe. Data reprodud courtesy J. Efecfrochem.Soc.. 100. 41 I (1953)
In order to account for the relatively high potential required to maintain the current it was suggested by Setoh and MiyataI3 that a thin barrier-foyer, similar t o that formed in non-solvent electrolytes, is present below the porous layer. This view has been supported by later work involving capacity and voltage-current measurements, which have allowed the thickness of the barrier-layer to be computed 14. As in the case of electrolytes which produce barrier films, the thickness has been found to be proportional to the anodising voltage, but is lower than the limiting growth rate of 1*4nm/V, and varies with the anodising conditions (Table 15.4). The structure of the anodic film, according t o present views, is shown diagrammatically in Fig. 15.1. *The limiting thicknessexpressedin nm/V is of some practicalvalue. but has little theoretical significance-at constant potential the rate of growth, although extremely small, i s still finite.
COATINGS PRODUCED BY ANODIC OXIDATION
15:7
The more or less regular pattern of pores imposes a cellular structure on the film, with the cells approximating in plan to hexagons, each with a central pore, while the bases which form the barrier-layer, are rounded. The metal surface underlying the film, therefore, consists of a close-packed regular array of nearly hemispherical depressions which increase in size with the anodising voltage. The thickness of the individual cell walls is approximately equal to that of the barrier-layer ". Table 15.4 Barrier-layer thickness in various electrolytes Electrolyte
Temp. ("C)
Unit barrier-layer thickness (nm/V)
15% sulphuric acid 3% chromic acid 2% oxalic acid
10 38 24
1.00 1.25 1.18
Note. Data reproduced courtesy J. Electrochem. Soc.. 101, 481 (I954)l4
In view of its position in the e.m.f. series ( E 0 N 3 + / ~ = I - 1 -66V (SHE)), aluminium would be expected to be rapidly attacked even by dilute solutions of relatively weak acids. In fact, the rate of chemical attack is slow, owing to the presence on the aluminium of a thin compact film of air-formed oxide. When a voltage is applied to an aluminium anode there is a sudden initial surge of current, as this film is ruptured, followed by a rapid fall to a lower, fairly steady value. It appears that this is due to the formation of a barrierlayer. Before the limiting thickness is reached, however, the solvent action of the electrolyte initiates a system of pores at weak points or discontinuities in the oxide barrier-layer .
Fig. 15.1 Diagrammatic cross-section of porous anodic oxide film
The formation of pores appears to start along the sub-grain boundaries of the metal, followed by the development of additional pores within the subgrains. Growth of oxide continues on a series of hemispherical fronts centred on the pore bases, provided that the effective barrier-layer thickness between the metal surface and the electrolyte within the pores, represented by the hemisphere radius, is less than 1.4 nm/V. As anodic oxidation proceeds at
15:8
COATINGS PRODUCED BY ANODIC OXIDATION
a uniform rate, a close-packed hexagonal cell-pattern is produced, the downward extension of the pore due to solution of oxide keeping pace with the downward movement of the oxidelmetal interface, as shown by the arrows in Fig. 15.1. It is fairly clear that the thickness of the individual cell walls cannot exceed the thickness of the barrier-layer if columns of unchanged metal are not to be left behind in the anodic film. The inverse relationship between number of pores and anodising voltage also implies that cells with much thinner walls cannot be formed. Growth of pores in excess of the limiting number appears to be inhibited at an early stage of development, but the actual mechanism is still in doubt. Radiochemical studies” indicate that the pore base is the actual site of formation of aluminium oxide, presumably by transport of aluminium ions across the barrier-layer, although transport of oxygen ions in the opposite direction has been postulated by some authorities I . The downward extension of the pore takes place by chemical solution, which may be enhanced by the heating effect of the current and the greater solution rate of the freshly formed oxide, but will also be limited by diffusion. It has been shown that the freshly formed oxide, y’-A1203, is amorphous and becomes slowly converted into a more nearly crystalline modification of y-A120:6. Prolonged action of the acid electrolyte on thick films may cause the pores to become conical in section, widening towards the upper surface of the film. This will impose an upper limit on film thickness in solvent electrolytes, as found in practice. Although it might seem at first sight that dyestuffs are merely held mechanically within the pores, and this view is probably correct in the case of inorganic pigments, there is some support for the opinion that only those dyestuffs which form aluminium/metal complexes produce really light-fast colorations. The effect of hot water sealing is to convert anhydrous y-A1203into the crystalline monohydrate, A1203.H,O, which occupies a greater volume and blocks up the pores, thus preventing further absorption of dyes or contaminants. The monohydrate is also less reactive.
Properties of Coatings Composition The main constituent of the film is aluminium oxide, in a form which varies in constitution between amorphous A1203and y-Al,O, , together with some monohydrate, Al,O,.H,O. In the presence of moisture, both the anhydrous forms are gradually transformed into the monohydrate, and the water content of as-formed films is, therefore, somewhat variable. After sealing in boiling water, the composition of the completely hydrated film obtained when using sulphuric acid approximates to: Al,O, 70 H*O 17
so,
13
It is probable that the SO, is combined with the aluminium as a basic sulphate.
COATINGS PRODUCED BY ANODIC OXIDATION
15:9
Films produced in oxalic acid contain smaller amounts (about 3%) of the electrolyte and only traces of chromium are found in chromic acid films. Sealed films show the electron diffraction pattern of the monohydrate, bohmite.
Density Owing to the variable degree of porosity of the anodic film, it is only possible to determine the apparent density, which varies with the anodising conditions and also with the film thickness. 3.2
“E
”
3-0
\
LII
>-
!=
In
5
2.8
n
5
10
15
FILM THICKNESS ( p m )
Fig. 15.2 Apparent density of anodic film as a function of film thickness (courtesy Aluminium, Berl.. 32, 126 (1938))
Fig. 15.2, taken from a paper by Lenz”, shows the variation in density with thickness for steam-sealed anodic films produced in sulphuric acid on aluminium of 99.99% and 99.5070 purity. A mean figure of 2.7 g/cm3 for sealed, and 2.5 g/cm3 for unsealed films is accepted by the British Standard for anodised aluminium ’*.
Hardness It is not possible to obtain a reliable figure for the hardness of anodic coatings with either the indentation or scratch methods, because of the influence of the relatively soft metal beneath the anodic film, and the presence of a soft outer layer on thick films. On Moh’s Scale, the hardness of normal anodic films lies between 7 and 8, i.e. between quartz and topaz. Methods are available for the determination of relative abrasion resistance using either a mixed jet of air and abrasive, as recommended in the appropriate British Standard’*or an abrasive wheel or disc. Owing to variations in the quality of the abrasive, and the performance of individual jets, a standard comparison sample is included in each batch. The hardness of the film is markedly affected by the conditions of anodising. By means of special methods involving dilute electrolytes at low temperatures and relatively high voltages*, with or without superimposed alternating current, it is possible to produce compact abrasion-resistant films with thicknesses of 50-75 pm and hardnesses of 200-500VPN, for special applications. Flexibility The normal anodic film begins to crack if subjected to an extension exceeding about 0.5%. Thinner films up to 5 pm in thickness appear to withstand a greater degree of deformation without obvious failure, and are often used for dyed coatings on continuously anodised strip from which
15: 10
COATINGS PRODUCED BY ANODIC OXIDATION
small items may be punched or stamped. Continuously anodised wire can be bent round a radius of 10-15 times its diameter without visible crazing. A greater degree of flexibility is also shown by the more porous coatings produced in 20-25% V.V. sulphuric acid at 35-40°C, while hard films are much less flexible. Unsealed films are only slightly more flexible than films sealed in water or dichromate solution.
Breakdown voltage The breakdown voltage of an anodic film varies with the method of measurement and conditions of anodising, and shows fluctuations over the surface. In the case of unsealed films, breakdown voltage also depends on the relative humidity at the time of measurement. It is normally measured by applying a slowly increasing alternating voltage between a loaded hemispherical probe on the upper surface of the film, and the underlying metal, contact to which may be established by removing a portion of the film I*. The breakdown voltage/thickness relationship for sealed films up to about 20 pm is approximately linear, and the slope of the curve for sulphuric acid films varies from 30 to 40V/pm. These results were obtained with a relatively high loading on the probe*; with reduced load (approx. 60 g and below on a hemispherical probe of 1 -6 mm radius) values of 60-100 V/pm can be reached. The higher figures probably represent limiting values which will apply to the conditions between adjacent laps or turns on coils wound from anodised strip or wire. Resistance The specific resistance of the dry anodic film is Dielectric constant The dielectric constant of anodic oxide films has been found to be 5-0-5.9 for sulphuric films, and 7-8 for oxalic films. A mean value of 7.45 has been quoted for barrier-layer films', but more recent work favours a value of 8 ~ 7 ~ ' . Thermal expansion The thermal expansion of the film is only about onefifth that of aluminium', and cracking or crazing is observed when anodised aluminium is heated above 80°C. The fine hair-cracks produced do not seem to impair the protective properties of the coating if anodising conditions have been correct. Heat conduction The heat conductivity of the film is approximately onetenth that of aluminium2. Heat resistance Apart from hair-cracks, little change is observable in the anodic film on heating up to 300-35OoC, although some dyed finishes may change colour at 200-25OoC, but at higher temperatures up to the melting point of the metal, films may become opaque or change colour, owing to loss of combined water, without losing their adhesion. Emissivity Table 15.5 shows the total heat emissivity of various aluminium surfaces, as a percentage of that of a black body. The figures have been recalculated from the data of Hase". The emissivity of anodised aluminium rises rapidly with film thickness up to 3 pm after which the rate of increase diminishes. *Several hundred grams, BS 1615 suggests 50-75 g.
COATINGS PRODUCED BY ANODIC OXIDATION
15:ll
Table 15.5 Relative heat emissivity of various aluminium surfaces ~
~~
Heat emissivity Surface
(%o) ~~
Highly polished Etched Bright roll finish Matt roll finish Aluminium paint Diecast Sandcast Anodised, according to film thickness Black body
4.3-6.4 6.4-8.5 5.3-7.4 8.5-16 17-32 16-26 26-36 38-92 100
Heat reflectivity The heat reflectivity of as-rolled aluminium is about 95%, but this high value may not be maintained for long in a corrosive atmosphere, although it is less affected by surface finish than is optical reflectivity. Anodising reduces the heat reflectivity, owing to absorption by the oxide layer; this effect increases with film thickness. There is a deep absorption trough in the region corresponding to a wavelength of 3 pm; this is probably due to the-OH grouping in the hydrate, the effects of which may be minimised by sealing the heated film in oil instead of waterz2.This treatment is particularly valuable for heat reflectors in apparatus using sources running at 900-1 OOO'C, which show a peak emission in the 2-3 pm region. Fig. 15.3 90
-
80
s t 2 I-
;
70
60
LL
% k
50
W I
5
10
15
20
FILM THICKNESS ( p m ]
Fig. 15.3 Heat reflectivity of anodised aluminium
shows the heat reflectivity of anodised super-purity aluminium for a source of this type23, plotted against film thickness. The benefits of the modified sealing treatment are obvious.
Refractive index The refractive index of the clear anodic film produced on aluminium of the highest purity in sulphuric acid is 1 59 in the as-formed condition, rising to 1 -62 after Reflectivity The total and specular reflectivities of an anodised aluminium surface are controlled by both the condition of the metal surface, polished
15: 12
COATINGS PRODUCED BY ANODIC OXIDATION
or matt, and the absorption or light-scattering properties of the oxide layer. Total reflectivity may be defined as the percentage of the incident light reflected at all angles, while specular reflectivity is that percentage reflected within a relatively narrow cone with its axis along the angle of reflection. For many years the standard instrument for measuring specular reflectivity has been that designed by Guild2*, but more recently a modified gloss head giving rather greater discrimination has been described by Scott 26. Other instruments, while placing a number of surfaces of varying specularity in the same relative order, may give different values for the specular reflectivity. The general brightness of a surface is chiefly dependent upon the total reflectivity T, while specular reflectivity S controls the character of the reflected image. In assessing the subjective brightness of a surface the eye tends to be influenced more by the S/Tratio or image clarity than by the total reflectivity. For a high degree of specularity, the metal surface must be given a high polish by mechanical means; this may be followed (or replaced) by electrochemical or chemical brightening. When such a brightened surface is protected by anodising. however, insoluble impurities (mainly iron and silicon) present in the aluminium will be incorporated in the anodic film and will increase its tendency to absorb or scatter light. Only metal of the highest purity, 99-99% AI, produces a fully transparent oxide film, while lower purities show decreased total reflectivities and S / T ratios after anodising because of the increased opacity of the anodic film. Table 15.6 Effect of metal purity and anodic film thickness on reflectivity ~
Film thickness (pm) Metal purity (TO)
99.5
99.8 99.99 (super purity) Super purity + 0.5 Mg Super purity + I .25 Mg Super purity + 0.7 Mg,0.3 Si, 0.25 Cu
2
IO
5
T
S/T
T
80 82 84 84 83 82
0-84
0.95 0.99 0.98 0.99 0.99
79 83 84 84 83 79
S/T
T
S/T
0.83
77
-
0.78
0-95 0-99
84 83 82
0.99
0.98 0.99 0.98
-
-
0.97 0.99
-
. ~~
Nofes. I . T =
total, S = specular reflectivity. 2. Data reproduced courtesy Mer. Rev.. 2 No. 8 (1957f3.
Table 15.6, taken from a monograph by Pearson and Phillips23,demonstrates these effects. The figures were obtained using the Guild meter on electrobrightened and anodised metal. Effect of anodising on mechanical properties The tensile strength of thin sections may be somewhat reduced by anodising, owing to the brittleness of the coating, but this effect is normally very slight. Thin sheet, less than about 0.6 mm with a relatively thick anodic coating, also has a tendency to break more easily on bending. The incompressibility of the anodic film on the inside of the bend probably enhances this effect, which is also seen on anodised wire.
COATINGS PRODUCED BY ANODIC OXIDATION
15: 13
Anodising should be used with caution on components likely to encounter high stresses, owing to the deterioration in fatigue properties liable to result under these conditions, but under light loading and with the thinner coatings, the reduction is negligible. In some cases2', an actual improvement has been reported. Friction The coefficient of friction of the sealed anodic film is 0.76, falling to 0.19 after impregnation with silicone oilz8. These results were obtained with anodised wire. Measurement of film thickness The thickness of an anodic film may be determined by a variety of non-destructive methods. Some of these are capable of a high degree of precision, while simpler methods are available for rough sorting. A number of instruments employing the eddy-current principle, with which, after prior calibration, a rapid estimate of film thickness may be made, are now available. With the best instruments, an accuracy of i 1 pm can be obtained. For approximate determinations of thickness, the breakdown voltage of the film may be measured. Breakdown voltage shows wide variations with anodising conditions and metal or alloy composition. A separate calibration curve is, therefore, needed for each treatment. Accuracy is comparatively low, rarely being greater than *2Vo of the total film thickness. For control or calibration purposes, film thickness can be determined by mounting a sectioned specimen and measuring the oxide film thickness directly on the screen of a projection microscope at a known magnification. Alternatively, the loss in weightI8 of an anodised sample of known area may be found after the film has been stripped in a boiling solution made up as follows: 3.5% v/v Phosphoric acid (s.g. 1.75) Chromic acid in distilled water 2.0% w/v Immersion for 10 min is usually sufficient to remove the film without the metal being attacked.
Corrosion Resistance Since the natural passivity of aluminium is due to the thin film of oxide formed by the action of the atmosphere, it is not unexpected that the thicker films formed by anodic oxidation afford considerable protection against corrosive influences, provided the oxide layer is continuous, and free from macropores. The protective action of the film is considerably enhanced by effective sealing, which plugs the mouths of the micropores formed in the normal course of anodising with hydrated oxide, and still further improvement may be afforded by the incorporation of corrosion inhibitors, such as dichromates, in the sealing solution. Chromic acid films, in spite of their thinness, show good corrosion resistance. The protective action of sulphuric films is mainly controlled by the anodising conditions, compact films formed at temperatures below 2OoC in 7% v/v sulphuric acid being more resistant than the films formed at higher temperatures in more concentrated acid. The wider pores of the latter result in less
15: 14
COATINGS PRODUCED BY ANODIC OXIDATION
protection but these films are more readily dyed. Greater protection is also given by thicker films, and a thickness of about 25 pm is generally considered adequate for architectural work in a normal urban environment. In a heavily polluted industrial area, even thicker films may be desirable, while in rural areas some reduction would be permissible. Bright anodised motorcar trim is generally given a film thickness of about 7 pm. Alumina monohydrate in the mass is very unreactive, being rapidly attacked only by hot sulphuric acid or caustic soda solutions, and the anodic coating shows similar characteristics to some degree. The presence in the film of macropores due to localised impurities or imperfections in the metal and overlying oxide can bring about rapid penetration, owing to the concentration of attack at the few vulnerable points. Metal of good quality specially produced for anodising should therefore be used in order to ensure that such weak points are absent. For vessels and tanks for holding liquids, it may be preferable to use unanodised aluminium, and to accept generalised corrosive attack rather than run the risk of perforation, which may occur with anodised metal. For ordinary atmospheric exposure, it is usually possible to arrange that thin spots of the film, such as the contact points of the anodising jigs, are located in relatively unimportant positions on the article and are hidden from view. Since the corrosion resistance of anodic films on aluminium is markedly dependent on the efficacy of sealing (provided the film thickness is adequate for the service conditions), tests for sealing quality are frequently employed as an index of potential resistance to corrosion. While it is admitted that an unequivocal evaluation of corrosion behaviour can only be obtained by protracted field tests in service, accelerated corrosion tests under closely controlled conditions can also provide useful information in a shorter time within the limitations of the particular test environment employed. Tests for sealing include dye staining tests such as that specified in BS 1615: 1972*, Method F, involving preliminary attack with acid, followed by treatment with dye solution. Nitric acidz9or a sulphuric acid/fluoride mixture may be used for the initial attack, and a rapid spot test” has been developed using the acid/fluoride mixture, followed by a solution of 10 g/l Aluminium Fast Red B3L W. Poor sealing is revealed by a deep pink to red spot, while good sealing gives nearly colourless to pale pink colorations. The test can be applied to architectural or other material on site. Physical tests of film impedance” using an a.c. bridge have also been recommended, although the correlation with corrosion resistance is necessarily empirical. Film impedance increases at an approximately linear rate with sealing time and film thickness. Exposure of the samples to a controlled moist atmosphere containing sulphur dioxide, as recommended in BS 1615 : 1972, Method If,is an example of a test bridging the gap between sealing tests and accelerated corrosion tests. After exposure for 24 h at 25 f 2”C, poorly sealed films show a persistent heavy white bloom, while good sealing produces at the most a slight superficial bloom. A rapid immersion test in a hot aqueous solution containing sulphur *A revised version BS 1615 : 1987 is now available.
COATINGS PRODUCED BY ANODIC OXIDATION
15: 15
dioxide has also been developed by Kape3* and is specified in BS 1615: 1972, Method E. Results are similar to those obtained in the preceding test, Method H. The method can also be made quantitative by measuring the weight loss. The accelerated corrosion test in most general use is the CASS test33in which the articles are sprayed intermittently with a solution made up as follows: 50 g/l NaCl 0.26 g/l CUCI,, 2HzO Acetic acid to pH 2.8-3-0
The specimens are clamped at an angle of 15" to the vertical in a baffled enclosure maintained at 5OoC, and the exposure time is 24-96 h. Corrosive attack of inadequately sealed or thin films is shown by pitting. An interesting derivative of the CASS test, known as the Ford Anodised Aluminium Corrosion Test (FACT)34 has been developed in the U.S.A. This makes use of a controlled electrolytic attack using the CASS solution. The electrolyte is contained in a glass test cell and clamped against the anodised surface with a Neoprene sealing gasket. A d.c. voltage of 200 V in series with a high resistance is maintained between an anode of platinum wire and the aluminium test piece as cathode. The integrated fall in potential across the cell over a fixed period of 3 min as corrosion proceeds and an increasing current flows, is taken as a measure of the corrosion resistance. A British version of this test using simplified circuitry for the integration is available commercially as the Anodisation Comparator*. Remarkably good correlation has been obtained between the readings of this instrument and the amount of pitting after exposure at a number of outdoor sites35.Comprehensive reviews of sealing techniques including test methods and corrosion behaviour have been published by and Wood3'. The behaviour of samples under the actual conditions of service is the final criterion, but unfortunately such observations take a long time to collect and assess, and the cautious extrapolation of data from accelerated tests must be relied on for forecasting the behaviour of anodised aluminium in any new environment. A tmospberic Exposure
Table 15.7 shows the effects of thin anodic oxide films on the resistance to industrial and synthetic marine atmospheres (intermittent salt spray) of three grades of pure aluminium. The results are taken from a paper by Champion and S ~ i l l e t tand ~ ~ show how relatively thin films produce a marked improvement in both environments. In an industrial atmosphere, an anodic film only 6-5 p m thick provides a two-fold increase in life over unprotected metal, and the effect under saltspray conditions is even greater. It is interesting to note that both the industrial atmosphere and salt-spray results show parallel trends. A similar improvement in expectation of life for thin anodic coatings has *SIBA Ltd., Camberley, Surrey.
15: 16
COATINGS PRODUCED BY ANODIC OXIDATION
been reported by Phillips39for 99.5070 AI, and for alloys of the following compositions: Al-1-25 Mn; AI-2 Mg-1 Mn; AI-1 Mg-I Si. The results for a high-copper alloy were less good, An interesting paper by Lattey and Neunzig4 shows that the better the surface finish of the aluminium the thinner the coating required for protection. Neunzig4' has also studied the effect of the hair-cracks produced by heating or bending on corrosion resistance. Although pitting was initiated by such cracks in thin films (5 pm), serious pitting in thicker films (15 pm) was observed only if anodising had been carried out at 25°C; films produced at 16-17"C were more resistant to corrosive attack. This re-emphasises the importance of maintaining correct anodising conditions for maximum corrosion resistance. More recently, results of exposure tests for 10 years in a severe industrial environment at Stratford, London, have been reported by the Fulmer Research Institute4'. A range of pure and alloy specimens, anodised to a maximum film thickness of about 25 pm, was exposed at an angle of 45'.
Table 15.7
Corrosion tests on unprotected and anodised pure aluminium
Corrosive eflect
Grade 1B (99.5%)
Grade 1A (99-8%)
Super purity
Film thickness
Film thickness
Film thickness
(pm)
(pm)
(pm)
0 Appearance* (life in years) Industrial Mechanical atmosphere propertiest (7 years (life in years) exposure) Pittingt (depthinmm)
6.5
0
4
6.5
(99.99%)
0
4
2.5
5
2.5
5
5
3.5
6
2.75
5.5
3
4.5
6
3
5
0.18
0.20 0.18 0.25
0.25
0.20
0.13
3
4
Appearance* Marine (life in years) < I 4 1 4 5 atmosphere Mechanical ( 1 1 years propertiest 5 >I1 8 7 >I1 exposure) (life in years) Pitting$ (depth in mm) 0.30 0.18 0.15 0.33 0.15
>I1 0.15
>I1 0.08
*No. of years to deterioration of surface appearance to a fixed arbitrary level. tNo.o f years to deterioration of mechanical properties to a fixed arbitrary level fMean depth of pitting obtained statistically.
Corrosion was assessed visually, by determination of weight loss after cleaning, and by reflectivity measurements. All specimens showed signs of pitting, and there was a considerable loss of reflectivity, the under surface being more affected than the upper. A striking feature of the results was the accelerating rate of deterioration in the last five years of exposure. Although none of the samples was completely protected, results were better for the purer specimens and the thicker films.
COATINGS PRODUCED BY ANODIC OXIDATION
15: 17
Maintenance
In architectural work, particular care must be taken to avoid destructive attack of the anodic film by alkaline mortar or cement during erection, and temporary coatings of spirit-soluble waxes, or acetate-butyrate lacquers are frequently applied to window frames and the like to protect against mortar splashes, which in any event should be removed at the earliest possible moment. The resistance of properly anodised aluminium exposed to the weather can be considerably enhanced by correct and regular cleaning. Deposits of soot and dirt should be removed by washing with warm water containing a nonaggressive detergent; abrasives should not be used. For window frames this washing may conveniently be carried out when the glass is cleaned in the normal way. In such circumstances the life of the coating may be prolonged almost indefinitely, as exemplified by the good condition of the chromicanodised window frames of Cambridge University Library which were installed in 1933, and of the sulphuric-anodised window frames of the New Bodleian Library, Oxford University, installed in 1938.
Recent Developments Practice of Anodising
Although there have been few changes in the basic anodising practices, and sulphuric acid is the electrolyte used in most plants, there have been many developments in the pretreatment, colouring and sealing processes associated with anodising. The trend in architectural applications has been towards more matt finishes, and the sodium hydroxide-based etchants used frequently contain additives such as sodium nitrate or nitrite or sodium fluoride. Chelating agents such as gluconates, heptonates or sorbitol are added to complex the aluminium produced, and other additives such as sulphides may be present in the etchant to complex zinc dissolved from the alloy, and allow it to be used continuously without dumping43. In terms of anodising itself, the introduction of a standard for architectural applications of anodised aluminium", and the European development of the Qualanod quality labelling scheme for architectural a n ~ d i s i n g ~ ' ~ ~ ~ , have been significant factors in the general improvement in the standard of anodising. Both of these standards require the use of thick coatings (20 or 25pm), which are sealed to a high quality level. The production of such coatings requires good control of operating parameters, particularly the anodising electrolyte temperature, which should be below 21T4'. The field of colour anodising has changed considerably since the late 1 m s . At that time the integral colour anodising processes were dominant in architectural applications, and electrolytic colouring was relatively new. Now, mainly because of the high energy costs associated with integral colour processes, electrolytic colouring is by far the most widely used technique.
15: 18
COATINGS PRODUCED BY ANODIC OXIDATION
In order to produce colour by this method, the anodised work is rinsed and transferred to a suitable metal salt solution. The process is electrolytic, and a.c. is passed between the work and a metal or graphite counter-electrode, causing the metal present in the solution to be deposited at the base of the pores of the anodic coating4*. The height of the metal deposited in the pores controls the depth of colour, and a range of shades is produced by varying the applied voltage and time. Ranges of bronze and black finishes are produced in nickel-, cobalt- or tin-based electrolytes, and pink, maroon or black finishes in electrolytes based on copper. The electrolytes usually contain the appropriate metal sulphate, with many other additives present to adjust or contol pH, to improve throwing power, or to make dark colours easier to produce. Nickel and cobalt electrolytesare used at pH values of 4-6, and tin and copper electrolytes at pH values of 1-2; an e.m.f. of the order of 10-20V and a current density of about 30-50 A/m2 are normally required. The finishes produced have very good light fastness and corrosion resistance, and, unlike integral colour finishes, the shade is largely independent of the aluminium alloy and the anodic film thickness used. The whole range of shades can be produced on films as thin as 5 pm, so the finishes are also being used in trim application^^^. Many patents and publications in the electrolytic colouring field now exist and they have been reviewed by many
author^".^'. In order to obtain a wider range of coloured finishes, electrolytic colouring processes have been combined with conventional dyeings2. The work is anodised normally to the required film thickness, electrolyticallycoloured in a cobalt- or tin-based electrolyte to a light bronze shade, and then overdyed in an appropriate dyestuff to give muted shades of red, blue, yellow or brown. Again the main application is architectural, and the finishes have good light fastness and durability. An alternative approach to widening the colour range with electrolyticcolouring has been the development of finishes based on optical interference effectsj3, whereby quite different colours can be produced in the same electrolyte. An intermediate treatment in a phosphoric acid anodising electrolyte is normally required, between anodising and electrolytic colouring, to produce these effects. With the increasing use of colour anodised finishes, sealing quality has become very important, and seal quality tests and standards have all improved. Sealing smut is more visible on coloured than on clear anodised surfaces, and it has become common practice to try to eliminate this chemically, rather than removing it by hand wiping. Approaches to this include dipping in mineral acids after sealings4,and adding surface active agents which prevent smut f ~ r r n i n g ’ ~ . ~ ~ . Sealing is normally carried out in boiling water and the high energy costs involved have led to the development of alternative, lower-energy methods. Approaches have included the use of boehmite accelerators such as triethanolamines to shorten the sealing time”, and the use of so-called ‘cold’ sealing systems. These latter approaches have mainly been developed in Italy5’, and are based on the use of nickel salts in the presence of fluorides. They are used at a temperature of about 3OoC for a time of 15 min, and are claimed to give good corrosion resistance.
COATINGS PRODUCED BY ANODIC OXIDATlON
15: 19
Mechanism of Anodising
The development of sophisticated electron-optical techniques now allows the direct observation of the barrier layer and the pore structure of all types of anodic coating. Much of the most relevant work has been carried out at the University of Manchester Institute of Science and Technology, starting with the work of O'Sullivan and Woods9,and most recently summarised by Thompson and Wood@. The very early stages of pore growth have been extensively studied, and the importance of surface topography and flaw sites in the pre-existing oxide established. Anion incorporation in the film is another important factor affecting film characteristics, and it has been shown that distribution of the anion within the cell wall structure varies from one electrolyte to another. The mechanism of colouring with integral colour finishes has been shown to depend on the presence of free metallic aluminium in the film, as well as on the inclusion of intermetallic constituents6'. With electrolytic colouring processes, colour is produced by light scattering effects, with the tiny metallic deposits within individual pores acting as light scattering centres6*. Distribution of metal in the pores varies from one electrolyte to another, and this can affect the corrosion resistance of the final The mechanism of sealing has been shown to involve an initial dissolution and reprecipitation of hydrated aluminium oxide on the pore walls, pseudoboehmite gel formation within the pores, and conversion of this to crystalline boehmite at the film surfaceM. The presence of an intermediate layer close to the film surface, in which the identity of the original pores has been lost, has also been r e c ~ g n i s e d ~ ~ . Properties of Coatings
The hardness and abrasion resistance of anodic coatings have never been easy properties to measure, but the development of a British Standard on hard anodising& has made this essential. Film hardness is best measured by making microhardness indents on a cross-section of a film67*68, but a minimum film thickness of 25pm is required. For abrasion resistance which moves measurements, a test based on a loaded abrasive backwards and forwards over the film surface, has improved the sensitivity of such measurements. Corrosion Resistance
Tests for quality of sealing of anodic coatings have become internationally standardised. They include dye spot tests with prior acid treatment of the surface (IS0 2143:1981 and BS 6161:Part 5:1982), measurement of admittance or impedance (IS0 2931:1983 and BS 6161:Part 6:1984), or measurement of weight loss after acid immersion (IS0 3210:1983 and BS 6161:Part 3:1984, and I S 0 2932:1981 and BS 6161:Part 4:1981). Of these the chromic-phosphoric acid immersion test (IS0 3210) has become the generally accepted reference test.
15 :20
COATINGS PRODUCED BY ANODIC OXIDATION
The recent revision of the main anodising standard (BS 1615:1987) has changed it from a ‘specification’ to a ‘method for specifying’, but it provides all the information necessary to write an appropriate specification for any anodised product. The atmospheric corrosion performance of the newer colour anodised finishes is of interest, and several authors have reported Longterm weathering of dyed finishes has also been described and this has led to the recommendation of a limited range of special dyes for architectural application^^^. Good performance of the combined anodised and electrophoretically deposited clear lacquered finishes, now used very widely in Japan, has also been together with details of the vertical lines used to produce them7’.
P.G. SHEASBY B. A. SCOTT REFERENCES
I . Young, L.. Anodic Oxide Film. Academic Press, New York (1%1) 2. Schenk, M., WerkstoffAluminiumundseine AnodkcheOxydation, Francke, Berne (1948) 3. Wernick, S. and Pinner, R., The Sugace Treatment and Finishing of Aluminium and its Alloys. Robert Draper, Teddington. 3rd edn (1964) 4. Processes for the Anodic Oxidation of Aluminium and Aluminium Alloy Parts, DTD. 91W, H.M.S.O., London (1951) 5. Kaiser Aluminum Co., US Pat. 3 031 387 (7.12.59) 6. Alcoa. U S Pat. 3 227 639 (24.10.61) 7. Aluminium Laboratories Ltd., UK Pat. 970 500 (29.3.62) 8. Campbell, W. J., Conference on Anodising Aluminium, A.D.A., Nottingham. Paper I I . Sept. (1961); Csokan, P., Metalloberfiache, 19 No. 8 , 252 (1965) and Trans. Inst. Met. Fin., 41, 51 (1964) 9. Tsuji. Y.. Trans. Inst. Met. Fin.. 40, 225 (1963) IO. Scott, B. A., Trans. Inst. Met. Fin., 43, I (1%5) 11. Edwards, J . D. and Keller, F., Trans. Amer. Inst. Min. (Metall.) Engrs., 156, 288 (1944) 12. Keller, F., Hunter, M. S. and Robinson, D. L.. J. Electrochem. Soc.. 100, 411 (1953) 13. Setoh, S. and Miyata. A., Sci. Pap. Imt. Phys. Chem. Res. Tokyo, 17, 189 (1932) 14. Hunter, M. S. and Fowle, P., J. Electrochem. Soc., 101, 481 (1954) 15. Lewis, J. E. and Plumb, R. C.. J. Electrochem. SOC.,105.4% (1958) I . W.. Z. Kristallogr., 91, 65 (1935) 16. Verwey, E. . 17. Lenz, D., Aluminium, Bed., 32, 126 (1956) 18. Anodic Oxidalion Coatings on Aluminium, British Standard 1615: 1972 19. Franckenstein, G.. Ann. Phys., 26, 17 (1936) 20. van Gee!, W. Ch. and Schelen, B. J. J., Philips Res. Rep., 12, 240 (1957) 21. Hase. R.. Aluminium, Berl.. 24, 140 (1942) 22. Gwyer, A. G. C. and Pullen, N. D., Metallurgia, Munch., 21, 57 (1939) 23. Pearson, T. G. and Phillips, H. W. L., Metallurg. Rev., 2 No. 8, 348 (1957) 24. Edwards, J. D., Mon. Rev. Amer. Electropl. SOC..26, 513 (1939) 25. Guild, J., J. Sci. Inst., 17, 178 (1940) 26. Scott, B. A., J. Sci. Inst., 37, 435 (1960) 27. Stickley, G. W. and Howell. F. M., Proc. Amer. SOC. Test. Mat.. 50, 735 (1950) 28. Vevers. H.H..Conference on Anodising - Aluminium. A.D.A., Nottingham. Discussion on Section 4, sedt. (1961) 29. Neunzig, H. and Rohrig. V.. Aluminium, 38 No. 3. I50 (1%2); Sacchi. F. and Paolini, G.. Aluminio. 6. 9 (1%1) 30. Scott, B. A., Electroplating and Metal Finishing, Feb. (1%5) 31. Wood, G. C., Trans. Inst. Met. Fin., 41. 99 (1964) 32. Kape. J. M.. Metal Industry, 95 No. 6, 1 I5 (1959) 33. ASTM Method B368
COATINGS PRODUCED BY ANODIC OXlDATlON
15:21
34. Quality Laboratory and Chem. Eng. Physical Methods. MA-P, BQ7-1, Ford (USA), Feb. ( 1970) 35. Carter, V. E. and Edwards, J., Trans. Inst. Met. Fin., 43, 97 (1965)and Carter, V. E., Ibid., 45, 64 (1967) 36. Thomas, R. W.. Symposium on Protecting Aluminium, Aluminium Federation, London (1970) 37. Wood, C. C., Trans. Inst. Met. Fin., 36, 220 (1959) 38. Champion, F. A. and Spillett, E. E.,Sheet Metal Ind., 33,25 (1956) 39. Phillips, H.W. L., Institute of Metals Monograph, No. 13 (1952) 40. Lattey, R. and Neunzig, H.. Aluminium. B e d , 32, 252 (1956) 41. Neunzig, H.. Aluminium, Bed., 34. 390 (1958) 42. Liddiard, E. A. G.. Sandersen. G. and Penn, J. E., Annual Technical Conference, Institute of Metal Finishing, Brighton, 28th May (1971) 43. Kape, J. M., Trans. Inst. Met. Fin., 49, 22 (197I) 44. Anodised Wrought Aluminium for External Architectural Applications, British Standard 3987:1974 45. Qualanod, Spec@cationsfor the Quality Sign for Anodic Oxidation Coatingson Wrought Aluminium for Architectural Purposes, Zurich ( I 983) 46. Carter, V . E., Trans. Inst. Met. Fin., 55, 9 (1977) 47. Architectural Anodising: Sulphuric Acid Anodic Film Quality, British Anodising Association (1981) 48. Sheasby, P. G. and Cooke. W. E., Trans. Inst. Met. Fin., 52, 103 (1974) 49. Short, E.P., Fern, D. and Kellermann, W.M., Paper 830389,SAE Conference, Detroit (1983) 50. Brace, A. W. and Sheasby, P. G., The Technology of Anodising Aluminium, Technology Ltd., UK (1979) 51. John, S., Balasubramanium. V. and Shenoi, B. A.. Fin. Ind., 2, 32 (1978) 52. Grossman. H.and Speier, C.Th., Aluminium, 55, 141 (1979) 53. Sheasby, P. G., Patrie, J., Badia, M. and Cheetham, G . , Trans. Inst. Met. Fin., 58, 41 ( 1980) 54. Aluminium Co. of America, US Patent 3 822 156 (2.7.74) 55. Gohausen, H.J., Gulvanotechnik. 69. 893 (1978) 56. Speiser, C.Th., Afuminium, 59, E350 (1983) 57. O h Mathieson Chemical Corp., US Patent 3 365 377 (23.1.68) 58. Strazzi, E., Alluminio, 50, 4% (1981) 59. O’Sullivan. J. P. and Wood, G. C., Proc. Royal Soc.. A317, 511 (1970) 60. Thompson, G . E.and Wood, G. C.. ‘Anodic Films on Aluminium’, in Corrosion: Aqueous Processes and Passive Films, by J . C. Scully (ed.), Academic Press (1983) 61. Wefers, K. and Evans, W. T., Plating ond Surf. Fin., 62, 951 (1975) 62. Goad, D. G. W. and Moskovits, M., J. Appl. Phys., 49,2929 (1978) 63. Sheasby, P. G., Paper presented at Aluminum Finishing Seminar, St. Louis (1982) 64. Wefers, K.. Aluminium. 49, 553 (1973) 65. Thompson, G. E.,Furneaux. R. C. and Wood, G.C.. Trans. Inst. Met. Fin., 53.97 (1975) 66. Hard Anodic Oxide Coatings on Aluminium for Engineering Purposes, British Standard 5599:1978 67. Vickersand Knoop Micro Hardness Tests, British Standard 541 1:Part 61981 68. Thomas, R. W.. Trans. Inst. Met. Fin.. 59. 97 (1981) 69. Gohausen, H.J., Trans. Inst. Met. Fin., 56. 57 (1978) 70. Faller, F. E.,Aluminium, 58, E8 (1982) 71. Knutsson, L. and Dahlberg, K., Trans. Inst. Met. Fin., 54, 53 (1976) 72. Patrie, J., Trans. Inst. Met. Fin., 53, 28 (1975) 73. Speiser, C.Th. and Schenkel. H.. Aluminium. 50. 159 (1974) 74. Patrie, J.. Revue de L’Aluminium, 448. 77 (1976) 75. Shibata, K., Light Metal Age, 41, 22 (1983)
15.2
Phosphate Coatings
Introduction The use of phosphate coatings for protecting steel surfaces has been known for over 60 years, and during this period commercial utilisation has steadily increased until today the greater part of the world production of motorcars, bicycles, refrigerators, washing machines, office furniture, etc. is treated in this way. By far the greatest use of phosphate coatings is as a base for paint, although other important applications are in conjunction with oil, grease, wax and spirit stains to provide a corrosion-resistant finish, with soaps to assist the drawing and pressing of steel, and with lubricating oil to decrease the wear and fretting of sliding parts such as piston rings, tappets and gears.
Applications Phosphate treatments are readily adaptable to production requirements for articles of all sizes, and for large or small numbers. Economical processing can be achieved, for example, by treating thirty car bodies per hour in a conveyorised spray or immersion plant, or by immersion treatment of small clips and brackets. Mild steel sheet is the material most frequently subjected to phosphate treatment, but a great variety of other ferrous surfaces is also processed. Examples include cast-iron plates and piston rings, alloy steel gears, high-carbon steel cutting tools, case-hardened components, steel springs and wire, powdered iron bushes and gears, etc. Phosphate treatments designed for steel can also be used for the simultaneous treatment of zinc die-castings, hot-dipped zinc, zinc-plated and cadmium-plated articles, but if there is a large quantity of these non-ferrous articles it is more economical to phosphate them without the steel. Phosphate solutions containing fluorides are used for processing steel, zinc and aluminium when assembled together, but chromate solutions are generally preferred when aluminium is treated alone. The increasing use of cathodic electrophoretic painting on steel, however, has led to a reassessment of the basic processes and formulations that might be most effective.
Methods The usual method of applying phosphate coatings is by immersion, using a sequence of tanks which includes degreasing and phosphating stages, with 15:22
I5 :23 their respective rinses. The treatment time ranges from 3 to 5 min for thin zinc phosphate coatings up to 30 to 60 min for thick zinc, iron, or manganese phosphate coatings. The accelerated zinc phosphate processes lend themselves to application by power spray, and the processing time may then be reduced t o 1 min or less. Power spray application is particularly advantageous for mass production articles such as motorcars and refrigerators, as the conveyor can run straight through the spray tunnel, which incorporates degreasing, rinsing, phosphating, rinsing and drying stages. Flow-coating and hand spray-gun application is sometimes employed where a relatively small number of large articles has to be phosphated. PHOSPHATE COATINGS
Mechanism of Phosphate Coating Formation All conventional phosphate coating processes are based on dilute phosphoric acid solutions of iron, manganese and zinc primary phosphates either separately or in combination. The free phosphoric acid in these solutions reacts with the iron surface undergoing treatment in the following manner ’: Fe 2H,PO, Fe(H,PO,), + H, . .(15.1) thus producing soluble primary ferrous phosphate and liberating hydrogen. Local depletion of phosphoric acid occurs at the metal/solution interface. As the primary phosphates of iron, manganese and zinc dissociate readily in aqueous solution, the following reactions take place: Me ( H2PO,) e MeHPO, H, PO, . . .(15.2) . . .(15.3) 3MeHP0, Me, (PO,), H3P0, 3Me(H2P0,), Me,(PO,), 4H,PO, . . .(15.4)
+
.
-+
*
+ + +
The neutralisation of free phosphoric acid by reaction 15.1 alters the position of equilibrium of equations 15.2, 15.3 and 15.4 towards the right and thereby leads t o the deposition of the sparingly soluble secondary phosphates and insoluble tertiary phosphates on the metal surface. As reaction 15.1 takes place even when the phosphating solution contains zinc or manganese phosphate with little or no dissolved iron, it will be seen that the simple or ‘unaccelerated’ phosphate treatment gives coatings which always contain ferrous phosphate derived from the steel parts being processed. After prolonged use, a manganese phosphate bath often contains more iron in solution than manganese and produces coatings with an iron content two or three times that of manganese. The relation between free phosphoric acid content and total phosphate content in a processing bath, whether based on iron, manganese or zinc, is very important; this relation is generally referred to as the acid ratio. An excess of free acid will retard the dissociation of the primary and secondary phosphates and hinder the deposition of the tertiary phosphate coating; sometimes excessive loss of metal takes place and the coating is loose and powdery. When the free acid content is too low, dissociation of phosphates (equations 15.2, 15.3 and 15.4) takes place in the solution as well as at the metaVsolution interface and leads to precipitation of insoluble phosphates as sludge. The free acid content is usually determined by titrating with sodium
15 :24
PHOSPHATE COATINGS
hydroxide to methyl orange end point, and the total phosphate by titration with sodium hydroxide to phenolphthalein end point. Using this test, nonaccelerated processes operated near boiling generally work best with a freeacid titration between 12-5 and 15% of the total acid titration. A zinc phosphate solution tends to produce coatings more quickly than iron or manganese phosphate solutions, and dissociation of primary zinc phosphate proceeds rapidly through reaction 15.2 to 15.3 or directly to tertiary zinc phosphate via reaction 15.4. Even so, a processing time of 30 min is usual with the solution near boiling. Another factor in the initiation of phosphate coating reaction is the presence in the processing solution of tertiary phosphate, either as a colloidal suspension or as fine particles'. This effect is most apparent in zinc phosphate solutions, which produce good coatings only when turbid. The tertiary zinc phosphate particles can be present to a greater extent in cold processing solutions and act as nuclei for the growth of many small crystals on the metal surface, thereby promoting the formation of smoother coatings. Similarly, the ferric phosphate sludge formed during the processing of steel in a zinc phosphate solution can play a useful part in coating formation3. The solubility of ferric phosphate is greater at room temperature than at elevated temperatures, and is increased by the presence of nitrate accelerators. To allow for saturation at all temperatures it is desirable always to retain some sludge in the processing bath. Coatings with optimum corrosion resistance are produced when the temperature of the bath is rising and causing super-saturation of ferric phosphate. With zinc/iron/phosphate/nitrate baths the iron content of the coating comes predominantly from the processing solution and very little from the surface being treated4. This greatly diminished attack on the metal surface by accelerated baths has a slight disadvantage in practice in that rust is not removed, whereas the vigorous reaction of the non-accelerated processes does remove light rust deposits. The solution of iron represented in equation 15.1 takes place at local anodes of the steel being processed, while discharge of hydrogen ions with simultaneous dissociation and deposition of the metal phosphate takes place at the local cathodes'. Thus factors which favour the cathode process will accelerate coating formation and conversely factors favouring the dissolution of iron will hinder the process. Cathodic treatment in a phosphating solution exerts an acceleratingaction as the reaction at all cathodic areas is assisted and the formation of a phosphate layer is speeded accordingly. Conversely, anodic treatment favours only the solution of iron at local anodes and hinders phosphate coating formation. An oxidisingagent acts as an accelerator by depolarisation of the cathodes, raising the density of local currents so that rapid anodic passivation of active iron in the pores takes place. This inactivation of local anodes favours the progression of the cathodic process. The accelerating effect of alternating current is explained by the practical observation that the cathodic impulse acting protectively greatly exceeds in its effect the anodic impulse which dissolves iron. In a similar manner the electrolytic pickling of iron with alternating current can dissolve iron at a slower rate than when no current is used.
15 :25
PHOSPHATE COATINGS
Reducing agents have the same ultimate effect as cathodic depolarisation in that they convert anodic regions to cathodic and increase the ratio of cathodic to anodic areas. Nitrogenous organic components such as toluidine, quinoline, aniline, etc. all act as inhibitors to the anodic reaction between metal and acid and thereby favour the cathodic reaction and accelerate the process.
Accelerators The majority of phosphate processes in use today are 'accelerated' to obtain shorter treatment times and lower processing temperatures. The most common mode of acceleration is by the addition of oxidising agents such as nitrate, nitrite, chlorate and hydrogen peroxide. By this means, a processing time of 1 to 5 min can be obtained at temperatures of 43-71 "C. The resultant coatings are much smoother and thinner than those from unaccelerated processes, and, while the corrosion resistance is lower, they cause less reduction of paint gloss and are more suited to mass-production requirements. Table 15.8 Amount and composition of the gases evolved on phosphating of I m z of sheet metal for deep drawing
Manganese phosphate Zinc phosphate Manganese phosphate (accelerated with nitrate) Zinc phosphate (accelerated with nitrate) Zinc phosphate containing 1.5-2 g/l iron (accelerated with nitrate)
-
30 40
60 30
7000 2540
87.5 II'4t 1 . 1 92.7 6.4t 0.9
30
I5
3500
84.6 9.1
1.3
5.0
70
5
78
16.7 75.3
8.0
-
70
5
85
32.1 57.0
1.6
9.3
A measure of the total of a phosphating solution. as indicated by the number of ml of 0 . I N sodium hydroxide ( 4 . 0 g/l) needed to neutralise IO ml of the phosphating solution to phenolphthalein. t Presumably from nitrides present in the steel.
The presence of nitrate as acelerator has a pronounced effect on the amount and composition of gas evolved from the work being treated' (Table 15.8). It will be observed that hydrogen evolution drops to a very low figure with the zinchitrate baths. The formation of nitrite arises from decomposition of nitrate by reaction with primary ferrous phosphate to form ferric phosphate: 2Fe2+ + NO;
+ 3H+
2Fe3++ HNO, + H,O In an acid solution sodium nitrite acts as a strong oxidising agent by the following reaction: 2NaN0,
+ 2H,PO,
+
-,2NaH2P0, + H 2 0 + N 2 0 + 2 ( 0 )
A slight degree of acceleration can be obtained by introducing traces of metals which are more noble than iron, for example nickel, copper, cobalt, silver and mercury. These metals are deposited electrochemically over the
15 :26
PHOSPHATE COATINGS
iron surface undergoing treatment, thereby providing more active cathodic centres and promoting phosphate deposition. This method of acceleration has the disadvantage of leaving minute particles of the noble metal in the coating, and, in the case of copper, this can seriously inhibit the drying of some types of paint coatings. Copper also forms local cells with the iron and so reduces corrosion resistance. Acceleration by addition of reducing agents, organic compounds, or by application of a cathodic or alternating current, is not nowadays used to any great extent. This situation may change if ways of controlling the P / (P + H)ratio become important (see later).
Nature of Coatings Effect of Metel Surface
The state of the metal surface has a pronounced effect on the texture and nature of phosphate coating produced by orthodox processes. Heavily worked surfaces tend to be less reactive and lead to patchy coatings. Grit blasting greatly simplifies treatment and gives uniform phosphate coatings. Accidental contamination of sheet steel with lead has been shown to have an adverse effect on the corrosion resistance and durability of phosphate coatings and paint’. Cleaning operations which make use of strong acids or strong alkalis tend to lead to the formation of excessively large phosphate crystals which do not completely cover the metal surface and therefore show inferior corrosion resistance; this is particularly serious if rinsing is inadequate between the preparatory treatment and the phosphating. Adherent dust particles can also lead to the formation of relatively large phosphate crystals, and surfaces which have been wiped beforehand show much smoother and more uniform phosphate coatings. On the other hand, the provision of vast numbers of minute nuclei assists the phosphate coating reaction to start at a multitude of centres, resulting in a finely crystalline coating. This effect can be obtained chemically by a predip in a solution of sodium phosphate containing minutely dispersed traces of titanium or zirconium salts6 or in weak solution of oxalic acid. This type of pre-dip entirely eliminates any coarsening effect due to previous treatment in strong alkalis or acids. Effect of Phosphate Solution
Improved nucleation within the phosphate solution itself can produce smoother coatings without the necessity of recourse to preliminary chemical treatment. This may be accomplished by introducing into the phosphating bath the sparingly soluble phosphates of the alkaline earth metals or condensed phosphates such as sodium hexametaphosphate or sodium tripolyphosphate. Such modified phosphating baths produce smoother coatings than orthodox baths and are very much less sensitive to cleaning procedures. Very thin coatings of ‘iron phosphate’ can be produced by treatment with solutions of alkali metal phosphate. These serve a useful purpose for the
PHOSPHATE COATINGS
15 :27
treatment of office furniture, toys, etc. where a high degree of protection is not required, and also as a base for phenolic varnishes, or resin varnishes requiring stoving at over 204°C. The coating is of heterogeneous nature and contains less than 35% iron phosphate (FeP0,.2H20) with the remainder probably yFe,O:. Thin phosphate coatings can be formed by application of phosphoric acid solution alone, Le. not containing metallic phosphates, to a steel surface, sufficient time being allowed after application to enable complete reaction to take place. In this way a thin film of iron phosphate can be formed. In practice it is difficult to obtain complete conversion and the remaining traces of phosphoric acid can cause blistering of paint coatings. This effect may be insignificant on rough, absorbent steel surfaces, e.g. ship’s plating, where heavy coats of absorbent paint are applied, and under these circumstances the treatment can enhance the corrosion resistance of the finishing system. Chemical Nature of Coatings
The simplest phosphate coating, that formed from solution containing only ferrous phosphate and phosphoric acid, consists of dark grey to black crystals of tertiary ferrous phosphate, Fe, (PO,), , and secondary ferrous phosphate, FeHPO, , with a small proportion of tertiary ferric phosphate, FePO, . Coatings formed from manganese phosphate solutions consist of tertiary manganese phosphate, and those from zinc phosphating solutions consist of tertiary zinc phosphate. With both the manganese and zinc type of coating, insoluble secondary and tertiary iron phosphates, derived from iron present in the bath, may be present in solid solution. Iron from the surface being treated can also be present in the coating, particularly at the metaVphosphate interface. The PO:- content of coatings may vary from 33 to 50%, whereas the theoretical PO:- content is lowest, at 41%, in Zn,(P0,),*4H20 and highest, at 63%, in FePO,. Crystal Structure
It has been suggested that the zinc phosphate coating has the composition Zn,(PO,), .Zn(OH), , but X-ray diffraction studies have given very good correlation between Zn,(PO,), * 4H20 and the zinc phosphate coatings on steel‘. Zn,(PO,), -4H,O appears in three crystal forms, a-hopeite (rhombic plates), p-hopeite (rhombic crystals), and p-hopeite (triclinic crystals). Their transition points are at 105, 140 and 163°C respectively. It has been observed’ that zinc phosphate coatings heated in the absence of air lose their corrosion resistance at between 150 and 163°C. Manganese phosphate coatings heated in the absence of air lose their corrosion resistance at between 200 and 218°C. At these temperatures, between 75 and 80% of the water of hydration is lost and it is assumed that this results in a volume decrease of the coating which causes voids and thereby lowers the corrosion resistance. Fig. 15.4 shows the loss of water of hydration from zinc, iron and iron-manganese phosphate coatings.
Table 15.9 &OCf?SS*
Main cation in phosphate bath Method of application Duration of treatment (min) Change in weight on hosphating (g/mZ) Coating weight (g/m ) PO$- (g/m2) Moisture (mg/m2) PO:- content of coating (%) Moisture content of coating (%) Hygroscopicity of coating (@lo) Absorption value (diacetone alcohol (g/m *)
P
' The letters u
d for designation indicate proprietary process. Data reproduced counesy J.I.S.I.. 170. I I (1952).
Analytical tests o n industrial phosphate coatings
P
S
T
Q
V
R
Fe
Mn
Zn
Zn
Zn
Zn
Immersion
Immersion
Spraying
Immersion
Immersion
Immersion
15
30
1-5
4
5
12
-26- 1 14.2 7.0 81.5 49.0 0-6
-26.4 21.2 8.9 76.1 42.0 0.4 0.2 10.9
3.37 5.43 2.07 396.6 38.0 6.9 1.3 13.04
1 *63 3.48 I a20 173.9 34.0 5.0 1 -0 10.87
5.87 12-28 4.46 771.7 36.0 6.4 1.5 11-96
0.3 11.4
2-61 4.46
1.96 152-2
44.0 3.4 1-2 10.9
PHOSPHATE COATINGS
-s1 - 6 I
P W %
c)
$
ri //-J
0: 4
u
z
* *3
15 :29
2-
C
0
HEATING TEMPERATURE ("C)
Fig. 15.4 Effect of heating o n phosphate coatings for 16 h at various temperatures, showing loss of water of hydration. Curve A zinc phosphate, B iron phosphate and C iron manganese phosphate (courtesy J.I.S.I., 170, 11 (1952))
The heating of phosphate coatings in the absence of air provides conditions similar to those prevailing during the stoving of paint on phosphated articles, but in general the paint stoving temperatures and times are well below those at which damage to zinc phosphate coatings takes place. The loss of water from conventional zinc and managanese phosphate coatings heated in air is from 10 to 20% higher than the loss on heating in the absence of air. It is thought that this greater loss may be due to oxidation of the iron phosphate present in the coatings. The most important uses for phosphate coatings entail sealing with oil or paint and it is therefore of interest to study absorption values. Table 15.9 compares the absorption of diacetone alcohol into coatings of widely differing thicknesses and composition; despite these differences, values of 10.812-9g/m2 are obtained throughout. It is therefore evident that absorption is predominantly a surface effect and not appreciably influenced by coating thickness.
Rinsing After phosphating, thorough rinsing with water is necessary in order to remove soluble salts which would otherwise tend to promote blistering under a paint film. Care should also be taken to ensure that the water supply itself is sufficiently free from harmful salts. Experience has shown that a water supply is potentially injurious if it exceeds any one of the three foliowing limits: 1. 70 p.p.m. total chlorides and sulphates (calculated as C12. 200 p.p.m. total alkalinity (calculated as CaCO,). 3. Maximum of 225 p.p-m. of (1) and (2) together.
+ SO:-).
15 :30
PHOSPHATE COATINGS
Improved corrosion resistance and reduced tendency to blistering can be obtained by treating the final rinse with chromic acid, or preferably with phosphoric and chromic acids combined. Normally a total acid content of 0-05% is used. Higher concentrations of chromic acid in the rinse will increase corrosion resistance, partly by passivation of any bare metal or pores in the phosphate coating, but mainly by absorption into the coating I O p 1 ' . The corrosion resistance rises steadily with increase of chromic acid strength, but above 0.2% chromic acid the phosphate coating tends to dissolve. Absorbed chromic acid is removed only with difficulty by hot or cold water rinsing and is not affected by trichlorethylene vapour treatment, Advantage may be taken of the higher corrosion resistance given by chromic acid, whether or not the metal is to be painted, but care must be taken with white finishing paints, as chromic acid residues may cause local yellowing of the paint in the form of streaks. British Standard requirements for chromic rinsing are shown in Table 15.10. Table 15.10 Concentration of chromate solution (BS 3189: 1973) ~~
Nature of phosphate coating and of sealing cout
I . Phosphate coatings of all classes to be sealed with paint, varnish or lacquer 2. Zinc phosphate coatings to be sealed with oil or grease 3. Manganese and/or iron phosphate coatings to be sealed with oil or grease
Concentrotion in terms of Crof (TO)
Min.
Max
0.0125
0-05
0.0125
0.25
0.0125
0.5
' The substitutionof an equal weight of phosphoric acid for up to one half of the chromic acid is permissible.
In recent years there has been a great increase in the use of demineralised water for rinsing, especially before electrophoretic painting. The demineralised water is generally applied by misting jets at the end of all other pretreatment stages and allowed to flow back into the last rinse tank. In certain cases rinsing may be dispensed with after non-accelerated phosphate treatment, but blistering of paint due to local concentration of solution in seams and crevices may occur. Rinsing is generally applied, regardless of the type of phosphate process employed Recent trends are away from rinses containing Cr(v1) and more towards those containing Cr(rrr) for health and safety reasons.
Corrosion Protection The corrosion protection provided by phosphate coatings without a sealing treatment is of a low order; their value when sealed is considerably greater. Unsealed corrosion tests are therefore of little value except perhaps for studying porosity or efficiency of coatings destined to be sealed only with oil. Mention has been made of the necessity for controlling the acid ratio of phosphating baths, particularly those of iron, manganese and zinc operating
15:31
PHOSPHATE COATINGS
Table 15.11 Typical phosphate coating processes Phosphate coating solution
Accelerator
Immersion time
(min)
lron hodmanganese Manganese Zinc Zinc Sodium/ammonium
None None Nitrate Nitrate Nitratehitrite or chlorate None
30 30 15 15
3 1-2 (spray)
TYP of coating
Heavy Heavy Heavy Medium Light Very light
Coating weight (g/m ')
10'87-32.61 10.87-32'61 8.70-32.61 3.26-32.61 1.09- 6.52 0.22- 0.65
near boiling point to produce heavy coatings. At a 'pointage' (see Table 15.8) of 30 in these solutions the free acidity is usually maintained between 1 2 - 5 and 15%; above this figure coatings with progressively lower corrosion resistance are obtained. Heavy phosphate coatings do not necessarily have better corrosion resistance than lighter coatings. Even with a single process, e.g. zinc/iron/ phosphatehitrate, no consistent relationship has been found between corrosion resistance and either coating weight or weight of metal dissolved. Phosphate processes containing little or no oxidising agent and based on manganese or zinc tend to accumulate iron in solution from the work being processed. With a manganese content of from 0.2% to 0.5% it is best to control the iron at from 0.2 to 0.4%; a higher iron content reduces the corrosion resistance and may lead to the formation of thin powdery coatings, while a lower iron content gives soft coatings. Similarly, a zinc process operates best with 0.15-0.5% zinc and 0.4-0-5% iron. Again, with a higher iron content corrosion resistance falls off and powdery coatings may be formed, and soft coatings result from a lower iron content. Jaudon13tested phosphate coatings with and without paint and found the salt-spray resistance, as judged by the first appearance of rust, to be as follows: Bare steel Few minutes Phosphated steel 12 h average Painted steel 150h 300 h Phosphated and painted steel Table 15.12 Typical uses of phosphate coatings on steel Coating weight
(g/m
21.74-32.61 10-87-21.74 5.43-10.87 2.17- 2-72 1.63- 2.17 0.22- 0-65
For corrosion resistance
For wear prevention and metal forming
-
Critical cold extrusion Normal cold extrusion 'Running in' treatment for piston rings, gears and tappets Wire and tube drawing Sheet steel pressing Light metal pressing
Military equipment, etc. requiring oil or grease finish Nuts, bolts, clips, brackets Cars, refrigerators, washing machines Steel drums, bicycles, office machinery Toys, office furniture Strip steel, for painting and forming
15 :32
PHOSPHATE COATINGS
Within broad limits, phosphate processes can be classified according to the main metallic radical of the processing solution and the type of accelerator used; typical processes are given in Table 15.1 1. The selection of process and of coating weight is mainly dependent on the end-use of the article being processed; the general requirements for corrosion resistance and wear prevention are given in Table 15.12. (See later for comments on P / ( P + H) ratio.)
Testing Heavy phosphate coatings are generally used as protection against corrosion in conjunction with a sealing film of oil or grease. The porosity or free pore area of these coatings should be kept to a minimum. MachuI4 devised a method of examination based on the quantity of electricity necessary to effect passivation of the bare steel and used this to determine the ‘free pore area’ which, in the phosphate coatings tested, varied from 0.27 to 63%. Attempts to use this method for the evaluation of the more widely used thin zinc phosphate coatings have not been successful, as these coatings show a porosity of less than 1- 5 % and the technique of measurement was not adequate for this range”. A method for making rapid measurements of the electricai resistance of phosphate coatings has been described by Scott and ShreirI6. Akimov and Ulyanov” proposed an acidified copper sulphate spot test for assessing the corrosion resistance of phosphated articles by timing the colour change from blue to light green, yellow or red owing to the precipitation of copper. The assumption was that the longer this change took to occur, the higher the corrosion resistance. The test has been thoroughly examined in this country and rejected because of variation in results and poor correlation with corrosion resistance. Sherlock and Shreir consider that the hydrogen permeation technique could provide a useful means of studying and evaluating the porosity of phosphate coatings. The most widely used accelerated tests are based on salt spray, and are covered by several Government Specifications. BS 1391:1952’* (recently withdrawn) gives details of a hand-atomiser salt-spray test which employs synthetic sea-water and also of a sulphur-dioxide corrosion test. A continuous salt-spray test is described in ASTM B 117-61 and BS AU 148: Part 2( 1969). Phosphate coatings are occasionally tested by continuous salt spray without a sealing oil film and are expected to withstand one or two hours spray without showing signs of rust; the value of such a test in cases where sealing is normally undertaken is extremely doubtful. The main value of salt-spray tests is in the evaluation of the effectiveness of phosphate coatings in restricting the spread of rust from scratches or other points of damage in a paint film. This feature is of particular interest to the motorcar industry, as vehicles are often exposed to marine atmospheres and to moisture and salt when the latter is used to disperse ice and frost from road surfaces. Great care is needed in the interpretation of a salt-spray test, as it has been found to favour thin iron phosphate coatings more than is justified by experience with natural weathering. In the motorcar industry the present custom is to use zinc phosphate coatings on the car bodies and all other parts exposed to the outside atmosphere. Humidity tests are generally of more practical use than salt-spray tests, particularly where painting is employed, as the thoroughness of rinsing may be checked by this means. The use of contaminated water can leave
15:33
PHOSPHATE COATINGS
water-soluble salts in the phosphate coating and lead to blistering of the paint film under humid conditions, as paint films are permeable to water vapour. Immersion in water, or subjection to high humidity in a closed cabinet, will generally show any defects of this kind within a few days. The British Automobile Standard specifies freedom from blistering after 200 h in distilled water at 100°F (38°C). Table 15.13 Weights of phosphate coatings (Defence Specification DEF-29) ~
cia I I1
111
TYP
Minimum coating weight (R/m ')
Mn or Fe Zn. etc.
7.6 4.3 1.6 0.5*
-
' A lower range of 0.5 10 1.6 s/m2 may be permitted where thin sections are lo k fabricated or formed after the application of paint. varnish or lacquer.
The texture or crystal size of phosphate coatings can conveniently be recorded by making an impression on clear cellulose tape moistened with acetone. Uniformity of crystal size is of importance for coatings which are to resist wear and assist metal working. Surface roughness may also be studied by means of a 'Talysurf meter. Phosphate coating weight determinations are generally performed by dissolving the coating from weighed panels by immersion in a solution of 20 g/l of antimony trioxide in concentrated hydrochloric acid at a temperature of 13-2l0Cl9. The solution is used once only. Thin iron or zinc phosphate coatings can be removed for weight determination by immersion in S% chromic acid solution at 70"C, but this solution should also be used once only, as the presence of more than a trace of phosphate leads to pitting of the steel and false results. Zinc phosphate coatings can be removed by immersion in 10% sodium hydroxide at boiIing temperature, aided by rubbing during rinsing. The Ministry of Defence requirements for phosphating are covered by Defence Specification DEF-29 and are divided into three classes as shown in Table 15.14 Salt-spray resistance
of phosphate coatings under various finishes (Defence Specification DEF-29) Finish
Oil Shellac Lanolin Air-drying paint Stoving lacquer Stoving paint
Period of test (days)
I 1 1
3 6 6
15 :34
PHOSPHATE COATINGS
Table 15.13. This specification follows good industrial practice, with additional safeguards in rinsing t o remove residues to treatment solutions. Nonaccelerated treatments must be followed by a single rinse which may contain chromate; accelerated treatments must be followed by three rinses -cold water, hot water and a final chromate rinse. Table 15.14 shows the salt-spray test requirements for phosphate coatings with various finishes without formation of rust; the paints and lacquer have the additional requirement that no rust shall be visible beyond 0.2 in (5 mm) from the deliberate scratches and no blistering, llfting or flaking beyond 0.05 in (1 *27mm) from the original boundaries of the scratches. The American Aeronautical Material Specification AMS 2480 A calls for 150h salt-spray test without rusting extending more than 0.125in (3- 175 mm) on either side of scratch marks, using a black enamel finish for the phosphate coating. Table 15.15 Weights of phosphate coatings (BS 3189:1973) Coating weight (g/m Closs of phosphate process
Min.
Max.
7.61 7-61
-
A I . Heavyweight (Mn or Fe) A 2. Heavyweight (Zn) B Medium weight (Zn.etc.) C Lightweight (Zn, etc.)
4.34 1.09
D
0.33
Extra lightweight (Fe)
-
4.34 1-09
British Standard 3189: 1973l9 contains valuable iRformation on the operation of phosphate processes t o obtain optimum results, and on the testing of phosphate coatings. The classification of coatings according to composition and weight is shown in Table 15.15. Recommendations for chromate rinsing are given in Table 15.10. The inspection and testing includes determination of coating weight, freedom from corrosive residues as shown by a humidity test, and resistance to corrosion by salt spray. British Standard 5493: 197720is also a valuable source of information.
The PAP + HI Ratio In recent years considerable interest has been focused on the so-called P / ( P + H) ratio in predicting the performance of phosphated steel when coated with cathodic electroprimer and paint2’-26. In this context, P is defined as the intensity of X-rays diffracted from the (100) plane of a 0) at an interplanar spacing d, of phosphophyllite (FeZn,(PO,), .4H2 0-884nm, and H is defined as the intensity of X-rays diffracted from the (020) plane of hopeite (Zn,(P0,),-4H20) at d = 0 . W n m . Initial work suggests that high values of this ratio (referred to as the ‘The Ratio’) are synonymous with good corrosion performance”. Later work 22-26 indicates tha the situation is much more complex than first thought and that many other factors also need to be considered such as method of application,
PHOSPHATE COATINGS
15:35
working temperature, bath chemistry and after-treatment, to name just a few. Reproducible values for The Ratio can be obtained, providing extensive multiple readings are taken in order to take into account topographical variations. Performance tests” show that although the high values of ‘The Ratio’ appear to be synonymous with good performance, this effect is masked by the use of Cr-containing after-treatments which result in superior corrosion resistance. Similarly, although dip application shows an overall superiority to sprayldip treatments, good results can be obtained with the latter. Indeed, there may be a reversal of The Ratio trend in this instanceu, i.e. performance is slightly inferior at relatively high values of The Ratio. It is also possible to find zinc phosphate coatings exhibiting good performance and high iron content even though X-ray diffraction studies may reveal no phosphophyllite present or crystalline species other than hopeitez’-26.This may be because corrosion resistance is related to a low proportion of hopeite, rather than phosphophyllite, in the coating. Other factors to be considered include the need for homogeneous phosphate layers of controlled thickness, the direct attachment of the primer to a coherent layer (primary phosphate) and the level of interlayer cohesion within the coating. Some papers” indicate that adhesion failure results from internal fracture of the phosphate coating and that it is concentrated at the junction between a primary microcrystalline or even amorphous layer close to the metal substrate and a secondary layer exhibiting relatively coarse crystallinity. The primary layer is comprised essentially of a zinc phosphate material and the Zn/P ratio in the retained primary layer after fracture is lower than that in the detached material, though close enough to be considered essentially similar. As already mentioned, acidic chromium-containing rinses for phosphate coatings considerably improve the resistance of paint to water soak and humidity testing. Some authors suggest that the main action is therefore not just the passivating of the regions of steel surface left active after the phosphating process, but could be due to action on the coating itself. If the coating is considered to be a solid alkali comprised of tertiary inorganic phosphates, then it is possible for amorphous phases containing Cr,O, or CrPO, to bind the secondary and primary layers together. Similarly improvements in cathodically electropainted systems, in terms of their resistance to water soak tests, are said%to be obtained by post-rinsing the phosphate surface with dilute acids or even alkalis. This latter effect is only obtained when at least 20-30% of the phosphate surface is actually removed. Thus in such circumstances the secondary phosphate layer is sufficiently depleted to allow the electroprimer direct access to the primary layer. These observations lend support to the notion that, although the potential for good corrosion resistance is greatest with cathodic electroprimer (compared with anodic), the risk of adhesion failure due to internal fracture of the phosphate coating is quite high. How far the formulation of a phosphating bath influences The Ratio is not entirely clear. Nitrite alone or in combination with chlorate has been the most widely used accelerator system for many years but more recently nitrite-free chlorate/organic systems have been increasingly favoured. Low zinc systems in which the bath is ‘starved’of zinc to promote a high iron content in the coating, originally introduced in Japan, have become widespread.
15 :36
PHOSPHATE COATINGS
Similarly in Japan there has been a strong move towards full dip treatment and over 50% of car body lines now employ this method. In Europe, while there are some dip-only plants, the majority of recent installations have presprays prior to the dip tank. In the USA spray-only plants still predominate. Zinc phosphate processes normally operate in the range 50-60°C. Low temperature processes operating at 25-35OC are widely used in the UK and Italy but have not been extensively adopted elsewhere. One area in which there is sometimes confusion is in appreciating exactly what The Ratio signifies. As mentioned above, this is an arbitrary ratio based on intensities of X-rays at very specific diffraction angles. Thus it can be very misleading to assume that the figures quoted are related in some way to the volume or weight fractions of actual hopeite and phosphophyllite crystals present. In extreme circumstances the occurrence of the X-ray peaks may actually move to other diffraction angles. Furthermore, if there is crystal orientation present in a sample then a wide scatter in The Ratio figures will result from alignment problems in the X-ray diffractometer. Finally, it is worth noting that the quality of the steel substrate can have an effect on the corrosion resistance promoted by any subsequent treatment by phosphating and painting. Indeed, it has been reported” that interesting results are obtained when cold-rolled steel panels, with different amounts of surface contamination, are zinc phosphated then coated with anionic or cationic electrocoat primers followed by a conventional filler-topcoat system. In salt spray, scab and filiform corrosion tests it is apparently possible to distinguish between different surface contamination levels and primer coatings. Carbonaceous residues on the steel can have a detrimental effect, and this can be confirmed in the case of anionic primer during salt spray tests. In the scab corrosion and filiform corrosion tests, however, anionic primer performance actually increases with surface contamination. It can be concluded that the steel condition and the type of coating affect the corrosion resistance of the entire system by inducing changes in the phosphate layer. With the current low level of surface contamination of commercial steels and the highly resistant modern coating formulations it is suggested” that the phosphate layer is the weakest link in the entire system.
M. 0. W. RICHARDSON R. E. SHAW
REFERENCES
.,
1. Machu. W Die Phosphatierung- Wissenschqftiiche Grundlagen und Technik. Verlag Chemie, Weinheim (1950) 2. Wusterfeld, H., Arch. Metullk., 3. 233 (1949) 3. ‘Determination of the Solubility of Ferric Phosphate in Phosphating Solutions Using Radioiron’, US Department of Commerce, Office of Technical Services, Rep. No. PB 1 1 1 , 399 (1953) 4. ‘RadiometricStudy of Phosphating Problems’. US Department of Commerce, Office of Technical Services, Rep. No. PB 111, 3% (1951) 5 . Wirshing, R. J. and McMaster, W. D.. Cunod. Point Vorn. Mag., 30, No. 5.42, 55 (1956) 6. Jernstedt, G., Chem. Engng. News, 21, 710 (1943); Trans. Electrochem. SOC., 83, 361 (1943) 7. ‘A Radiometric Study of the Iron Phosphating Process’, U S Dept. of Commerce, Office of Technical Services, Rep. No. PB 1 1 1 , 400 (1953)
PHOSPHATE COATINGS
15:37
8. ‘X-ray Diffraction Study of Zinc Phosphate Coatings on Steel’, US Department of Commerce, Office of Technical Services, Rep. No. PB I 1 1. 486 (1954) 9. Doss, J., Org. Finish., 17, 8 , 6 (1956) 10. B.S.I. Phosphate Coatings (Drafting) Panel, ‘Phosphate Coatings as a Basis for Painting Steel’, J. Iron St. Inst., 170. 10 (1952) 11. ‘Radiometric Evaluation of the Effectiveness of the Chromic Acid Rinse Treatment for Phosphated Work’, US Department of Commerce, Office of Technical Services, Rep. No. PB 111,397 (1952); ‘A Study of the Effect of Chromic Acid and Chromic-phosphoric Acid Rinse Solutions upon the Subsequently Applied Paint Coatings’. US Dept. of Commerce, Office of Technical Services, Rep. No. PB I 1 I , 578 (1954) 12. B.S.I. Phosphate Coatings (Drafting) Panel, ‘Phosphate Coatings as a Basis for Painting Steel’, J. Iron St. Inst., 170. 13 (1952) 13. Jaudon, E., feint.-Pigm.-Vernis.,25, 224 (1949) 14. Machu. W..Korros. Metallsch.. 20, 1 (1944) 15. ‘Development of Accelerated Performance Tests for Paint-Phosphate-MetaI Systems. Surface Preparation of Metals’. Aberdeen Proving Ground, USA (1955) 16. Scott, J. W. and Shreir, L. L.,Chem. and fnd. (Rev.),807 (1957) 17. Akimov, G. V. and Ulyanov, A. A., C. R. Acad. Sci. U.R.S.S., 50, 271 (1945) 17(a). Sherlock. J. C. and Shreir. L. L., Corros. Sci.. 11. 543 (1971) 18. Performance Testsfor Protective Schemes used in the Protection of Light Gauge Steel, BS 1391:1952; British Standards Institution, London 19. Phosphate Treatment of Iron and Steel for Protection aganst Corrosion, BS 3189:1973, British Standards Institution, London 20. Code of Practice for Protective Coating of Iron and Steel Structures Against Corrosion, BS 5493: 1977, British Standards Institution. London 21. Miyawaki, T.. Okita, H., Umehara, S. and Okabe. M..Proc. Interfinish.. 80,303 (1980) 22. Richardson, M. 0. W., Freeman, D. B., Brown, K. and Djaroud, N., Trans. I.M.F., 61, 155 (1983) 23. Freeman, D. B., Brown, K., Richardson, M. 0. W and Tiong, H., Finishing, 27-28, Aug. (1984) 24. Cooke, B. A., ‘Aspects of Metal Pretreatment before Painting’, Proc. IX fnt. Conf. Org. Sci. and Tech., Athens, 29-46 (1983) 25. Kojima, R., Nomura, K. and Ujahira, Y . , J. Jap. Soc. Col. Mat., 55, 6 365-373 (1982) 26. Richardson, M. 0. W. and Freeman, D. B., ‘Pretreatment and Cathodic Electropaint Performance-Use of the P/P + H Ratio, Ann. Tech. Conf. IMF, Bournemouth, April (1985) 27. Soepenberg, E. N., Vrijburg, H. G., van Ooij, W. J. and Vries, 0. T., The Influence of Steel Quality. Pretreatment and Coating Systems on the Corrosion of Automotive Steel, pp. 381-393 (1984)
15.3 Chromate Treatments Introduction The addition of chromates to many corrosive liquids reduces or prevents attack on metals, and chromates are often added to waters in contact with metals as corrosion inhibitors. Under atmospheric exposure an alternative method is used; this consists of depositing on the metal a chromate film which acts as a reservoir of soluble chromate. Although the quantity of chromate which can be held in this way at the metal surface is small, the film nevertheless improves the performance of metals with a high intrinsic corrosion resistance, e.g. cadmium, copper and some aluminium-base materials. With metals which are more liable to corrode, however, such as magnesium alloys and high-strength aluminium alloys, chromate films are used primarily for improving the adhesion of paint, their own inhibiting action making a useful contribution to the total protection. Chromate treatments can be applied to a wide range of industrial metals. They are of two broad types: (a) those which are complete in themselves and deposit substantial chromate films on the bare metal; and (b) those which are used to seal or supplement protective coatings of other types, e.g. oxide and phosphate coatings. Types of treatment for various metals are summarised in Table 15.16.
Principles of Chromate Treatment Chromate ions, when used as inhibitors in aqueous solutions, passivate by maintaining a coherent oxide film on the metal surface. Passivation is maintained even in a boiling concentrated chromic acid solution*, in which many of the oxides in bulk form are soluble. The passivity breaks down rapidly, however, once the chromate is removed. In order that a chromate film may be deposited, the passivity which develops in a solution of chromate anions alone must be broken down in solution in a controlled way. This is achieved by adding other anions, e.g. sulphate, nitrate, chloride, fluoride, as activators which attack the metal, or by electrolysis. When attack occurs, some metal is dissolved, the resulting hydrogen reduces some of the chromate ion, and a slightly soluble goldenbrown or black chromium chromate (Cr,O,- CrO, -xH,O) is formed. *Vigorous attack can occur with industrial-grade chromic acid, which can contain sulphuric acid as an impurity.
15 :38
15:39
CHROMATE TREATMENTS
This compound is deposited on the metal surface unless the solution is sufficiently acid to dissolve it as soon as it is formed. The film also usually contains the oxide of the metal being treated, together with alkali metal (when this is present in the treatment solution) perhaps in the form of a complex basic double chromate analogous to zinc yellow. Table 15.16 Summary of types of chromate treatment Type of treatment
Metal
Type of deposit
Solution, radicals ~
Aluminium and its alloys
(a) Alkaline dip
Alkaline chromate
(6) Acid dip, 1
Acid chromate/ fluoride/phosphate
Acid dip, 2 (c) Acid pickle
(d)Sealing of anodic films
Acid chromate/ fluoriddnitrate Acid chromate/ sulphate and chromate/phosphate Chromate/ dichromate in pH range 5 to 7
~
~
Oxide/hydroxide with perhaps some chromate Phosphate with perhaps some chromate Not known, but contains substantial chromate Very thin, may contain chromate Blockage of pores with hydroxide/ chromate
Cadmium and zinc
Acid dip
Acid chromate/ sulphate, sometimes with additions
Thin hydrated chromium chromate
Copper
Acid pickle
Acid chromate/ sulphate
Very thin, may contain chromate
Iron and steel
Rinse after phosphate treatment
Dilute acid chromate with or without phosphate
Probably some basic chromate left in the phosphate coating
Magnesium alloys
(a) Strongly acid
Acid chromate/ nitrate Chrornate/sulphate with buffer, pH 4 to 5 , also Dow No. 7 Chromate/sulphate, pH 6, at boiling or with anodic current Neutral chromate
Thin chromium chromate Thick chromium chromate
( 6 ) Anodic
Chromate and complexing salt Alkaline chromate
Very thin, may contain chromate Very thin, may contain chromate
Dip
Alkaline chromate
Very thin, may contain chromate
dip (b) Moderately acid dip (c) Slightly acid
dip
(d) Sealing of anodic films Silver
(0)
Dip
Thick chromium chromate Chromate retained by oxide, etc. coating
~
Tin
15:40
CHROMATE TREATMENTS
The stability of the natural oxide film reinforced by the chromate ion determines the conditions of pH, ratio of activating anion to chromate, and temperature at which the oxide is broken down and a chromate film deposited. Thus magnesium alloys can be chromate-treated in nearly neutral solutions, whereas aluminium alloys can be treated only in solutions of appreciable acidity or alkalinity. The same principle tends to apply to the protective efficiency of the chromate film, i.e. the greater the intrinsic corrosion resistance of the metal, the greater the protection conferred by the soluble chromate in the chromate film.
Aluminium Chromate Treatment of Aluminium
Several immersion treatments using solutions containing chromates ’ have been developed for aluminium. It is not always clear to what extent the films formed can properly be called chromate films, i.e. films containing a substantial amount of a slightly soluble chromium chromate, but even if the film consists largely of aluminium oxide or hydroxide or other salt with chromate physically absorbed, it will still provide a reservoir of soluble chromate at the metal surface. Treatments fall into two classes: alkaline and acid. The latter are of more recent development.
Alkaline treatments These are all based on the original Bauer-Vogel process in which a boiling solution of alkali carbonate and chromate is used. The best known is the Modified Bauer-Vogel process (DTD 913); others contain silicate (E. W. process), fluoride, chromium carbonate, and/or disodium phosphate (Pylumin processes). The films formed are light to dark grey in colour, depending on the process and the composition of the alloy being treated, and consist substantially of aluminium oxide or hydroxide and probably some soluble chromate, either combined or adsorbed. The protection against mild atmospheres is fair, and is improved by sealing in hot sodium silicate solution. The films produced provide a good basis for paint. Acid treatments The principal acid processes were developed in the USA under the name Alodine, and are marketed in the UK as Alocrom and under other names. The original solutions were based on acid solutions containing phosphate, chromate and fluoride ions. Immersion for up to 5 rnin in the cold or warm solution leads to the deposition of a greenish film containing the phosphates of chromium and aluminium, and possibly some hexavalent chromate. The more recent Alocrom 1 200 process uses an acid solution containing chromate, fluoride and nitrate. Room-temperature immersion for 15 s to 3 min deposits golden-brown coatings which contain chromate as a major constituent. The success of the Alocrom 1 200 process has prompted the introduction of several other commercial processes which deposit similar substantial chromate-bearing films.
CHROMATE TREATMENTS
15:41
Acid pickles Some of the acid pickles used to clean and etch aluminium alloy surfaces and remove oxide and anodic films, such as the chromic/ sulphuric acid pickle (method 0of DEF STAN 03-2)and other chromic-acid bearing pickles (App. Fof DEF-151) probably leave on the surface traces of absorbed or combined chromate which will give at least some protection against mild atmospheres.
Sealing of Anodic Films
In view of the porous nature of anodic films, especially those produced by the sulphuric acid process (Section 15.1), sealing treatments have been developed in an attempt to improve their protective value. Although not very effective on the relatively dense films produced by the chromic acid process, the sealing treatments enhance the protection afforded by films produced by the sulphuric acid process. For conferring protection against corrosion the most effective treatment is immersion for 5-15 min in a boiling chromate/ dichromate solution just on the acid side of pH 8, Le. at a pH value at which aluminium oxide and hydroxide just begin to be slightly soluble. Defence Specification DEF-15 1 quotes two solutions, one a 7-10% dichromate/ chromate solution at pH 6-7, and the other a 5% dichromate solution containing a small amount of chromate to bring the pH from 4 (dichromate only) to between 5.6 and 6. The chromate sealing treatment imparts to the anodic film a distinct yellow to brown colour, which is probably due to a basic aluminium chromate or alkali chromate adsorbed on to aluminium hydroxide. The film gives appreciable protection against marine exposure.
Chromate Passivation of Cadmium and Zinc Cadmium and zinc coatings are widely used to protect steel from rusting, and for preventing accelerated corrosion when two dissimilar metals, e.g. copper and aluminium are in contact. It is important that zinc and cadmium should themselves be preserved from corroding, so that they may give protection by physical exclusion and sacrificial action. The durability of cadmium and zinc coatings depends on their thickness and their intrinsic corrosion resistance under any given exposure. On close-tolerance parts, the thickness is of necessity limited to 25pm or often appreciably less. Zinc corrodes quite rapidly in humid and marine conditions, and cadmium, though more resistant, is not immune. Both metals are attacked by the organic vapours emitted by some plastics and paints, and by wood2. It is therefore often highly desirable to apply a protective coating. The best protection is given by paint. An etch-primed paint scheme can be applied directly to the metal; for other paints an inorganic treatment must be given to ensure good adhesion. Of the two classes of inorganic treatment, phosphate treatment has little protective value in itself, but chromate passivation gives appreciable protection and in mildly corrosive surroundings may be sufficient in itself.
15 :42
CHROMATE TREATMENTS
The most commonly used chromate passivation process is the Cronak process developed by the New Jersey Zinc Co. in 1936, in which the parts are immersed for 5-10 s in a solution containing 182 g/l sodium dichromate and 6 ml/l sulphuric acid. A golden irridescent film is formed on the zinc or cadmium surface. Many variants (all fairly acidic) have been developed subsequently; all are based on dichromate (or chromic acid) with one or more of the following: sulphuric acid, hydrochloric acid (or sodium chloride), nitric acid (or nitrate), phosphoric acid, formic acid and acetic acid. A survey by Biestek3shows that several of these variants are as good as the Cronak process, although none is superior. Practical details of the Cronak process are given in Specification DEF-130, and a comprehensive account of the process as applied to zinc plate has been published by Clarke and Andrew'. Fig. 15.5 shows the loss of zinc and the
E
.-
d
-AW E
c u
0; $2
> x O W
EE
alL d
0
c
al
I
Time o f
immersion ( 5 )
Fig. 15.5 Effect of sulphuric acid concentration on chromate passivation of zinc. Solution: 182gA of Na2Cr207.2Hz0 HzS04as indicated; temp. 18T; 1.0 x IO-' rng Zn/crn2 = 0.145 Nrn thickness
+
weight of film deposited as a function of immersion period and of variation of sulphuric acid content above and below the normal 6mM. The curves show that in the normal bath the weight of film deposited is equal to the weight of zinc dissolved, and that as the acid is consumed, the solution becomes more efficient in converting metal to film. The curves also show that the dissolution of zinc during film formation is small, less than 0.25 pm, which is an important consideration when small parts such as nuts and bolts are being treated. (On such parts, for reasons of tolerance, relevant specifications are forced to allow minimum thicknesses of down to 4 pm of cadmium and zinc plate.) Claims for other passivation solutions should always be considered in relation to the quantity of metal consumed, unless, of course, the solutions are intended solely for use on zinc-base die-castings, where tolerance on thickness is unimportant. The chromate film deposited by the Cronak process on zinc consists largely of a hydrated chromium chromate and contains some 10% by weight
CHROMATE TREATMENTS
15 :43
of hexavalent chromium, equivalent to 20% of CrOi-. At least part of this chromate is soluble in water and available for protecting the underlying zinc or cadmium; on account of this solubility, passivated parts should not be washed in very hot water. Heating at 100°C or higher tends to dehydrate the film and render the chromate in it insoluble, with consequent reduction in protective value; any heat treatment after plating, e.g. for de-embrittlement, should therefore be completed before chromate passivation. If the yellow colour of the chromate film is considered undesirable, treated parts can be subjected to an aqueous extraction ‘bleaching’ treatment, but much of the protective value will be lost thereby. The quantitative results quoted above all refer to zinc surface. it is likely that the behaviour of cadmium would be similar; in view of the fact that the equivalent weight of cadmium is double that of zinc, it is even more important that the passivation solution shall not attack and dissolve the metal to any appreciable extent.
Cleaning Etch for Copper and its Alloys Copper and its alloys can be cleaned and brightened by immersion in solutions of substantial quantities of dichromate with a little acid (see, for instance method Q of DEF STD 03-2/1). Such solutions impart some resistance to tarnishing, ascribed to the formation of very thin chromate films. Clarke and Andrew have developed a similar solution further activated by addition of chloride ions which deposits more substantial films shown to contain hexavalent chromium. The films give appreciable protection against salt spray and tarnishing by sulphur dioxide.
Iron Chromate Treatment
In spite of the effectiveness of chromates in stopping the rusting of steel in aqueous solutions, no successful chromate filming process has been developed for this purpose. Chromate Rinsing of Phosphated Steel
The protective value of a phosphate coating is enhanced by a dip or rinse in an acid chromate solution. Joint Service Specification DEF-29 makes such a rinse mandatory for steel parts treated by an accelerated process, and optional after treatment by a non-accelerated process. Details of rinses are given in Section 15.2 (Table 15.10, p. 15:30).
Magnesium Alloys (See also Section 4.4.)
15:44
CHROMATE TREATMENTS
Chromate Treatments
Chromates are very effective inhibitors of the corrosion of magnesium alloys by saline and other waters, and many treatments have been developed by means of which substantial films containing slightly soluble chromate are formed in the metal surface. Except on parts which are to be exposed only to a rural atmosphere, chromate treatment must be supplemented by paint, for which it provides a good base. Magnesium is a relatively reactive metal, and can be chromated in nearly neutral solutions as well as in acid solutions. The range of treatments possible illustrates well the r61e of pH, activating anion, temperature and duration of treatment in promoting the breakdown of passivity in the chromate solution and the consequent formation of a chromate film.
Strongly acid bath This class is represented by a treatment developed in Germany over 60 years ago and widely used since. The solution contains 15% sodium or potassium dichromate and 20-24% V.V. concentrated nitric acid. Parts to be treated are immersed for 30 s to 2 min at room temperature and then allowed to drain for 5 s or more before being washed. Most or all of the film formation occurs during the draining period and the chief function of the immersion period is to clean the surface by etching. The film is thin and of a golden or irridescent grey colour. The process is not suitable for close tolerance parts and is mainly used for protection during storage prior to matching. The process is used in the UK as bath (iv) of DTD 91IC, in the USA as the Dow No. 1 treatment and in the USSR as treatment MOKH-1.
Medium acid baths, pH 4-5 At this acidity a dichromate solution plus sulphate ion as activator is sufficient to deposit chromate films in 30 min or so at room temperature or in a few minutes at boiling point. Unfortunately, a solution of alkali dichromate and alkali sulphate is quite unbuffered, and other substances must be added to give the bath a useful life over the working pH range. Acetates have been used successfully, but salts of aluminium, chromium, manganese and zinc have been more commonly employed. The pH of the solution rises slowly during use until basic chromates or sulphates begin to precipitate. The solution can then be rejuvenated by the addition of chromic or sulphuric acid or acid salts. A successful bath of this class is the Magnesium Elektron Chrome Manganese Bath, bath (v) of DTD 911C, which contains 10% sodium dichromate, 5% magnesium sulphate (as a source of sulphate) and 5% manganese sulphate (as a source of sulphate and as a buffering agent). Treatment is by immersion for up to 2 h at room temperature or up to 10 min at boiling point, the treatment being continued until the appearance of the deposited film has passed the thin golden stage and reached the dark brown to black stage. A second bath of this class is the Dow No. 4 which contains sodium dichromate and potassium chrome alum; this solution is used at boiling point. The Dow No.7 treatment, popular in the USA,also falls within this class. The process differs from other chromate treatments in that the activator, magnesium fluoride, is formed on the metal surface by immersion in 20% hydrofluoric acid solution, the parts then being immersed in a 10-15% alkali dichromate solution with or without sufficient alkaline earth fluoride to saturate it. A slow action occurs on the surface and the fluoride film is replaced by a chromate or mixed chromate/fluoride film.
CHROMATE TREATMENTS
15:45
The dichromate solution is quite unbuffered over the working pH range of 4-0-5.5, but the degree of attack on the metal is so slight that in practice appreciable surface areas can be treated before readjustment of the pH by addition of chromic acid becomes necessary. The process is used in the USSR under the code name MFKH-I . Slightly acid baths, pH 6 At this pH, a boiling temperature must be used to aid the activation of the sulphate present; alternatively, activation can be accomplished by use of an anodic current. The R.A.E. ‘hot half-hour bath’, bath (iii) of DTD 911C falls into this class. The solution contains 1.5% each of ammonium and alkali dichromates, 3% ammonium sulphate, and enough ammonia to raise the pH from 4 (dichromate stage) to 6. Parts to be treated are immersed in the boiling solution for 30min. The solution is well buffered against rise of pH due to magnesium dissolving in the solution, partly by the chemical reaction dichromate chromate, and partly by loss during boiling of ammonia. A closely similar process is used in the USSR under the code name MOKH-6. In the USA the process is applied after a hydrofluoric acid pretreatment, either as above (Dow No. 8 treatment and USSR MFKH-3) or at 50-60°C with the aid of the galvanic current generated when the parts under treatment are connected electrically to the steel tank or to steel cathodes in the solution (Dow No. 9 treatment). A cold treatment relying on electric current for activation has been developed at R.A.E.; the solution consisted of 15% sodium dichromate and 5% potassium permanganate with added caustic soda to bring the pH to the lower end of the treatment range of 6-0-7-1. Parts were made anode at a current density of 30-80 A/dm2 for a treatment time of 20 min. +
Chromate Sealing A large number of electrolytic treatments of magnesium, anodic or ax.,
have been developed, in which adherent white or grey films consisting of fluoride, oxide, hydroxide, aluminate or basic carbonate are deposited from alkaline solutions containing caustic alkali, alkali carbonates, phosphates, pyrophosphates, cyanides, aluminates, oxalates, silicates, borates, etc. Some films are thin, and some are relatively thick. All are more or less absorbent and act as good bases for paint, though none contributes appreciable inhibition. All can, however, absorb chromates with consequent improvement of protective efficiency. The simplest method of chromate sealing involves immersion in a dilute alkali chromate or dichromate solution followed by washing; retained chromate imparts a yellow colour to the film. More substantial amounts of slightly soluble chromate can be deposited in the thicker type of absorbent anodic film by a method developed by Dr. L. Whitby at High Duty Alloys Ltd6. In this, anodised parts are immersed first in a boiling 30% solution of sodium chromate and then in a boiling 2% solution of zinc nitrate. Residues of the first solution in the film react with the second solution to give a substantial yellow deposit of a basic zinc chromate, probably similar in composition to zinc yellow.
15 :46
CHROMATE TREATMENTS
Silver Chromate treatments have been developed for protecting silver against sulphide tarnishing by the deposition of very thin films which are assumed to contain chromate. A Dutch-American immersion treatment’ uses a chromate solution and a complexing agent, e.g. cyanide, ammonia or E.D.T.A. Working pH values depend on the nature of the agent and lie within the range pH 1-12. Another treatment consists of making the silver parts cathode in an alkaline chromate solution.
Tin Alkaline chromate treatments for tin, e.g. the Protecta-Tin processes*, have been developed by the Tin Research Institute. The solutions resemble the M.B.V. compositions for treating aluminium, but are more alkaline. Thin invisible films which resist staining by heat and sulphur-bearing compounds and give protection against humid atmospheres at pores are deposited.
Etch Primers While etch primers, also known as pretreatment primers and wash primers, can be regarded as priming paints which promote their own adhesion by etching the metal surface, they may also be regarded as phosphate/chromate etching treatments which leave an organic residue on the surface to form the basis of the subsequent paint scheme. A detailed account of the etch primers has been given by Coleman’. The standard etch primer (WP-1,DEF-1408)consists of two solutions, one containing polyvinyl butyral resin and zinc tetroxychromate in ethyl alcohol with n-butanol, and the second containing phosphoric acid and ethyl alcohol. It is essential that a small critical amount of water be present in the latter. The two solutions are mixed in appropriate ratio for use; the mixture deteriorates and should be discarded when more than 8 h old. Single-pack etch primers of reasonable shelf life are available but contain less phosphoric acid than the above and are not considered to be so effective. The reactions which take place when the mixed etch primer is applied to a metal are complex. Part of the phosphoric acid reacts with the zinc tetroxychromate pigment to form chromic acid, zinc phosphates and zinc chromates of lower basicity. The phosphoric acid also attacks the metal surface and forms on it a thin chromate-sealed phosphate film. Chromic acid is reduced by the alcohols in the presence of phosphoric acid to form chromium phosphate and aldehydes. It is believed that part of the chromium phosphate then reacts with the resin to form an insoluble complex. Excess zinc tetroxy chromate, and perhaps some more soluble less basic zinc chromes, remain to function as normal chromate pigments, i.e. to impart chromate to water penetrating the film during exposure. Although the primer film is hard
CHROMATE TREATMENTS
15 :47
enough for over-coating after drying for 1 h, the above reactions continue in the nominally dry film for two to three days, during which time the film remains rather sensitive to water. Etch priming is widely used on aluminium alloy, and is particularly effective on cadmium and zinc. The adhesion to stainless steel and titanium is good. It has also been used quite widely on bare steel and on magnesium alloy, but on these metals its performance is not, in the opinion of some investigators, always quite reliable. For best protection the etch primer coating is followed with a full paint scheme.
Recent Developments A comprehensive review by Biestek and Weber lo covering earlier work on the technology of chromating and the properties of chromate coatings on a range of metals, has been published. The advent of surface sensitive techniques such as X-ray photoelectron spectroscopy has enabled advances to be made regarding the composition of films formed. Treverton and Davies ' I found that the chromate conversion coating on aluminium consisted mainly of Cr(rI1) with aluminium oxides and fluorides present at the film/substrate interface. Matienzo and Holub '* have shown that a chromic acid rinse after conversation coating, introduced additional chromium which was incorporated in the coating as CR(VI).The acid rinse also eliminated fluorides whilst forming a thicker protective layer. Using scanning electron microscopy, Arrowsmith et af.13 have shown that coatings on aluminium are produced by precipitation of spherical particles which merge and form successive layers. Film growth was maintained by transport of solution through open channels rather than by migration of ions through a continuous layer. In the case of galvanised steel surfaces, DuncanI4 investigated the composition of the chromate layer after heating, immersion in water and outdoor exposure. Recent developments in zinc coating technology have resulted in the availability of a wide range of compositions. It is known that the formation of suitable conversion coatings on Zn-Ni and Zn-Co electrodeposits is difficult and there is only a limited amount of published information, although several commercial systems are in use. Similarly, zinc coatings produced by hot-dipping, eg. Zn-AI, pose particular problems. Initial work on the conversion coating of different types of zinc substrate prior to powder coating has been reported". Since chromates are highly toxic there has been concern over their discharge into the environment and in the handling of chromate compounds and treated components. This has resulted in increasingly more stringent environmental and factory regulations. The precipitation of Cr(v1) from effluent solutions is difficult and can only be overcome by expensive multistep treatments. Attempts to comply with legislation have resulted in the development of various new techniques and formulations, but the mechanisms of reaction are still similar to those outlined above. Formulations suggested by Barnes et af.I6 use nitrates, hydrogen peroxide or persulphates to initiate metal dissolution, with nitrate being preferred. Successful film-forming compounds include trivalent chromium ions
15 :48
CHROMATE TREATMENTS
provided by the leather tanning salt Chrometan, and aluminium ions provided by aluminium sulphate. In the case of chromate solutions a more stable film was produced if a complexant such as sodium hypophosphite was present in solution. LeRoy ” has developed solutions based on thioglycollic esters where the thioglycolate grouping HS-CH,-C(0)-0- is very reactive with zinc, provided that the zinc surface is clean. Solutions have to be emmulsifiedto stop setting of the polymer and the best coatings are produced when solution pH and temperature are in the range 2-8 and 5O-8O0C, respectively. The nature of conversation coatings produced on tin and zinc from molybdate and tungstate solutions has been determined and compared with those produced when using chromate solutions (see for example Wilcox and Gabe”). No-rinse treatments of the organic and inorganic types have been discussed by Matienzo and Holub”. Dissolution of the metal occurred at low pH followed by deposition of a polymer complex or silica as the pH at the surface increased. Chromium is still introduced from the solution as Cr(II1) and Cr(vI), but no acidic rinse followed the formation of the coating. N. R. SHORT H. G. COLE REFERENCES 1. Wernick, S. and Pinner, R., Surface Treatment and Finishing of Aluminium Alloys, Robert Draper, Teddington (1956)
2. Rance, Vera E. and Cole, H. G.. Corrosion of Metals by Vapoursfrom OrganicMaterials, London, H.M.S.O. (1958) 3. Biestek, T . , Prace Inst. Mech.. 6 (3/1956), 39 (1957) 4. Clarke, S. G. and Andrew, J. F., J. Electrodep. Tech. Soc., 20, 119 (1945) 5. Clarke, S. G. and Andrew, J. F.. Proceedings of the First International Congress on Metallic Corrosion, London, 1961. Butterworths. London, 173 (1962) 6. High Duty Alloys Ltd., UK Pat. 570 054 (1945) 7. N. America Phillips Co. Inc., US Pat. 2 850 419 (1958) 8. Britton, S. C. and Angles, R . M.,J. Appl. Chem., 4, 351 (1954) 9. Coleman, L. J., J. Oil Col. Chem. Ass.. 42, 1, 10 (1959) 10. Biestek. T. and Weber, J.. Electrolytic and Chemical Conversion Coatings, pp. 1-127, Portcullis Press, Redhill (1976) 11. Treverton. J. A. and Davies N. C. ‘XPS Studies of a Ferricyanide Accelerated chromate Paint Pretreatment Film on an Aluminium Surface’, Surf. Interfacial Anal., 3, 194-200 (1981) 12. Matienzo, L. J. and Holub, K. J. ‘Surface Studies of Corrosion-preventing Coatings for Aluminium Alloys’, App/ic. Surf. Sci., 9, 47-43 (1981) 13. Arrowsmith, D. J.. Dennis, J. K.and Sliwinski, P. R., ‘Chromate Conversion Coatings on
Aluminium: Growth of Layers of Spherical Particles. Trans. Inst. Met. Fin., 62, 117-120 ( 1984) 14. Duncan, J. R., Electron Spectroscopy of Chromated Galvanized Steel Sheet after Heating. Immersion in Water or Outdoor Weathering. Surface Tech., 17, 265-276 (1982) 15. Short, N. R.,Dennis, J. K. and Agbonlahor, S. O., ‘Conversion Coating of Zinc Coated Substrates Prior to Powder Coating’, Trans. Insf. Met. Fin., 66. 107-111 (1988) 16. Barnes, C., Ward, J. J. B., Sehmbi. T. S. and Carter, V. E., Non-chromate passivation treatments for zinc. Trans. Inst. Met. Fin., 60, 45-48 (1982) 17. LeRoy. R. L.. ‘Polythioglycolate Passivation of Zinc’, Corrosion., 34. 113-1 19 (1978) 18. Wilcox, G . D. and Gabe, D. R., Passivation Studies Using Group VIA Anions. Br. Corr. J., 19, 1%-u)o (1984)
16
MISCELLANEOUS COATINGS
16.1 Vitreous Enamel Coatings 16.2 Thermoplastics
16:3 16:13
16.3 Temporary Protectives
1694
16: 1
16.1 Vitreous Enamel Coatings
Nature of Vitreous Enamels A vitreous enamel coating is, as the name implies, a coating of a glassy substance which has been fused onto the basis metal to give a tightly adherent hard finish resistant to many abrasive and corrosive materials. The purpose of modern vitreous enamels is twofold, i.e. to confer corrosion protection to the metal substrate and at the same time to provide permanent colour, gloss and other aesthetic values. Most of the corrosion resistance, and indeed other properties of the finish, are determined by the composition of the vitreous enameller’s raw material frit, although other factors can influence them to a minor degree. Frit, for application to sheet and cast iron, is essentially a complex alkali-metal alumino borosilicate and is prepared by smelting together at temperatures between 1 100 and 1450OC an intimate mixture of refractory materials such as silica, titania, felspar, china clay, etc. with fluxes exemplified by borax, sodium silicofluoride and the nitrates and carbonates of lithium, sodium and potassium. The smelting continues until all the solid matter has interreacted to form a molten mass, but unlike true glass this liquid does contain a degree of bubbles. At this stage the melt is quenched rapidly by either pouring into water or between water-cooled steel rollers to form ‘frit’ or ‘flake’. Frit may be milled dry or wet. The long established dry process is used for cast iron baths and for chemical plant. Vitreous enamel application by a dry electrostatic method is being used on an increasing scale. In these cases, the frit is milled alone, or with inorganic colouring or refractory additives. This is achieved in cylinders using balls of porcelain, steatite or more dense alumina, or with pebbles of flint, to produce a fine powder of predetermined size. In the more common wet process the frit is milled with water, colloidal clay, opacifier, colouring oxide, refractory and various electrolytes in a ball mill to a closely controlled fineness or coarseness. Typical frit and mill formulae are given in Table 16.1. Frits are tailormade for each application so that the most desired properties are at their maximum in each case and thus the formulae presented must be regarded as examples of general composition.
16:3
16:4
VITREOUS ENAMEL COATINGS
Table 16.1 Typical enamel frit compositions (olo) and a mill addition Chemical plant
Na20 LizO CaO BaO CaF, Na2SiF, Ai2 O3 BZ O3 SiO, Ti02
coo
NiO MnO Sb2 0 5
15.8
-
1.2
-
3.4
Sheet iron (white)
Sheet-iron
(groundcoats) 17.5
-
6.0
21.8
-
5.5
16.0
-
-
5.5
4.0
-
-
-
-
2.9 0.9
5.0 20.0
9.0 18.2
1.0 25.0
60.0
50.0 0.4 0.5 0.6 -
44.0
47.0 -
15.8
-
-
-
0.3
0.6 0.6 -
0.5 0.5
0.5
-
Sheer iron (acid resistant black)
Cast iron
(semiopaque)
7.0
16.0
17.5
5.0 1.0
1.0
-
-
5.5 2.5 15.0
46.0 18.0
-
-
-
3.0
3.0 6.0
-
2.0 2.0
-
1.0 7.0
4.5 7.5
53.0 8.0
43.5
0.4 0.6 -
2.0
13.5
-
8.5
Sheet iron white mill addition
Frit Water Titania Clay Bentonite Sodium nitrite Potassium carbonate
Grind to fineness of 1 g residue on 200 mesh sieve (50 ml sample) 0.3
0.05 0.1
Metal and Metal Preparation To obtain a defect-free finish it is essential L a t the basis meta is of 1 le correct composition and suitably cleaned. Cast iron
For cast iron enamelling the so-called grey iron is preferred. Its composition varies somewhat depending upon type and thickness of casting, but falls within the following limits: 3.25-3.60Vo total C, 2-80-3.20% graphitic C, 2-25-3.00% Si, 0-45-0.65Vo Mn, 0.60-0.95% P and 0-05-0*10% S . The standard method of cleaning cast iron for enamelling is by grit or shot blasting which may be preceded by an annealing operation. Steel
Two general types of sheet steel are in current use, viz. cold-rolled mild steel and decarburised steel. A typical analysis for cold-rolled steel is 0.1% C, 0.5% Mn and 0.04% S. It can be obtained in regular, deep drawing or extra-deep drawing grades. This type of steel is normally used with a groundcoat including cobalt and nickel, as shown in Table 16.1.
VITREOUS ENAMEL COATINGS
16:5
Decarburised steel is a mild steel that has undergone a heat treatment in
a controlled atmosphere to reduce the carbon content to about 0.005%. This type of steel can be used for white or coloured enamel direct to steel. Sheet steel is normally prepared for application of enamel by a sequence of operations including thorough degreasing, acid pickling and neutralisation. A nickel dip stage is often included to deposit a thin, porous layer of nickel applied at about 1 g/m2, especially when conventional groundcoat is not used (see Section 13.7). Enamel Bonding
For effective performance the enamel must be firmly bonded to the underlying metal and this bond must persist during usage. The bond is formed by the molten enamel flowing into the ‘pits’ in the metal, Le. mechanical adhesion, and by solution of the metal in the glass, Le. chemical adhesion. The coefficient of thermal expansion of the enamel in relation to the cast iron or sheet steel and enamel setting temperature determines the stress set up in the coating. As enamel, like glass, is strongest under compression, its thermal expansion should be slightly less than the metal. En8mel Application 8nd Fusion
Vitreous enamel is normally applied to the prepared metal or over a groundcoat by spraying or dipping. Alternative wet techniques are used, of which the most common has been electrostatic wet spraying. Electrophoretic deposition from the slurry has been found to be highly suitable for some components. On sheet iron a groundcoat, including cobalt and nickel, is generally used, but for mass production (e.g. cookers) use of decarbonised steel and direct application of colours is more common. This involves a more complex steel pretreatment. After drying the applied slurry, the enamel is fused onto sheet steel at about 800-850°C for about 4-5 min. For cast iron a longer time and lower temperature are normal. The old dry process enamelling of cast iron (baths etc.) is no longer widely used. The method consisted of sieving finely powdered frit onto the preheated casting and inserting the casting back into a furnace at about 900°C to produce the smooth finish. In recent years increasing use has been made by many manufacturers, who require a limited range of colours, of the electrostatic application of a dry powder spray. Dry electrostatic finishes are fused at temperatures in the same range as conventional ones.
Properties of Enamel Coatings Affecting Corrosion Mechanic8l Properti8s
This group includes such items as surface hardness, i.e. scratch and abrasion resistance, adhesion and resistance to chipping, crazing and impact. All of
16:6
VITREOUS ENAMEL COATINGS
these and other properties depend upon the adhesion between the vitreous enamel layer and the metal being good and r e m a h n g so. There is no single test that will give a quantitative assessment of adhesion, and those which have been proposed all cause destruction of the test piece. It has already been stated that this property is dependent upon mechanical and chemical bonds between the enamel and the metal. One must, however, also consider the stresses set up at the interface and within the glass itself during cooling after fusion or after a delayed length of time. The coefficient of thermal expansion is primarily determined by the frit composition, although mill additions can have a minor influence. As a general rule, superior acid and thermal shock resistance obtain with low expansion enamel, and the skill of the frit manufacturer is to obtain good resistance and also to maintain a sufficiently high expansion to prevent distortion of the component (pressing or casting). Several workers have produced a set of factors for expansion in relation to the enamel oxides that constitute the frit, which provides a guide to the frit producer. However, as these factors are derived from a study of relatively simple glasses smelted to homogeneity it must be emphasised that they are only a guide. The effect of substituting certain oxides for others in a standard titanium superopaque enamel is given in Table 16.2. The use of a nickel dip improves adhesion by minimising iron oxide formation, but it should be noted that some iron oxide formation is necessary to produce enameVmeta1 adhesion. In the commonest methods of testing for adherence to sheet iron, the coated metal is distorted by bending, twisting or impact under a falling weight. In the worst cases the enamel is removed leaving the metal bright and shiny, but in all others a dark coloured coating remains with slivers of fractured enamel adhering to a greater or lesser degree. With cast iron enamelling it is not possible to distort the metal and in this case an assessment of adhesion is obtained by dropping a weight on to the enamel surface and examining for fractures. Erroneous results can obtain in that often thicker enamel coatings appear to be better bonded and resistant to impact, whereas in fact the converse is true. Providing the bond is adequate this test really gives an indication of the strength of the enamel itself.
Table 16.2 Effect of frit ingredients on enamel expansion
Constituent varied
Expansion change
Increase alkali metal Replace Na, 0 by Li,O Replace Na,O by K,O Increase fluorine Increase B, 0, Replace SiO, by TiO, Increase TiO, Replace SiO, by A1,0, Introduce P,05 Introduce BaO Increase SiO,
Increase Increase Decrease Decrease Decrease Increase Slight increase Slight increase Slight increase Increase Decrease
-
VITREOUS ENAMEL COATINGS
16:7
According to Andrews' a typical sheet iron groundcoat has a tensile strength of about 10 kg/mm2. In small cross section, however, the tensile strength of glass is improved and fine threads, e.g. as in glass fibre, are quite strong. Enamels under compression are 15-20 times stronger than an equal thickness under tension. The hardness of an enamel surface is an important property for such items as enamelled sink units, domestic appliances, washing machine tubs which have to withstand the abrasive action of buttons, etc. On Moh's scale most enamels have a hardness of up to 6 (orthoclase). There are two types of hardness of importance to users of enamel, viz. surface and subsurface. The former is more important for domestic uses when one considers the scratching action of cutlery, pans, etc. whereas subsurface hardness is the prime factor in prolonging the life of enamelled scoops, buckets, etc. in such applications as elevators or conveyors of coal and other minerals. Of the several methods of measuring this property those specified by the Porcelain Enamel Institute and the Institute of Vitreous Enamellers are the best known and most reliable. They both consist of abrading a weighed enamel panel with a standard silica or other abrasive suspended in water and kept moving on an oscillating table with stainless steel balls. The loss in weight is measured periodically and a graph of time versus weight loss indicates both the surface and subsurface abrasion resistance. Pedder' has quoted relative weight loss figures for different types of enamel and they are shown in Table 16.3. Fine bubbles uniformly distributed throughout the coat improve elasticity and thus mill additions and under and over firing influence this property. The greatest effect on elasticity is enamel thickness and most developments are aimed at obtaining a satisfactory finish with minimum thickness. Appen et ai. have produced factors for calculating the elastic properties of enamel. Table 16.3 Comparison of abrasion resistance of different enamels* Types of enamel
Average loss in weight (g) t
Acid resisting titania based Acid resisting non-titania Antimony white cover coat High refractory enamel Plate glass
56 x 1 0 - ~ 342 X 582 x 129 x 1 0 - ~ 70 x 1 0 - ~
' Table after Pedder2. tOverall figure for lesls under slandardised conditions for each grade of enamel.
Thermal Properties
These properties are made use of in many applications ranging from domestic cookers to linings which must withstand the heat from jet engines. There is simple heat resistance, i.e. the ability of the enamel to protect the
16:8
VITREOUS ENAMEL COATINGS
underlying metal from prolonged heat and also thermal shock resistance, which is the ability to resist sudden changes in temperature without failure occurring in the coating. These thermal properties depend upon the relative coefficient of thermal expansion of enamel and metal, enamel setting point, adhesion, enamel thickness and geometry of the shape to which the finish is applied. It is obvious that the adhesion must be good in order to prevent rupture at the enamel/metal interface during heating and cooling. Thick coatings are liable to spall when subjected to thermal change due to differential strain set up within the enamel layer itself, caused by the poor heat conductivity of the glass. Thus again thin coatings are desirable. Compressive forces on enamel applied to a convex surface are less than when a concave surface is coated, and it is therefore apparent that the sharper the radius of the metal the weaker the enamel applied to it will be. This fact is also relevant to mechanical damage. Thermal shock resistance is important for gas cooker pan supports and hotplates where spillage is liable to occur, but in oven interiors heat resistance is more relevant. The softening point of conventional cast and sheet iron enamels is about 5OO0C, but special compositions are obtainable which operate successfully at 60OOC. Other more specialised enamels withstand service conditions ranging from being in excess of dull red heat, e.g. as obtained in fire backs, to those capable of enduring short exposure to temperatures of around 1 O O O T , e.g. in jet tubes, after burners, etc.
Chemical Resistance
That examples of glass and glazes manufactured many centuries ago still exist is an indication of the good resistance of such ceramics to abrasion, acids, alkalis, atmosphere, etc. In this section, chemical resistance will be divided into three parts, viz. acid, alkali (including detergents) and water (including atmosphere). Normally an enamel is formulated to withstand one of the corrosive agents more specifically than another, although vitreous enamel as a general finish has good 'all round' resistance, with a few exceptions such as hydrofluoric acid and fused or hot concentrated solutions of caustic soda or potash. Acid resistance This property is best appreciated when the glass structure is understood. Most enamel frits are complex alkali metal borosilicates and can be visualised as a network of SO, tetrahedra and BO3 triangular configurations containing alkali metals such as lithium, sodium and potassium or alkaline earth metals, especially calcium and barium, in the network interstices. Fused silica may be regarded as the ultimate from the acid resistance aspect but because of its high softening point and low thermal expansion it cannot be applied to a metal in the usual manner. Rupturing or distorting this almost regular SiO., lattice makes the structure more fluid. Thus to reduce its softening point B , 0 3 is introduced whereby some of the Si-0 bonds are broken and an irregular network of
16:9
VITREOUS ENAMEL COATINGS
\ 0
\ / B I
/ 0.
\ and
0
1
/ 0
0
/
\ /
/si\
0‘
0
\
is formed. Further distortion of the network is obtained by introducing alkali and alkaline earth metals into the lattice. If fluorine is included in the linking two frit, more bonds are broken; in this case an oxygen atom (-0-) silicon or boron atoms is replaced by a fluorine atom (F-) which being monovalent cannot joint two Si or B atoms, hence causing bond rupture. A study of the relevant phase diagrams and eutectics proves useful in formulating low firing enamels. Thus all frit ingredients act as either network formers or modifiers and with the principal exception of silica, titania and zirconia, all cause a diminution in acid resistance. The reacting acid causes an exchange between metal ions in the network modifier of the glass and hydrogen ions from the acid. This naturally occurs at the enamel surface, but as the etching or leaching reaction proceeds, a resulting thin layer of silica-rich material inhibits further reaction. Thus acid attack is dependent upon enamel composition and pH, with time and temperature playing a part. Sodium oxide and boric acid are both leached out by acid attack, and it has been found that the Na,O/B,O, ratio is constant for any one enamel and is dependent upon enamel composition. An increase in titania content of the frit acts in a similar way to increasing silica in enhancing acid resistance with the added advantage that the coefficient of expansion is also raised slightly and the glass viscosity not increased as much as by the equivalent SiO, increment. This only applies to the titania remaining in solution in the glass and does not necessarily hold when the frit is supersaturated with TiO,, which occurs with the modern opaque sheet iron covercoats when some of the pigment recrystallises and causes opacification on cooling from the firing process. In formulating holloware enamels the degree of acid resistance required is less than for chemical plant, e.g. reaction vessels, and consequently the ROz (SiO, and TiO,) is lower thus permitting increased quantities of fluxes to be incorporated which confer improved ‘workability’. Furthermore, they can be fired at lower temperatures and have superior chip resistance. Conversely, chemical plant enamels are higher in silica and dissolved titania and require harder firing. An example of such an enamel is shown in Table 16.1. The acid resistance called for on domestic appliances varies with the particular component, e.g. the oven interior of a gas cooker necessitates a higher resistance than the outside sides -the former being at least Class A using 2% sulphuric acid while the latter can have a lower grading based on the less aggressive citric acid tests. These tests are detailed in BS 1344:Part 3 (IS0 8290) and BS 1344: Part 2 (IS0 2722), respectively. The enamel mill addition, degree of firing and furnace atmosphere all affect acid resistance. An increase in clay and alkaline electrolyte detracts from this property and underfiring also has an adverse effect. The use of organic suspending agents is thus preferable to clays, from this aspect, but
16: 10
VITREOUS ENAMEL COATINGS
other factors must also be considered. Similarly the replacement of 1070 milling clay by % mo of the more colloidal bentonite is beneficial. Large additions of quartz at the mill improve heat resistance and, provided the firing temperature is increased to dissolve a sufficient quantity of this silica in the glass, the acid resistance is also enhanced. In the glass-bottle industry the bottles can be cooled in a dilute SO,/SO, atmosphere to increase chemical resistance. A similar effect has been noted with vitreous enamel. It has been postulated that a thin layer of -OH groups or -OH-H,O (hydronium) ions is adsorbed on the surface of a fired enamel. These ions are transformed into -0S0, or -0S0, in the presence of oxides of sulphur which are more resistant to further acid attack. It is known that the acid resistance of a recently fired enamel improves on ageing, probably due to the enamel reaction with S 0 2 / S 0 3in the atmosphere and it is quite common for the grading to improve from Class A to Class AA (BS 1344). In enamels for chemical plant such as autoclaves it is not only the degree of acid resistance which is important but also the freedom of the finish from minute flaws detectable by high frequency spark testing or chemical methods. The chemical methods depend upon a colour change when the reagent such as ammonium thiocyanate reacts with the iron exposed at the bottom of the pinhole or flaw in the finish. Alternatively, an electric cell can be formed via the exposed iron in the flaw and detected chemically. In general, strong mineral acids are more severe in their attack on enamel than weak organic acids. Vargin3 has stated that the severity of action of organic acids on enamel increases with the increase in the dissociation constant of the acid. Temperature plays a major part in acid resistance, the nearer the boiling point the greater the rate of attack. It is more significant than acid concentration. It is recognised that vitreous enamel possesses good acid resistance, but an exception occurs with hydrofluoric acid. This is due to the relative ease of reaction between this acid and the silica (which is the largest constituent in the frit) to form silicon tetrafluoride. This reaction is made use of in some ‘de-enamelling’plants.
Alkali and detergent resistance The usual method of de-enamelling sheet iron is by immersion in fused or hot strong aqueous solutions of caustic soda when the silica network is broken down to form sodium silicate. However, in spite of this fact, enamels are capable of withstanding detergents and mild alkalis and this finish is often used very successfully in washing machines, baths, sink units, etc. where alkaline conditions prevail. Such enamels are usually higher in alumina than acid-resistingenamels and often contain zirconia in the frit. Other elements which aid alkali resistance are barium, calcium, lead and zinc’ and their function in this context is to increase the bond with the essentially silica network and form insoluble silicates which act as a protective coating slowing down the formation of soluble sodium silicate. The necessity for alkali resistance is relatively limited when compared with detergent resistance and it has been shown that whilst these two properties are similar, a finish resistant to one is not necessarily as resistant to the other. The Institute of Vitreous Enamellers produced a report on detergent
VITREOUS ENAMEL COATINGS
16: 11
resistance in 195g5 and the following facts are taken from it: 1. Semi-opaque acid-resistant titania enamels and alkali-resistant frit
2.
3. 4.
5.
generally have good detergent resistance whereas non-acid-resistant sign -based finishes have poor resistance. enamels and A1,0, /B20, /Pz05 Initially, detergent attack is accompanied by a deposit on the enamel surface which can be abraded off resulting in an apparently unaffected glossy appearance. This contrasts with acid attack when a progressive weight loss occurs and original gloss cannot be restored once it has been lost or diminished. After more prolonged detergent attack it is not possible to restore the original high gloss. The rate of attack is very dependent upon temperature, that at boiling being several times greater than that at room temperature. An increase in milling clay has a marked effect on improving this property. Increased detergent concentration, coarser grinding of the frit and nonstandard firing all cause minor deterioration in resistance.
In the design of an enamel for a washing machine tub, detergent resistance alone is not sufficient and the enamel must also be capable of withstanding the possible abrasive action of buttons, zip fasteners, etc. Resistance to water and atmosphere These properties are of particular importance in enamelled signs, architectural panels, cooking utensils and hospital ware subjected to repeated sterilisation. That such enamelled signs as ‘Stephen’s Inks’, etc. are still in existence and in good condition after many years outside exposure coupled with the fact that the use of vitreous enamel as a finish for architectural panels is growing are ready pointers to the good water and atmospheric resistance of enamel. Enamelled hospital utensils such as kidney bowls score over organic finishes because of their ease of sterilisation and also because they are less accommodating to germs, bacteria, etc. on account of their lower electrostatic type attraction for such microbes. The action of water on enamel is in many ways similar to that of acids in that the network modifier is the weak link and through hydrolysis can be removed from the glass system resulting in loss of gloss and a porous surface. As with acids and alkalis, the attack on the glass by water can be continued in extreme cases, by an attack on the inorganic colouring matter initially liberated or made more active. In an enclosed system the soluble salts first leached out from the enamel by water become in turn the corrosive element and further attack is dependent upon the pH of such a salt, or, for example, on the Na20/B20, ratio. The introduction of divalent calcium and barium oxides into frits in preference to monovalent sodium and potassium generally increases water resistance. Furthermore, oxides of tetravalent and pentavalent metals have a favourable effect on the resistance of glasses and enamels to water. The influence of B,O, and fluorine in the frit upon chemical resistance is variable and is dependent upon the content of them and the balance of the frit constituents, but they usually cause a diminution in resistance. In general, mill-added clay, silica and opacifier increase water resistance provided the firing or fusing of the enamel is at the optimum.
16: 12
VITREOUS ENAMEL COATINGS
As is expected, atmospheric resistance is related to water and the acid formed from C02, SO2, SO,, etc. The action of ultraviolet light has no apparent effect on vitreous enamel unlike the case with organic finishes. There is good correlation between atmospheric resistance and acid resistance, and this fact is helpful to manufacturers of architectural panels who can easily and quickly determine the latter property and not have to carry out lengthy exposures to the relatively unpolluted air. An exception, however, occurs with reds and yellows where a strict correIation is not always true, and in these cases a test based upon exposure to a saturated copper sulphate solution under illumination by a white fluorescent light has been advocated. In the main the comments recorded in this section apply to enamels fused onto sheet and cast iron. Enamel is, however, applied to aluminium, stainless steel, copper and noble metals on account of its aesthetic value and also to confer durability to the base metal. With low melting point metals such as aluminium it is obvious that superb resistance to chemicals is not so feasible as if iron was the base. Nevertheless, such metals are vitreous enamelled in growing quantities and sold, indicating that the range of colour and durability obtained is superior to that possible with alternative finishes. It can justly be claimed that a vitreous enamel coating applied to sheet or cast iron (or indeed any other metal) will confer to the basic shape colour, gloss, texture and a high degree of resistance to corrosive influences.
N. S.C. MILLAR C. WILSON REFERENCES 1. Andrews, A. I., Porcelain Enamels, Gerrard Press, Champaign, Ill., USA 2. Pedda. J. W. G . . ‘Wear and Tear of Enamelled Surfaces’, Inst. Vit. Enam., 9 No, 9, May (1959) 3. Vargin, V. V. (Ed.), 2chnologV ofEnamels. Maclaren & Sons Ltd., 31 and 78 (1967) 4. Krauter, J. C. and Kraaijveld, Th. B., ‘The Corrosion Resistance of Enammelled Articles’, Ins/. Vit. Enam., 21 No. 2. Summer (1970) 5. I.V.E. Technical Sub-committee Report, ‘An Investigation into the Effect of Detergents on Vitreous Enamel’, Bull. Inst. Vi/. Enam., 10, 285 (1960)
BIBLIOGRAPHY Hughes, W., ‘A Report on the Status of Electrodeposition for Porcelain Enamels’, Inst. Vit. Enam.. 20 No. 2. Summer (1%9) Maskell, K. A., ‘Practical Experiences with Electrocoating of Vitreous Enamel’, Inst. Vir. Enam., 20 No. 3, Autumn (1969) Vitreous Enamels, Borax Consolidated Ltd, London SWI (1965)
16.2 Thermoplastics
Introduction There has been considerable growth in the use of thermoplastics as corrosion-resistant coatings in the last 30 years. In the 1950s a few hundred tons per year were being applied by techniques such as fluid-bed coating, plastisol dipping and solution spraying. Since then a large number of other metal finishing technologies have been introduced, including coil coating and extrusion coating. The current tonnage of thermoplastics used in Europe must by now be some tens of thousands of tons. Thermoplastics which are used for corrosion protection can be applied in coatings as thin as 0.025 mm by solution techniques and in excess of 5 mm by extrusion or plastisol dipping. They are used where environmental resistance, chemical resistance, abrasion resistance, sound deadening or cushioning are required. They are used in those market areas that necessitate metallic mechanical strength plus thermoplastic corrosion resistance.
Substrate Preparation Whatever application method is used, the maximum corrosion resistancecan only be achieved if the metalwork is properly prepared. This preparation consists of dressing, blasting and conversion coating.
Dressing Sharp edges must be removed. Thermoplastics have a greater coefficient of thermal expansion than metals. They therefore shrink onto the metal and if sharp edges are present then these will cut through the coating and become exposed. These exposed edges will start to corrode and this will inevitably result in underfilm creep corrosion. Welds should be continuous and porous-free and dressed to remove lumps and weld spatter. Degreasing Mild steel is generally given a temporary protective coating of oil which must be removed. This is done in a vapour degrease tank using chlorinated solvents such as l , l , 1-trichloroethane or trichloroethylene. Alternatively, an aqueous alkaline degreasing solution can be used. It is beneficial to use the former prior to grit blasting and the latter prior to conversion coating. 16:13
16:14
THERMOPLASTICS
Shot or grit blasting Blasting is used to remove rust and to increase the surface area and hence increase apparent adhesion. A variety of abrasives is available, including chilled iron grit and aluminium oxide. The selected abrasive is fired under pressure at the metalwork to create the desired result.
Conversion coating Conversion coatings are chemical solutions which react with the metal surface to create a corrosion-resistant layer onto which the coating can bond. For mild steel iron phosphate is used to attain good adhesion, but it does not give the underfilm corrosion resistance which can be obtained using zinc phosphate. Zinc coatings can be treated with either zinc phosphate or chromate. Aluminium is usually treated with chromate2*
’.
Application Methods The application methods will be categorised by the physical form of the thermoplastic, e.g. liquid, powder, granule. Liquid Application Methods
Spraying Thermoplastics solutions such as those based on p.v.c./p.v.a. copolymers may be applied by conventional paint spraying equipment. Because they are thermoplastic they do not require heat to crosslink them, but they may require some heat to evaporate off the solvents. When the solubility of the thermoplastic is poor at room temperature it may be possible to produce a dispersion in a mixture of diluents and latent solvents. This dispersion may be applied by conventional paint spray equipment. The coated item is placed in an oven where the diluents evaporate off. The latent solvents then dissolve the thermoplastic and evaporate from this solution at a controlled rate, thus producing a continuous film. P.V.F., and p.v.d.f. and p.t.f.c.e. coatings are produced from dispersions of this type. Solvent-free P.V.C. plastisol may be spray applied. P.V.C. spray coatings are currently used extensively by the automotive industry for undersealing of vehicles to prevent corrosion. The plastisol, being resilient, is not cracked or abraded by stone chippings. P.V.C. plastisols have a high viscosity compared with solution and other dispersion systems. Therefore, they have to be applied by airless spray or air-assisted airless spray equipment. P.V.C. coatings must be heated to produce a solid tough coating on cooling. The reasons for this are discussed later in the materials section. Dipping P.V.C. plastisols are used for corrosion protection of pipes, tanks etc. against aqueous chemicals and slurries at temperatures up to 60°C. They are used for the coating of plating jigs to prevent the jigs from being plated and also to prevent corrosion caused by the various acid etching solutions used in the plating process. The coating technique starts by applying a solvent-based adhesive on to a previously pretreated metal substrate. The item is then preheated to 200-25OoC, the exact time and temperature depending on the metal thickness. It is then dipped in the plastisol which partly gels owing to the
THERMOPLASTICS
16: 15
heat radiating from the item. It is then raised out of the plastisol and placed in an oven for final gelation, when, its optimum physical properties and full chemical resistance will be attained.
Coil coating Coil coating is the technique of depositing a fiim of liquid on to a continuously moving thin steel or aluminium sheet. The sheet is uncoiled from a roll at the start of the process and recoiled at the end. The coils are then cut to length and formed into the required shape. During the process the sheet will pass through pretreatment tanks. It is coated with adhesive primers and top coats. Stoving is usually necessary after application of each coat. When P.V.C. is applied at thicknesses in excess of 100pm the coating can be embossed to produce a variety of textured finishes, for example a leather grain effect. The coil coating industry in Europe is using about 50 OOO t of paint per year. This figure includes a significant quantity (between 5 OOO-10 OOO t ) of P.V.C. applied as plastisol and some p.v.d.f. applied from a dispersion. P.V.C. is used extensively in the building industry for external cladding and internal partitions. It is used because it has excellent weathering properties and will protect the substrate against corrosion for periods in excess of 10 years. When it is applied at a thicknessesof about 200 pm it can withstand the hard handling techniques often associated with building sites. P.V.D.F. is used where very high UV resistance is required, e.g. external building cladding in tropical countries. Powder AppLk8tion Methods
Thermoplastics can be produced in the form of a powder by grinding extrusion-compounded granules. The grinding can be carried out at ambient temperature when rotating blade or rotating disc mills are used. Alternatively, those thermoplastics which are heat sensitive or very tough at ambient temperature may be cryogenically ground on a pin-disc mill. Whichever technique is employed, the correct particle size distribution is obtained either by the use of an air classifier or by conventional screen mesh sieving. Fluidised bed The fluidised bed consists of two boxes on top of one
another. The top and larger one contains the powder, and the lower one is separated from it by metal mesh and a semipermeable membrane. Air is pumped under pressure into the lower compartment and then diffuses through the membrane and through the powder. The powder particles are lifted and separated by the air. This results in a considerable reduction in the bulk density so that the item to be coated can easily be submerged in the powder. The pretreated metalwork to be coated is heated in an oven to a temperature of between 260 and 36OoC, depending on the metal thicknesses and the coating to be applied. It is then withdrawn from the oven and dipped into the fluidised powder. Here the fine powder particles are blown onto the hot metal where they melt. After a few seconds (5-10s is normal), the item is removed from the powder and the unfused outer particles are allowed to fuse. Then either the item is allowed to air cool or it is water quenched. The cooling method can affect crystal structure and hence surface finish and
16: 16
THERMOPLASTICS
physical properties. If there is insufficient heat content in the metal further heating may be necessary to fuse the coating fully and produce an acceptable surface finish. The fluidised bed coating technique is used extensively for wirework items such as dish drainer racks, vegetable racks, office trays etc. The technique is also widely used for street furniture e.g. metal lampposts, signposts and balustrading, and for metal office furniture and domestic garden furniture. It also provides chemical corrosion resistance on valves, pipes, couplings etc. Plastics used for fluidised bed powder coatings include polyethylene, P.v.c., nylon, p.v.f.2, p.e.c.t.f.e. and a variety of polyolefins and their copolymers.
Electrostatic powder spraying In the electrostatic powder spraying process plastic powder is blown under pressure from a hopper through a gun. The gun has a barrel 15-45 cm long and 3-5 cm in diameter. At the end of the gun is a charged point. The charge is between 10 and 20 kv and may be positive or negative. The powder picks up the charge and is attracted to the pretreated metal object which is earthed. The item is then placed in an oven to fuse the powder into a smooth coating. The powder particle size should be 20-75 pm. Particles smaller than 20 pm are too light to be transported by the compressed air and form a charged cloud through which further powder does not pass easily. If particles are too large, the charge-to-mass ratio is too low and the particles tend to fall before reaching the earthed metal item. Most thermoplastics are not suitable for spraying because they are too tough. If they were brittle enough to be economically ground to the required fine particle size the physical properties of the coating would be poor. Also, for optimum charge retention the volume resistivity of the powder should be at least Most thermoplastics fall below this. However, Nylon 11 powders are available for general use and P.V.C. powders are used for coating continuous galvanised wire mesh for fencing. Guns have been developed that generate the electrostatic charge by friction rather than by electric high voltage. These are the turbo-electric guns. Their advantage over the electric type is safety. Their disadvantage is lack of control. Flame spraying In flame spraying applications the pretreated items should be heated by passing the flame gently over the metal surface. A skin temperature of 60-100°C is usually sufficient. This ensures that the molten droplets will flow out and fuse together to give a smooth finish with good adhesion to the substrate. The powder is then blown through a very hot flame, melts and is deposited as molten droplets onto the item to be coated. The gases used to produce the flame should not produce an oxidising atmosphere since this will dramatically reduce the physical and chemical resistant properties of any thermoplastic applied. The particle size of the powder should be 150-300 pm. If the particles are too big they will not completely melt and a poor surface finish will result. The flame will inevitably cause some degradation to the surface of the particles. Since the surface area to mass ratio increases as the particle size decreases, very fine particles should be avoided. The process is not widely used in factories but has found a niche in coating large external structures, e.g. large security gates. It can also be used for the
THERMOPLASTICS
16: 17
repair of coatings which have suffered on-site damage. The major concern with this technique is that the polymer will be degraded by the very high temperatures employed. In addition, the process is very operator dependant. To become, more acceptable, a great deal more work needs to be done in equipment design and material technology. Cascade coating The cascade coating technique is used extensively for the external coating of metal pipes with polyethylene to convey natural gas throughout Europe. There are several ways of using this technique but in all cases the pipe is evenly heated to a surface temperature of 250-350°C. Powder is then poured from above, ‘cascaded’, onto the rotating pipe. A second heating operation may be necessary to completely fuse the powder. There are two common variants of the coating method. In the first, the complete length of pipe is heated either in an oven or over a bank of gas burners. The pipe is then moved to an area where the powder is cascaded on to the rotating pipe from a hopper which extends the full length of the pipe. The excess powder is collected in a trough below and recirculated to the hopper. In the second method the pipe rotates and moves laterally through a bank of gas burners or an induction heater, then through a continuous, but narrow, cascade of powder. The cascade comes from a hopper which is at right angles to the direction of movement of the pipe. The pipe continues to travel through a second bank of gas burners where complete fusion of the powder takes place. The coating is applied to protect the steel from corrosion due to the acid or alkaline condition of the soil surrounding the pipe in service. Usually, the process requires three layers. First, an epoxy powder is applied to achieve adhesion to the pretreated metal and therefore resistance to cathodic disbondment. Second, a ‘tie’ layer of polyolefin copolymer is applied and third a thick layer of polyethylene is cascaded, which in effect protects the epoxy from physical damage. Rotational lining The rotational lining technique is derived from the rotational moulding technique, the mould being replaced by the item to be coated. The technique may be used for coating the inside of a11 kinds of cylinders and has found particular favour among the makers of fire extinguishers. A special self-adhesive stress crack-resistant grade of polyolefin is used in the majority of water-based fire extinguishers in the UK. The rotational lining technique consists of pouring a predetermined weight of polymer powder into the preheated cylinder. The cylinder is then rotated in two perpendicular axis while the outside of the cylinder is heated. The heat may be from direct radiant burners or the complete rig may be positioned in an oven. The item must be rotated during the cooling cycle to prevent sagging. To reduce the possibility of polymer degradation and to optimise cycle time, it is essential that the powder is heated to the minimum temperature that will ensure the production of a porous-free, uniformally thick, coating inside the cylinder. Miscellaneous powder coating methods Apart from the coating techniques described briefly above, the jobbing or custom coater has a whole armoury of other methods which are more or less related to those described above.
16: 18
THERMOPLASTICS
Channelling This technique is used for coating the inside of a pipe. The pipe, which is continuously rotated, is heated over a bank of heaters stretching the length of the pipe. The required amount of thermoplastic powder is weighed and put into a metal channel. The channel is then put inside the pipe, inverted to empty it, and withdrawn. The skill is in removing the channel without badly scoring the coated surface. When full fusion of the powder has occurred, the heat is turned off and the pipe continues to rotate until the coating has solidified. Flock spraying This technique is used where electrostatic spraying is inappopriate, e.g. where thick coatings are required. The pretreated metal is heated and the powder is blown onto the workpiece from a flocking gun which is similar to a conventional wet paint gun but with no needle and with the nozzle 1-2.5cm in diameter. The metal should be preheated to a temperature sufficient to fuse the powder without further heating, but occasionally it may be necessary to apply a naked flame over the surface to ensure a good finish. This technique can be used for coating the flange ends of pipes which have been lined by channelling. Granular Application Methods The two major plastics processing techniques of extrusion and injection moulding are used for coating metals.
Extrusion In very simple terms the extruder is a heated cylinder containing a rotating screw. There is a hopper at one end to supply the plastic granules and a die at the other through which the molten polymer is extruded. The technique is widely used for producing garden hose, automotive trim, window profiles, plastic films etc. But it is also used for the corrosion protection of metal tube, rod and wire. Fencing wire is coated in PVC using this technique. The wire may then be woven into chainlink mesh fencing. However, there is normally no adhesion between the coating and the wire. Adhesion can be achieved if the fluidised bed process is used. Injection moulding The injection moulder is a machine which first melts a thermoplastic and then injects that molten polymer into a mould. Such items as baskets, bowls, bins, telephones and electronic housings are produced by this technique. It can be used for lining valves. In this case the valve would be used as part of the mould. Very thick coatings are produced which give chemical resistance to the valve. At the same time, the metal valve housing will protect the valve from mechanical damage. The polymers used for this process include polyethylene, polypropylene and p.v.d.f.
Materials Liguids
PVC/PVA copolymer solutions Polyvinyl chloride/polyvinyl acetate copolymers can be readily dissolved in blends of aromatic hydrocarbon,
THERMOPLASTICS
16: 19
ketone and ester solventsto produce solution vinyls. Terpolymerscontaining acid groups can be blended with the copolymer to enhance adhesion to metal substrates. Plasticisers can be added to improve flexibility and conventional P.V.C. stabilisers are used where thermal or UV resistance is required. They are applied by wet paint spray techniques and have the advantage, over other paint systems, of long-term flexibility. Conventional alkyd systems may have an initial degree of flexibility, but within 12 months outside become rigid and then crack due to thermal expansion and contraction of the substrate. This phenomenon is less likely to occur with a well formulated vinyl solution.
P.V.D.F.
Polyvinylidene fluoride (p.v.d.f. or p.v.f.2) dispersions are applied by the coil-coating process. They are blends of p.v.d.f. resin and acrylic. The combination produces a system which has excellent weatherability and which can be bonded via an adhesive primer to a galvanised steel or aluminium substrate. They are used where prolonged exposure to high UV resistance is required, such as prestige building cladding in tropical and subtropical climates.
P.V.C. plastisols P.V.C. plastisols are liquids which contain little or no solvent/diluent. They consist of a blend of polyvinyl chloride (P.v.c.) resins, plasticisers, stabilisers, viscosity depressants, pigments and sometimes fillers. Whatever application method is used, there is always a heating step. When P.V.C. plastisol is heated to over 100°C the P.V.C. resin which is suspended in plasticiser stabiliser etc. starts to dissolve in the plasticisers. When solution is complete the system is cooled to room temperature and a solid homogeneous coating results. The thermal and U V resistance will depend on the stabiliser systems used. The hardness of the coating will depend on the amount and type of plasticiser used. Correct selection of the plasticiser can permit the use of the plastisols at high or low temperatures, provide fire resistance or oil resistance. Plastisols can be produced in a range of gloss levels from 80 units down to 10 gloss units. The application method used depends on the intended use of the item. Spraying is used by the automotive industry to underseal the substructure of vehicles to provide corrosion resistance. Plastisol coatings are tough enough to resist mechanical damage from stones and other objects thrown up from roads. Coil coating is used to coat galvanised steel sheet. The building construction industry uses this for the exterior cladding and roofing of buildings. Lifetimes of 15 years and more can be expected before first maintenance. Internal partitioning is produced by the same process. Shelving and electronic equipment housing are also produced from coil coated steels. Dipping is used to apply coatings of 1-6 mm thick to pipes, tanks, vessels, etc. in a wide range of uses: 1. water cooling pipework in power stations; 2. pipes, tanks, extraction hoods and ducting in the chemical industry for many acid, alkaline and neutral solutions up to 60°C; 3. pipework in the water section of oillwater separation plants on offshore oil platforms;
16:20
THERMOPLASTICS
4. hoppers and stillages to reduce noise and damage to components in the
engineering industry. Powders
Polyethylene Polyethylene is one of the lowest cost thermoplastic materials. Hence when looking for a coating or lining it is generally considered first. Three types of polyethylene are available:
a high-pressure high-temperature reaction process. This creates a molecule with a high degree of random branching. Thus crystallinity and hence density are low. 2. High density polyethylene produced by a low-pressure low-temperature process involving Ziegler-Natta catalysts. This creates low levels of branching and hence a high degree of crystallinity. 3. Linear low density polyethylene is also produced by the low-pressure low-temperature Ziegler-Natta catalyst route. Other monomers are incorporated such as butene or octene, which disrupt the crystallinity and reduce density. 1. Low density polyethylene produced by
All polyethylenes are soft, flexible and resistant to acids and alkalis up to 60°C. They retain this flexibility down to -40°C. Hence they have good resistanceto impact even at low temperatures. However, unless correctly formulated they can suffer from environmental stress cracking (ESC), poor adhesion and UV degradation. ESC is the phenomenon which occurs when a thermoplastic is put under stress, e.g. bent, in a particular environment and prematurely cracks or crazes. Alcohol and detergent are examples of agents that can cause ESC in polyethylenes.
Fluidised bed coating Unmodified polyethylenes are used for coating wirework items such as vegetable racks, record racks etc. Light stabilised grades are used for coating garden wirework such as compost bins or hanging baskets. Highly modified systems containing adhesion promoters are used for chemical resistant applications such as coating pipes, valves, etc. Polyolefin copolymers Although there is a wide variety of these available, the only one currently commercially available as a compounded powder is saponified EVA. This is reported to have good weatherability and will not suffer from ESC. One major advantage this coating has is that it can be applied by the fluidised bed process at low temperatures and this offers the possibility of coating temperature-sensitive metals such as galvanised steel. Polyolefin alloys Plascoat Systems Ltd. has developed a range of polyolefin alloys in its Performance Polymer Alloy (PPA) range. The exact compositions of these are secret. These products have been tailor-made to meet the needs of specific markets, e.g. (a) Lining the inside of aqueous-based fire extinguishers. This requires a coating material which will adhere to the inside of the fire extinguisher. It is applied by a rotational lining technique and must not melt and sag
THERMOPLASTICS
16:21
during the curing of the epoxy powder paint used on the outside. Furthermore, it must not suffer from stress cracking in service. (b) The lining of hot water cylinders. This requires a coating which will adhere well to metal. It must have good resistance to water at 80°C and be largely impermeable to water to prevent corrosion of the metal substrate. (c) Coating of bus-bars. The coating must have excellent electrical resistance. It must be capable of being applied at thicknesses of up to 2 mm. In this case there is no adhesion so that the coating can easily be stripped off to allow contacts to be made after installation if necessary.
P.V.C. P.V.C. powders are blends of P.V.C. resin, plasticisers, stabilisers and pigments. The plasticisers soften the coating and increase impact strength. The amount normally used in P.V.C. powder creates a coating with a Shore A hardness of 80-90 units. With this level of hardness the coating will be resistant to impact damage down to -5°C and at the same time will not be so soft as to significantly affect resistance to impact penetration at higher temperatures. The stabilisers are selected to give adequate thermal stability during processing and excellent U V resistance in service. Correct plasticiser selection can decrease the water permeability of the coating and increase the long-term adhesion and hence corrosion resistance. The pigments are present to give aesthetic appeal, but they must be correctly selected for optimum resistance to the effects of weathering. After metal pretreatment it is essential that a suitably formulated adhesive primer is used, because P.V.C. does not itself adhere to metals. FIuidised coating In the UK P.V.C. powders are widely used for coating street furniture and fencing posts. Street furniture includes road signposts and brackets, lampposts, balustrading and seating. In the UK and the rest of Europe P.V.C. coatings are used for welded wire mesh used for fencing. In the USA P.V.C. (vinyl) is a general coating material and is used for coating, for example, dishwasher baskets. Electrostatic spraying PVC can be applied by the electrostatic process to continuous galvanised wire mesh.
Nylon 11 Nylon 11 is a hard abrasion-resistant, scuff-resistant coating. When correctly formulated and applied, it can be used for exterior application. It has good resistance to solvents and to a range of alkalis and salt solutions up to 80°C. If water quenched, the coating has excellent impact strength. However, Nylon 11 is crystalline and pull-back from sharp edges can be a problem. It is therefore essential that metal work is well radiused. Nylon 11 is applied using a fluidised bed process to a wide variety of substrates including metal chair frames, door furniture and wire dishwasher baskets. It can also be applied by electrostatic spraying, but generally only where the application is decorative and where the metal work is thin, Le. less than 0.2 mm. P.V.D.F.
Polyvinylidene difluoride is a coating which offers resistance to
Table 16.4 Properly
Relative density Impact strength Tensile strength Hardness Abrasion Taber (H18 load 5008) External weathering Chemical resistance Acid Alkali Solvent
(g/cm 3, (J)
(MPa) (Shore A) (rng/lOOO cycles)
Properties of thermoplastic powder coatings Nylon
P. V.D.F.
1.04 4.5
1.78
40 98
51
Polyethylene
Polyolefin copolymer
Polyolefin PPA 65 alloys
0.93 2 10.3 10
0.33 13 95
1.03 1.32 13 95
I .26 1.7 17 85
415
-
210
50
Poor*
Good
Poor
Excellent
33 Good
Fair Fair Poor
Fair Good Poor
Good Good Poor
Good Fair Poor
Good Good
-
P. V.C.
Poor
-
99
Excellent Excellent Excellent Fair
4
B
E
r
2
=!
8
THERMOPLASTICS
16 :23
chemicals up to 90°C.It is more resistant to stronger acids and alkalis than the above-mentioned coating materials. It is also a hard abrasion-resistant coating. It is applied using a fluidised bed process, generally in two coats. A precompounded blend of p.v.d.f., corrosion-inhibitive pigments and adhesive components is applied first, followed by a top coat of pure p.v.d.f. The primer coat protects the metal and the top coat protects the primer coat from attack by the chemicals. The properties of the thermoplastic powder coatings are summarised in Table 16.4. Granules
The range of thermoplastic materials that can be extruded or injection moulded is too large and varied for coverage in this book.
W. G. O’DONNELL
16.3 Temporary Protectives
Definition Many metal articles have to be transported and stored, sometimes for long periods, and are then used with their working surfaces in the bare state. Unless these surfaces are protected between manufacture and use, most of them will rust and corrode due to the effect of humidity or atmospheric pollution. The materials used for such protection are called temporary protectives as they provide protection primarily for the transportation and storage period. The significance of the term temporary lies not in the duration of the efficacy of the protective, but in the fact that it can easily be removed, so that the protected surfaces, can if necessary, be restored to their original state. They provide a water and oxygen-resistant barrier by reason of their blanketing effect and/or because of the presence of naturally occurring or added inhibitors which form an adsorbed layer on the metal surface.
Types of Temporary Protectives There are many temporary protectives on the market and it would be impracticable to describe them individually. However, they may be classified according to the type of film formed, i.e. soft film, hard film and oil film; the soft film may be further sub-divided into solvent-deposited thin film, hot-dip thick film, smearing and slushing types. All these types are removable with common petroleum solvents. There are also strippable types based on plastics (deposited by hot dipping or from solvents) or rubber latex (deposited from emulsions); these do not adhere to the metal surfaces and are removed by peeling. In addition there are volatile corrosion inhibitors (V.C.I.) consisting of substances, the vapour from which inhibits corrosion of ferrous metals. Soft-film Materials
Those deposited in the cold from a solvent usually consist of lanolin or petrolatum mixtures in such solvents as white spirit or coal tar naphtha. The film is thinner than other soft films deposited by different methods. 16:24
TEMPORARY PROTECTIVES
16:25
Materials applied by dipping the article to be protected in the hot molten material are usually based on petrolatum. Corrosion prevention depends largely on the barrier provided by the film, but for improved protection, corrosion inhibitors are added. The film may be relatively hard and waxy or quite soft like pharmaceutical petroleum jelly. The smearing types of material are usually lubricating grease compositions, i.e. blends of soaps and lubricating oil, but may be mixtures containing petrolatum, oil, lanolin or fatty material. They are softer than the hot-dip materials to permit cold application by smearing. The slushing compounds are a variant of the smearing types, and possess some flow properties at room temperature so that brush marks produced during application are reduced. Some materials contain solvent, so that they are free-flowing as applied, but stiffen when the solvent evaporates. Had-tilm Materials
These were developed to facilitate handling after treatment and to avoid contamination of adjacent components. The films are deposited in the cold and should be tough and neither sticky nor brittle. The deposited films may be plasticised resins, bitumens, etc. which are varied according to the subsidiary properties required, such as transparency and colour. The solvents used vary according to the solubility of the ingredients, drying time requirements, flammability and permissible toxicity in given circumstances. As with the soft-film solvent-deposited materials, the surface coverage is large, and for this reason, and because they can be applied at room temperature, hard and soft-film solvent-deposited protectives are widely used. Oil-type Materials
These are usually mineral oils of medium or low viscosity, which contain specific corrosion inhibitors and anti-oxidants. In spite of the relatively low protective properties of the fluid films, which are not nearly so great as those of the previously described solid films, these materials have an established field of use on the internal surfaces of tanks and assembled mechanisms, and where solid material or solvent cannot be tolerated. Strippable Coatings
The most important of these to date are those applied by hot dipping. Many are based on ethyl cellulose and the dipping temperature is comparatively high (about 1 9 0 O C ) . They rely mainly on the thickness (= 2mm) and toughness of the coatings for their extremely good protective properties, and they have the added advantage of giving protection against mechanical damage so that little added packaging is required for transport. Re-use of the material is frequently possible. The disadvantages are the necessity for special dipping tanks and cost; this latter may, however, be offset by saving in packaging materials.
16:26
TEMPORARY PROTECTIVES
The strippable films deposited from solvents in the cold are much thinner 0.05-0.25 mm) than those from the hot-dip materials, and their protective properties are not nearly so good. A possible difficulty which must be watched for is the development of brittleness on ageing and consequent difficulty of stripping. Latex films containing inhibitors such as sodium benzoate have been found to deteriorate under tropical conditions, but may have a use in more temperate climates. ( 5
Specbl hl0difi;cationsof the Afarementioned Types
These have been developed for special uses. For example, since petroleumbased materials harm natural rubber, a grease based on castor oil and lead stearate is available for use on the steel parts of rubber bushes, engine mountings, hydraulic equipment components, etc. (but not on copper or cadmium alloys). Some soft-film solvent-deposited materials have water-displacing properties and are designed for use on surfaces which cannot be dried properly, e.g. water-spaces of internal combustion engines and the cylinders or valve chests of steam engines. A recent application of this type of fluid is assistance in the removal of ingested salt spray from jet aircraft compressors and the neutralisation of corrosive effects. Other types of water-displacing fluids are claimed to have fingerprint neutralising properties or to be suitable for use on electrical equipment. Some oil-type materials serve temporarily as engine lubricants and contain suitable inhibitors to combat the corrosive products of combustion encountered in gasoline engines. Volatile corrosion inhibitors (see also Section 17.1) are a special type of protective, which when present as a vapour inhibit the rusting of ferrous metals. They are generally used as an impregnant or coating on paper or synthetic film; as a powder, either loose or in a porous container; or in the form of a 5% w/v solution in non-aqueous solution (e.g. methylated spirits) with application by either swab or spray. Their effectiveness in preventing corrosion depends not only upon the inherent activity of the material but also upon their volatility and rate of release from the supporting medium. Being volatile, some form of enclosure is necessary for continued effectivenesswhether it is the closing of orifices with bungs or overwraps when protecting internal surfaces, or by sealing the outer container for other packed stores. Volatile corrosion inhibitors should be used with caution in the presence of non-ferrous metals which may be attacked, particularly in the presence of free water. Care should also be taken with painted surfaces and with some plastics and other organic materials which may become discoloured or damaged. The types of temporary protectives in general use are given in Table 16.5.
General Scope of the Materials Temporary protectives against corrosion should be used only where removal is subsequently necessary for the fitting or the working of surfaces to which they are applied.
16: 27
TEMPORARY PROTECTIVES
Table 16.5 Types of temporary protectives in general use Type of protective
Typical ingredients'
Solvent-deposited hard film ((I) ordinary grade
( a ) Plasticised bitumens, plasticised resins, white spirit, coal tar naphtha, chlorinated solvents (b) water-displacing (b) As (a) above grade together with waterdisplacing agents Solvent-deposited soft film ( a ) ordinary grade ( a ) Lanolin, petrolatum, with and without specific corrosion inhibitors and anti-oxidants, white spirit, coal tar naphtha, chlorinated solvents (b) water-displacing (b) As ( a ) above grade together with waterdisplacing agents Hot-dipping soft Petrolatum, lanolin, film with and without specific corrosion inhibitors Smearing
Metallic soap and mineral oil, soft petrolatum, lanolin (castor oil/lead stearate for rubbercontaining components)
Method of application
Properties offilm
Dipping spraying, brushing
Solid, thin, tough, non-sticky, removable by wiping with solvent
Dipping, spraying, brushing
Solid, thin, greasy, removable by wiping with solvent
Dipping in molten material
Solid, thick, waxy or greasy, removable by wiping with solvent or immersing in hot oil Solid, thick, greasy, removable by wiping with solvent
Smearing, brushing
These coatings are designed to protect packaged engineering materials against corrosion due to a humid atmosphere, in both rural and general industrial conditions, during transit and storage in temperate and tropical climates. Where conditions are severe, extra packaging may be required or, in the case of thick soft-film materials, extra thicknesses may be applied. The coatings are also often used to protect unpackaged spares during shelf storage. In normal thicknesses, temporary protectives are unsuitable for outdoor exposure and they should be protected against gross liquid water by coverings or wrappings. The petrolatum-based thick-film material and some greases, however, will give adequate protection outdoors if they are applied extra thickly. Protection cannot be expected if the surfaces remain in contact with waterlogged packing material. Corrosion preventives should be applied to surfaces which are clean and dry or corrosion may well continue beneath the coating. Materials with
16:28
TEMPORARY PROTECTIVES
Table 16.5 (continued) T Y P of
protective Slushing
Oil Strippable (a) hot-dipping grade
(b) cold applied grade
Volatile corrosion inhibitor (V.C.I.)
Typical ingredients*
Method of application
Properties o/J[rn
Smearing, brushing
As for smearing protective
Dipping rinsing, spraying
Liquid, thin, oily
(a) Ethyl cellulose. cellulose acetate butyrate, mineral oil. plasticiser. resins, stabilisers
(a)Dipping in molten material
(0)
(b) Vinyl copolymer resins, plasticisers, stabilisersflammable or nonflammable solvents Organic amino salts (e.g. dicyclohexylamine nitrite, cyclohexylamine carbonate)
(b) Spraying, dipping
Metallic soap and mineral oil, oil-softened petrolatum, lanolin, small amounts of solvent Mineral oil, specific corrosion inhibitors and anti-oxidants
From solution by spraying, as a powder by sprinkling, by wrapping with V.C.1.-impregnated paper
Solid, tough, non-adherent, often leaves oily film with lubricating properties; film removed by stripping (b) Solid, tough, non-adherent film, removed by stripping
Adsorbed, nonvisible film
*Some details of typical compositions. where these are available, are given in Petroleum. Oils and Lubricants (POL) and Allied Producls. Defence Guide DG-12. Section IV. Ministry of &fence, H.M.S.O.. London (1968).
special properties such as water displacement or the ability to neutralise fingerprints should not be used in place of drying and clean handling, but only where the application demands it.
Causes of Failure It practice it is usually difficult t o establish the reasons for failure as a number of factors may be simultaneously responsible, such as (a) application of the protective to dirty surfaces, (b) carelessness in application, (c) inherent inadequacy of the material, ( d ) exposure to unreasonably severe conditions, (e)inevitable difficulties in application. Point (c) includes inadequacy not only in protective properties but, in the case of the hard-film materials, in certain physical properties, e.g. the film may become brittle and flake when handled, may remain too sticky and become contaminated with dirt or adhere to the
TEMPORARY PROTECTIVES
16:29
wrapping paper more strongly than to the surface to be protected, may age to form an insoluble material and become difficult to remove, or may not remain flexible and adherent at low temperatures. Point ( d ) includes, for example, the use of soft-film materials in hot conditions at temperatures too near to their melting point. As regards (e), it may be difficult to avoid thin places in the film arising from contact with other surfaces during the process of application, drying-off of the solvent, or cooling; when such thinning occurs, good surface-active properties are advantageous. In this connection, it may be pointed out that scraping in transit and stacking, and local thinning due to grit, dirt, etc. are common; it follows therefore that shelf storage of unpacked items should be avoided if possible.
General Comments on Application Application by dipping gives the most complete film, is the most economical in material, and is usually the quickest for large quantities of articles. This method should be chosen whenever possible. Spraying is the next best. Brushing and hand-smearing should be adopted only when dipping or spraying is not feasible. During the dipping process, articles with recesses should be rotated in the bath so that air can escape. Dipping baths should be kept covered when not in use to prevent contamination, and, in the case of solvent-containing materials, to prevent concentration by evaporation of the solvent, as this would lead to excessivefilm thicknesses and long drying times. The composition of a bath of solvent-containingmaterial should be checked periodically. Unaided evaporation of the solvent from solvent-deposited films is usual, but the process can be speeded up by blowing air over the articles or by gentle warming; the heating, however, should not be excessive. During hot-dipping in petrolatum-based materials, film thickness can be varied by altering the temperature of dipping and the duration of immersion. The petrolatum will first chill on to a cold article put in the bath, the solid coating bridging small crevices. This may give sufficient protection, but it may be desirable for the article to attain the temperature of the bath so that the molten petrolatum will penetrate into all the crevices, e.g. between the ball and race of a rolling bearing. The article may then be withdrawn, allowed to cool and given a quick dip to build up the film thickness.
Choice of Temporary Protective Hard-film protectives can be applied to most types of single articles and are especially suitable in mass-production systems. They should not be applied to assemblies because the hard film is liable to cement mating surfaces together and considerable difficulty may arise in the removal of the protective film. This type of protective should be removed before the article is put into use. The soft-film solvent-depositedtype can be used broadly for the same purposes as the hard-film type. A grease-resistant wrapping is required as an inner wrapping (as for all soft-film types) in packaging. Grades of this
16 :30
TEMPORARY PROTECTIVES
material, consisting essentially of lanolin in a solvent, have been found to give better protection to packaged articles than some of the best available hard-film materials, and are to be preferred for articles with very high precision surfaces. The film is usually dispersable in lubricating oil and it is therefore not so important to remove it from surfaces when an article comes into use except when it has become contaminated with grit and dirt. The thick soft films produced by hot dipping are suitable for highly finished as well as normal machined surfaces. Grades with drop-points substantially higher than 50°C are preferable for tropical storage as otherwise marked softening and possible thinning of the protective film is likely to occur. These films can be applied to many types of assemblies, the chief exceptions being assemblies with inaccessible interiors that cannot readily be blanked-off and fine mechanisms where any residue might interfere with the free movement of parts or their subsequent lubrication with low viscosity oil. These films can also be used on parts which might be affected by the solvent from the thin or soft film protectives, but they should not be applied to items having plastics or leather components. Greases are usually applied by brush or smearing; the brush must be sufficiently stiff to give intimate contact with the surface yet not so stiff as to leave deep brush marks. Greases should not be melted and therefore cannot be applied by dipping or spraying; also, no attempt should be made to dissolve them in a solvent for application. They are particularly useful where only part of the surface of the item requires protection, because of the ease of application by cold smearing. They can be used in this way also in conjunction with solvent-deposited protectives for assemblies of a low degree of complexity, by coating screw threads and filling clearance spaces before dipping the article in the solvent-containing protective. Grease films can be made thick enough to give the desired level of protection. Wrapping is desirable to protect the very soft film. Removal before use is chiefly for the purpose of removing grit and dirt. The slushing material finds its most useful application on big machinery requiring protection of large areas during storage or during intervals of idleness in machine shops. The effect of dust and dirt contamination should therefore be considered an important factor in assessing the quality of these materials. The lower protective quality of oil-type materials largely restricts their use on internal surfaces of, for example, internal combustion engine cylinders, and gear-box and back-axle assemblies of motor vehicles. Such materials are widely used to fulfil the simultaneous function of a protective and a lubricating oil; e.g. in sewing machines the protective can also serve as a lubricant during its initial period of use. The functions of corrosion inhibitor and hydraulic oil are also often combined. Oil-type materials are also used on small nuts, screws and washers which cannot easily be protected by solid-film materials; in this case protection must be reinforced by good packaging. The hot-dip strippable coating is applicable when a high standard of protection from corrosion and mechanical damage is required, as on gauges and tools which so often have their working surfaces facing outwards. Assemblies must have orifices plugged so that molten material cannot penetrate during the dipping.
TEMPORARY PROTECTIVES
16:31
Volatile corrosion inhibitors are particularly useful when oil, grease or other adherent films are unsuitable. They should be used in conjunction with a primary wrap which should form as close an approach to a hermetically-sealed pack as possible. They are widely used to provide protection to precision tools, moulds and dies, and also on a larger scale to car body components.
General Remarks The listing of so many types of protective might indicate some complication in use. It should, however, be realised that the materials are to some extent interchangeable, and in most works it is seldom necessary to have more than two or three materials. It is emphasised that protection should be given by the manufacturer of the article as soon as possible after its fabrication; if stocks have to be held in a part-finished state, protection should also be given during this period. This is important for cast iron because corrosion once started is difficult to stop. If the conditions at the receiver’s works or depot are particularly severe, the maker’s protective processes should be appropriately supplemented. The bibliography given below is classified according to the aspect of the subject mainly dealt with, but some references, of course, deal with several aspects. In addition there is a considerable body of patent literature concerning specific inhibitors.
Recent Developments Strippable coatings based on such resins s vinyl, acrylic and polyethylene are finding increasing favour for applying to finished products to protect them during transit, the coating being left on the product until it reaches the dealers showroom or, in some cases, the consumer. These coatings offer excellent temporary protection against moisture, chemicals and weathering and some stand up well to such fabricating techniques as bending and deep forming. The coatings are easy to apply and some remove simply by piercing the film and peeling it off, others by washing away by applying an alkaline solution or solvent. The toxicity of lead-containing greases has led to alternative products being used for the protection of components where the product is likely to come in contact with rubber. Of those products considered silicone-based greases have been found to be particularly suitable and their application to hydraulic equipment components such as brake cylinders, where they can provide internal protection against corrosion both during transit and use, has been found particularly beneficial. Corrosion-inhibited petroleum-based waxes deposited from solvent are finding application in both the automotive and aircraft industries for the supplementary protection of hollow sections of the finished product. These waxes are applied by airless or air-assisted pressure-feed spraying techniques
16:32
TEMPORARY PROTECTIVES
to clean and dry, but often painted, surfaces to provide increased protection against corrosion dur to humid and corrosive atmospheres during both transit and use. T. N. TATE D. R.A. SWYNNERTON E. W. BEALE BIBLIOGRAPHY
General Description of Types and Mode of Use Albin, J., Iron Age.. 155 No. 23, 52 (1945) Anon., Mod. Packag.. 1764 (1944)' Bayliss, D.A. J., Prot. Coat. Linings, 1 No. 3 (1984) Boyer, J. R. C.. Steel, 116 No. 24, 129 and 176 (1945) Brookman, J.. Anti. Corros, Methods Mater., 31 No. 7 (1984) Brookman, J., Anti. Corros. Methods Mater., 32 No. 4 (1985) Carpenter, H.B., Iron Steel Engng.. 24 No.9,13 (1947) Elgar, D.,J. Finis. I d . , 1N No. 1 1 (1977) Could, B., Iron Age, 155 No.24, 66 (1945). Houghton, E. F. et al., Sfeel, 116 No. 14. 106 and 149 (1945) Larson, C. M.. Nut. Petrol. News, 37. R609 (1945) Lurchek, J. G.,Iron Steel Engng.. 26 No.5, 82 (1949) Maim, C. J.. Nelson. H. B. and Hiatt, G. D., Industr. Engng. Chem., 41, 1065 (1949). Mock, J. A.. Mater. Eng., 90 No. 3 (1979) Petroleum, Oils and Lubricants (POL)and Allied Products, Defence Guide DG-12, Ministry of Detence, H.M.S.O. London (1968) Pohl, W., Erdol u. Kohle. 8, 552 (1955) Prince, W. H., Mod. Plasf., 22, 116 (1944); Rhodes, C. M. and Chase, G. F., Mod. Packag., 18, 117 (1945) Sellei. H. and Lieber, E., Corros. Mat. Prof., 5, 10-12 and 22 (1948) Shearon, W. H.and Horberg, A. J., Industr. Engng. Chem., 41, 2 672 (1949) Smith, T., Anti. Corros. Methodr Mater., 31 No. 3 (1984) Stroud, E. G. and Vernon, W. H. J., J. Appl. Chem., 2, 173 (1952)t Temporary Protection of Metal Surfaces Against Corrosion (During Transport and Storage), BS 1 133: Section 6: 1965 (also deals extensively with testing) Trabanelli, G., Proc. Eur. Fed. of Corros 74fh Manifestation, Budapest (1974) Waring, C. E., Mod Packag., 19, 143 and 204 (1946). Zorll, U., Adhesion, 19 No.9 (1975)
Clarke, S. G. and Longhurst, E.E.,Selected Government Research Reports (London), 3: Protection and Electrodeposition of Metals, 135. H.M.S.O., London (195111 Hickel. A. E., Petrol Refn., 27, 424 (1948) Inst. Petrol. Protectives Panel, J. Inst. Petrol., 40,32 (1954) McConville, H. A., Gen. Elect. Rev., 49 No. 10, 30 (1946) Schwiegler, E. J. and Berman, L. U., Lubric. Engng., 11, 381 (1955) Stroud, E. G. and Rhoades-Brown, J. E., J. Appl. Chem., 3, 281 (1953)l Symposium on the Testing of Temporary Corrosion Preventives (15 authors), J . Inst. Petrol., 36, 423(1950) Walters, E. L. and Larsen, R. G., Corrosion, 6, 92 (1950) Wright, W. A. S., Amer. Soc. Test. Mater., Spec. Tech. Pub. No. 84. 18 (1948)
*Hotdip strippablc coatings tRubbcr-latex-bad strippable coatings !%Lanolin solutions
TEMPORARY PROTECTIVES
16:33
Investigations of Mode of Action Baker, H. R., Jones, D. T. and Zisman, W. A,, Industr. Engng. Chem., 41, 137 (1949) Baker, H. R., Singleterry. C. R. andSolomon, E. M., Indusrr. Engng. Chem., 46, I 035 (1954) Baker, H . R. and Zisman. W. A.. Industr. Engng. Chem.. 40, 2 338 (1948) Barnum, E. R., Larsen, R. G. and Wachter, A., Corrosion, 4, 423 (1948) Bigelow, W. C., Pickett, D. L. and Zisman. W. A., J . Colloid Sci., 1, 513 (1946) Cessina, J. C., fndustr. Engng. Chem., 51. 891 (1959) Hackerman, N. and Schmidt, H. R., Corrosion, 5, 237 (1949) Hackerman, N. and Schmidt, H. R., J. Phys. Chem.. 53, 629 (1949) Kaufman. S. and Singleterry, C. R., J . Colloid Sci.. 7 , 453 (1952) Kaufman, S. and Singleterry, C. R., J . Colloid Sci., 10, 139 (1955) Pilz, G. P. and Farley, F. F., Industr. Engng. Chem., 38, 601 (1946) Van Hong. Eisler. L., Bootzin, D. and Harrison, A., Corrosion, 10, 343 (1954)
17
CONDITIONING THE ENVIRONMENT
17.1 Conditioning the Atmosphere to Reduce Corrosion 17.2 Corrosion Inhibition: Principles and Practice 17.3 Mechanism of Corrosion Prevention by Inhibitors 17.4 Boiler and Feed-Water Treatment
17:3 17:lO 17:40 17:66
17.1
Conditioning the Atmosphere to Reduce Corrosion
The impurities normally present in uncontrolled atmospheres are capable of producing serious corrosion on many metals and alloys which do not corrode significantly in clean, dry air (Section 2.2). It is therefore in principle possible to prevent corrosion by purifying the atmosphere, or by using a volatile corrosion inhibitor. In extreme cases, pure, dry nitrogen under positive pressure can be used. These methods will seldom be practicable with working equipment, but they may offer the most attractive solution in transport or storage, especially since they are often very effective against the particular hazards of these conditions. Temporary protectives (Section 15.3) may also be used. The most important corrosive agents to be considered are water vapour, acid fumes (particularly sulphur dioxide) salts and hydrogen sulphide. Water plays an essential part in stimulating attack by all the other agents, except hydrogen sulphide, so that drying the atmosphere is the most important single means of preventing corrosion. Control of other contaminants will, however, be important where satisfactory drying is not practicable.
Control of Relative Humidity At high relative humidity the common corrosive agents produce a film of aqueous electrolyte on exposed metal surfaces. No significant corrosion results on iron, zinc, aluminium, copper or their alloys (apart from tarnishing by hydrogen sulphide), unless the relative humidity is above 60% (Section 2.2). In packaging and storage, the relative humidity is usually kept below 50%. Packages are most conveniently protected with desiccants, but for larger volumes, drying by cooled surfaces may be used, and in storerooms, the relative humidity can be kept down by heating. Desiccants and Desiccated Packages
Desiccating agents used in corrosion prevention must be cheap, easy to handle and non-corrosive. These requirements rule out many of the familiar laboratory desiccants, and in practice the most common packaging desiccants are silica gel, activated alumina and quicklime (calcium oxide). Activated 17:3
17 :4
CONDITIONING THE ATMOSPHERE TO REDUCE CORROSION
clays are sometimes also used, and for very low relative humidities, molecular sieves. Silica gel and activated alumina present few practical problems. They are easily reactivated after use by heating in a ventilated oven, to 130-300°C for silica gel, and 150-700°C for activated alumina. British standard specifications have been published for desiccants for packaging ',', which regulate the contents of soluble chloride and sutphate, dust content and absorptive capacity. Quicklime is less easy to handle, and swells considerably on hydration. It is cheap, however, and is often used on open trays to protect process equipment, machinery, furnaces, etc. during shut-down periods. If it is accidentally flooded with water, the slurry of hydrated lime provides an alkaline medium in which uncoated steel surfaces will remain without rusting. Packages intended for use with desiccants must have low permeability to water vapour. It is therefore necessary to consider the design of the package in relation to the storage life required. This subject is beyond the scope of the present work, and guidance should be sought from standard textbooks on packaging3. The B.S.I. Packaging Code4 includes sections on desiccants, temporary protectives and the use of various types of packaging materials. The following formulae are used for calculating the weight of desiccant required for a given package: 1. For tropical storage with average water-vapour pressure 3 2 kN/m2 W = 40ARM + Dunnage Factor 2. For temperate storage with average water-vapour pressure 1 - 0 kN/M* W = 1 lARM Dunnage Factor 3. For completely impervious packages:
-
+
V
+
W = - Dunnage Factor 6 Where W = weight (g) of 'basic desiccant' (Le. one which absorbs 27% of its dry weight of moisture in an atmosphere maintained at 50% r.h. at 25"C), A = area (m') of the surface of the desiccated enclosure, D = weight (g) of hygroscopic blocking, cushioning and other material inside the barrier (including cartons, etc.), M = maximum time of storage (months), R = water-vapour transmission rate of the barrier (g m-' d-') measured at 90% r.h. differential and 38°C and V = volume (litre) of the air inside the barrier.
Dunnage Factor is D/5 for timber with moisture content higher than 14%, 0 / 8 for felt, carton board and similar materials, and D/10 for plywood and
timber with moisture content less than 14%. Rates of transmission may be affected by creasing, scoring, etc. especially for waxed papers, and of course also strongly depend on thickness. Information can be obtained from suppliers of materials, or measurements can be made according to a method given in BS 3 177:19596which includes a table
CONDITIONING THE ATMOSPHERE TO REDUCE CORROSION
17 :5
of representative values. General guidance on materials is also given in Sections 7 and 21 of the B.S.I. Packaging Code. Air transport may set up pressure differencesthat disrupt the water vapour barrier of a package, expelling some of the air present at ground level and chilling the contents. Admission of warm, moist air on landing may produce heavy condensation on the contents. BS 1133, section 20, advocates the use of pressure-relief values for packages for air freight4. With desiccants with absorptive capacities differing from 27%, the weight calculated from these formulae will need to be proportionately adjusted. Packs of desiccant are obtainable commercially containing quantities stated in terms of basic desiccant. Dry Storage and Dry Rooms
Storage rooms are similar in principle to packages, but the rate of entry of moisture is less predictable. Replacement of the air and diffusion of water vapour will have a considerable effect on the atmosphere with building materials other than glass and metals, and will vary markedly with weather conditions. Desiccating agents can be exposed on open trays in store rooms, but in some cases, continuous circulation of the air through the desiccant may be preferable. Finely divided desiccant should be prevented from reaching exposed metal surfaces. In most cases, however, the air is dried by condensation on a cooled surface, or the relative humidity is lessened without actually removing water vapour by heating the store (Section 2.2). Some practical points need to be considered in these cases: 1. Ventilation is necessary in heated stores, even if the heaters do not themselves produce water vapour, for otherwise the relative humidity will probably rise because water vapour is desorbed from building materials. Ventilation is even more important if gas or kerosine heaters are used. 2. The relative humidity of the air must be measured in relation to the temperature of the metal surfaces to be protected. If incoming air at 83% relative humidity at 13°C is heated to 18"C, its relative humidity will fall to 60070,but if it then comes into contact with surfaces at 10°C or below, condensation will occur until their temperatures rise sufficiently to prevent it. This situation can arise with massive metallic objects during a sudden change in the weather, or if temperature is allowed to fluctuate between day and night. It may thus be necessary to keep a store heated in summer as well as in winter, and to heat suficiently to keep the average relative humidity as low as 30% if the maximum is not often to exceed 50%. The relative humidity and temperature of the store should be measured and recorded regularly if this method of preventing corrosion is to be operated economically and effectively. 3. Condensation may lead to corrosion when components are placed in relatively impervious wrappings in warm and humid workrooms or stores and then transferred to cold surroundings, and this should be taken into account in choosing the packaging technique.
17:6
CONDITIONING THE ATMOSPHERE TO REDUCE CORROSION
Elimination of Contaminants Many common materials are not severely corroded even at high relative humidity so long as the surfaces are clean, and dust particles and gaseous contaminants are eliminated from the air. It is seldom practicable to rely entirely on this method ofprotection, although copper and silver can be protected from tarnishing by wrappings impregnated with salts of copper, lead or zinc6, which react with hydrogen sulphide. Elimination of contaminants is nevertheless desirable, since it will minimise damage if other measures (such as desiccation) become ineffective during storage, and also because it will often improve the performance of the object in its ultimate application. Surface cleaning as a preparation for coatings is discussed in Sections 11.1 and 1 1.2. It is important to control degreasing baths to prevent accumulation of water and formation of corrosive products which will contaminate the atmosphere as well as the objects being degreased. In the case of trichlorethylene, stabilisers are added to prevent formation of hydrochloric acid'. Exclusion of dust is beneficial, and may necessitate filtering the air or use of a temporary protective. Sweat residues These contain fatty acids and sodium chloride, and increase the risk of corrosion after handling. Components should be washed in a solution of 5 % water in methanol. Packaging materials Materials to be used in contact with metals should be as free as possible from corrosive salts or acid. BS 1133, Section 7: 1967 gives limits for non-corrosive papers as follows: chloride, 0.05% (as sodium chloride); sulphate, 0.25% (as sodium sulphate) and pH of water extract 5.5-8.0. Where there is doubt, contact corrosion tests may be necessary in conditions simulating those in the package. Organic materials Corrosive vapours are sometimes emitted by organic materials used either in packaging or in the manufactured article, and may be troublesome in confined spaces. Some woods, particularly unseasoned oak and sweet chestnut, produce acetic acid (see Section 18.10), and certain polymers used in paints, adhesives and plastics may liberate such corrosive vapours as formic acid and hydrogen sulphide'. It may be necessary to carry out exposure trials, particularly where materials capable of liberating formaldehyde or formic acid are involved. Most corrosion problems of this kind can be prevented by using desiccants, and in many cases they are confined to imperfectly cured materials. For an excellent review see Reference 9.
Volatile Corrosion Inhibitors Atmospheric corrosion can be prevented by using volatile inhibitors which need not be applied directly to the surfaces to be protected. Most such inhibitors are amine nitrites, benzoates, chromates, etc. They are mainly used with ferrous metals. There is still some disagreement as to the mechanism of action. Clearly, any moisture that condenses must be converted to an inhibitive solution. There is no doubt that the widely used volatile inhibitors are effective in aqueous solutions containing moderate
CONDITIONING THE ATMOSPHERE TO REDUCE CORROSION
17: 7
concentrations of chloride and sulphate, and it appears that in most cases, the effective inhibitor could equally well be applied as an ester or the sodium salt. On this view, amine salts would be useful in practice for avoiding acid conditions, or because their volatility makes them convenient (see below), rather than for any specific effect of the amine, e.g. in preventing adsorption. Certain free amines have considerable effect as volatile inhibitors. It should be said, however, that a large variety of substances, such as S-naphthol or rndinitrobenzene, have some inhibitive action" and some of these may act by hindering wetting of the metal surface. A more recent development is the use of compounds containing reducible nitro groups, which are thought to act by stimulating the cathodic process, thus assisting anodic polarisation. An inhibitor of this type, hexamethyleneimine 3 5 dinitrobenzoate, is said to be in use in the CIS, and appears to be effective with a wide range of metals ' I . Commercially available inhibitors differ in respect of volatility, the pH of the aqueous solution, and in attacking some metals while protecting others. The choice of inhibitor may therefore involve a compromise. In order to secure protection in rather aggressive conditions, it may be necessary to choose a relatively volatile inhibitor, which is quickly transferred into the vapour, so that condensed moisture is made innocuous as it forms. This will be particularly necessary with large structures. Such a material, however, will also be quickly lost from the enclosure compared with one less volatile and therefore slower acting. It may be advantageous to use a more alkaline inhibitor where there is contamination by acid fumes, and mixed inhibitors have been employed on this basis. It has been suggested that inhibitors could be designed to control volatility, alkalinity, etc. It is extremely difficult to devise a laboratory test for volatile corrosion inhibitors in conditions simulating those in a typical package, and convincing evidence is seldom available to show that a new formulation is superior to the commercially available materials. Dicyclohexylammonium nitrite'* (DCHN) has a solubility of 3.9 g in lOOg of aqueous solution at 25"C, giving a solution pH of about 6.8. Its vapour pressure at 25°C appears to be about 1.3 x 10-3N/mZ,but the value for commercial materials depends markedly on purity. It may attack lead, magnesium, copper and their alloys and may discolour some dyes and plastics. Cyclohexylammonium cyclohexyl carbamate (the reaction product of cyclohexylamine and carbon dioxide, usually described as cyclohexylamine carbonate or CHC)'3v'4 is much more volatile than DCHN (vapour pressure 53 N/mz at 25"C), and much more soluble in water (55 g in lWcm3 of solution at 25"C, giving a pH of 10.2). It may attack magnesium, copper, and their alloys, discolour plastics, and attack nitrocellulose and cork. It is said to protect cast iron better than DCHN, and to protect rather better in the presence of moderate concentrations of aggressive salts. Both these materials are available commercially as powders and in impregnated wrapping papers and bags. Various modified inhibitors are also available, containing mixtures of the two, or more alkaline materials such as guanidine carbonate. Other proprietary inhibitors contain volatile amines, e.g. morpholine, combined with solution inhibitors. Certain solution inhibitors have been reported to act to some extent as volatile inhibitors,
17 :8
CONDlTlONlNG THE ATMOSPHERE TO REDUCE CORROSION
e.g. sodium nitriteI4. On the whole, the use of these materials appears to be consistent with the principles stated above, and they provide a very convenient means of protection, particularly for complex, not-too-large equipment, where the surfaces are not too heavily contaminated, and conditions of enclosure are reasonably good. Dosages of 35g/m3 of free space, or 11 g/m2 of surface have been recommended for packages. CHC may have some advantage in large, impervious structures, such as boilers, box girders, etc. if openings can be fitted with caps. Volatile inhibitors containing borate (for zinc) and chromate (for copper and its alloys) have been discussed in the literature, but little commercial development appears to have taken place in the UK. A review of inhibitors against atmospheric corrosion is given by Rosenfel'd and Persiantseva Is. Volatile inhibitors can be applied as loose powder in trays, by insufflation, in sachets, in tapes, in applicators containing impregnated foam, in sprays, or in impregnated wrappings. They have the obvious advantages that the packaging can be less elaborate than that required with desiccants and that equipment can be used immediately on opening the package, without the need for cleaning or stripping temporary protectives. Also, since the inhibitor may be effective at high relative humidity, or even under gross wetting, the protection may persist for a time even if the package is damaged. The application needs to be carefully considered in the light of the design and materials of construction of the equipment and its package and the cleanliness of the surfaces. Commercial suppliers recommend precautions against breathing the vapour or dust, skin contact and ingestion in food etc., and against ignition of dust or vapour from heated surfaces. Acknowfedgement
Extracts from the British Standards Packaging Code BS 1133, Section 7:1%7 and Section 19:1968 quoted in this section are reproduced by permission of the British Standards Institution, 2, Park Street, London, W l A 2BS,
from whom copies of the complete standard may be obtained. G . O . LLOYD REFERENCES 1. Silica Gel for Use as a Desiccant for Packages, BS 2 540:1960, British Standards Institution, London 2. Activated Alumina for Use as a Desiccantfor Packages. BS 2 541: 1960, British Standards Institution. London 3. Fundamentals of Packaging, Institution of Packaging and Blackie. London ( 1981 ) 4. Packaging Code, BS 1 133. Section 61966. Temporary Protection of Metal Surfaces Against Corrosion (During Transport and Storage); Section 7: 1967. Paper and Board Wrappers, Bags and Containers; Section 19:1986, Use of Desiccants in Packaging; Section 2 0 1973. Packaging for Air Freight Section 21: 1976, Transparent Cellulose Films, Plastics Films, Metal Foil and Flexible Laminates. British Standards Institution, London 5 . Permeability to Water Vapour of Flexible Sheet Materials, BS 3 177:1959, British Standards Institution, London 6. Chemistry Research 1957. Report of the Director of the Chemical Research Laboratory, D.S.I.R., H.M.S.O., London, 17 (1958)
CONDITIONING THE ATMOSPHERE TO REDUCE CORROSION
17 :9
7. Trichlorethylene, BS 580:1%3, British Standards Institution, London 8. Rance, V. E. and Cole, H. G., Corrosion of Metals by Vapours from Organic Materials, H.M.S.O. London (1958) 9. Donovan, P. D. and Stringer, J., British Corrosion Journal, 6 , 132 (1971) 10. Rajagopalan, K. S., Subramanyan, N. and Sundaram, N., Proc. 3rd Int. Cong. Metallic Corrosion, Moscow 1%6. 2, Mir, Moscow, 179 (1%9) 1 I . Rozenfel’d, I . L., Persiantseva, V. P. and Terentiev, P. B., Corrosion, 20, 222t (1964) 12. Shell VPI, Technical Bulletin FC 70:55:TB, Shell Fine Chemicals, London (1977) 13. Machinery, 85, London, 630 (1954) 14. Gars, I . and Schwabe, K., Werkstoffe u. Korr., 14, 842 (1963) 15. Rozenfel’d, I. L. and Persiantseva, V. P., ZaschitaMetallou., 2 , 5 (1966); English Translation: Protection of Met&, 2 , 3, Scientific Information Consultants, London (]%a)
BIBLIOGRAPHY Paine, F. A. (Ed.), Packagingfor Environmental Protection, Newnes-Butterworths (1974)
17.2 Corrosion Inhibition: Principles and Practice Introduction Corrosion may be described* as ‘the undesirable reaction of a metal or alloy with its environment’ and it follows that control of the rate of process may be effected by modifying either of the reactants. In ‘corrosion inhibition’, additions of certain chemicals are made t o the environment, although it should be noted that an aqueous environment can, in some cases, be made less aggressive by other methods, e.g. removal of dissolved oxygen or adjustment of pH. Environments are either gases or liquids, and inhibition of the former is discussed in Section 17.1. In some situations it would appear that corrosion is due to the presence of a solid phase, e.g. when a metal is in contact with concrete, coal slurries, etc. but in fact the corrosive agent is the liquid phase that is always present’. Inhibition of liquid systems is largely concerned with water and aqueous solutions, but this is not always so since inhibitors may be added to other liquids to prevent or reduce their corrosive effectsalthough even in these situations corrosion is often due to the presence of small quantities of an aggressive aqueous phase, e.g. in lubricating oils and hydraulic fluids (see Section 2.1 1). The majority of inhibitor applications for aqueous, or partly aqueous, systems are concerned with three main types of environment: 1. Natural waters, supply waters, industrial cooling waters, etc. in the near-neutral (say 5-9) pH range. 2. Aqueous solutions of acids as used in metal cleaning processes such as
pickling for the removal of rust or rolling scale during the production and fabrication of metals, or in the post-service cleaning of metal surfaces. 3. Primary and secondary production of oil and subsequent refining and transport processes. Following a brief discussion of inhibitor classifications and of types of chemicals used as inhibitors, the principles and practice of inhibition are * A more precise definition of corrosion is provided by IS0 in IS0 8044 (see Reference 120). ‘Attack of metal surfaces by the mechanical action of solid materials is properly described as erosion and is not discussed here.
17: 10
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17:ll
considered in terms of the principal factors affecting inhibitor performance (Principles) and the systems in which inhibitors are used (Practice).
Inhibitor Classifications A number of methods of classifying inhibitors into types or groups are in use but none of these is entirely satisfactory since they are not mutually exclusive and also because there is not always general agreement on the allocation of an inhibitor to a particular group. Some of the main classifications-used particularly for inhibitors in near-neutral pH aqueous systems - are as follows.
‘Safe’or ‘dangerous’inhibitors Each inhibitor must be present above a certain minimum concentration for it to be effective (see Principles), and this classification relates to the type of corrosion that will occur when the concentration is below the minimum, or critical, value. Thus, when present at insufficient concentration a ‘safe’ inhibitor will allow only a uniform type of corrosion to proceed at a rate no greater than that obtaining in an uninhibited system, whereas a ‘dangerous’ inhibitor will lead to enhanced localised attack, e.g. pitting, and so in many cases make the situation worse than in the absence of an inhibitor. Anodic or cathodic inhibitors This classification is based on whether the inhibitor causes increased polarisation of the anodic reaction (metal dissolution) or of the cathodic reaction, Le. oxygen reduction (near-neutral solutions) or hydrogen discharge (acid solutions). Oxidising or non-oxidising inhibitors These are characterised by their ability to passivate the metal. In general, non-oxidising inhibitors require the presence of dissolved oxygen in the liquid phase for the maintenance of the passive oxide film, whereas dissolved oxygen is not necessary with oxidising inhibitors. Organic or inorganic inhibitors This distinction is based on the chemical nature of the inhibitor. However, in their inhibitive action many compounds that are organic in nature as, for example, the sodium salts of carboxylic acids, often have more similarities with inorganic inhibitors. Other classifications Authors ‘in the former Soviet Union have classified inhibitors as Type A to include film-forming types, or Type B which act by de-activating the medium, e.g. by removal of dissolved oxygen. Type A inhibitors are then further sub-divided into A(i) inhibitors that slow down corrosion without suppressing it completely, and A(ii) inhibitors that provide full and lasting protection. From the practical aspect, a useful classification is perhaps one based on the concentration of inhibitor used. It is usually the case that inhibitors are used either at low concentrations, say less than approximately 50 p.p.m., or at rather higher levels of greater than 5 0 0 p.p.m. The determining factors in the selection of the concentration used, and hence the type of inhibitor, are the economics, disposal (effluent) problems, and the facilities available for monitoring the inhibitor concentration.
17: 12
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
Types of Chemicals Used as Corrosion Inhibitors Before discussing the nature of chemicals that are used specifically as corrosion inhibitors, reference must be made to two methods of water treatment that are sometimes included in descriptions of inhibitive treatments. These are, respectively, de-aeration techniques and pH control. Since the presence of dissolved oxygen is necessary to sustain the corrosion process in most aqueous systems the removal of this gas by mechanical or chemical methods is an obvious method of corrosion control. The chemicals commonly used are sodium sulphite or hydrazine. There are two distinct mechanisms involved in controlling corrosion by controlling the pH. Firstly, the pH is adjusted to ensure that the metal is exposed to a solution of a pH value at which corrosion is minimal. In the case of ferrous metals corrosion tends to decrease with pH values higher than approximately 9.0. Hence, simple additions of alkali, such as caustic soda, lime, soda ash, etc. can reduce the corrosion rate of iron and steel. On the other hand such treatment will increase the corrosion rate of other metals, particularly of aluminium and its alloys, and so pH adjustment is not advisable in mixed metal systems. Secondly, the pH is adjusted to give deposition of thin protective carbonate scales from waters of suitable composition. For water saturated with calcium bicarbonate a rise in pH will cause precipitation of calcium carbonate. The pH adjustment to achieve this can be determined from the Langelier (see Section 2.3) or Ryzner Stability Indices; these require a knowledge of the pH of the actual system and of the pH of the water when it is saturated with calcium carbonate. It must be emphasised that such calculations measure only the scale forming propensity. of the water, and are not direct measurements of the extent of corrosion reduction since other factors can influence the degree of protection afforded by the scale. A common feature of both these methods is that the quantity of treatment chemical can be calculated from stoichiometric relationships* in the reactions involved. This is not so with conventional inhibitor treatments. With these the concentration of inhibitive chemicals can only be determined on the basis of experimental laboratory studies, service trials and overall practical experience. The scientific and technical corrosion literature has descriptions and lists of numerous chemical compounds that exhibit inhibiting properties. Of these only a very few are ever actually used in practical systems. This is partly due to the fact that in practice the desirable properties of an inhibitor usually extend beyond those simply relating to metal protection. Thus cost, toxicity, availability, etc. are of considerable importance as well as other more technical aspects (see Principles). Also, as in many other fields of scientific development, there is often a considerable time lag between laboratory development and practical application. In the field of inhibition the most notable example of this gap between discovery and application is the case of sodium nitrite. Originally reported in 1899’ to have inhibitive properties, it remained effectively unnoticed until the i940s3; it is now one of the most widely employed inhibitors. Some examples from recently published review papers will indicate the ‘In practice an excess over the stoichiometric requirement, e.g. of sulphite for de-aeration, is used.
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17: 13
wide range of chemicals that show inhibitive properties. Hersch et uL4 in an extensive laboratory study examined over 70 compounds many of which were good inhibitors. Trabanelli et ul.’ in discussing organic inhibitors list and discuss some 150 compounds. Extensive reviews by Indian workers include those on inhibitors for aluminium and its alloys6 (225 references) and for copper’ (93 references). Corrosion inhibitors in industry have been reviewed by Rama Char8 (134 references). More detailed studies of the properties and uses of individual inhibitors also yield much useful data as, for example, that by Walker’ who gives 92 references in discussing the use of benzotriazole as an inhibitor of copper corrosion. In addition to reviews of this type there are a number of books entirely devoted to the subject of corrosion inhibition, of which two have been available since the early 1960s’*’O. For near-neutral aqueous solutions the function of inhibitors of the anodic class is generally considered to be that of assisting in the maintenance, repair or reinforcement of the natural oxide film that exists on all metals and alloys. Typical examples of such inhibitors for mild steel include the soluble chromates, dichromates, nitrites, phosphates, borates, benzoates and salts of other carboxylic acids. Some (nitrites and chromates) are oxidising compounds, whereas others show no oxidising capability. The ‘safe’ or ‘dangerous’ aspect of these inhibitors varies considerably and depends very much on circumstances. In the presence of aggressive ions, i.e. those that oppose the action of inhibitors (see The Composition of the Liquid Environment), the oxidising type tend, when present in insufficient quantity for complete protection, to give localised attack. However, the non-oxidising type, e.g. benzoate”, can also show this type of behaviour but to a less marked extent. Other compounds used in near-neutral aqueous solutions include polyphosphates, silicates, zinc ions, tannins and soluble oils. These are usually assigned to the cathodic class although some are reported to affect the anodic reaction. Their function is to precipitate thin adherent films on cathodic areas of the corroding metal surface thus preventing access of oxygen to these sites. Zinc ions can react with cathodically produced hydroxyl ions to produce insoluble hydroxides that are partially protective. Similar reactions lead to the formation of films incorporating phosphates and silicates. In general these cathodic inhibitors are considered safe, Le. not giving rise to localised attack in non-protective conditions. The extent of inhibition afforded to metals other than mild steel depends on the metal and the inhibitor (see The Nature of the Metal, and Dissimilar Metals in Contact). The cathodic type of inhibitor is perhaps less susceptible than the anodic type to the nature of the metal. However, cathodic inhibitors are usually less efficient (although performing quite satisfactorily in many systems) in terms of reduction in corrosion rate, than are anodic inhibitors. The latter, when used in adequate concentrations, can often achieve 100% protection. In a very few cases there are inhibitors that have been developed for the protection of specific metals, e.g. sodium mercaptobenzothiazole and benzotriazole for preventing the corrosion of copper. In acid conditions oxide films are not usually present on the metal surface and the cathodic reaction is primarily that of hydrogen discharge rather than oxygen reduction. Thus, inhibitors are required that will adsorb or bond directly onto the bare metal surfaces and/or raise the overpotential for hydrogen ion discharge. Inhibitors are usually organic compounds
17: 14
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
having N, S or 0 atoms with free (donor) electron pairs. There are exceptions to this bonding principle: some quaternary ammonium compounds with no donor electrons have inhibitive properties in acid solutions. In modern practice, inhibitors are rarely used in the form of single compounds- particularly in near-neutral solutions. It is much more usual for formulations made up from two, three or more inhibitors to be employed. Three factors are responsible for this approach. Firstly, because individual inhibitors are effective with only a limited number of metals the protection of multi-metal systems requires the presence of more than one inhibitor. (Toxicity and pollution considerations frequently prevent the use of chromates as ‘universal’ inhibitors.) Secondly, because of the separate advantages possessed by inhibitors of the anodic and cathodic types it is sometimes of benefit to use a formulation composed of examples from each type. This procedure often results in improved protection above that given by either type alone and makes it possible to use lower inhibitor concentrations. The third factor relates to the use of halide ions to improve the action of organic inhibitors in acid solutions. The halides are not, strictly speaking, acting as inhibitors in this sense, and their function is to assist in the adsorption of the inhibitor on to the metal surface. The second and third of these methods are often referred to as synergised treatments.
Principles The nature of the metal Since the majority of inhibitors are specific in their action towards particular metals, an inhibitor for one metal may have no effect and even an adverse effect on other metals. Table 17.1 is a general guide to the effectiveness of various inhibitors for metals in the near-neutral pH
Table 17.1 General guide to the effectiveness of various inhibitors in the near-neutral pH range Inhibitor Metal
Chromates Nitrites
Benzoates Borates Phosphates Silicates
Tannins
~
Mild steel Effective
Effective
Effective
Cast iron
Effective
Ineffective Variable Effective
Effective
Zinc and Effective zinc alloys Copper Effective and copper alloys AluEffective minium and aluminium alloys Lead-tin soldered joints
Effective Effective
Ineffective Ineffective Effective
-
Reasonably effective Reasonably effective Reasonably effective Reasonably effective
Reasonably effective Reasonably effective Reasonably effective Reasonably effective
Partially effective
Partially effective
Effective Effective
Partially effective
Partially effective
Variable Variable
Reasonably Reasonably effective effective
-
Reasonably Reasonably effective effective
Aggressive Effective
-
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17: 15
range. In addition, the compound dodecamolybdophosphate is reported” as approaching chromates in its ability to prevent the corrosion of a number of metals. However, there is at present only one reported application in practical systems (see Inhibitors in Practice: Central Heating Systems). It must be emphasised that anions usually considered aggressive towards some metals can actually reduce or even prevent corrosion of other metals in certain situations, thus effectively becoming inhibitors. For example, although nitrates ‘ I - 13* I4 can prevent the inhibitive action of benzoate, chromate, nitrite, etc. towards mild steel they can be incorporated into some inhibited antifreeze formulations to reduce the corrosion of aluminium alloys. Nitrates have also been reportedI5 as the only inhibitors capable of preventing the stress-corrosion cracking of type 304 stainless steel. On the other hand inhibitors are necessary to prevent the corrosion of mild steel in ammonium nitrate solutions 16. Sulphates generally behave as aggressive ions towards mild steel and other metals in waters, but can inhibit the chloride-induced pitting of stainless steels and caustic embrittlement in boilers.
Dissimilar metals in the same system Because of the specific action of many inhibitors towards particular metals, problems arise in systems containing more than one metal. In the majority of cases these problems can be overcome by the choice of a formulation incorporating inhibitors for the protection of each of the metals involved. With this procedure it is necessary not only to maintain an adequate concentration of each of the inhibitors but also to ensure that they are present in the correct proportion. This is because of two effects: firstly, failure to inhibit the corrosion of one metal may intensify the attack on the other metal; the best example of this is with aluminium and copper in the same system, and failure to inhibit copper corrosion-usually achieved with sodium mercaptobenzothiazole or benzotriazole -can lead to increased corrosion of the aluminium as a result of deposition of copper from copper ions in solution on to the aluminium surface. Secondly, an inhibitor of the corrosion of one metal may actually intensify the corrosion of another metal. Thus, benzoate is usually used to prevent the corrosion of soldered joints by nitrite inhibitor added to protect cast iron in the same system. A benzoate:nitrite ratio of greater than 7:l is necessary in these cases. Inhibitors can also lead to the co-called ‘polarity-reversal’ effects. In corrosive environments the zinc coating on galvanised steel acts sacrificially in preventing the corrosion of any exposed steel. However, in the presence of sodium benzoate’8*’9or sodium nitrite” steel exposed at breaks in the zinc coating may corrode quite readily. Nature of the metal surface Clean, smooth, metal surfaces usually require a lower concentration of inhibitor for protection than do rough or dirty surfaces. Relative figures for minimum concentrations of benzoate, chromate and nitrite necessary to inhibit the corrosion of mild steel with various types of surface finish have been given in a recent laboratory These results show that benzoate effectiveness is particularly susceptible to surface preparation. It is unwise, therefore, to apply results obtained in laboratory studies with one type of metal surface preparation to other surfaces in practical conditions. The presence of oil, grease or corrosion products on metal surfaces will also affect the concentration of inhibitor required with the
17: 16
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
added danger of a marked depletion of inhibitor during service as a result of its chemical reaction with these contaminants. It is thus advisable to remove such contaminants before commencing inhibitor treatment. This can be done mechanically, but chemical cleaning may often be necessary. A particular method of preparing rusted surfaces is that involving the phosphate-delayed-chromate (P.D.C.) technique21*’2,in which the system is first treated with an acid phosphate solution to remove rust prior to the introduction of chromate inhibitor. The latter can then be used at a lower concentration than might otherwise have been necessary.
Nature of the environment This is usually water, an aqueous solution or a two- (or more) component system in which water is one component. Inhibitors are, however, sometimes required for non-aqueous liquid systems. These include pure organic liquids (AI in chlorinated hydrocarbons); various oils and greases: and liquid metals (Mg, Zr and Ti have been added to liquid Bi to prevent mild steel corrosion by the latterz3). An unusual case of inhibition is the addition of NO to N,O, to prevent the stress-corrosion cracking of Ti-6A1-4V fuel tanks when the N’O, is pressurisedZ4. In at least one case water may itself act as an inhibitor, as in the corrosion of titanium by methanol”. In all circumstances it is important to ensure that the inhibitor is chemically compatible with the liquid to which it is added. Chromates, for example, cannot be used in glycol antifreeze solutions since oxidation of glycol by chromate will reduce this to the trivalent state which has no inhibitive properties.
Composition of the liquid environment The ionic composition, arising from dissolved salts and gases, has a considerable influence on the performance of inhibitors. In near-neutral aqueous systems the presence of certain ions tends to oppose the action of inhibitors. Chlorides and sulphates are the most common examples of these aggressive ions, but other ions, e.g. halides, sulphides, nitrates, etc. exert similar effects. The concentration of inhibitor required for protection will depend on the concentrations of these aggressive ions. Laboratory tests ‘ 1 * 1 3 s 1 4 9 2 6have given some quantitative relationships between inhibitor (Ci) and aggressive ion (C,) concentrations that will provide protection for mild steel. These are of the form log Ci = K log C,
+ constant
where K is related to the valencies of the respective ions. Although halide ions are aggressive in near-neutral solutions they can be used to improve the action of inhibitors in acid corrosion (see Practice: Acid Solutions). Variations exist among the halides, e.g. chloride ions favour the stress-corrosion cracking of Ti in methanol whereas iodide ions have an inhibitive action”. Dissolved solid and gaseous impurities can also affect the pH of the system and this may often lead to decreased inhibitor efficiency. In industrial plant, cooling waters can take up SO,, HzS or ammonia and pH control of inhibited waters will be necessary. The leakage of exhaust gases into engine coolants is an example in which corrosion can occur despite the presence of inhibitors.
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17: 17
p H of the system All inhibitors have a pH range in which they are most effective and even in nominally ‘neutral’ solutions close pH control is often necessary to ensure the continued efficiency of inhibitive treatments. Nitrites lose their effectiveness below a pH of 5-5-6-0; polyphosphates should be used between pH 6.5 and 7 - 5 ; chromates, although less susceptible to pH changes, are generally used at about pH 8.5; silicates can be used over a wide pH range but the Na,O:SiO, ratio depends on the pH value of the water.
Temperature of the system When inhibitors are used in the 0-100°C range it is usually found that higher concentrations become necessary at the higher temperatures ‘ L ’ ~ j.4 . Other inhibitors can lose their effectiveness altogether as the temperature is raised. A prime example of this is the polyphosphate type of inhibitor. This is effective in circulating systems at temperatures below about 40°C, but at higher temperatures reversion to orthophosphate can occur and this species is ineffective at the concentrations at which it will then be present. If calcium ions are present, additional loss of inhibitor will occur due to calcium phosphate precipitation. Inhibitor concentration To be fully effective all inhibitors require to be present above a certain minimum concentration. In many cases the corrosion that occurs with insufficient inhibitor may be more severe than in the complete absence of inhibitor (see ‘Safe’ and ‘Dangerous’ inhibitors). Not only is the initial concentration of importance but also the concentration during service. Inhibitor depletion may occur for a variety of reasons. In the initial stages of use, i.e. after the first application, the inhibitor concentration may fall off rapidly due to its reaction with contaminants in the system and also as a result of protective film formation. Thus, initial concentrations of inhibitor are often recommended to be at higher levels than those subsequently to be maintained. Losses may also occur due to mechanical rather than chemical effects as, for example, with windage losses in cooling towers, blow-down in boilers, and leakages generally. Maintenance of a correct inhibitor concentration (level) is particularly important where low-level treatments, e.g. less than 100 p.p.m. are used. Such treatments are, however, usually applied (for economic and effluent reasons) in large capacity systems, and plants of this nature will usually have skilled personnel available for control purposes. In smaller closed systems, e.g. automobile engines, higher concentrations of more than approximately 0.1 Yo are commonly used, but in these applications there is usually a good reserve of inhibitor allowed for in the recommended concentration and routine checking is of less importance. Nevertheless, since these inhibitors are often of the ‘dangerous’ type, gross depletion may lead to enhanced corrosion. Inhibitor control can be effected by conventional methods of chemical analysis, inspection of test specimens or by instrumentation. The application of instrumental methods is becoming of increasing importance particularly for large systems. The techniques are based on the linear (resistance) polarisation method and the use of electrical resistance probes. They have the advantage that readings from widely separated areas of the plant can be brought together at a central control point. (See Section 18.1.)
17: 18
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
Mechanical effects Corrosion can often be initiated or intensified by the conjoint action of mechanical factors. Typical examples include the presence of inherent or applied stresses, fatigue, fretting or cavitation effects. Inhibitors that are effective in the absence of some or all of these phenomena may not be so in their presence. In fact it may not always be possible to use inhibitors successfully in these situations and other methods of corrosion prevention will be required. Aeration and movement of the liquid For the majority of inhibitors in near-neutral aqueous systems an adequate supply of dissolved oxygen is necessary for them to function properly. The dissolved oxygen present in solutions that are in equilibrium with atmospheric air is adequate for this purpose, but in systems that have become de-aerated the non-oxidising type of inhibitor may not be fully effective. Even in aerated systems the transport of oxygen and inhibitor t o the metal surface is assisted by the movement of the solution. In fact, quiescent solutions may require higher concentrations of inhibitor than do circulation systems. Butler** has shown, for example, that polyphosphates (normally applied only to flowing solutions) can inhibit under quiescent conditions but at much higher concentrations. However, there are reported instances of excessive aeration having an adverse effect on inhibitor performance*. The action of tannins is partly associated with their effects at the metal surface, Le. as conventional inhibitors, and partly with their ability to react with and remove dissolved oxygen. In heavily aerated systems these inhibitors may be less effective due to depletion by this latter effect. Presence of crevices, dead-ends, etc. Effective protection by inhibitors relies on the continued access of inhibitor to all parts of the metal surface (see Aeration and Movement of the Liquid). It frequently happens that this condition is difficult to achieve due to the presence of crevices at joints, deadends in pipes, gas pockets, deposits of corrosion products, etc. Corrosion will then occur at these sites even though the rest of the system remains adequately protected. Effects of micro-organisms There are three main effects that can arise as a result of the presence of micro-organisms in aqueous solutions: (a) direct bacterial participation in metal corrosion usually due to the action of sulphate-reducing bacteria in anaerobic conditions or of the Thiobucillus and Ferrobacillus genera in aerobic conditions; the action of these organisms can lead to the accumulation of large amounts of corrosion product and pitting of the metal; (b) accumulation of flocculent fungal growths that can impede water flow and (c) breakdown and hence depletion of inhibitors by bacterial attack. Many inhibitors will lose their effectiveness in the presence of one or more of these effects. Indeed inhibitors may act as nutrient sources for some microbial organisms. In these circumstances it will be necessary to incorporate suitable bactericides in the inhibitor formulations. *Apart from extreme cases involving Cavitation effects.
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17: 19
Scale formation Controlled scale deposition by the Langelier approach or by the proper use of polyphosphates or silicates is a useful method of corrosion control, but uncontrolled scale deposition is a disadvantage as it will screen the metal surfaces from contact with the inhibitor, lead to loss of inhibitor by its incorporation into the scale and also reduce heat transfer in cooling systems. Apart from scale formation arising from constituents naturally present in waters, scaling can also occur by reaction of inhibitors with these constituents. Notable examples are the deposition of excess amounts of phosphates and silicates by reaction with calcium ions. The problem can be largely overcome by suitable pH control and also by the additional use of scale-controlling chemicals. Toxicity, disposal and effluent problems With the increasing awareness of environmental pollution problems, the use of and disposal of all types of treated waters is receiving greater attention than ever before. This often places severe restrictions on the choice of inhibitor, particularly where disposal of large volumes of treated water is involved. The disposal of chromate and phosphate inhibitor formulations is important in this respect and there is an increasing move towards the low-chromate-phosphate types of formulation. In fact for some applications even this approach is not acceptable and inhibitor formulations containing bio-degradable chemicals are being introduced. Other considerations In addition to the above general factors affecting inhibitor application and performance, there will be other special effects relating to particular types of systems, e.g. in oil-production technology. Some of these are referred to in appropriate cases in the following section.
Inhibitors in Practice A difficulty arises in describing the precise chemical nature of many inhibitor formulations that are actually used in practice. With the advancing technology of inhibitor applications there are an increasing number of formulations that are marketed under trade names. The compositions of these are, for various reasons, frequently not disclosed. A similar problem arises in describing the composition of many inhibitor formulations used in the former Soviet Union. Here the practice is to use an abbreviated classification system and it is often difficult to trace the actual composition, although in many cases a judicious literature search will provide the required information. The following discussion is thus restricted to inhibitor formulations that can be described in chemical terms.
Aqueous Solutions and Steam Potable Waters
In these waters there is a severe limitation on inhibitor choice because of the potability and toxicity factors. As pointed out by Hatch”, the possibilities
17 :20
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
are limited to calcium carbonate scale deposition, silicates, polyphosphates and zinc salts. Silicates do not prevent corrosion completely and their inhibitive effect is more marked in soft waters. The molar ratio Na,O:SiO, is important. For example, Stericker3’ has proposed Na20:3.3Si0, at 8 p.p.m. for most waters, but NazO:2-1Si02is preferred if the pH is below 6-0. Concentrations of 4-10 p-p-m. are recommended, and the method of application is often by by-passing part of the flow through a silicate (waterglass) reservoir, the slow dissolution giving the required inhibitor concentration in the main flow. With polyphosphates the most efficient inhibition is obtained in the presence of divalent ions such as Ca’+ or ZnZ+;in fact the Ca2+:polyphosphate ratio is more important than the actual concentration. A minimum value of 1 :5 has been given for this ratio with an overall concentration of up to 10 p.p.m. The optimum pH is in the 5-7 range and the inhibitive action is often improved by the addition of zinc salts. Hatchz9 points out that the treatment concentration depends on the nature of the water distribution system. Thus, with small towns a feed of 5 p.p.m. is needed to provide a residual of 0-5-1 p.p.m. whereas for the more compact systems in cities a feed of 1 p.p.m. is often sufficient. The action of the inhibitor is affected by existing deposits in the mains, and higher initial doses of about 10 p.p.m. are often required. Even higher dosages, say 50-100 p.p.m., can be used for cleaning old mains. Cooling Systems
For the purposes of corrosion inhibition these may be broadly divided into three types: (a) ‘Once-through’, in which the cooling water runs continually to waste as in the condenser systems using seawater; (b) ‘open’, in which cooling towers are used to dissipate heat taken up by the cooling water elsewhere; (c) ‘closed’in which the cooling water is retained in the system, the heat being given up via a heat-exchanger as in refrigeration plant, vehicle cooling systems, etc. Systems (a)and (b) are generally much larger in terms of watercapacity and metal area than those of type (c).
Once-through systems Where mild steel is the primary metal of construction, this usually being so for low-chloride waters, simple treatments with lime and soda can be effective in making the water less aggressive. Of the conventional inhibitors, polyphosphates at 2-10 p.p.m. with small amounts of zinc ions will reduce tuberculation but not necessarily the overall corrosion rate”. The use of 9 p.p.m- of an organo-activated zinc-phosphate-chromate inhibitor has been described3’ and this can be replaced, although with some loss in effectiveness, by 10 p.p.m. of polyphosphate if effluent problems exist. Effluent and economic problems in fact limit the choice of inhibitors, and the solution to the corrosion problem may lie in selecting a more suitable material of construction. Mild steel is often avoided and non-ferrous alloys such as the cupro-nickels and aluminium brasses are employed. These alloys are normally resistant even in aerated saline waters but corrosion problems can arise. Small amounts of iron, arising from the alloy or from elsewhere in the system, contribute towards the resistance of these alloys. Bostwick” showed the advantage of adding FeSO, to seawater condenser systems, and recently confirmed that 1 p.p.m. of this chemical added three times daily
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17 :21
for about 1 h to power-house intakes has contributed 25-30% to the life of the condenser tubes. More recently high-molecular-weight water-soluble polymers of, for example, the non-ionic polyacrylamide type have been described 35 for inhibiting the corrosion of cupro-nickel condenser tubes. Open recirculating systems These are more amenable to inhibition since it is possible to maintain a closer control on water composition. Corrosion inhibition in these systems is closely allied to a number of other problems that have to be considered in the application of water treatment. Most of these arise from the use of cooling towers, ponds, etc. in which the water is subject to constant evaporation and contamination leading to accumulation of dirt, insoluble matter, aggressive ions and bacterial growths, and to variations in pH. A successful water treatment must therefore take all these factors into account and inhibition will often be accompanied by scale prevention and bactericidal treatments. The controlled deposition of thin adherent films of calcium carbonate is probably the cheapest method ofreducing corrosion, but may not always be entirely satisfactory because local variations in pH and temperature will affect the nature and extent of film deposition. Treatment with conventional inhibitors is very much governed by environmental constraints. In the mid-twentieth century the choice was probably restricted to chromate or nitrite. For chromates Darrin36 emphasised the need for a high initial dosage of 1 OOO p.p.m. subsequently lowered to 300-500 p.p.m. The principal drawbacks of this method are the possibility of localised attack if chloride or sulphate contents rise during operation and the environmental problems. Sodium nitrite used at about 500 p.p.m. is also susceptible to chloride and sulphate and the pH control (7-0-9.0) is probably more important than with chromates. Nitrite is susceptible to bacterial decomposition and can give rise, particularly if reduced to ammonia, to stress-corrosion cracking of copper-base alloys. However, nitrite is used with success in cooling tower systems. Bacterial decomposition of nitrite can be controlled with bactericides. In air-cooling systems Conoby and Swain3’ quote the use of a shock treatment of 2,2’methylene bis (Cchloro-phenol) at 100 p.p.m. and a weekly addition of sodium penta-chlor phenate to control algae formation. On the low-level treatment side, polyphosphates, variously described as glassy phosphates, hexametaphosphate, etc. have been used as corrosion inhibitors. The concentrations recommended are somewhat above those used in ’threshold treatment’ to control scale deposition. The most effective protection is obtained in the presence of an adequate quantity of calcium, magnesium or zinc ions. In general a polyphosphate: calcium ratio expressed as P,O,:Ca of not greater than 3.35:l is recommended. The overall concentration will vary with conditions, but for cooling towers this falls in the 15-37 p.p.m. (as P,Os) range. When starting treatment, a higher initial dosage is required, this may be as high as 100 p.p.m. Fisher32suggests an initial dosage of 20 p.p.m. for a corroded steel cooling system dropping to 10 p.p.m. after one week’s operation. The application and properties of poly-phosphates have been reviewed by Butler3*and by Butler and I ~ o n ~ ~ . Polyphosphate inhibitors are subject to some limitations that are mainly concerned with reversion to orthophosphate and subsequent scale deposition
17 :22
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
if the calcium concentration is high. Some difficulties with their use have been summarised by Beecher et a1.*. Silicates at about 20-40 p.p.m. are also used in cooling-water treatment although the build-up of protection can be slow. At higher temperatures calcium silicate may be deposited from hard waters. Modern practice, on grounds of economy and avoidance of pollution, is towards the use of a combination of inhibitors at low concentration levels. Four main types of compound are involved, viz. chromates, polyphosphates, zinc salts and organic materials, and these are used in various combinat i o n ~ ~ ' .The ~ ' . principle involved is to combine a cathodic with an anodic type of inhibitor, e.g. zinc ions and/or polyphosphate with chromate. These mixed inhibitor systems usually require an operating pH of 6-7'"' and thus should only be used where pH control facilities are available. Typical formulations include 10-12 p.p.m. of a 1:4 Na,Cr,O,:Zn mixture43which provides good inhibition of copper as well as of steel corrosion, and 35 p.p.m. of a zinc-chromate-organic mixture. The latter introduces 12 p.p.m. of Cr0:- and 3-5 p.p.m. of Znz+ (added as ZnSO,) into the water, the organic compound is described as a powerful surface-active agent *. The zinc-dichromate method is further improved by adding phosphate and sometimes organic compounds such as lignosulphonates and synthetic polymers. Comeaux4' has listed the constituents of nine commercially available inhibitors. Each of these contains chromate and zinc with the Cr0,:Zn ratio varying from 0.92 to 30-0,five contain phosphate with Zn:P04 from 0.1 to 3-24 and three contain organic compounds. In some formulations of the zinc-phosphate type the organic compound will be of the mercaptobenzothiazole type to inhibit corrosion of copper3' . Five to ten p.p.m. of poly-phosphate is said3' to assist the inhibitive action of 20-40 p.p.m. of silicate but it is still important to avoid calcium silicate deposition on heat transfer surfaces. However, in recommending a silicate-complex phosphate inhibitor (25 p.p.m. at pH 6-5-8-0) Ulmer and Wood" state that scale formation is not a problem if the silicate is below 100 p.p.m., except if film boiling occurs, when scaling would occur in any case. The use of 100 p.p.m. of orthophosphate plus 40 p.p.m. of chromate plus 10 p.p.m. of polyphosphate has also been recommended4' As anti-pollution requirements become more demanding the use of even these low-level chromate-phosphate treatments is not always approved. New inhibitor formulations employ more acceptable bio-degradable organic compounds, often in conjunction with zinc ions. A formulation consisting of an organic heterocyclic compound plus zinc salt plus an 'alkalinity stabilising agent' has been described* applied initially at 500 and then at 100 p.p.m. Organic phosphorus-containing compounds have been introduced for scale control but also for corrosion inhibition. These are salts of aminomethylenephosphonic acid In conjunction with zinc salts they can be used in place of other treatments and have the advantage that close pH control is not required. Corrosion inhibitors compounded from zinc salts and derivatives of methanol phosphonic acid are described in a US patent4*.Although AMP was one of the first phosphonic acids to be introduced for scale and corrosion inhibition there are now a number of related compounds available and in wide use. These include 1-hydroxyethylidene 1, 1'-diphosphonic acid (HEDP), nitrilo-tris-phosphonic acid (NTP), phosphono-butane-tetra carboxylic acid (PBTC), etc.
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17 :23
Closed recirculating systems This type of system is most commonly encountered in the cooling of internal combustion engines. Inhibitors are required for engine coolants in order to prevent corrosion of the constructional metals, to prevent blockage of coolant passages by corrosion products and to maintain heat-transfer efficiency by keeping metal surfaces free from adherent corrosion products. The problem is often associated with inhibition of antifreeze solutions which are almost invariably ethanediol*-water solutions. When uninhibited these can become acid due to oxidation of the ethanediol in operating conditions. However, inhibition is also important with water coolants that are used in the summer months4'. The best practice is to use inhibited antifreeze throughout the year changing it annually.
Road vehicles Numerous formulations exist for coolant inhibition in road vehicles. The inhibitors most frequently encountered are nitrite, benzoate, borax, phosphate, and the specific copper inhibitors sodium mercaptobenzothiazole (NaMBT) and benzotriazole. Various combinations of these are in use. In the UK three compositional British Standards namely BS 3 150, 315 1 and 3152 were in use for many years. However, advances in other formulations and a general move towards performance rather than compositional specifications have resulted in the withdrawal of BS 3 150-2. Nevertheless, a brief description of these formulations is given, as they illustrate various aspects of inhibitor properties and use. BS 3150 contains triethanolammonium orthophosphate (T.E.P.), which is prepared by neutralising 0-9-1.0% H,P04 with triethanolamine so that the pH of a 50% aqueous solution is 6.9-7.3,and NaMBT (0.2-0-3Vo).This formulation was based on the original work of Squires''. The T.E.P. protects ferrous metals and aluminium alloys and the NaMBT protects copper and copper alloys. In the absence of NaMBT corrosion of copper can occur leading to marked attack on aluminium alloys. The NaMBT concentration becomes depleted with time, but experience indicates that with normal usage in road vehicles an annual replacement of the whole coolant will give satisfactory results. BS 3151 contained 5 -0-75% sodium benzoate plus 0.45-0.55Vo sodium nitrite in the undiluted ethanediol and is based on the original work of Vernon etaL5'. The nitrite is for protection of cast iron with the benzoate to protect other metals, but more importantly to protect soldered joints against the adverse action of nitrite. The nitrite concentration depletes in service, but again a one year period of satisfactory inhibition is provided. BS 3152 contained borax (2.4-3-OVoNa2B40, 10H20). Some controversy exists as to the efficiency of borax used alone, particularly with aluminium alloys. Nevertheless, borax has been much used and service experience has shown satisfactory inhibition. More recent formulations include other inhibitors, e.g. 3% borax plus 0.1070mercaptobenzothiazole plus 0.1% sodium metasilicate (Na,SiO, -5H20)plus 0.03070lime (Ca0)the percentages being '70by weight of the ethanediol which is then used at 33 vol. '5'0 dilutions2. In the UK, a standard for describing minimum requirements for inhibited engine coolants is provided by BS 658OS3.(Test methods are described in BS 5117.) Alkali-metal phosphates have been incorporated in antifreeze solutions but there are indications of unfavourable behaviour with aluminium alloys
-
*Ethylene glycol.
17:24
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
under heat-transfer conditions. Soluble oils have also been used as inhibitors but they can cause deterioration of rubber hose connections.
Locomotive diesels As larger volumes of coolant are required in railway locomotives than in road vehicles, the cost of inhibition is proportionally greater. An additional factor is the possibility of cavitation attack of cylinder liners. These considerations place a restriction on the choice of inhibitors. In the past, chromates have been used at concentrations of up to 0.4%, but their use presents handling and disposal problems. Chromates cannot be used with ethanediol antifreeze solutions. A 15:1 borate-metasilicate at a concentration of 1% has been used in the UK. Nitrate is added to this to improve inhibition of aluminium alloy corrosion. Tannins and soluble oils are also used, but probably to a lesser extent than in the past. The benzoatenitrite formulation (formerly BS 315 1) is effective and has been used by continental railwayss4. Marine diesels Again a wide number of formulations are in use. The inhibitors commonly employed include nitrites, borates and phosphates. Typical formulations include a 1: 1 nitrite:borax mixture at 1250-2000 p.p.m. and pH 8.5-9.0; and 1250-2 0oO p.p.m. of nitrite with addition of tri-sodium phosphate to give phenolphthalein alkalinity. The factors affecting railway diesels apply also to marine diesels but with the additional restriction that the inhibitors must not present a toxicity hazard when the cooling system is associated with equipment for producing drinking water. This is because of the possibility of accidental leakage between the two systems. Central Heating Systems
The principal components in these systems are a cast iron or steel boiler, copper or steel pipework, pressed steel or cast-iron radiators and a copper hot-water tank or calorifier to supply heat to domestic water. If systems are properly designed, installed and maintained, the concentration of dissolved oxygen in the circulating water-which should be subject to little makeup - is low and corrosion is minimal. Nevertheless, corrosion problems occasionally arise in these systems. Often these are associated with ingress of oxygen but this is not always so. The main problems are the perforation of pressed-steel radiators and the necessity for the frequent release of hydrogen gas from radiators. The latter effect is associated with the production of excess amounts of magnetite (Fe,04) as a result of the Schikorr reaction: 3Fe(OH), Fe,04 H, 2H,O +
+
+
Thin adherent films of magnetite form on the steel surfaces in the initial stages of operation of the system, but in troublesome situations the magnetite becomes non-adherent and in extreme cases can lead to pump blockages. These difficulties can often be overcome with inhibitive treatments although the procedure is not acceptable where there is any possibility of inadvertent mixing of the heating water with domestic water. The excess magnetite problem has been associated with the catalytic action of copper ions on the Schikorr reactionss*s6.Hence, inhibitors that will
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17 :25
prevent copper dissolution should reduce magnetite formation. For this purpose 0.01% benzotriazole can be added to the water. For general corrosion inhibition a mixture of 1.O%, sodium benzoate with 0.1 Yo sodium nitrite has been successfully used in a number of installation^^'"^. Sodium metasilicate has been used with success, but usually in softened waters. Other workers suggest that it is not reliable due to the possibility of localised attack (results with Na2Si20,)58and because of possible pipe and pump blockage by gel-formation or precipitation of hardness salts. The use of a silicate-tannic acid treatment has also been described. A further development is the introduction, based on test rig results, of a four-component formulation containing sodium benioate, sodium nitrite, sodium dodecamolybdophosphate and benzotriazolem. Somecorrosion inhibitors can encourage the production of fungal growths in the relatively static cold water in the header tanks of these systems. Biocides will then often need to be included in the inhibitor formulation. Steam-condensate Lines
The causes and inhibition of corrosion in steam-condensate lines have been reviewed by Obrecht". The major causes of corrosion are carbon dioxide and oxygen and the problems are associated not only with damage to the pipes, which may be of steel but often of copper-base alloys, but also with iron and copper pick-up which will be deposited elsewhere in the circuit. Neutralisation treatments can be employed to keep the pH in the 8 . 5 - 8 - 8 region62.Typical compounds used for this purpose are ammonia, cyclohexylamine, morpholine and benzylamine. An important requirement is that these agents should condense at the same rate as the steam. This is not necessarily so with ammonia and pockets of unneutralised condensate may occur. Furthermore, ammonia can cause attack of copper-base alloys. The amines, except at high concentrations, are less aggressivein this respect, they have better distribution characteristics and condense at the same rate as the steam. An important disadvantage with these materials is their cost, since about 3 p.p.m. are needed per p.p.m. of carbon dioxide" and so they tend to be used only in high recovery systems. Inhibitive (as opposed to purely neutralisation) techniques now employ long-chain aliphatic amines with alkyl groups containing 8-22 carbon atoms"'-63. The most effective are the straight-chain aliphatic primary amines with Clo-,8,the best known example being octadecylamine and its acetate salt. They are used at a total concentration of only 1-3 p.p.m. and are effective against carbon-dioxide and oxygen-induced corrosion. They function by producing a non-wettable film on the metal surfaces. The acetate salt is used to facilitate dispersion and solubilisation. The most effective distribution is achieved by injection into the boiler or the main steam header. The protective film ceases to form at about 200°C63and in a condensing turbine system inhibition will be provided through the feed system up to the point where the feed reaches this temperature. These inhibitors have been successfully applied to prevent exfoliation of 70 Cu-30 Ni tubes 61. Contrary to the method of application of inhibitors to water systems, the
17:26
CORROSION INHIBITION: PRlNClPLES AND PRACTICE
initial addition of filming amine should be at a lower concentration than that subsequently used. This is because the surface-active nature of the amine will loosen and remove existing corrosion products and these will accumulate elsewhere in the system. A cleaning-up phase of up to a month may therefore be necessary to avoid these effects. High-chloride Systems
(sea-water, desalination, refrigerating brines, road de-icing salts, etc.) Complete inhibition of corrosion in waters containing high concentrations of chloride is difficult, if not impossible to achieve economically. Despite this, many such systems make use of inhibitors to give marked reductions in corrosion rates. In refrigerating brines, chromates at a pH of about 8-8 5 have been widely used. Concentrations recommended are between 2 0oO and 3 300 p.p.m. corresponding to the 125 and 2001b (56.7 and 90.7 kg) of sodium dichromate per 1 OOO ft3(28.32 m3) for calcium or sodium chloride brines, respectively, recommended by the American Society of Refrigerating Engineers. In diluted sea-water high concentrations of sodium nitrite can bring about a reduction in the corrosion of steel. For example, corrosion in 50% seawater can be inhibited with 10% sodium nitrite@, and 3-7% of this inhibitor has been recommended for preventing the corrosion of turbine journals due to sea-water ingress6’. The beneficial effects of mixtures of chromates and phosphates have been reportedM. Combinations of these inhibitors have been examined with respect to preventing corrosion in desalination plant. Oakes et a/.67have reported good results with 5 p.p.m. of chromate plus 30-45 p.p.m. of Na,HPO,. Legault et a/.68conclude that three mixtures are effective for mild steel in oxygen-saturated sea-water at 121OC, Le. dichromate plus phosphate at 50 p.p.m., chromate plus phosphate plus zinc plus iodideat 100 p.p.m. and chromateplus phosphateat 50p.p.m. Thechromate: phosphate ratio is usually 1:l with Na3P0, as the phosphate. Various opinions exist as to the value of inhibitor^^^ in road de-icing salts. Chromates have been advocated but to a limited extent because of the toxicity effects. In general, the most widely used inhibitors are the polyphosphates, either alone or in conjunction with other inhibitors. The use of polyphosphates on roads in Scandinavia has been reported7’, although difficulties arise with loss of inhibitor by absorption in the sand that is mixed with the salt. Extensive laboratory tests conducted in the UK showed” that polyphosphates were more effective in preventing further rusting of damaged painted panels than in preventing the corrosion of bare steel. A further development is to compound the polyphosphate with an organic-type inhibitor 72*73. Acid Solutions
Probably the major use of inhibitors in acid solutions is in pickling processes. The chief requirements of the inhibitor are that it should not decompose during the life of the pickle, not increase hydrogen absorption by the metal
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17 :27
and not lead to the formation of surface films with electrically insulating properties that might interfere with subsequent electroplating or other surface treatments. A wide variety of compounds are used in acid inhibition. These are now mainly organic compounds usually containing N, S or 0 atoms, although inorganic arsenic and antimony compounds have been used in the past. In general, for the pickling of steel, as pointed out for example by M a ~ h u ~ ~ , sulphur-containing compounds are preferred for sulphuric acid solutions and nitrogen-containing compounds for hydrochloric acid solutions. Every and Riggs” list 76 individual compounds and 32 mixtures that were subjected to laboratory tests and concluded that often a mixture of N- and Scompounds was better than either type alone. The superiority of S- compounds for inhibition in sulphuric acid is borne out in the list of 112 compounds quoted by Uhlig76. Twelve of the fourteen most effective compounds listed contain S atoms in the molecule. Typical of these are phenylthiourea, di-ortho-tolyl-thiourea, mercaptans and sulphides, 90% protection being provided with 0.003-0-01 070 concentrations. N- compounds used as acid inhibitors include heterocyclic bases, such as pyridine, quinoline and various amines. Carassiti7’ describes the inhibitive action of decylamine and quinoline, as well as phenylthiourea and dibenzylsulphoxides for the protection of stainless steels in hydrochloric acid pickling. Hudson et a1.78,79 refer to coal tar base fractions for inhibition in sulphuric and hydrochloric acid solutions. Good results are reported with 0-25 vol. ‘70 of distilled quinoline bases with addition of 0 . 0 5 ~sodium chloride in 4N sulphuric acid at 93°C. The sodium chloride is acting synergistically, e.g. 0 . 0 5 ~NaCl raises the percentage inhibition given by 0.1% quinoline in 2N H,S04 from 43 to 79%. Similarly, potassium iodide improves the action of phenylthiourea*’. Acetylenic compounds have been described for inhibition in acid s o l ~ t i o n s ~Typical ~ - ~ ~ . inhibitors include 2-butyne-1 ,4-diol, I-hexyne-3-01 and 4-ethyl-I-octyne-3-01. An exception to the ‘lone pair’ or ‘donor’ electron requirement of organic inhibitors is provided by the quaternary ammonium compounds. M e a k i n ~ ~ reports ’~’ the effectiveness of tetra-alkyl ammonium bromides with the alkyl group having C 2 10. Comparative laboratory tests of commercial inhibitors of this type have been describedB6.The inhibiting action of tetra-butyl ammonium sulphate for iron in H,S-saturated sulphuric acid has been described, better results being achieved than with mono-, di- or tri-butylamines 87. In the former Soviet Union much use is made of industrial by-products to prepare acid inhibitors. The PB class is obtained by treating technical butyraldehyde with ammonia and polymerising the resulting aldehydeammonia. PB-5,for example, with 0*01-0.15% of an arsenic salt is used in 20-25% HCl. A mixture of urotropine (hexamethyleneimine, hexamine) with potassium iodide, a regulator and a foaming agent is the ChM inhibitor. BA-6 is prepared from the condensation product of hexamine with aniline. A more recent development is the Katapin series which consists of p-alkyl benzyl pyridine chlorides; Katapin A, for example, is the p-dodecyl compound. The beneficial effect of chloride ions on inhibitor action is brought out in
17 :28
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
the acid descaling of ships’ tanks while at seaE8.Inhibited 0.75% sulphuric acid prepared with sea-water can be used for this process at ambient temperatures, the chloride present in the sea-water acting synergistically with the inhibitor. In the practice of acid pickling a foaming agent is often added to the pickling bath in order to facilitate penetration of the rust and scale by the inhibited acid and also to provide a foam blanket to prevent spray coming from the bath. After removal from the bath the metal is rinsed well, finally in hot water and then often dipped in a mild alkali, phosphate or chromate bath, to provide short term protection before the next operation. A suggested variation on this procedure is to follow the acid pickling by a hot water rinse in a bath with a 35 mm layer of stearic acid on the bath surface. As the metal is withdrawn through this a water repellent film is left on its surface. The Oil Industry
Corrosion problems requiring the application of inhibitors exist in the oil industry at every stage of production from initial extraction to refining and storage prior to use. Comprehensive reviews of these inhibitors have been given by Bregman10g90. Four main processes are involved. (a) Primary production, (b)secondary production, (c) refining and (d)storage, and each of these may be further subdivided.
Primary production Although the technology of the process has many variations, the common factor is that oil flows from the deposit through steel tubing to the surface. Corrosion problems arise due to the presence of water that invariably accompanies the oil. It has been shown” that corrosivity is related to the water content which can vary over a wide range. This water can contain a variety of corrosive agents including carbon dioxide, hydrogen sulphide, organic acids, chlorides, sulphates, etc. Wells containing H,S are referred to as sour and those free from H,S as sweet; the former are the more corrosive. In some sweet wells the crude oil itself can provide protection of the metal if the oil: water ratio is suitable, but this effect will not be found in sour wells. Most of the inhibitors in use are organic nitrogen compounds and these have been classified by BregmanW as (a) aliphatic fatty acid derivatives, (b) imidazolines, (c) quaternaries, (6)rosin derivatives (complex amine mixtures based on abietic acid); all of these will tend to have long-chain hydrocarbons, e.g. C18H,,as part of the structure, (e) petroleum sulphonic acid salts of long-chain diamines (preferred to the diamines), (f)other salts of diamines and (g) fatty amides of aliphatic diamines. Actual compounds in use in classes (a) to (d) include: oleic and naphthenic acid salts of n-tal1owpropylenediamine;diaminesRNH(CH,),NH, in which R is a carbon chain of 8-22 atoms and x = 2-10; and reaction products of diamines with acids from the partial oxidation of liquid hydrocarbons. Attention has also been drawn to polyethoxylated compounds in which the water solubility can be controlled by the amount of ethylene oxide added to the molecule. The method of inhibitor application varies considerably since so many factors have to be considered. These include the oil: water ratio, the types
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17 :29
of oil and the water composition, the fluid velocity, temperature, type of geological formation, emulsion formation, economics, method of well completion, solubility and specific gravity of the inhibitor, etc. It has been statedg2that there are over 20 methods of introducing an inhibitor into the well to ensure that it enters the producing stream. These include: ‘slug’treatment in which regular injections of inhibitors are made with automatic injection equipment; ‘batch’ treatments in which sufficient inhibitor is added to last for a month or longer; ‘weighted’ treatmentsg3 in which organic weighting agents can raise the density of the inhibitor formulation thus assisting its dispersal; ‘micro-encapsulation’% methods with a liquid inhibitor weighted and coated with a water-soluble sheath to give controlled release at a given temperature; ‘squeeze’9*-% technique where the inhibitor is displaced under pressure into the geological formulation whence it is absorbed into the rock and then gradually desorbed into the deposit. All these and other methods are subject to their particular advantages and disadvantages which are discussed in the relevant technical literature. Secondary recovery In this, water is forced down into the strata to displace further quantities of oil. This water can be that initially obtained from the well or it can be taken from other convenient sources. In either case the probability is that the water will be of an aggressive nature. As the water is now being forced down into the deposit there is the danger of blockage of the geological formation by corrosion products and this is an added reason for inhibition. Apart from the presence of dissolved salts there are the major problems of the oxygen and bacterial contents. Sulphite additions may be made to deal with dissolved oxygen but the method is not so straightforward9’ as, for example, in boiler-water treatment. Thus, care is required in brines containing H,S as the catalyst* may be precipitated as sulphide. The sulphite may be lost in the deposition of calcium sulphite hemihydrate if the calcium concentration is high. As in primary production, organic nitrogen compounds are often used since many of these have dispersant properties that will prevent the formation of adherent depositsw. It has been suggested that dissolved oxygen can prevent long-chain amines from being fully effective as corrosion inhibitors. Nevertheless some inhibitors of this type appear to be immune to this effect, for example see the results of Jones and Barrett9*. Oxygen removal has been combined with long-chain amine treatment by using a 40% methanol solution of the oleyldiamine adduct of SO, -the so-called ODASA method*. A concentration of 25 p.p.m. of this inhibitor is quoted for the scavenging of 1 p.p.m. of oxygen and field trials have shown reductions in oxygen from 0.5 to less than 0.1 p.p.m. Inorganic inhibitors can also be used in waterflood treatment to limit oxygen corrosion; a zinc-glassy phosphate type at 12-15 p.p.m. and pH 7 . 0 - 7 - 2 has been described’w. Silicates at 100 p.p.m. have also been used. A particular problem in oil recovery arises in the acidising process for stimulating well production in limestone formation^'^. IO2 . For many years 15% hydrochloric acid for this process has been successfully inhibited with commercially available organic inhibitors to minimise attack on the *Small amounts (less than 1 p.p.m.) of cobalt salts are usually added to the sulphite to catalyse its reaction with oxygen.
17:30
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
steel equipment. Sodium arsenite with a surfactant has been although problems can occur subsequently at the refining stage due to catalyst poisoning. In a discussion of acetylenic-type inhibitors Tedeschi ela/.'O3 show that the action of compounds such as hexynol and ethyl octynol for this type of application can be improved by the use of nitrogen synergists such as ethylene diamine, dimethyl formamide, urea or ammonia. However, with the advent of deeper wells this concentration of acid is not so effective and 28-30% concentrations become necessary. These higher concentrations, and the higher temperatures at the well-bottom, together place a limitation on the existing inhibitors. Research is active in this field, e.g. in one case a survey of some 20000 compounds was made from which it was concluded that acetylenic compounds and some nitrogen compounds offered promise"'. Russian workers'04 have described the inhibitor ANP-2 for use with 20% HCl in acidising. This is the HCl salt of aliphatic amines with an amine number of 15.75 obtained from the nitration of paraffins. At 0.1-0- IS%, ANP-2 reduced the corrosion of steel at 43°C in 20% HCl by 20 times.
Refining Inhibitors are necessary in the processing of crude oil - particularly where steel is involved -since many of the process fluid constituents are corrosive. Copper-bearing alloys, e.g. admiralty metal, are also used and the problem of controlling steel corrosion is often made more difficult by the need to use methods that will not enhance the corrosion of non-ferrous parts of the system. In general the corrosive agent is the water in the oil stream and its corrosivity is increased by the presence of H,S, C 0 2 , O,,HCl and other acids (naphthenics can be a source of corrosion). As in so many other situations the problem of inhibition cannot be considered in isolation. Problems concerned with fouling and scaling must be taken into account and comprehensive reviews of these problems have been published90-'0s.Since many of the corrosion problems are due to the presence of acids one remedy is to adjust the pH to 7.0-7-5 by adding sodium hydroxide, sodium carbonate or ammonia. At higher pH values ammonia can lead to corrosion, and possibly stress-corrosion cracking of copper-base alloys. The neutralisation of hydrochloric acid with ammonia will produce ammonium chloride and deposits of solid NH4C1 can be corrosive even in the absence of waterIM. Other disadvantages of alkali treatment are those of expense and the necessity for pH control to prevent scale formation. Nitrogen-containing organic inhibitors, often in conjunction with ammonia, are now widely used. These compounds are usually of similar types to those for primary production, because, as Bregman has pointed out, the corrosive agents are often the same in the two cases. This author has reviewed the compounds used and points out that most are imidazoline derivatives. He cites Brooke'07 for specific applications. Thus, 6 p.p.m. of an imidazoline with ammonia to pH 7 . 5 added to the overhead of a crude topping unit increased the length of a run from 6 to 18 months. In another application, corrosion of overhead condensers and the top tray of a distillation column was prevented by the use of 4 p.p.m. of an amino alkyl aryl phosphate soluble in light hydrocarbons. This had to be changed to 4 p-p.m. of a methylene oxide rosin amine type of inhibitor after the phosphate was found to cause deposits when the produced fuel was used in internal
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17:31
combustion engines. This last observation is a further example of factors other than those relating only to the metal-environment reaction influencing the selection of an inhibitor. The possible adverse effect of inhibitors on process catalysts in refineries must also be considered. Storage Corrosion is again mainly associated with the presence of water which separates at the bottom of storage tanks. Inhibition in this water layer can be achieved with the highly soluble inorganic inhibitors. Nitrites, silicates, polyphosphates, etc. have been used as well as oil-soluble inhibitors. Organic inhibitors include imidazolines alone or with other inhibitors, itaconic salts, oleic acid salts of various amines and polyalkene glycol esters of oleic acid. Again there are other requirements that must be fulfilled apart from prevention of corrosion.
Reinforced Concrete
Inhibitors to prevent or retard the corrosion of steel reinforcing bars in concrete have been discussed on a number of occasions. Treadaway and Russell IO8 consider the important considerations to be (a) the extent of inhibition, (b) the rate of inhibitor consumption, (c) the type of attack if inhibition fails and (d)the effect of the inhibitor on concrete strength. Of these (d)is of considerable importance'08*'09.The best practice appears to be the coating of the bars with a strong inhibitive slurry rather than a general incorporation into the concrete mix as a whole. The inhibitors generally considered applicable are sodium benzoate"' (2-1070 in a slurry coating); sodium and sodium benzoate plus sodium nitrite'08. A mixture of grease with Portland cement, sodium nitrite, casein and water applied as a 2 1mm layer coating for reinforcing bars has been described1I3.Sodium mercaptobenzothiazole, stannous chloride and various unidentified proprietary compounds have also been described for inhibition in concrete. Laboratory tests have been reported by Gouda et ai. 'I4. Miscetaneous
In nitrogenous fertiliser solutions of the NH,NO,-NH, -H,O type corrosion of steel can be prevented by 500 p.p.m. of sulphur-containing inhibitors, e.g. mercaptobenzothiazole, thiourea and ammonium thiocyanate. However, these inhibitors are not so effective where most of the NH, is replaced by urea. For these solutions phosphate inhibitors such as (NH,),HPO, and polyphosphates were more effective'''. In the hydraulic transport of solids through steel pipelines, inhibitors of the sodium-zinc-phosphate glass type have been shown IL6 to be effective. In the case of coal slurries the polyphosphate type was rejected because the de-oxygenating action of the coal lowered the inhibitor effectiveness. Hexavalent chromium compounds at 20 p.p.m. were more effective"'. In gas reforming plants, e.g. the hot potassium carbonate process for C 0 2 removal, sodium metavanadate is used to prevent mild-steel corrosion'". Banks reports ' I 9 that this treatment does not reduce the rather high corrosion rate of Cu-30Ni in these plants.
17 :32
CORROSION 1NHIBlTlON: PRINCIPLES AND PRACTICE
Recent Developments Terminology The International Standards Organization has recently ''O defined a corrosion inhibitor as a 'chemical substance which decreases the corrosion rate when present in the corrosion system at a suitable concentration, without significantly changing the concentration of any other corrosive agent .'This last point is significant since it excludes chemicals employed for deaeration or pH control from the definition of a corrosion inhibitor. On the other hand, it should be noted that the inhibitor is '. . . present in the corrosion system . . .', and thus arsenic when added to brasses to prevent dezincification may be classified as an inhibitor. Literature Recent additions to the literature on the principles and practice of inhibition include books concerned with the subject as a whole, and reports of conferences and papers, or reports concentrating on particular aspects of the subject. Books include the volume by the late Professor I. L. Rozenfel'd''' and collected data in the form of references, patents etc. from various sources122-'ZS. Conferences include the recent quinquennial events at the University of Ferraralz6, IZ7, each providing substantial contributions to all aspects of corrosion inhibition. The uses of molybdates as inhibitors have been reviewed by Vukasovich and Farr in a paper with 221 references'" and test methods for inhibitors in a report from the European Federation of Corrosion with 49 references General considerations The principles and practice of corrosion inhibition in recent years have become increasingly dominated by health and safety considerations. These relate to all aspects of inhibitor practice, i.e. to handling, storage, use and disposal. The problem becomes particularly important when there is the possibility of contact of inhibited media with the environment. Thus, it can be argued that the 'safety' requirements for inhibitors for solar heating systems should be more rigid than those for engine cooling systems since in the former case small leaks in heat exchangers could lead to contamination of domestic water. Open Recirculating Systems
The matter of environmental protection has been recognised for some years in the inhibition of open recirculating systems, although the technical advances that have been made in this period in inhibitor formulation have also been prompted by other requirements, for example, the need to avoid scale deposits. Since the 1940s there has been a move away from the use of chromates at relatively high concentrations to formulations with lower concentrations and eventually to non-chromate treatments. This situation has come about as a result of a better understanding of the role of other types of chemicals in the inhibition process, notably phosphorus-containing compounds, organics (e.g. triazoles), zinc salts and polymeric compounds. The sequence of formulation types may be deduced from p. 17:21 and may be summarised as: high levels of chromate; low levels, of chromate, but with polyphosphates; further reductions in chromate by the introduction of zinc ions; and polyphosphate instability overcome by the use of phosphates and
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17:33
metal-free, Le. no chromium or zinc, formulations achieved by the use of polymeric chemicals. Organic inhibitors of the triazole or mercaptobenzothiazole classes are added when protection of copper is a particular requirement. Theapolymeric compounds were introduced to assist in controlling scale deposition and generally for keeping systems free from deposits and suspended matter. These materials are long-chain polymers containing repeating carboxyl groups with molecular weights of several thousands. Examples include polyacrylates, polymaleics and various copolymers. The mechanism of their action in contributing to corrosion inhibition is not fully understood but must be associated in some way with the maintenance of clean surfaces. Although chemicals in closed circulation systems do not generally come into contact with the environment - except perhaps on disposal - problems can exist with safety in handling. A particular example is the need for caution in the mixing of coolants containing nitrites with those containing amines because of the possible production of carcinogenic nitrosoamines. This caution has been expressed in other fields of use of inhibitors (see below).
Solar Heating Systems
The requirements for inhibitors in solar heating, systems are in some ways similar to those for domestic central heating systems but in other respects to those for engine cooling systems. A significant difference from the latter is the need, in many parts of the world, for the presence of an anti-freeze agent. Since ethanediol is toxic, the more acceptable propanediol (propylene glycol) tends to be used together with relatively non-toxic corrosion inhibitors such as silicates, phosphates, BTA, etc. A particular requirement is the need for high-temperature stability since surface temperatures of panels exposed to sunlight can be well in excess of 100°C. Polymeric compounds have also been put forward as inhibitors for solar heating systems as described, for example, in a patent application in 1982 for a polymerisable acid graft copolymer, e.g, acrylic acid-(oxyethylene-oxypropylene) copolymer together with a silicate’3’.
Refrigerating Brines
For many years such media have been based on strong salt solutions, e.g. calcium chloride brines. Sodium dichromate has been used (see p. 17:26), but recently other inhibitors have been claimed to be effective. One patent quotes N-alkyl-substituted alkanolamines, e.g. 2-ethyl ethanolamine + BTA at pH 9.0’3’.A mixture of hydrazine hydrochloride + BTAI3’ has been claimed as well as a mixture of gelatin + triethanolamine + potassium dihydrogen phosphate’”. Other examples are to be found in the patent literature and the above are quoted to illustrate the diversity of chemicals that may be used. Absorption-type refrigerators operating with strong lithium bromide solutions can also be inhibited by a number of chemicals. Thus, a mixture of lithium hydroxide + BTA + sodium molybdate has been reported’34.
17:34
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
+
Elsewhere, in a series of Japanese patents, mixtures of resorcinol sodium nitrate, glycerine sodium nitrate, lithium hydroxide tungstate, etc., have been claimed to be effective. An example of the use of inhibited cooling mixtures of low toxicity is provided by a patentI3’ which describes a mixture of silicate+ polyphosphate a saccharide, e.g. sucrose or fructose, as the inhibitor formulation in a propylene glycol + potassiumhydrogen-carbonate mixture used in aluminium cooler boxes for ice-cream.
+
+
+
Metal Working Coolants
There has been much activity in this field of corrosion inhibition in recent years which appears to have been prompted by health and safety requirements. As with engine coolants, the use of nitrites, particularly where amines may also be present, needs to be considered carefully. Nitrites have been widely used in cutting, grinding, penetrating, drawing and hydraulic oils. Suggested replacements for nitrites and/or amines make use, inter alia, of various borate compounds, e.g. monoalkanolamide borates. Molybdates have also been proposed in conjunction with other inhibitors, e.g. carboxylates, phosphates, et^'^^. Water-based metalworking fluids usually contain other additives in addition to corrosion inhibitors, e.g. for hard-water stability, anti-foam, bactericidal proderties and so on. Thus, claims are made for oil-in-water emulsions with bactericidal and anti-corrosion properties. Oil and Gas Production, Transport and Storage
This is a prolific field for inhibitors although the main types remain as grouped by Bergmanm (see p. 17). In this application of inhibitors, probably more than in any other, the methods of introducing the inhibitor into the corrosive environment receive as much attention as the nature of the inhibitor. The most severe conditions are those met in acidising treatments, typically with 15-35070 hydrochloric acid at high downhole temperatures. Compounds with triple bonds, Le. acetylenic compounds, continue to receive attention. Patents have been filed for mixtures of propargyl alcohol with, for example, cellosolve a phenol formaldehyde resin tar bases 13’ heterocyclic nitrogen compounds acetylenic dialkylthiourea or a quaternary antimony oxide’39. With the increasing development of sour, i.e. H,S-containing wells there is a need to assess the performance of inhibitors in such contaminated environments. Reports of inhibitor performance often make special reference to performance in the presence of H,S, which can be accompanied by CO, . Schmitt has emphasised the need for assessing the effects of corrosion inhibitors on the hydrogen uptake by the metal as well as the retardation of metal dissolutionIz6.For example, in discussing, inhibitors of the quaternary ammonium type, he and his co-workers point out that, depending on the inhibitor, the H,S present could increase or decrease the efficiency of the inhibitor in blocking, hydrogen absorption. For 10% formic acid good results have been reported with p-dodecylbenzylquinolinium chloride and benzylquinolinium iodide la.
+
+
+
+
+
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17:35
Corrosion inhibitors in gasoline are present to provide protection to the fuel distribution system which operates at ambient temperature. It is particularly important that the inhibitors do not adversely affect other requirements of the fluid, for example, in carburettor detergency. The most effective inhibitors appear to be salts, or esters, of carboxylic or phosphoric acids, often associated with long-chain radicals. Test methods for inhibition of water contaminated gasoline include NACE TM-01-72 and ASTM D6651 (IP-135). Gasohol is formed by mixing 96-95 volumes of gasoline with 5-10 volumes of ethanol. Although substantially anhydrous, the presence of small quantities of water, say up to 0.3070,can lead to corrosion. Various triazoles with polyisobutylene have been proposed 14’, e.g. 3-amino-lH-1,2,4-triazole and maleic acid anhydride. Other formulations are based on reaction products of carboxylic acids with amines. Problems associated with alcoholgasoline corrosion have been described 142.
Checklist of Steps to Minimise Corrosion by the Use of Inhibitors 1. The practice of corrosion inhibition requires that the inhibitive species should have easy access to the metal surface. Surfaces should therefore be clean and not contaminated by oil, grease, corrosion products, water hardness scales, etc. Furthermore, care should be taken to avoid the presence of deposited solid particles, e.g. stones, swarf, building materials, etc. This ideal state of affairs is often difficult to achieve but there are many cases where less than adequate consideration has been given to the preparation of systems to receive inhibitive treatment. Acid treatments, notably with 3-5% citric acid, with or without associated detergent washes, are often recommended and adopted for cleaning systems prior to inhibition. However, it is not always appreciated that these treatments will not remove particulate material particularly when, as is often the case, the material is insoluble in acids. 2. Even with adequate cleaning procedures it is still necessary to ensure that the inhibitor reaches all parts of the metal surfaces. Care should be taken, particularly when first filling, a system, that all dead ends, pockets, crevice regions, etc., are contacted by the inhibited fluid. This will be encouraged in many systems by movement of the fluid in service but in nominally static systems it will be desirable to establish a flow regime at intervals to provide renewed supply of inhibitor. 3. Inhibitors must be chosen after taking into account the nature and combinations of metals present, the nature of the corrosive environment and the operating conditions in terms of flow, temperature, heat transfer, (see ‘Principles’ p. 17:14). 4. Inhibitor concentrations should be checked on a regular basis and losses restored either by appropriate additions of inhibitor, or by complete replacement of the whole fluid-as recommended, for example, with engine coolants.
17:36
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
5 . Where possible, some form of continuous monitoring should be
employed, although it must be remembered that the results from monitoring devices, probes, coupons etc., refer to the behaviour of that particular component at that particular part of the system. Nevertheless, despite this caution, it must be recognised that the monitoring of the corrosion condition of an inhibited sysem is a well-established procedure and widely used (10:26-10:32).
Conclusions The principles and practice of corrosion inhibition have been described in terms of the factors affecting inhibitor performance and selection (principles) and the more important practical situations in which inhibitors are used (practice). For the latter a brief account is given of the nature of the system, the reasons for inhibitor application and the types of inhibitor in use. Tabulated results have been avoided since these are either obtained from carefully controlled laboratory tests or from specific systems and would thus require much qualification before their application to other systems. The very wide use of inhibitors is obvious, but emphasis must always be placed on the factors affecting their performance and on the specific circumstances and other requirements relating to particular applications. A. D. MERCER
REFERENCES
V. P., MetaNic Corrosion Inhibitors, Pergamon, Oxford (1960) Proc. Royal Artillery Assoc. (Woolwich), 26 No. 5 (1899); Moody, G. T., Proc. Chem. Soc., 19, 239 (1903) Wachter, A. and Smith, S . S., Industr. Engng. Chem., 35, 358 (1943) Hersch, P., Hare, J. B., Robertson, A. and Sutherland, S. M., J. Appl. Chem., 11, 246 (1961) Trabanelli, G. and Carassiti, V.,Advances in Corrosion Science and Technology,Vol. I , edited by Fontana, M. G. and Staehle, R. W.,Plenum Press, New York-London (1970) Desai, M. N . , Desai, S. M., Gandhi, M. H. and Shah, C. B., Anti-corrosionMethods and Materials, 18 No. 4, 8-13 (1971); ibid. No. 5, 4-10 (1971) Desai, M. N., Rana, S. S. and Gandhi, M. H., ibid., 18 No. 2, 19-23 (1971) Rama Char, T. L. and Padma, D. K., Trans. Inst. Chem. Engnrs., 47, T177-182 (1969) Walker, R., Anti-corrosion Methods and Materials, 17 No. 9, 9-15 (1970) Bregman, J. J., Corrosion Inhibitors, MacMillan, New York-London (1%3) Brasher, D. M. and Mercer, A. D., Br. Corrosion J., 3 No. 3, 120-129 (1968) Brasher, D. M. and Rhoades-Brown, J. E., ibid.. 4, 74-79 (1969);8, 50 (1973) Mercer, A. D. and Jenkins, I. R., ibid., 3 No.3, 130-135 (1968) Mercer, A. D., Jenkins, I. R. and Rhoades-Brown, J. E., ibid., 136-144 (1%8) Couper, A. S., Mat. Prof., 8 No. 10, 17-22 (1969) Gherhardi, D., Rivola, L., Troyli, M. and Bombara, G., Corrosion, 20 No. 3, 731-79t
1. Putilova. I. N., Balezin, S. A. and Barannik,
2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
(1%)
17. Rozenfel’d, 1. L. and Maksimchuk, V. P., Proc. Acad. Sci. (USSR), Phys. Chem. Section, 119 No. 5, 986 (1958) 18. Wormwell, F. and Mercer, A. D., J. Appl. Chem., 2 , I50 (1952) 19. Gilbert, P. T. and Hadden, S. E., ibid., 3, 545 (1953) 20. Thomas, J . G . N. and Mercer, A. D., 4th Int. Cong. on Metallic Corrosion, Amsterdam, 1969, N.A.C.E., Houston, 585 (1972)
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17:37
21. Ride, R. N., Symposium o n Corrosion (Melbourne Univ.), 267 (1955-56); J. Appl. Chem., 8, 175 (1958) 22. Edwards, W. T., Le Surf, J. E. and Hayes, P. A,, 2nd European Symp. on Corrosion Inhibitors, 1965; Univ. Ferrara, 679-700 (1966) 23. Spinedi, P. and Signorelli, G., 1st European Symp. o n Corrosion Inhibitors, Ferrara, 1960; Univ. Ferrara, 643-652 (1961) 24. Vance, R. W., Proc. 1st Joint Aerospace and Marine Corrosion Techn. Seminar, 1968; NACE, 34-35 (1%9) 25. Mansfeld, F., J. Elecfrochem. SOC., 118 No. 9, 1 412-1 415 (1971) 26. Matsuda, S. and Uhlig, H. H., J. Elecfrochem. Soc., 111, 156 (1954) 27. Mazza, F. and Trasatti, S., 3rd European Symp. on Corrosion Inhibitors., Ferrara, 1970; Univ. Ferrara, 277-291 (1971) 28. Butler, G. and Owen D., Corrosion Science, 9, 603-614 (1969) 29. Hatch, G. B., Mat. Prof.,8 No. 11, 31 (1969) 30. Stericker, W., Indusfr. Engnr. Chem., 37, 716 (1945) 31. Cone, C. S . , Mat. Prof. and Perf., 9 No. 7, 32-34 (1970) 32. Fisher, A. O., Mar. Prof., 3 No. 10, 8-13 (1964) 33. Bostwick, T. W., Corrosion, 17 No. 8 , 12 (1%1) 34. Brooks, W. B., Mat. Prof., 7 No. 2, 24-26 (1968) 35. Edwards, B. C., Corrosion Science, 9 No. 6. 395-404 (1969) 36. Darrin, M., Indusfr. Engng. Chem., 38, 368 (1946) 37. Conoby, J. F. and Swain, T. M., Mar. Prof., 6 No. 4, 55-58 (1967) 38. Butler, G., as Ref. 27, 753-776 (1971) 39. Butler, G . and Ison, H. C. K., Corrosion and its Prevention in Wafers. Leonard Hill,
London (1966) 40. Beecher, J. and Savinelli, E. A., Mat. Prof., 3 No. 2, 15-20 (1964) 41. Comeaux, R. V., Hydrocarbon Processing, 46 No. 12, 129 (1967) 42. Sussman, S., Mor. Prof., 3 No. 10, 52 (1964) 43. Hatch, G. B., Mat. Prof.. 4 No. 7, 52 (1965) 44. Ulmer, R. C. and Wood, J . W., Corrosion, 8 No. 12, 402 (1952) 45. Verma, K. M., Gupta, M. P. and Roy, A. K., Technology Quat. (Bull. Fertiliser Corp. India), 5 No. 2, 98-102 (1%8) 46. Hwa, C . M., Mat. Prof. and Perf., 9 No. 7, 29-31 (1970) 47. Schweitzer, G. W., Paper at the Int. Water Conference, Pittsburgh (1969) 48. US Patent No. 3 532 639 (4.3.68) 49. Mercer, A. D. and Wormwell, F., J. Appl. Chem., 9, 577-594 (1959) 50. Squires, A. P. T. B., The Protecfion of Motor Vehiclesfrom Corrosion, SOC.Chem. Ind. Monograph, No. 4 (1958) 51. Vernon, W. H. J., Wormwell, F., Ison, H. C. K. and Mercer, A. D., Motor Ind. Res. Assocn. Bulletin, 4th quarter, 19-20 (1949) 52. Dulat, J.. Brit. Corrosion J., 3 No. 4, 190-196 (1968) 53. BS 6580, British Standards Institution,Milton Keynes (1985) 54. Cavitation Corrosion and its Prevention in Diesel Engines, Symposium, 10th Nov. 1965, British Railways Board (1966) 5 5 . Cotton, J. B., Chem. and Indusf., 11th Feb., 214 (1967) 56. Cotton, J. B. and Jacob, W. R., Br. Corrosion J., 6 No. 1, 42-44 (1971) 57. Spivey, A. M., Chem. and Indusf.. 22nd April, 657 (1967) 58. Venczel, J. and Wranglen, G., Corrosion Science, 7, 461 (1967) 59. Drane, C. W., Br. Corrosion J., 6 No. 1, 39-41 (1971) 60. von Fraunhofer, J . A., ibid., No. 1, 28-30 (1971) 61. Obrecht, M. F.. 2nd International Congress on Metallic Corrosion, New York, 1%3; NACE, 624-645 (1966) 62. Maase, R. B., Mat. Prof., 5 No. 7, 37-39 (1966) 63. Streatfield, E. L., Corrosion Technology, 4 No. 7, 239-244 (1957) 64. Hoar, T. P., J. SOC.Chem. Indusf., 69, 356-362 (1950) 65. Bowrey, S. E., Trans. Inst. Marine Eng., 61 No. 3 , 1-9 (1949) 66. Palmer, W. G., J. Iron and Sfeel Inst., No. 12, 421-431 (1949) 67. Oakes, B. D.. Wilson, J. S. and Bettin, W. J., Proc. 26th NACE Conf, 1%9; NACE, 549-552 (1970)
68. Legault, R. A. and Bettin. W. J., Mar. Prof. and Perf., 9 No. 9, 35-39 (1970) 69. Mofor Vehicle Corrosion and Influence of De-Icing Chemicals, OECD Report. Oct. ( 1969)
17:38
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
70. Asanti, P., ‘Corrosion and its Prevention in Motor Vehicles’, Proc. Inst. Mech. Engns., 182, Part 35, 73-79 (1%7-68) 71. Bishop, R. R. and Steed, D. E., ibid., 124-129 (1967-68) 72. B.P. Applic. No. 61748/69 73. Rood Research 1970, Dept. of the Environment, Road Research Laboratory, H.M.S.O., London (1971) 74. Machu, W., as-Ref. 27, 107-119 (1971) 75. Every, R. L. and Riggs, 0. L., Mar. Prof., 3 No. 9. 46-47 (1964) 76. Uhlig, H. H., The Corrosion Handbook, John Wiley and Sons, Inc., 910 (1948) 77. Carassiti, V., Trabanelli, G. and Zucchi, F., as Ref. 22, 417-448 (1966) 78. Hudson, R. M., Looney, Q, L.and Warning, C. J., Brit. Corrosion J., 2 No. 3, 81-86 ( 1967) 79. Hudson, R. M. and Warning, C. J., Mut. Prot., 6 No. 2, 52-54 (1967) 80. Alfandary, M., eta/., as Ref. 22, 363-375 (1966) 81. Froment, M. and Desestret, A., as Ref. 22, 223-236 (1966) 82. Funkhouser, J.G., Corrosion, 17, 28% (1961) 83. Putilova, 1. N. and Chislova, E. N., Zerhchitu Metullov, 2 No. 3, 290-294 (1966) 84. Meakins, R. J., J. Appl. Chem., 13, 339 (1963) 85. Meakins, R. J., Br. Corrosion J., 6 No. 3, 109-113 (1971) 86. Riggs, 0. L. and Hurd, R. M., Corrosion, 24 No. 2, 45-49 (1968) 87. Rozenfel’d, I. L., Persiantseva, V. P. and Damaskina, T. A., Zushchita Mefullov., 9 NO. 6, 87-690 (1973) 88. Geld, I. and Acampara, M. A., Mat. Prot., 3 No. 1, 42-46 (1964) 89. Geld, I. and D’Oria, F., Mar. Prot., 6 No. 8 , 42-44 (1%7) 90. Bregman, J. I., as Ref. 27, 339-382 (1971) 91. Greenwell, H. E., Corrosion. 9, 307-312 (1953) 92. Waldrip, H. E., Mar. Prof., 5 No. 6, 8-13 (1966) 93. Patton, C. C., Deemer, D. A. and Hilliard, H. M., Mat. Prot., 9 No. 2, 37-41 (1970) 94. Haughin, J. E. and Mosier, B., Mot. Prof.. 3 No. 5 , 42-50 (1964) 95. Poetker, R. H. and Stone, J. D., Pefr. Eng., 28 No. 5, B-29-34 (1956) 96. Kerver, J. K. and Morgan, F. A., Mut. Prof., 2 No. 4, 10-20 (1%3) 97. Templeton, C. C., Rushing, S. S. and Rodgers, J . C., Mat. Prof., 2 No. 8 , 42-47 (1963) 98. Jones, L. W. and Barrett, J. P., Corrosion, 11, 217t (1955) 99. Dunlop, A. K., Howard, R. L. and Raifsnider, P. J., Mut. Prot., 8 No. 3.27-30 (1969) loo. Hatch, H. B. and Ralston, P. H., Mat. Prot., 3 No. 8 , 35-41 (1964) 101. Coulter, A. W. and Smithey, C. M., Mat. Prot., 8 No. 3, 37-38 (1969) 102. McDougall, L. A., Mat. Prot., 8 No. 8 , 31-32 (1969) 103. Tedeschi, R. J., Natali, P. W. and McMahon, H. C.. Proc. NACE 25th Conf., 1969; NACE, 173-179 (1970) 104. Rybachok, I. N., Mikhailov, M. A. and Tarasova, N. A., Korroziya i Zashchitu v Nefreguzovoi Prom., No. 7, 7-10 (1971) 105. Nathan, C. C., Mat. Prot. und Perf., 9 No. 11, 15-18 (1970) 106. Carlton, R. H., Mat. Prot., 2 No. 1, 15-20 (1963) 107. Brooke, J. M., Hydrocurbon Processing, Jan., 121-127 (1970) 108. Treadaway, K. W. J. and Russell, A. D., Highwuys und Public Works, 36 No. 1704, 19-21; NO. 1705, 40-41 (1%8) 109. Arber, M.G. and Vivian, H. E., Ausfruliun J. Appl. Science, 12 No. 4, 435-439 (1961) 110. North Thames Gas Board, British Patent No. 706 319 (31.3.54) 111. Moskin, V. M. and Alexseev, S. N., Beton i Zhelezbeton, No. 1, 28 (1957) 112. Alexseev, S. N. and Rozenfel’d, L. M., ibid., No. 2, 388 (1958) 113. Dilaktorski, N. L. and Oit, L. V., Tr. Nuuchn.-Issled Inst. Befonu i Zhelez Befona, Akud. Sfroit. i Arkhitekt. SSR, No. 22. 54-60 (1%1) 114. Gouda, V. K., Br. Corrosion J . , 5 No. 5, 198-208 (1970) 115. Banks, W. P., Mat. Prof., 7 No. 3, 35-38 (1%8) 116. Anon., Mat. Prof., 6 No. 2, 61 (1%7) 117. Swan, J. D., Bamberger, D. R. and Barthauer, G. L., Mot. Prot.,2 No. 9.26-34 (1963) 118. Bienstock, D. and Field, J.M., Corrosion, 17, 571t-574t (1961) 119. Banks, W. P., Mot. Prot., 6 NO. 11, 37-41 (1%7) 120. Glossary of Termsfor Corrosion of Metuls und Alloys, BS 6918:1987, I S 0 8044-1986 121. Rosenfeel’d, 1. L. Corrosion Inhibitors, McGraw-Hill, (1981)
CORROSION INHIBITION: PRINCIPLES AND PRACTICE
17 :39
122. Ranney, M. W., Corrosion Inhibitors: Manufacture and Technology, Chemical Technology Review No 60,Noyes Data Corporation, New Jersey and London (1976) 123. Robinson, J. S., Corrosion Inhibitors: Recent Developments, Chemical Technology Review No 132, Noyes Data Corporation, New Jersey (1979) 124. Collie, M. J. (ed.), CorrosionInhibitors:Developments Since 1980, Chemical Technology Review No 23, Noyes Data Corporation, New Jersey (1983) 125. Flick, E. W., Corrosion Inhibitors: An Industrial Guide, Noyes Data Corporation, New Jersey (1987) 126. Proc. 6th European Symposium on Corrosion Inhibitors (6 SEIC) Ann. Univ. Ferrara, N.S., Sez. V. Suppl. N8 (1985) 127. Proc. 7th European Symposium on Corrosion Inhibitors (7SEIC) Ann. Univ. Ferrara, N.S.,Sez. V. Suppl. N9 (1990) 128. Vukasovich, M. S. and Farr, J . P. G., Polyhedra 5 No. 1/2), 551-559 (1986) 129. Mercer, A. D., ‘Test Methods for Corrosion Inhibitors’, Br. Corros. J., 20 No. 2, 61-70 (1985) 130. Mohr, P. andMatulewicz, W. N., (Union CarbideCorp.), U S . Patent 4404 114,13 Sept. 1983, Appl. 389 394, 17 Jan. 1982 131. Katayama Chemical Works Co. Ltd, Jap. Patent 5891, 176, [appl. 26 Nov 19811 quoted in Chem. Abs., CA 99: 179935 132. Katayama Chemical Works Co. Ltd, Jap. Patent 57209981 [appl. 19 June 19811 quoted in Chern. Abs., CA 98: 145894 133. Gisela, S. et ai. (VEB Fahlberg-List), Ger. (East) DD 211, 357 appl. 10 Nov 1982 [appl. 6 May 19861 quoted in Chem. Abs., CA 102: 97635 134. Itoh, M. et al. Corrosion Engineering, 36, 139-45 (1987) 135. Hanazaki, M., Harada, N and Shimada, K (Nippon Light Metal Company Ltd), US Patent 4655951 [appl. 6 March 19861 quoted in Chem. Abs. CA107: 10537 136. Vukasovich, M. S.. Lubrication Engineering. 40, No. 8, 456-462 (1984) 137. Biggs, G. L. (Hughes Tool Co.), U.S.Patent 4 698 168, 29 Aug. 1986 138. Nicols, J. D., (Air Products and Chemicals Inc..) U.S.Patent 4 557 838, Apr. 1982 139. Lewis, M. (Halliburton Co.), EP 130 006 (first application USA 1983) 140. Schmitt, G., Werh. Korroz., 35, 3, 107-110 (1984) 141. Sung, R. L. (Texaco) U.S.Patent 4 282 007, 1981 142. 7th Symposium Int. Carbur. Alcool., (1986)
17.3 The Mechanism of Corrosion Prevention by Inhibitors The mechanisms of corrosion inhibition will be described separately for acid and neutral solutions, since there are considerable differences in mechanisms between these two media. Definitions and classifications of inhibitors are given in Section 17.2 and by Fischer'.
Inhibitors for Acid Solutions The corrosion of metals in aqueous acid solutions can be inhibited by a very These include relatively simple substances, such wide range of as chloride, bromide or iodide ions and carbon monoxide, and many organic compounds, particularly those containing elements of Groups V and VI of the Periodic Table, such as nitrogen, phosphorus, arsenic, oxygen, sulphur and selenium. Organic compounds containing multiple bonds, especially triple bonds, are effective inhibitors. Organic compounds of high molecular weight, e.g. proteins and polysaccharides, also have inhibitive properties. The primary step in the action of inhibitors in acid solutions is generally agreed to be adsorption on to the metal surface, which is usually oxide-free in acid solutions. The adsorbed inhibitor then acts to retard the cathodic and/or anodic electrochemical processes of the corrosion. The factors influencing the adsorption and the electrochemical reactions will be considered in turn. (See also Section 20.1 .) Adsorption of Corrosion Inhibitors onto Metals
Measurements of the adsorption of inhibitors on corroding metals are best carried out using the direct methods of radio-tracer detection4-' and solution depletion measurements '-lo. These methods provide unambiguous information on uptake, whereas the corrosion reactions may interfere with the indirect methods of adsorption determination, such as double layer capacity measurements ' I , coulometry", ellipsometry l2 and reflectivityI2. Nevertheless, double layer capacity measurements have been widely used for the determination of inhibitor adsorption on corroding metals, with apparently consistent results, though the interpretation may not be straightforward in some cases. 17 :40
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17: 41
Direct measurements on metals such as iron, nickel and stainless steel have shown that adsorption occurs from acid solutions of inhibitors such as iodide ionsI4, carbon monoxideI5 and organic compounds such as 10. 16, thioureas9. I6-I8, sulphoxides 19-", sulphides2'*22and mercaptansi6. These studies have shown that the efficiency of inhibition (expressed as the relative reduction in corrosion rate) can be qualitatively related to the amount of adsorbed inhibitor on the metal surface. However, no detailed quantitative correlation has yet been achieved between these parameters. There is some evidence that adsorption of inhibitor species at low surface coverage 8 (for complete surface coverage 8 = 1) may be more effective in producing inhibition than adsorption at high surface coverage. In particular, the adsorption of polyvinyl pyridine on iron in hydrochloric acid at f3 < 0.1 monolayer has been found to produce an 80% reduction in corrosion rate lo. In general, the results of direct adsorption measurements provide a basis for the widely used procedure of inferring the adsorption behaviour of inhibitors from corrosion rate measurements. This involves the assumption that the corrosion reactions are prevented from occurring over the area (or active sites) of the metal surface covered by adsorbed inhibitor species, whereas these corrosion reactions occur normally on the inhibitor-free area. The inhibitive efficiency is then directly proportional to the fraction of the surface covered with adsorbed inhibitor. This assumption has been applied to deduce the effects of concentration on the adsorption of inhibitors, and to compare the adsorption of different inhibitors (usually related in structure) at the same concentration. On the whole, the interpretation in this way of the efficiency of inhibitors in terms of their adsorption behaviour has given consistent results, which have clarified the factors influencing inhibition and adsorption. However, some qualifications are necessary in this approach, since this simple relationship between inhibitive efficiency and adsorption will not always apply. As mentioned above, at low surface coverage (8 < 0- 1 ) . the effectiveness of adsorbed inhibitor species in retarding the corrosion reactions may be greater than at high surface coverage'0*16*23. In other cases, adsorption of inhibitors, e.g. t h i ~ u r e a s " ~ ~ ~ and amine^^'^' from solutions of low concentration may cause stimulation of corrosion. Furthermore, in comparing the inhibitive efficiency and adsorption of different inhibitors, possible differences in the mechanism and effectiveness of retardation of the corrosion reactions must be considered2*. The information on inhibitor adsorption, derived from direct measurements and from inhibitive efficiency measurements, considered in conjunc-'~, that tion with general knowledge of adsorption from s ~ l u t i o n ~ ~indicates inhibitor adsorption on metals is influenced by the following main factors. Surface charge on the metal Adsorption may be due to electrostatic attractive forces between ionic charges or dipoles on the adsorbed species and the electric charge on the metal at the metal/solution interface. In solution, the charge on a metal can be expressed by its potential with respect to the zero-charge potential (see Section 20.1). This potential relative to the zerocharge potential, often referred to as the +potential3', is more important with respect to adsorption than the potential on the hydrogen scale, and indeed the signs of these two potentials may be different. As the +potential
17: 42 THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
becomes more positive, the adsorption of anions is favoured and as the +-potential becomes more negative, the adsorption of cations is favoured. The +-potential also controls the electrostatic interaction of the metal with dipoles in adsorbed neutral molecules, and hence the orientation of the dipoles and the adsorbed molecules. For different metals at the same 4potential, electrostatic interactions should be independent of the nature of the metal, and this has been used as a basis to compare adsorption of inhibitors on different metal^^'"^. Thus Antropov” has shown that the adsorption of some inhibitors on mercury can be related to their adsorption and inhibitive effect on iron, when considered at the same +-potentials for both metals. The differences in behaviour of an inhibitor on various metals can also in some cases be related to the differences in +-potentials at the respective corroding potentials. The functional group and structure of the inhibitor Besides electrostatic interactions, inhibitors can bond to metal surfaces by electron transfer to the metal to form a coordinate type of link. This process is favoured by the presence in the metal of vacant electron orbitals of low energy, such as occur in the transition metals. Electron transfer from the adsorbed species is favoured by the presence of relatively loosely bound electrons, such as may be found in anions, and neutral organic molecules containing lone pair electrons or ?r-electronsystems associated with multiple, especially triple, bonds or aromatic rings. In organic compounds, suitable lone pair electrons for co-ordinate bonding occur in functional groups containing elements of Groups V and VI of the Periodic Table. The tendency to stronger co-ordinate bond formation (and hence stronger adsorption) by these elements increases with decreasing electronegativity in the order 0 < N < S < Se35‘37,and depends also on the nature of the functional groups containing these elements. The structure of the rest of the molecule can affect coordinate bond formation by its influence on the electron density at the functional The effects of substituents in related inhibitor molecules, e.g. a n i l i n e ~ ~ ~ . ~aliphatic ’ - ~ ’ , amines”, amino acids*, benzoic pyridines and aliphatic sulphides”, on the inhibitive efficiencies have been correlated with changes in electron densities at functional groups, as derived from nuclear magnetic resonance measurements 39, values of Hammett constants (aromatic or Taft constants (aliphatic molec u l e ~ ) ~ ’ * *or * ~from ~ , quantum mechanical calculations 38,4’948. The results of these investigations generally indicate that the electron density at the functional group increases as the inhibitive efficiency increases in a series of related compounds. This is consistent with increasing strength of coordinate bonding due to easier electron transfer, and hence greater adsorption. An analogous correlation has been demonstrated by Ha~kerman”.~’ between inhibitive efficiencies in a series of cyclic imines (CH2),NH and changes in hybrid bonding orbitals of the electrons on the nitrogen atom making electron transfer and coordinate bond formation easier. 38,39*43948,
Interaction of the inhibitor with water molecules Due to the electrostatic and co-ordinate bond interactions described under the previous two headings, the surfaces of metals in aqueous solutions are covered with adsorbed water molecules. Adsorption of inhibitor molecules is a displacement
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :43
reaction involving removal of adsorbed water molecules from the surface. During adsorption of a molecule, the change in interaction energy with water molecules in passing from the dissolved to the adsorbed state forms an important part of the free energy change on adsorption. This has been shown to increase with the energy of solvation of the adsorbing species, which in turn increases with increasing size of the hydrocarbon portion of an organic molecule 36. Thus increasing size leads to decreasing solubility and increasing adsorbability. This is consistent with the increasing inhibitive efficiency observed at constant concentrations with increasing molecular size in a series of related corn pound^^^^^^^^^.
Interaction of adsorbed inhibitor species Lateral interactions between adsorbed inhibitor species may become significant as the surface coverage, and hence the proximity, of the adsorbed species increases. These lateral interactions may be either attractive or repulsive. Attractive interactions occur between molecules containing large hydrocarbon components, e.g. n-alkyl chains. As the chain length increases, the increasing van der Waals attractive force between adjacent molecules leads to stronger adsorption at high coverage. Repulsive interactions occur between ions or molecules containing dipoles and lead to weaker adsorption at high coverage. The effects of lateral interactions between adsorbed inhibitors on inhibitive efficiency have been discussed by Hoar and Khera26. In the case of ions, the repulsive interaction can be altered to an attractive interaction if an ion of opposite charge is simultaneously adsorbed. In a solution containing inhibitive anions and cations the adsorption of both ions may be enhanced and the inhibitive efficiency greatly increased compared to solutions of the individual ions. Thus, synergistic inhibitive effects occur in such mixtures of anionic and cationic inhibitor^^^'^^. These synergistic effects are particularly well defined in solutions containing halide ions, I-. Br-, C1-, with other inhibitors such as quaternary ammonium cations56, alkyl benzene pyridinium cations5’, and various types of It seems likely that co-ordinate-bond interactions also play some part in these synergistic effects, particularly in the interaction of the halide ions with the metal surfaces and with some Reaction of adsorbed inhibitors In some cases, the adsorbed corrosion inhibitor may react, usually by electro-chemical reduction, to form a product which may also be inhibitive. Inhibition due to the added substance has been termed primary inhibition and that due to the reaction product secondary i n h i b i t i ~ n ~ In ~ - ~such ~ . cases, the inhibitive efficiency may increase or decrease with time according to whether the secondary inhibition is more or less effective than the primary inhibition. Some examples of inhibitors which react to give secondary inhibition are the following. Sulphoxides can be reduced to sulphides, which are more efficient inhibitor^'^,^.^',^',^ . Quaternary phosphonium and arsonium compounds can be reduced to the corresponding phosphine or arsine compounds, with little change in inhibitive e f f i c i e n ~ y ~Acetylene ~’~. compounds can undergo reduction followed by polymerisation to form a multimolecular protective film66*67. Thioureas can be reduced to produce HS- ions, which may act as stimulators of corro~ion~~.~~’~~.
17 :44 THE
MECHANISM OF CORROSION PREVENTION BY INHIBITORS
Effects of Inhibitors on Corrosion Processes
In acid solutions the anodic process of corrosion is the passage of metal ions from the oxide-free metal surface into the solution, and the principal cathodic process is the discharge of hydrogen ions to produce hydrogen gas. In air-saturated acid solutions, cathodic reduction of dissolved oxygen also occurs, but for iron the rate does not become significant compared to the rate of hydrogen ion discharge until the pH exceeds about 3. An inhibitor may decrease the rate of the anodic process, the cathodic process or both processes. The change in the corrosion potential on addition of the inhibitor is often a useful indication of which process is retarded24*67. Displacement of the corrosion potential in the positive direction indicates mainly retardation of the anodic process (anodic control), whereas displacement in the negative direction indicates mainly retardation of the cathodic process (cathodic control). Little change in the corrosion potential suggests that both anodic and cathodic processes are retarded (see Section 1.4 for appropriate potential versus current diagrams). The effects of adsorbed inhibitors on the individual electrode reactions of corrosion may be determined from the effects on the anodic and cathodic A displacement of the polarisation curves of the corroding meta124~28~68*69. polarisation curve without a change in the Tafel slope in the presence of the inhibitor indicates that the adsorbed inhibitor acts by blocking active sites so that reaction cannot occur, rather than by affecting the mechanism of the reaction. An increase in the Tafel slope of the polarisation curve due to the inhibitor indicates that the inhibitor acts by affecting the mechanism of the reaction. However, the determination of the Tafel slope will often require the metal to be polarised under conditions of current density and potential which are far removed from those of normal corrosion. This may result in differences in the adsorption and mechanistic effects of inhibitors at polarised metals compared to naturally corroding metal^^^"^^'^. Thus the interpretation of the effects of inhibitors at the corrosion potential from applied current-potential polarisation curves, as usually measured, may not be conclusive. This difficulty can be overcome in part by the use of rapid polarisation method^^^'^'. A better procedure24 is the determination of ‘true’ polarisation curves near the corrosion potential by simultaneous measurements of applied current, corrosion rate (equivalent to the true anodic current) and potential. However, this method is rather laborious and has been little used. Electrochemical studies have shown that inhibitors in acid solutions may affect the corrosion reactions of metals in the following main ways.
Formation of a diffusion barrier The absorbed inhibitor may form a surface film which acts as a physical barrier to restrict the diffusion of ions or molecules to or from the metal surface and so retard the corrosion reactions. This effect occurs particularly when the inhibitor species are large molecules, e.g. proteins, such as gelatine, agar agar; polysaccharides, such as dextrin; or compounds containing long hydrocarbon chains. Surface films of these types of inhibitors give rise to resistance polarisation and also concentration polarisation affecting both anodic and cathodic reactions72.Similar effects also occur when the inhibitor can undergo reaction to form a multimolecular
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :45
surface film, e.g. acetylenic compounds67and sulphoxides ‘9-73. Blocking of reaction sites The interaction of adsorbed inhibitors with surface metal atoms may prevent these metal atoms from participating in either the anodic or cathodic reactions of corrosion. This simple blocking effect decreases the number of surface metal atoms at which these reactions can occur, and hence the rates of these reactions, in proportion to the extent of adsorption. The mechanisms of the reactions are not affected and the Tafel slopes of the polarisation curves remain unchanged. Behaviour of this type has been observed for iron in sulphuric acid solutions containing 2,6-dimethyl quinolinez4, P-naphthoquinolineW, or aliphatic ~ u l p h i d e s ~ ~ . It should be noted that the anodic and cathodic processes may be inhibited to different extent^'^*'^*^^. The anodic dissolution process of metal ions is considered to occur at steps or emergent dislocations in the metal surface, where metal atoms are less firmly held to their neighbours than in the plane surface. These favoured sites occupy a relatively small proportion of the metal surface. The cathodic process of hydrogen evolution is thought to occur on the plane crystal faces which form most of the metal surface area. Adsorption of inhibitors at low surface coverage tends to occur preferentially at anodic sites, causing retardation of the anodic reaction. At higher surface coverage, adsorption occurs on both anodic and cathodic sites, and both reactions are inhibited. Participation in the electrode reactions The electrode reactions of corrosion involve the formation of adsorbed intermediate species with surface metal atoms, e.g. adsorbed hydrogen atoms in the hydrogen evolution reaction; adsorbed (FeOH) in the anodic dissolution of i r ~ n ~ The ” ~ presence ~. of adsorbed inhibitors will interfere with the formation of these adsorbed intermediates, but the electrode processes may then proceed by alternative paths through intermediates containing the inhibitor. In these processes the inhibitor species act in a catalytic manner and remain unchanged. Such participation by the inhibitor is generally characterised by a change in the Tafel slope observed for the process. Studies of the anodic dissolution of iron in the presence of some inhibitors, e.g. halide ion^'^'^^-^^, aniline and its derivative^".^^, the benzoate ion7’ and the furoate ionw, have indicated that the adsorbed inhibitor Z participates in the reaction, probably in the form of a complex of the type (Fe-I),,,, or (Fe-OH-I),,,,. The dissolution reaction proceeds less readily via the adsorbed inhibitor complexes than via (Fee OH),,,, , and so anodic dissolution is inhibited and an increase in Tafel slope is observed for the reaction. Adsorbed species may also accelerate the rate of anodic dissolution of metals, as indicated by a decrease in Tafel slope for the reaction. Thus the presence of hydrogen sulphide in acid solutions stimulates the corrosion of iron, and decreases the Tafel slopeZS.s4*56. The reaction path through (Fee HS-),..,,. has been postulated to lead to easier anodic dissolution than that through (Fe.OH),,,, . This effect of hydrogen sulphide is thought to be responsible for the acceleration of corrosion of iron observed with some inhibitive sulphur compounds, e.g. t h i o u r e a ~ ~ ~at ~ ~low ~ *concentrations, *I, since hydrogen sulphide has been identified as a reduction product. However, the effects of hydrogen sulphide are complex, since in the presence of inhibitors such as aminess6, quaternary ammonium cationss6, thioureass4.*’,
17: 46 THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
synergistic enhancement of inhibition is observed due to interaction of adsorbed HS- ions with the adsorbed inhibitor. Inhibitors may also retard the rate of hydrogen evolution on metals by affecting the mechanism of the reaction, as indicated by increases in the Tafel slopes of cathodic polarisation curves. This effect has been observed on iron in the presence of inhibitors such as p h e n y l - t h i o ~ r e a ~ ~ acetylenic ’~~, h y d r ~ c a r b o n s ~ aniline ~ ’ ~ ~ , derivativesE4, benzaldehyde derivatives” and pyrilium saltsBS.According to AntropovS6 and G r i g o r y e ~ ~the ’ ~ ~rate , determining step (which depends on experimental conditions7’) for the hydrogen evolution reaction on iron in acid solutions (pH less than 2) is the recombination of adsorbed hydrogen atoms to form hydrogen molecules. Grigoryev &4v8S has shown that addition of anilines, benzaldehydes and pyrilium salts to hydrochloric acid tends to retard the discharge of hydrogen ions to form adsorbed hydrogen atoms on iron, so that this step rather than the recombination step tends to control the rate of the overall hydrogen evolution reaction. Some inhibitors, e.g. and sulphoxidesa7,which can add on hydrogen ions in acid solutions to form protonated species, may accelerate the rate of the cathodic hydrogen evolution reaction on metals, due to participation of the protonated species in the reaction. This occurs when the discharge of the protonated species to produce an adsorbed hydrogen atom at the metal surface occurs more easily than the discharge of the hydrogen ion. This effect becomes more significant as the hydrogen overvoltage of the metal increases, and so may be observed to a greater extent on zinc than on iron 33. Alteration of the electrical double layer The adsorption of ions or species which can form ions, e.g. by protonation, on metal surfaces will change the electrical double layer at the metal-solution interface, and this in turn will affect the rates of the electrochemical reactions 33,’4. The adsorption of cations, e.g. quaternary ammonium ions ” and protonated makes the potential more positive in the plane of the closest approach to the metal of ions from the solution. This positive potential displacement retards the discharge of the positively charged hydrogen ion. For the inhibition of iron corrosion by pyridines in acid solutions, Antropov” has calculated the theoretical inhibition coefficients of the hydrogen ion discharge reaction, due to the effect of adsorbed pyridine cations on the electrical double layer. The calculated values agreed well with observed values at low inhibitor concentrations, indicating that inhibition could be wholly attributed to electrostatic effects, and that blocking of the surface by adsorbed inhibitor is not important. Conversely, the adsorption of anions makes the potential more negative on the metal side of the electrical double layer and this will tend to accelerate the rate of discharge of hydrogen ions. This effect has been observed for the sulphosalicylate ions4and the benzoate ion”. Conclusions
Thus, inhibitors of corrosion in acid solution can interact with metals and affect the corrosion reaction in a number of ways, some of which may occur
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :47
simultaneously. It is often not possible to assign a single general mechanism of action to an inhibitor, because the mechanism may change with experimental conditions. Thus, the predominant mechanism of action of an inhibitor may vary with factors such as: its concentration, the pH of the acid, the nature of the anion of the acid, the presence of other species in the solution, the extent of reaction to form secondary inhibitors and the nature of the metal. The mechanism of action of inhibitors with the same functional group may additionally vary with factors such as the effect of the molecular structure on the electron density of the functional group and the size of the hydrocarbon portion of the molecule. However, the mechanisms of action of a number of inhibitors have now been identified and are beginning to be understood on the molecular level.
Inhibitors in Near-neutral Solutions The corrosion of metals in neutral solutions differs from that in acid solutions in two important respects (see Section 1.4). In air-saturated solutions, the main cathodic reaction in neutral solutions is the reduction of dissolved oxygen, whereas in acid solution it is hydrogen evolution. Corroding metal surfaces in acid solution are oxide free, whereas in neutral solutions metal surfaces are covered with films of oxides, hydroxides or salts, owing to the reduced solubility of these species. Because of these differences, substances which inhibit corrosion in acid solution by adsorption on oxide-free surfaces, do not generally inhibit corrosion in neutral solution*. Inhibition in neutral solutions is due to compounds which can form or stabilise protective surface films. The inhibitor may form a surface film of an insoluble salt by precipitation or reaction. Inhibitors forming films of this type include: ( a ) salts of metals such as zinc, magnesium, manganese and nickel, which form insoluble hydroxides, especially at cathodic areas, which are more alkaline due to the hydroxyl ions produced by reduction of oxygen; ( b ) soluble calcium salts, which can precipitate as calcium carbonate in waters containing carbon dioxide, again at cathodic areas where the high pH permits a sufficiently high concentration of carbonate ions; ( c ) polyphosphates in the presence of zinc or calcium, which produce a thin amorphous salt film. The mechanism of action of these inhibitors seems fairly straightforward". The salt films, which are often quite thick and may be visible, restrict diffusion, particularly of dissolved oxygen to the metal surface. They are poor electronic conductors and so oxygen reduction does not occur on the film surface; these inhibitors are referred to, therefore, as cathodic inhibitors. The mechanism of inhibition by polyphosphates is more complex, and the various theories of their action have recently been described by Butlerw. Another class of inhibitors in near-neutral solutions act by stabilising oxide films on metals to form thin protective passivating films. Such inhibitors are the anions of weak acids, some of the most important in practice being chromate, nitrite, benzoate, silicate, phosphate and borate. Passivating *Exceptions are organic compounds of high molecular weight, e.g. gelatine, agar and dextrin. Adsorption of these large molecules is partly effective in shielding the metal surface from reaction in neutral as well as acid solutions2.
17 :48
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
oxide films on metals offer high resistance to the diffusion of metal ions and the anodic reaction of metal dissolution is inhibited; thus these inhibitive anions are often referred to as anodic inhibitors, and they are more generally used than cathodic inhibitors to inhibit the corrosion of iron, zinc, aluminium, copper and their alloys, in near neutral solutions. The conditions under which oxide films are protective on these metals in relation to inhibition by anions may be characterised in terms of three important properties of the oxide film (see also Sections 1.4 and 1.5): 1. The Flade potential, which is the negative potential limit of stability of
the oxide film. At potentials more negative than the Flade potential the oxide film is unstable with respect to its reduction or dissolution, or both, since the rates of these two processes exceed that of film formation. 2. The critical breakdown potential, which is the positive potential limit of stability of the oxide film. At this potential and more positive potentials, the oxide film is unstable with respect to the action of anions, especially halide ions, in causing localised rupture and initiating pitting corrosion. 3. The corrosion current due to diffusion of metal ions through the passivating film, and dissolution of metal ions at the oxide-solution interface. Clearly, the smaller this current, the more protective is the oxide layer. All of these three properties of the oxide films on metals are influenced by the anion composition and pH of the solution. In addition the potential of the metal will depend on the presence of oxidising agents in the solution. Inhibition of corrosion by anions thus requires an appropriate combination of anions, pH and oxidising agent in the solution so that the oxide film on the metal is stable (the potential then lying between the Flade potential and the breakdown potential), and protective (the corrosion current through the oxide being low). Most of the information available on the mechanism of action of inhibitive anions relates to iron, which will be discussed in some detail, and followed by brief accounts of zinc, aluminium and copper.
Iron
The corrosion of iron (or steel) can be inhibited by the anions of most weak However, other anions, particularly acids under suitable those of strong acids, tend to prevent the action of inhibitive anions and stimulate breakdown of the protective oxide film. Examples of such aggressive anions are the halides, sulphate, nitrate, etc. Brasher” has shown that, in general, most anions exhibit some inhibitive and some aggressive behaviour towards iron. The balance between the inhibitive and aggressive properties of a specific anion depends on the following main factors (which are themselves interdependent). Concentration
Inhibition of iron corrosion in distilled water occurs only
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :49
when the anion concentration exceeds a critical At concentrations below the critical value, inhibitive anions may act aggressively and stimulate breakdown of the oxide films9'. Effective inhibitive anions have low critical concentrations for inhibition. Brasher92has classified a number of anions in order of their inhibitive power towards steel, judged from their critical inhibitive concentrations. The order of decreasing inhibitive efficiency is: azide, ferricyanide, nitrite, chromate, benzoate, ferrocyanide, phosphate, tellurate, hydroxide, carbonate, chlorate, o-chlorbenzoate, bicarbonate, fluoride, nitrate, formate. Thus, normally aggressive anions such as fluoride and nitrate may inhibit steel corrosion at sufficiently high concentrations. pH Inhibitive anions are effective in preventing iron corrosion only at pH values more alkaline than a critical value. This critical pH depends on the anion, e.g. approximate critical pH values for the inhibition of iron or steel in about 0.1 M solution of the anion increase in the order: chromatew, 1.0; a ~ e l a t e 4.5; ~ ~ , nitrite"*97, 5 . 0 - 5 . 5 ; 6.0; phosphate""', 7.2; hydroxide", =12 (not 0.1 M). The critical pH value for inhibition depends on the concentration of the inhibitive anion. In azelate" and nitrite% solutions, there are indications that t4e critical pH for inhibition decreases as the anion concentration increases. However, in benzoate solutionsw, increasing benzoate concentration displaces the critical pH to more alkaline values. Dissolved oxygen concentration and supply Inhibition of the corrosion of iron by anions requires a critical minimum degree of oxidising power in the solution. This is normally supplied by the dissolved oxygen present in airsaturated solutions. Gilroy and Mayne"' have shown that the critical oxygen concentration for inhibition of iron in 0- 1 M sodium benzoate (pH 7.0) is = O m 3 p.p.m., considerably less than the air-saturated concentration of "8p.p.m. As the oxygen concentration is reduced below this critical value, the rate of breakdown of the passivating oxide film increases. As the pH of 0.1 M sodium benzoate is reduced below 7.0, the critical oxygen concentration for inhibition increases IO2. The critical oxygen concentration for inhibition depends on the nature of the anion"'. If the inhibitive anion possesses oxidising properties, e.g. chromate93-IO3*IO4, nitrite", pertechnetatel@'-lM, then the presence of dissolved oxygen may not be necessary for inhibition. The critical oxygen concentrations for good inhibitive nonoxidising anions are low lo'. If the dissolved oxygen concentration is increased above that of the air-saturated solution, the inhibition of iron corrosion is facilitatedIo2, and inhibition may even be achieved in chloride ~olution"~.Similarly, increasing the oxygen supply to the iron surface by rapid stirring or aeration of the solution may favour inhibition, resulting in inhibition at lower critical anion concentration^^^"^, and again inhibition in chloride solutions may be obtainedIm. Addition of an oxidising agent may improve the efficacy of inhibitive anions, e.g. Mayne and Page''' have recently shown that the presence of hydrogen peroxide lowers the critical concentrations of sodium benzoate and sodium azelate required for inhibition of steel, and also lowers the critical pH values for inhibition. Aggressive anion concentration When aggressive anions are present in the
17: 50 THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
solution, the critical concentrations of inhibitive anions required for protection of iron are increased 108~111-116.Brasher and Mercer "*-'" have shown that the relationship between the maximum concentration of aggressive anion Cas, permitting full protection by a given concentration of inhibitive anion c i n h . is of the form where K is a constant dependent on the nature of the inhibitive and aggressive anions, and n is an exponent which is approximately the ratio of the valency of the inhibitive anion to the valency of the aggressive anion. This relationship indicates a competitive action between inhibitive anions and aggressive anions and its significance will be discussed below. In general, the more aggressive the anion, the smaller the concentration which can be ~ aggressive tolerated by an inhibitive anion. The order of t o l e r a n ~ e "of anions is, with certain exceptions, consistent with the order of aggressiveness of these anions as determined from their tendency to induce breakdown of the oxide film on iron in aerated solutions. Nature of the metal surface The critical concentration of an anion required to inhibit the corrosion of iron may increase with increasing surface roughness. Thus, Brasher and Mercer"' showed that the minimum concentration of benzoate required to protect a grit-blasted steel surface was about 100times greater than that required to protect an abraded surface. However, surface preparation had little effect on the critical inhibitive concentrations for ~ h r o m a t e "or ~ nitrite'I4 The time of exposure of the iron surface to air after preparation and before immersion may also affect the ease of inhibition by anions. There is evidence''. 'O'* 'I' that the inhibition by anions occurs more readily as the time of pre-exposure to air increases. Similarly, if an iron specimen is immersed for some time in a protective solution of an inhibitive anion, it may then be transferred without loss of inhibition to a solution of the anion containing much less than the critical inhibitive concentration9'. Temperature In general, the critical concentrations of anions, e.g. benzoate'083112, chromate".' and nitrite'I4, required for the protection of steel increase as the temperature increases. Passivating Oxide Films
Studies of iron surfaces inhibited in solutions of anions have shown by several independent techniques, e.g. examination of in situ and stripped and ellipsofilms by electron diffraction 118-124, cathodic reduction m) metry'", the presence of a thin film (thickness 2 3 x 10-6m to 5 x of cubic iron oxide (Fe,04 or y-Fe,O,), which is rather similar to the airformed oxide film 128-30. Immersion of iron bearing its air-formed oxide film into solutions of inhibitive anions usually results in a thickening of the oxide layer125,131, except at relatively low pH'". Oxide film growth on iron in inhibitive solutions of anions as well as in air, follows a direct logarithmic law, the rate constants being generally slightly greater in solution than '"3
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :5 1
in 127. 131.133-136 I . t is generally agreed that inhibition of the corrosion of iron by anions results from their effects on this oxide layer'37-'M.These effects are of several kinds, and they will now be discussed in relation to the theories of inhibition by anions. It seems probable, that there is no single mechanism of inhibition, but that a number of factors are involved, their relative importance depending on the nature of the anion and experimental conditions. Uptake of Anions by Oxide Fims
Early studies on oxide films stripped from iron showed the presence of chromium after inhibition in chromate solution '41 and of crystals of ferric phosphate after inhibition in phosphate solutions 1 2 ' . More recently, radiotracer studies using labelled anions have provided more detailed information on the uptake of anions. These measurements of irreversible uptake have and phosshown that some inhibitive anions, e.g. chromateW*'33.-'36'46,147 phate"7*'48,are taken up to a considerable extent on the oxide film. However, other equally effective inhibitive anions, e.g. benzoate 149,'sI pertechnetate is','5z and azelateLS3,are taken up to a comparatively small extent. Anions may be adsorbed on the oxide surface by interactions similar to those described above in connection with adsorption on oxide-free metal surfaces. On the oxide surface there is the additional possibility that the adsorbed anions may undergo a process of ion exchange whereby they replace oxide ions, which leave the oxide lattice for the solution. Adsorption and ion-exchange represent different aspects of the same process. However, it would be expected that an anion would be more firmly bound after ion exchange because of the greater interaction with neighbouring metal ions. Anions taken up by adsorption/ion exchange, e.g. phosphate'" and c h r ~ m a t e ' ~would ~, be expected to be distributed fairly uniformly over the surface, though binding energies would vary with different types of adsorption site. There is considerable evidence that uptake of anions may also be concentrated into particles of separate phases located in the main oxide film, e.g. phosphate 117*12', pertechnetate'" and azelate The formation of these particles of separate phase has been observed mainly when conditions are relatively unfavourable for inhibition, e.g. low pH thin oxide film due to short air e x p o s ~ r e "and ~ the presence of aggressive anions '54. This evidence of the uptake of inhibitive anions into oxide films forms the basis of the 'chemical' or 'pore plugging' theory of inhibition, associated originally with Evans'" etai. In this theory the rdle of the inhibitive anion is to promote the repair of weak points or pores in the oxide film, where corrosion has started, by reacting with dissolving iron cations to form insoluble products of separate phase, which plug the gaps. These insoluble products may contain the inhibitive anion either as a salt, e.g. phosphate'*', or a basic salt, e.g. or as an insoluble oxide, e.g. Cr203 from Precipitation of such solid products is favoured if the pH in the region of the pores does not become acid. Thus, on the basis of this theory, inhibition by anions such as phosphate, borate, silicate and carbonate, is enhanced by their buffer properties which serve to prevent a fall in pH in the anodic areas. Since ferric salts are usually more insoluble than
17 :52 THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
ferrous salts, the requirement of oxidising power in the solution for inhibition is explained as necessary for oxidation of ferrous to insoluble ferric compounds, either by dissolved oxygen or oxidising anions, e.g. chromate or nitrite. There is undoubted evidencethat pore plugging as described by this theory does occur, particularly when conditions are relatively unfavourable for inhibition. However, this theory does not provide a complete explanation of the action of inhibiting anions. Some inhibitive anions, e.g. azide" and pertechnetatelm, do not form insoluble salts with ferrous or ferric ions. Furthermore, the pertechnetate ion has negligible buffer capacityIw. Oxidising power is not necessarily a criterion of inhibitive efficiency, e.g. permanganate rapidly oxidises ferrous ions to ferric, but is a poor i n h i b i t ~ r ' ~Also, ~ . there is little correlation between the extent to which anions are incorporated into oxide films and their inhibitive efficiency 'I7. The inhibitive action of anions on iron is soon lost after transfer of the metal from the anion solution to water. The beha~iour"'*"~ of iron in solutions containing mixtures of inhibitive and aggressive anions indicates that there is a competitive uptake of the inhibitive and the aggressive anions. These facts strongly suggest that the inhibitive effect of anions is exerted through a relatively labile adsorption on the oxide surface, rather than irreversible incorporation into the oxide film. Effect of Inhibitive Anions on Formation of Passivating Oxide
Inhibitive anions can also contribute to the repair of weak points, pores or damage to the oxide film on iron by promoting the formation of a passivating film of iron oxide at such areas. This was put forward as a mechanism of action of inhibitive anions by Stern'", who proposed that the formation of passivating iron oxide was easier in the presence of such anions (due to an increase in the rate of the cathodic process, arising from either reduction of an oxidising anion or acceleration of oxygen reduction) so that a greater equivalent anodic current would be available to more easily exceed the critical current density for passivation. Stern also suggested that inhibitive anions might facilitate the anodic process of oxide formation by reducing the magnitude of the critical current density or by making the Flade potential more negative. Subsequent work has shown that inhibitive anions affect mainly the anodic process. Thus in solutions of the oxidising inhibitive anions chromate IO3*Is", nitrite 'sI and pertechnetateIw, reduction of dissolved oxygen is the predominant cathodic process. There is evidence ' O 3 ~la, that some anions can increase somewhat the rate of oxygen reduction, but the effects do not appear sufficiently large to be significant. However, anodic polarisation studies*'51 161-163 have shown that the critical current density for passivation is much smaller in the presence of inhibitive anions than aggressive anions. Comparing a number of inhibitive anions 15', the critical current densities for passivation have been found to increase in generally the same order as the inhibitive efficiencies decrease. In solutions of inhibitive anion^'^'-'^, as the pH becomes more acid the critical current density for passivation generally increases. In benzoate solutionIw, the presence of dissolved oxygen has been shown to reduce considerably the critical current ""9
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :53
density for passivation. However, in carbonate solution 163, dissolved oxygen has little effect. The effects of anions on the passivation reaction are related to their adsorption, since radiotracer measurements during passivation of iron in solutions of sodium phosphate and sodium iodohippurate (a substituted benzoate) have indicated that the greater the adsorption of the anion on the active iron surface, the smaller the critical current density for passivation. Brasher 166,'67 has found that a steel specimen which has begun to corrode in a solution of an aggressive anion can be inhibited by addition of a nonoxidising inhibitive anion only if the potential has not become more negative than a certain value, termed the 'critical potential for inhibition'. This critical potential depends on the nature of the anionI6', and on the relative concentrations of inhibitive and aggressive anions in the solution, becoming more positive as the concentration of aggressive anion increases Im. The critical potential for inhibition has been related to the effects of potential on the adsorption of and is probably the potential at which adsorption of the inhibitive anion on the active corroding areas has the minimum value necessary to reduce the metal ion dissolution rate to such an extent that oxide film formation can occur. In pure benzoate or phosphate solutions, the critical potentials for inhibition (benzoate -0.28 V, phosphate -0.43 V) are close to the critical passivation for iron. This is a further indication that inhibition under these circumstances occurs due to the formation of passivating iron oxide at the corroding areas. Effects of inhibitive Anions on the Dissolution of Passivating Oxide
The passivating oxide layer on iron should remain stable and protective provided its rate of formation exceeds its rate of dissolution. Dissolution of the outer layer of y-Fe,O, can occur in two ways. At potentials more positive than the Flade potential, dissolution occurs by passage of Fe3+ ions from the oxide surface into solution'69*170. The effect of anions on the rate of this process has not been systematically studied, but there is evidence that the rates are considerably smaller in solutions of chromatesN than of sulphate169.However, the rates in sulphate are slightly less than in solutions of phthalateI7', an inhibitive anion, which may be due to some complex formation. The dissolution rates in solutions of these anions decrease considerably as the pH increases. The thickness of the oxide film on iron also controls the Fe3+ dissolution rates, which decrease markedly as the oxide film thickness i n c ~ e a s e s ' ~Thus, ~ . under adverse conditions, i.e. relatively low pH, low inhibitive power of anions, low oxide thickness (especially at weak points in the film), on immersion of an iron specimen, an appreciable Fe3+ dissolution current could flow, which could depress the potential to the vicinity of the Flade potential. In this region, the rate of oxide dissolution increases I7O, due to the onset of reductive dissolution 10'*172, leading to passage of Fe2+ ions into solution. The dissolving Fe2+ ions derive from the reduction of Fe3+ions in the surface layer of y-Fe,O,, by electrons supplied from the oxidation of metallic iron to form cations. Gilroy and Mayne"' have shown that the rate of reductive dissolution of the oxide film
17 :54 THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
on iron is faster in solutions of aggressive anions than in solutions of inhibitive anions. The rate of dissolution increases as the dissolved oxygen content decreases. Gilroy and Mayne have shown further”, that the rates of oxidation in solution of Fe2+to Fe3+by dissolved oxygen are greater in the presence of inhibitive anions than of aggressive anions. They propose that a function of the inhibitive anion is to stimulate the oxidation by oxygen of any Fe2+ions produced in the surface of the y-Fe,O, film, thus retarding its dissolution. These effects of anions on the reductive dissolution of the oxide film should correspond to effects on the Flade potential, Le. a decrease in the rate of reductive dissolution should displace the Flade potential to more negative values, and vice versa. The effects of a number of anions at varying pH on the Flade potential have been described by Freiman and Kolotyrkin The reductive dissolution of the outer y-Fe,O, layer exposes the inner magnetite layer of the oxide film. In acid solutions (pH less than 4) the magnetite layer rapidly dissolves’”, but in near neutral solution it may be stable and protective, depending on the nature of the anion present and its concentration I”. The magnetite layer is stable in inhibitive solutions of anions, e.g. benzoate I”, carbonate 16,, hydroxide 16,, borate126(though not bicarbonate I”). The stability of the magnetite layer controls the inhibition of corrosion of iron when coupled to electronegative metals such as aluminium, zinc or cadmium”’. Thus inhibitive anions can retard the dissolution of both the y-Fe,O, and the magnetite layers of the passivating oxide layer on iron. This has the dual effect of preventing breakdown of an existing oxide film and also of facilitating the formation of a passivating oxide film on an active iron surface, as discussed in the previous section. Inhibitive Anions and Aggressive Anions
An important function of inhibitive anions is to counteract the effects of aggressive anions which tend to accelerate dissolution and breakdown of the oxide films. The relationships ‘Is (mentioned above) between the concentrations of inhibitive anions and aggressive anions, when inhibition is just achieved, correspond to competitive uptake of the anions by adsorption or ion exchange at a fixed number of sites at the oxide surface. The effects of the valencies of the competing anions are generally consistent with the total charge due to anion uptake being constant. Iron surfaces protected in solutions of inhibitive anions rapidly begin to corrode on addition of aggressive anions or on transfer to distilled water. All these facts indicate that inhibitive anions overcome the effects of aggressive anions through participation in a reversible competitive adsorption such that the adsorbed inhibitive anions reduce the surface concentration of aggressive anions below a critical value. The reasons why some anions exhibit strong inhibitive properties while others exhibit strong aggressive properties are not entirely clear. The principal distinction seems to be that inhibitive anions are generally anions of weak acids whereas aggressive anions are anions of strong acids. Due to hydrolysis, solutions of inhibitive anions have rather alkaline pH values and buffer capacities to resist pH displacement to more acid values. As discussed
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS 17 :55
above, both these factors are beneficial to the stability and repair of the oxide film. However, the primary difference between inhibitive and aggressive anions must arise from their effects on dissolution reactions at the oxide surconsiders that the face. For the various X0;- ions, Cartledge'40*143 difference between inhibitive and non-inhibitive anions is due to the contrasting internal polarity of the X + - 0 - bond creating different electrostatic interactions in the electrical double layer, thus affecting transfer of metal ions into solution. Another important factor is that the bonds formed between anions of weak acids and metal ions in the oxide surface are of a more coordinate character than those formed by anions of strong acids. The mechanism of dissolution of metal ions from the oxide surface is not well understood, but according to Heusler17' it proceeds by the passage of Fe(OH)*+ions into solution. It seems likely that dissolution of other anion complexes will occur, and it would appear that dissolution of the more coordinately bonded complexes with inhibitive anions occurs less readily than that of the more ionically bonded complexes with aggressive anions. In addition, the electron transfer to the ferric ion in the coordinate bonds with inhibitive anions will tend to stabilise the ferric state against reduction to the ferrous state, making the oxide more resistant to reductive dissolution. The inhibitive efficiency of anions tends to increase with size in a homologous series9', due probably to the increasing tendency to adsorption, and decreasing solubility of the ferric-anion complex. Zinc
The effects of inhibitive and aggressive anions on the corrosion of zinc are broadly similar to the effects observed with iron. Thus with increasing concentration, anions tend to promote corrosion but may give inhibition above a critical concentration 14'* l6O. '78. Inhibition of zinc corrosion is somewhat '~'*'~* more difficult than that of iron, e.g. nitrite'79*'80and b e n ~ o a t e ~ ~ *are not efficient inhibitors for zinc. However, inhibition of zinc corrosion is observed in the presence of anions such as ~ h r o m a t e " * ' ~ ~borate179 * ~ ' ~ , and nitr~cinnamate~'.''~, which are also good inhibitors for the corrosion of iron. Anions such as sulphate, chloride and nitrate are aggressive towards zinc and prevent protection by inhibitive anions I@'. The presence of dissolved oxygen in the solution is essential for protection by inhibitive anions. As in the case of iron, pressures of oxygen greater than atmospheric or an increase in oxygen supply by rapid stirring can lead to the protection of zinc in distilled water'83. Inhibition of zinc corrosion occurs most readily184in the pH range of 9 to 12, which corresponds approximately to the region of minimum solubility of zinc hydroxide. The ways in which inhibitive anions affect the corrosion of zinc are mainly similar to those described above for iron. In inhibition by chromate, localised uptake of chromium has been shown to occur at low chromate concentrations'w,'85 and in the presence of chloride ionsig5. Thus under conditions unfavourable for inhibition, pore plugging occurs on zinc. Inhibitive anions also promote the passivation of zinc, e.g. passivation is much easier in solutions of the inhibitive anion, than in solutions of the non-inhibitive anions, carbonate and bicarbonate IE9, A critical 14'9
17 :56 THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
inhibition potential, analogous to that on iron, has been observed for zinc in borate solutions Thus inhibitive anions promote repair of the oxide film on zinc by repassivation with zinc oxide. The requirement of dissolved oxygen for inhibition indicates that the passivating oxide is stabilised at potentials more positive than the Flade potential by the reduction in dissolution rate due to the inhibitive anion. The passivating film is ZnO. which dissolves as divalent cationsLgo,and there is no evidence of reductive dissolution. Thus, on zinc the inhibitive anion presumably stabilises the oxide by formation of an adsorbed complex with the zinc ion, the dissolution rate of which is less than that of analogous zinc complexes with water, hydroxyl ions or aggressive anions. Aluminium
When aluminium is immersed in water, the air-formed oxide film of amorphous y-alumina initially thickens (at a faster rate than in air) and then an outer layer of crystalline hydrated alumina forms, which eventually tends to stifle the r e a ~ t i o n ' ~ ' -In ' ~near-neutral ~. air-saturated solutions, the corroby anions which are inhibitive sion of aluminium is generally inhibited Lw*195 for iron, e.g. chromate, benzoate, phosphate, acetate. Inhibition also occurs in solutions containing sulphate or nitrate ions, which are aggressive towards iron. Aggressive anions for aluminium include the halide ions '92*L94-'%, F-, C1-, Br-, I-, which cause pitting attack, and anions which form soluble complexes with aluminium Iw, e.g. citrate and tartrate, which cause general attack. Competitive effect^'^^.'^', similar to those observed on iron, are observed in the action of mixtures of inhibitive anions and chloride ions on aluminium. The inhibition of aluminium corrosion by anions exhibits both an upper and a lower pH limit. The pH range for inhibition depends upon the nature of the anionLw. the oxide film on aluminium In near-neutral and de-aerated solutions is stable and protective in distilled water and chloride solutions, as well as in solutions of inhibitive anions. Thus the inhibition of aluminium corrosion by anions differs from that of iron or zinc in that the presence of dissolved oxygen in the solution is not necessary to stabilise the oxide film, Le. the Flade potential is more negative than the hydrogen evolution potential. Lorking and Mayne Iw* '% observed that inhibition of aluminium corrosion occurred only when the initial rate of dissolution of aluminium oxide in solutions of anions was less than a critical value. If this dissolution rate was decreased by presaturation of the solution with aluminium oxide, the corrosion of aluminium could be inhibited in normally aggressive solutions containing chloride or fluoride ions. The oxide film dissolves as A13+ions, the degree of hydrolysis and rate of dissolution depending on the pH 198-20'. There is no evidence of reductive dissolution. Thus, as with zinc, the inhibitive anions probably act by adsorption on to A1 3 + ions in the oxide surface to form a surface complex, which has a low dissolution rate. The formation of surface compounds by anions on aluminium oxide has been discussed by Vedder and Vermilyea202*203 in connection with the inhibition of hydration of anodic oxide films on aluminium. In corrosion inhibition by chromate ions, their interaction with the oxide film on aluminium has been
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :57
shown by Heine and Pryor2" to result in the formation of an outer layer of the film which is more protective due to its high electronic resistance and low dissolution rate. Chromate ions were also found to prevent the uptake and penetration of chloride ions into the aluminium oxide film2049205.
Copper
Little work has been carried out on the mechanism of inhibition of the corrosion. of copper in neutral solutions by anions. Inhibition occurs in solutions containing chromate9', benzoate9' or nitritezMions. Chloride ion^^^*^^^ and sulphidetm ions act aggressively. There is evidence2@that chloride ions can be taken up into the cuprous oxide film on copper to replace oxide ions and create cuprous ion vacancies which permit easier diffusion of cuprous ions through the film, thus increasing the corrosion rate. Copper corrosion can also be effectively inhibited in neutral solution by organic compounds of low molecular weight, such as benzotriazole2m.2102'2 Benzotriazole is particularly effective in and 2-mer~aptobenzothiazole~~. preventing the tarnishing and dissolution of copper in chloride solutions. In the presence of benzotriazole, the anodic dissolution reaction, the oxide film growth reaction and the dissolved oxygen reduction reaction, are all inhibited2m*2'2, indicating strong adsorption of the inhibitor on the cuprous oxide surface.
Conclusions
The mechanism of action of inhibitive anions on the corrosion of iron, zinc and aluminium in near-neutral solution involves the following important functions: 1. Reduction of the dissolution rate of the passivating oxide film. 2. Repair of the oxide film by promotion of the reformation of oxide. 3. Repair of the oxide film by plugging pores with insoluble compounds. 4. Prevention of the adsorption of aggressive anions. Of these functions, the most important appears to be the stabilisation of the passivating oxide film by decreasing its dissolution rate (Function 1). Inhibitive anions probably form a surface complex with the metal ion of the oxide, i.e. Fe3+,Zn2+,Al'+, such that the dissolution rate of this complex is less than that of the analogous complexes with water, hydroxyl ions or aggressive anions. For iron only, the special mechanism of reductive dissolution enables the ferric oxide film to dissolve more easily as Fe2+ ions. Inhibitive anions may retard this process by catalysing the re-oxidation by dissolved oxygen of any Fe2+formed in the oxide surface. Stabilisation of the oxide films by decrease of dissolution rate is also important with respect to repassivation by oxide formation (Function 2). The plugging of pores by formation of insoluble compounds (Function 3) does not appear to be an essential function, but is valuable in extending the range of conditions under which inhibition can be achieved. The suppression of the adsorption of
17 :58 THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
aggressive anions (Function 4) by participation in a dynamic reversible competitive adsorption equilibrium at the metal surface appears to be related to the general adsorption behaviour of anions rather than a specific property of inhibitive anions. The relative importance of these functions also depends to a considerable extent on the solution conditions. Under favourable conditions of pH, oxidising power and aggressive anion concentration in the solution, Function 1 is probably effective in preventing film breakdown. Under unfavourable conditions for inhibition, localised breakdown will occur at weak points in the oxide film, and Functions 2 and 3 become important in repairing the oxide film.
Recent Developments Recent developments in the mechanisms of corrosion inhibition have been discussed in reviews dealing with acid solution^^'^-^^^ and neutral solutions213,214~216.218’219 Novel and improved experimental techniques2I8, e.g. surface enhanced Raman spectroscopyu0, infrared spectroscopy221,Auger electron spectroscopy2”, X-ray photoelectron spectroscopy222and a.c. impedance analysis223,have been used to study the adsorption, interaction and reaction of inhibitors at metal surfaces.
.
Adsorption of Corrosion Inhibitors onto Metals
The bonding of adsorbed corrosion inhibitors onto metals has been described in terms of the concepts of ‘hard-soft acid and bases’215*2z4 and electrosorption valency225.Work has continued on the correlation of the effects of substituents in related molecules, e.g. aliphatic amines226, thiophene^^^', pyridines226-u8, b e n z o a t e ~ ~ ~anthranilates232, ~,~~’, thioglycolic acids234and b e n z ~ t r i a z o l e s ~on ~ ~inhibitive , efficiencies with electron densities at functional groups. These studies have generally confirmed that, in both acid and neutral solutions, substituents increase the inhibitive efficiences, probably because of stronger adsorption forces arising from increased electron density on the functional group due to a nucleophilic substituent, or the polar character of an electrophilic substituent. Considerable enhancement of adsorption and inhibition can occur with an inhibitor containing more than one functional (particularly if chelation is p o ~ s i b l e ~ ~or~ .because ~ ~ ~ ) , of synergistic interaction of two inhibitor^^^'^^^^. Inhibitive efficiencies have also been correlated with steric factors227*240 and Mechanisms in Acid Solutions
The four mechanisms discussed above, of the action of inhibitors remain essentially unchanged. Further work on acetylenic alcohols has indicated that barrier films can form owing to crosslinking by hydrogen bonding and synergistic interactions243.Theoretical treatments of the electrochemical
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :59
mechanisms of inhibition have shown that the contribution of each mechanism can be evaluated from electrochemical There is now considerable evidence that the corrosion of metals in acids can proceed also by a chemical me~hanism"~.Inhibitors can give different effects on the rates of chemical corrosion and electrochemical c o r r o ~ i o n ~ ~ ~ ~ ~ ~ . Mechanisms in Near-neutral Solutions
The mechanisms of action of inhibitors which form salt films on metals have been reviewed 218. Regarding inhibition of corrosion of iron and steel by anions, further evidence for anion incorporation into oxide films has been obtained using radiotracers248and Auger electron spectroscopy24g.Support for the poreplugging mechanism has been given by autoradiography which have demonstrated that in inhibitive solutions containing carboxylate anions, the extent of localised uptake of the anion decreases with increasing pH and increasing inhibitive efficiency of the anion. The anodic passivation of iron has been shown2'6*219.25@-254 to involve the formation of an oxide of lower valency state, possibly containing the anion, before the formation of the ferric oxide film. The effects of anions on these processes have been discussed in relation to the role of metal ion-anion Inhibitive anions may be divided into two main types2I6. Type I anions, which include particularly carboxylates such as phthalate230*2S5*256 and acetate257,have little or no inhibitive effect in deaerated solutions in retarding active dissolution and facilitating passivation. Dissolved oxygen above a critical concentration produces strong synergistic inhibitive effects, owing probably in part to the more alkaline pH produced at the metal surface by oxygen r e d u c t i ~ n ~ ' ~ * ~ ~ * . Type I1 anions, which include the more effective inhibitors, nitrite259, chromate252,m ~ l y b d a t e substituted ~ ~ ~ ~ ~ ~b ,e n z o a t e ~ ~ phenylanthrani~~,~~', late^^^^, have inhibitive properties in deaerated solution. The role of dissolved oxygen is then to act mainly as a redox system. The effects of inhibitive anions on the dissolution of the passivating oxide films are analogous 26'*262. Recent developments have also been reported in the inhibition of zinc 238, a l u m i n i ~ mand ~ ~ copper ~ * ~ ~220*265. J. G. N. THOMAS REFERENCES I . Fischer, H., Werksfoffe u. Korrosion, 23, 445, 453 (1972) 2. Putilova, J. N., Balezin, S. A. and Barannik, V. P.. Mefalfic Corrosion fnhibifors, Pergamon Press, London (1960) 3. Trabanelli, G. and Carassiti, V., Advances in Corrosion Science and Technotogy, 1, Plenum Press, New York, London, 147 (1970) 4. Lacombe, P., 2nd European Symposium on Corrosion Inhibitors, Ferrara 1965, University of Ferrara, 517 (1966) 5. Gileadi, E., ibid., 543 (1966) 6. Wormwell, F. and Thomas, J. G. N., SurfacePhenonlena of Metals, Society of Chemical Industry, London, Monograph No. 28, 365 (1968)
17 :60 THE MECHANISM
OF CORROSION PREVENTION BY INHIBITORS
7. Thomas, J. G. N., Werkstoffe u. Korrosion, 19, 957 (1968) 8. Conway, B. E. and Barradas, R. G., Transactions of the Symposium on Electrode Processes, John Wiley and Sons, New York, 299 (1961) 9. Cavallaro, L., Felloni, L. and Trabanelli, G., First European Symposium on Corrosion Inhibitors, Ferrara, 1960, University of Ferrara, 111 (1961) 10. Annand, R. R., Hurd, R. M. and Hackerman, N., J. Electrochem. SOC.,112, 138 (1965) 11. Gileadi, E., J. Electroanal. Chem., 11, 137 (1966) 12. Optical Studies of Adsorbed Layers at Interfaces, Symp. Faraday SOC.,4 (1970) 13. Epelboin, I., Keddam, M. and Takenouti, H., J. Applied Electrochem., 2, 71 (1972) 14. Heusler, K. E. and Cartledge, G. H., J. Electrochem. Soc., 108, 732 (1961) 15. Trabanelli, G., Zucchi, F. and Zucchini, G. L., Corrosion, Traitement,Protection, Finition, 16, 335 (1968) 16. Zucchini, G. L., Zucchi, F. and Trabanelli, G., 3rd European Symposium on Corrosion Inhibitors, Ferrara, 1970, University of Ferrara, 577 (1971) 17. Ross, T. K. and Jones, D. H., as Ref. 9, 163 (1961) 18. Fedorov, Y. V., Uzlyuk, M. V. and Zelenin, V. M., Protection of Metals, 6, 287 (1970) 19. Schwabe, K. and Leonhardt, W., Chem. Ing. Tech., 38, 59 (1966) 20. Baldi, L., Carassiti, V., Trabanelli, G., Zucchi, F. and Zucchini, G. L., Proc. 3rd International Congress Metallic Corrosion, Moscow, 1966, MIR, Moscow, 2, 127 (1969) 21. Trabanelli, G., Zucchini, G. L., Zucchi, F. and Carassiti, V., British Corrosion J . , 4, 267 ( 1969) 22. Schwabs, K., Reinhard, G., Fischer, M., Schaarschmidt, K., Werkstoffe u. Korrosion, 22, 302 (1971) 23. Fujii, S. and Aramaki, K., Proc. 3rd International Conference on Metallic Corrosion, Moscow. 1966, MIR. Moscow, 2, 70 (1%9) 24. Hoar, T. P. and Holliday, R. D., J. Applied Chem., 3, 502 (1953) 25. Makrides, A. and Hackerman, N., Ind. Eng. Chem., 47, 1773 (1955) 26. Hoar, T. P. and Khera, R. P., as Ref. 9, 73 (1961) 27. Felloni, L. and Cozzi, A., as Ref. 4, 253 (1966) 28. Donahue, F. M., Akiyama, A. and Nobe, K., J. Electrochem. Soc., 114, 1006 (1967) 29. Frumkin, A. and Damaskin, B. B., Modern Aspects of Electrochemistry, No. 3, Ed. J. O'M. Bockris, Butterworths, London, 149 (1964) 30. Electrosorption, Ed. E. Gileadi, Plenum Press, New York, (1967) 31. Damaskin. B. B., Petrii, 0.A. and Batrakov, V. V., Adsorption of Organic Compounds on Electrodes, Plenum Press, New York (1971) 32. Antropov, L. I., Proc. First International Congress on Metallic Corrosion, London, 1961, Butterworths, London, 147 (1962) 33. Antropov, L. I., Corrosion Science, 7, 607 (1967) 34. Fischer, H. and Seller, W., Corrosion Science, 6, 159 (1966) 35. Makrides, A. C. and Hackerman, N., Ind. Eng. Chem., 46, 523 (1954) 36. Blomgren, E., Bockris, J. O'M. and Jesch, C., J. Phys. Chem., 65, 2 OOO (1961) 37. Szklarska-Smialowska, Z. and Dus, B., Corrosion, 23, 130 (1967) 38. Ayers, R. C. and Hackerman, N., J . Electrochem. Soc., 110, 507 (1963) 39. Cox, P. F., Every, R. L. and Riggs, 0. L., Corrosion, 20, 299t (1964) 40. Donahue, F. M. and Nobe, K., J. Electrochem. Soc., 112, 886 (1965) 41. Grigoryev, V. P. and Osipov, 0. A., Proc. 3rdInternational Congresson Metallic Corrosion, Moscow, 1966, MIR, Moscow, 2, 48 (1969) 42. Altsybeeva, A. I., Levin, S. 2. and Dorokhov, A. P., as Ref. 16, 501 (1971) 43. Donahue, F. M. and Nobe, K., J. Electrochem. Soc., 114, 1012 (1967) 44. Grigoryev, V. P.and Ekilik, V. V., Protection of Metals, 4, 23 (1968) 45. Grigoryev, V. P. and Gorbachev, V. ., Protection of Metals, 6 , 282 (1970) 46. Grigoryev, V. P.and Kuznetzov, V. V., Protection of Metals, 5 , 356 (1%9) 47. Akiyama, A. and Nobe, K., J. Electrochem. Soc., 117, 999 (1970) 48. Vosta, J. and Eliasek, J., Corrosion Science, 11, 223 (1971) 49. Brandt, H., Fischer, M. and Schwabe, K., Corrosion Science, 10, 631 (1970) 50. Hackerman, N., Corrosion, 18, 332t (1962) 51. Aramaki, K. and Hackerman, N., J. Electrochem. Soc., 115, 1007 (1968) 52. Mann, C. A., Lauer, B. E. and Hultin, C. T., Industr. Engng. Chem., 28, 159, 1048 ( 1936) 53. Szlarska-Smialowska, Z. and Wieczorek, G., as Ref. 16, 453 (1971) 54. Iofa, Z. A., as Ref. 4, 93 (1966)
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
17 :61
55. 56. 57. 58.
Hackerman, N., Snavely, E. S. and Payne, J. S., J. Electrochem. Soc., 113, 677 (1966) Iofa, Z. A., Batrakov, V. V. and Cho-Ngok-Ba, Electrochimica Acta, 9, 1645 (1964) Fud Hasan, S. and Iofa, Z. A., Protection of Metals, 6 , 218 (1970) Cavallaro, L., Felloni, L., Trabanelli, G. and Pulidori, F., Electrochimica Acta, 9, 485
59. 60. 61. 62. 63. 64.
Murakawa, T., Naguara, S. and Hackerman, N., Corrosion Science, 7, 79 (1967) Hudson, R. M. and Warning, C. J., Materials Protection, 6 No. 2, 52 (1967) Hudson, R. M. and Warning, C. J., Corrosion Science, 10, 121 (1970) Horner, L. and Rottger, F., Korrosion, 16, Verlag Chemie, WeinheimIBergstr., 57 (1963) Ertel, H. and Horner, L., as Ref. 4, 71 (1966) Lorenz, W. J. and Fischer, H., Proc. 3rd International Congress on Metallic Corrosion, Moscow, 1966, MIR,Moscow, 2, 99 (1969) Horner, L., Werkstoffe u. Korrosion, 23, 466 (1972) Putilova, J. N., as Ref. 4, 139 (1966) Poling, G. W., J. Electrochem. SOC., 114, 1 209 (1967) Kaesche, H. and Hackerman, N., J . Electrochem. Soc., 105, 191 (1958) Kaesche, H., Die Korrosion der Metalle, Springer-Verlag, Berlin, 159 (1966) Kelly, E. J., J . Electrochem. SOC., 115, 1 111 (1968) Okamoto, G., Nagayama, M., Kato, J. and Baba, T., Corrosion Science, 2, 21 (1962) Machu, W., as Ref. 9, 183 (1961) Thibault. S. and Talbot, J., as Ref. 16, 75 (1971) West, J. M., J. Applied Chem., 10, 250 (1960) Kelly, E. J., J. Electrochem. Soc., 112, 124 (1965) Lorenz, W. J., Eichkorn, G., Albert, L. and Fischer, H., Electrochimica Acta, 13, 183
(1964)
65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76.
(1968)
Lorenz, W. J., Corrosion Science, 5, 121 (1965) Ammar, I. A., Darwish, S., Khalil, M. W., Electrochimica Acta, 12, 657 (1967) McCafferty, E. and Zettlemoyer, A. C., J. Phys. Chem., 71, 2 444 (1967) Vaidyanathan, H. and Hackerman, N., Corrosion Science, 11, 737 (1971) Iofa, Z. A., Protection of Metals, 6 , 445 (1970) Duwell, E. J., J. Electrochem. SOC., 109, 1013 (1962) Zucchi, F., Zucchini, G. L. and Trabanelli, G., as Ref. 16, 121 (1971) Grigoryev, V. P. and Ekilik, V. V., Protection of Metals, 4, 517 (1968) Grigoryev, V. P. and Osipov, 0. A., as Ref. 16, 473 (1971) Antropov, L. I. and Savgira, Y.A,, Protection of Metals, 3, 597 (1967) Davolio, G. and Soragni, E., as Ref. 16, 219 (1971) Evans, U. R., The Corrosion and Oxidation of Metals, Arnold, London, 134 (1960) Butler, G., as Ref. 16, 753 (1971) 90. Heyn, E. and Bauer, O., Mitteilungen Koniglichen MaterialprufungsamtBerlin-Dahlem,
77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89.
26, 74 (1908) 91. Hersch, P., Hare, J. B., Robertson, A. and Sutherland, S. M., J. Appl. Chem., 11,251, 265 (1961) 92. Brasher, D. M., British Corrosion Journal, 4, 122 (1%9) 93. Pryor, M. J. and Cohen, M., J. Electrochem. Soc., 100, 203 (1953) 94. Brasher, D. M., Beynon, J. G., Rajagopalan, K. S. and Thomas, J. G. N., British Corrosion Journal, 5, 264 (1970) 95. Mayne, J. E. 0. and Ramshaw, E. H., J. Appl. Chem., 10, 419 (1960) 96. Wachter, A., Industr. Eng. Chem., 37, 749 (1945) 97. Legault, R. A. and Walker, M. S., Corrosion, 20, 282t (1964) 98. Wormwell, F. and Mercer, A. D., J. Appl. Chem., 2, 150 (1952) 99. Davies, D. E. and Slaiman, Q. J. M., Corrosion Science, 11, 671 (1971) 100. Pryor, M. J. and Cohen, M., J. Electrochem. SOC., 98, 263 (1951) 101. Gilroy, D. and Mayne, J. E . O., British Corrosion J., 1, 102 (1965) 102. Slaiman, Q. J. M. and Davies, D. E.,as Ref. 16, 739 (1971) 103. Cartledge, G. H., J. Phys. Chem., 65, 1009 (1961) 104. Cartledge, G. H., J. Electrochem. Soc., 113, 328 (1966) 105. Sympson, R. F. and Cartledge, G. H., J. Phys. Chem., 60, 1 037 (1956) 106. Cartledge, G. H., J. Phys. Chern., 64, 1882 (1960) 107. Bengough, G. D. and Wormwell, F., Chem. and Ind., 549 (1950) 108. Bogatyreva, E. V. and Balezin, S. A., J. Applied Chem., U.S.S.R., 32, 1 094 (1959) 109. Wormwell, F. and Ison, H. C. K., as Ref. 107
17: 62 THE MECHANlSM OF CORROSION PREVENTION BY 1NHlBlTORS 110. 111. 112. 113. 114.
Mayne, J. E. 0. and Page, C. L., British Corrosion J., 5 , 93 (1970) Matsuda, S. and Uhfig, H. H., . I . Electrochem. SOC., 111, 156 (1964) Brasher, D. M. and Mercer, A. D., British Corrosion J., 3, 120 (1968) Mercer, A. D. and Jenkins, I. R., Brifish Corrosion J., 3, 130 (1968) Mercer, A. D., Jenkins, I. R. and Rhoades-Brown, J . E., Brirish Corrosion J . , 3, 136 (1968)
115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130, 131. 132. 133. 134. 135. 136. 137.
Brasher, D. M., Reichenberg, D. and Mercer, A. D., British Corrosion J., 3, 144 (1968) Legault, R. A., Mori, S. and Leckie, H. P., Corrosion, 26, 121 (1970) Thomas, J. G. N., British Corrosion J., 5 , 41 (1970) Mayne, J. E. 0. and Pryor, M. J., J. Chem. Soc., 1 831 (1949) Mayne, J. E. 0..Menter, J. W. and Pryor, M. J., J. Chem. SOC., 3 229 (1950) Mayne, J. E. 0. and Menter, J. W., J. Chem. Soc., 99 (1954) Mayne, J. E. 0. and Menter, J. W., J. Chem. SOC., 103 (1954) Cohen, M., J. Phys. Chem., 56, 451 (1952) Draper, P. H. G., Corrosion Science, 7 , 91 (1967) Foley, C. L., Kruger, J. and Bechtoldt. C. J., J. Electrochem. Soc., 114, 994 (1967) Hancock, P. and Mayne, J. E. O., J. Chem. Soc., 4 172 (1958) Nagayama, M. and Cohen, M., J. Electrochem. SOC., 109, 781 (1962) Kruger, J., J. Electrochem. Soc., 110. 654 (1963) Hancock. P. and Mayne, J. E. 0..J . Chem. Soc., 4 167 (1958) Sewell, P. B., Stockbridge, C. D. and Cohen, M., J. Electrochem. Soc., 108, 933 (1961) Dye, T. G., Fursey, A. and Lloyd, G. 0.. Mem. Scient. Revue Metall., 65, 383 (1968) Brasher, D. M., British Corrosion J.. 1, 183 (1966) Mayne, J. E. 0. and Page, C. L., British Corrosion J . , 7 , 111 (1972) Brasher, D. M. and Kingsbury, A. H., Trans. Faraday SOC., 54, 1214 (1958) Kubaschewski, 0. and Brasher, D. M., Trans. Faraday SOC.,55, 1 200 (1959) Brasher, D. M. and Mercer, A. D., Trans. Faraday SOC., 61, 803 (1965) Brasher, D. M., De, C. P. and Mercer, A. D., British Corrosion J., 1, 188 (1966) Hoar, T. P., Corrosion, A Symposium, Melbourne 1955, University of Melbourne, 124 (1956)
138. 139. 140. 141.
Brasher, D. M., as Ref. 9, 313 (1961) Mayne, J. E. O., as Ref. 9, 273 (1961) Cartledge, G. H., Corrosion. 18, 316t (1962) Brasher, D. M., Beynon, J. G., Mercer, A. D. and Rhoades-Brown, J. E., as Ref. 4, 559 (1966)
142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160.
Gilroy, D. and Mayne, J. E. 0.. as Ref. 4, 585 (1966) Cartledge, G. H., British Corrosion J., 1, 293 (1966) Brasher, D. M., Tribune de Cebedeau, No. 300, 1 (1968) Hoar, T. P. and Evans, U. R., J . Chem. Soc., 2 476 (1932) Cohen, M. and Beck, A. F., Z. Elektrochem., 62, 6% (1958) Cartledge, G. H.and Spahrbier, D. H., J. Electrochem. Soc., 110, 644 (1963) Pryor, M. J., Cohen. M. and Brown, F., 1. Electrochem. Soc., 99, 542 (1952) Brasher. D. M. and Stove, E. R., Chemistry and Industry, 171 (1952) Gatos, H. C., as Ref. 9, 257 (1961) Cartledge, G. H., J. Phys. Chem., 59, 979 (1955) Spitsin, V. I., Rozenfeld, I. L., Persiantseva, V. P., Zamochnikova, N. N. and Kuzina, A. F., Corrosion, 21, 211, (1965) Mayne, J. E. 0.and Page, C. L., British Corrosion J., 7 , 115 (1972) Mellors, G. W., Cohen, M. and Beck, A. F., J. Electrochem. SOC., 105, 332 (1958) Evans, U. R., J. Chem. Soc., 1 020 (1927) Cartledge. G. H., J. Electrochem. Soc., 114, 39 (1967) Stern, M., J . Electrochem. SOC., 105, 635 (1958) Thomas, J. G. N. and Nurse, T. J., British Corrosion J . , 2, 13 (1967) Rozenfeld, I. L., Dokl. Akad. Nauk SSSR., 78, 523 (1951) Gouda, V. K., Khedr, M. G. A. and Shams El Din, A. M., Corrosion Science, 7 , 221 (1 967)
161. 162. 163. 164. 165.
Hancock, P. and Mayne, J. E. O., J . Applied Chem., 9, 345 (1959) Freiman, L. 1. and Kolotyrkin, Y. M., Protection of Metals, 1, 135 (1965) Thomas, J. G. N., Nurse, T. J. and Walker, R., British Corrosion J., 5 , 87 (1970) Slaiman, Q. J. M. and Davies, D. E., Corrosion Science, 11, 683 (1971) Thomas, J. G. N., British Corrosion J . , 1, 156 (1966)
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS 17 :63
166. Brasher, D. M., Nature. Lond., 185, 838 (1960) 167. Brasher, D. M., h o c . 1st International Congress on Metallic Corrosion, 1961, Butterworths, London, 156 (1962) 168. Antropov, L. I. and Kuleshova, N. F., Protection of Metals, 3, 131 (1967) 169. Vetter, K. J., 2. Elektrochem., 59, 67 (1955) 170. Heusler, K. E., Ber. Bunsenges. Phys. Chem., 72, 1 I97 (1%8) 171. Weil, K. G. and Bonhoeffer. K. F., Z. Phys. Chem. N.F., 4, 175 (1955) 172. Pryor. M. J. and Evans, U. R., J. Chem. Soc., 1 259, I 266, I 274 (1950) 173. Gilroy, D. and Mayne. J. E. O., British Corrosion J..l, 107 (1965) 174. Freiman, L. I. and Kolotyrkin, Y. M., Protection of Metals, 5, I13 (1%9) 175. Vetter, K. J. and Klein, G., Z . Phys. Chem. N . F . , 31, 405 (1%2) 176. Heusler, K., Weil, K.G. and Bonhoeffer, K. F., 2. Phys. Chem. N.F., 15, 149 ( 1958) 177. Mercer, A. D. and Thomas, J. G. N., as Ref. 16,777 (1971) 178. Abu Zahra, R. H. and Shams El Din, A. M., Corrosion Science, 5,517(1965) and 6,349 (1966)
179. Wormwell, F., Chemistry and Industry, 556 (1953) 180. Thomas, J. G. N., Mercer, A. D. and Brasher, D. M., Proc. 4th International Congress on Metallic Corrosion, 1969, N.A.C.E., Houston, 585 (1972) 181. Gilbert, P.T. and Hadden, S. E., J . Applied Chem.. 3, 545 (1953) 182. Brasher, D. M. and Mercer, A. D., Proc. 3rd International Congress on Metallic Corrosion. 1966, MIR, Moscow, 2, 21 (1%9) 183. Evans, U. R. and Davies, D. E., J . Chem. SOC.,2607 (1951) 184. Lorking, K. F. and Mayne, J. E. O., Proc. 1st International Congress on Metallic Corrosion, 1961, Butterworths, London, 144 (1962) 185. McLaren, K. G., Green, J. H. and Kingsbury, A. H., Corrosion Science, 1, 161, 170 (1961)
186. El Wakkad, S . E. S., Shams El Din, A. M. and Kotb, H.. J. Electrochem. Soc., 105,47 (1958) 187. Davies, D. E. and Lotlikor, M. M., British Corrosion J., 1, 149 (1966) 188. Lotlikar, M. M. and Davies, D. E., Proc. 3rd International Congress on Metullic Corrosion, 1966, MIR, Moscow, 1, 167 (1969) 189. Kaesche, H., Electrochimica Acta, 9, 383 (1964) 190. Armstrong, R. D. and Bulman, G. M., J. Electroanal. Chem., 25, 121 (1970) 191. Hart, R. K., Trans. Faraday SOC., 53, 1 020 (1957) 192. Pryor, M. J., 2. Elektrochem., 62,782 (1958) 193. Godard, H. P. and Torrible, E. G., Corrosion Science, 10, 135 (1970) 194. Lorking, K. F. and Mayne, J. E. 0.. J. Appl. Chem., 11, 170 (1961) 195. Bohni, H.and Uhlig, H. H., J. Electrochem. Soc., 116, 906 (1%9) 1%. Lorking, K. F. and Mayne, J. E. O., British Corrosion J . , 1, 181 (1966) 197. Anderson, P. J. and Hocking, M. E., J. Appl. Chem., 8, 352 (1958) 198. Kaesche, H.,Werkstoffe u. Korrosion, 14, 557 (1%3) 199. Plumb, R. C., J. Phys. Chem.. 66, 866 (1962) 200. Straumanis, M. E. and Poush, K., J. Electrochem. Soc., 112, 1 I85 (1965) 201. Heusler, K. E. and Allgaier, W., Werkstoffe u. Korrosion, 22, 297 (1971) 202. Vedder, W. and Vermilyea, D. A., Trans. Faraday Soc., 65, 561 (l%9) 203. Vermilyea, D. A. and Vedder, W., Trans. Faraday Soc., 66, 2644 (1970) 204. Heine, M. A. and Pryor, M. J., J. Electrochem. Soc., 114, I O 0 1 (1967) 205. Heine, M. A., Keir, D. S. and Pryor, M. J., J. Electrochem. Sac., 112, 24 (1965) 206. Hoar, T. P., J. Soc. Chem. Ind. (Lond.), 69, 356 (1952) 207. Catty, 0.and Spooner, E. C. R., TheElectrode Potential Behauiour of Corroding Metals in Aqueous Solutions, Clarendon Press, Oxford, 199 (1938) 208. Bonora, P. L., Bolognesi, G. P., Borea, P. A . , Zucchini, G. L. and Brunoro, G., as Ref. 16,685 (1971) 209. North, R. F. and Pryor, M. J., Corrosion Science, 10, 297 (1970) 210. Cotton, J. B. and Scholes, I. R., British Corrosion J., 2, 1 (1%7) 211. Poling, G. W., Corrosion Science, 10, 359 (1970) 212. Mansfeld, F., Smith, T. and Parry, E. P., Corrosion, 27, 289 (1971) 213. Rozenfeld, I. L., Corrosion Inhibitors, McGraw-Hill, New York (1981) 214. Nathan, C.C.,(ed.) Corrosion Inhibitors, NACE, Houston (1973) 215. Homer, L., Chem. Zeilung, 100, 247 (1976)
17:64
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS
216. Thomas, J. G. N., 5th European Symposium on Corrosion Inhibitors, p. 453, Ferrara 1980, University of Ferrara (1980) 217. Antropov, L. I., Soviet Materials Science, 19, 79 (1983) 218. Hollander, O., Geiger, G.E., Ehrhardt, W.C. Corrosion/82, Houston 1982, NACE, Paper No 226, (1982) 219. Szklarska-Smialowska, Z., Passivity of Metals, p. 443, The Electrochemical Society, Princeton (1978) 220. Thierry, D., and Leygraf, C., J. Electrochem. Soc., 132, 1009 (1985) 221. Bockris, J. 0. M.,Habib, M.A. and Carbajal, J. L., J. Electrochem. SOC., 131, 3032 (1984); 132, 108 (1985) 222. Augustynski, J., Balsanc, L., Mod. Aspects Electrochem. 13, 251 (1979) 223. Mansfeld, F., Kendig, M.W. and Lorenz, W. J., J. Electrochem. SOC., 132, 290 (1985) 224. Aramaki, K., as Ref. 216, 267 (1980); J . Electrochem. SOC., 134, 1896 (1987) 225. Koppitz, F. D., Schultze, J. W. and Rolle, D., J. Electroanal. Chem., 170, 5 (1984) 226. Antropov, L. I., Ledovskikh, V. M. and Kuleshova, N. F., Protection of Metals, 8, 42 (1972); 9, 151 (1973) 227. Altsybeeva, A. J., Dorokhov, A. P. and Levin, S. Z., Protection of Metals, 10, 626 (1974) 228. Zhovnirchuk, V. M.,Skrypnik, Yu. G., Babei, Yu. J., Baranov, S. N. and Mindyuk, A. K., Protection of Metals, 18, 502 (1982) 229. Vosta, J., Eliasek, J. and Knisek, P., Corrosion, 32, 183 (1976) 230. Rozenfeld, I. L., Kuznetsov, Yu. I., Kerbeleva, I. Ya., Brusnikina, V. M., Bochorov, B. V., Lyashenko, A. A., Protection of Metals, 14, 495 (1978) 231. Kuznetsov, Yu. I., Kerbeleva, I. Ya, Brusnikina, V. M. and Rosenfeld, I. L., Soviet Electrochemishy, 15, 1460 (1979) 232. Kuznetsov, Yu. I., Rozenfeld, I. L., Kuznetsova, J . G. and Brusnikina, Soviet Electrochemistry, 18, 1422 (1982) 233. Szklarska-Smialowska, Z. and Kaminski, M., Corros. Sci., 13, 1 (1973) 234. Carroll, M.J. B., Travis, K. E. and Noggle, J. H., Corrosion, 31, 123 (1975) 235. Eldakar, N. and Nobe, K., Corrosion, 36, 271 (1981) 236. Ledovskikh, V. M.,Protection of Metals, 19, 245 (1983) 237. Zecher, D. C., Mater. Perf., 15 No. 4, 33 (1976) 238. Leroy, R. L., Corrosion, 34, 98, 113 (1978) 239. Ledovskikh, V. M.,Protection of Metals, 20, 45, 502 (1984) 240. Lawson, M.B., Corrosion/81, Toronto 1981, NACE, Paper No. 254 (1981) 241. Fokin, A. V., Pospelov, M.V., Levichev, A. V., Protection of Metals, 17,415 (1981); 19, 242 (1983) 242. Dupin, P., De Savignac, A., Lattes, A., Sutter, B. and Haicour, P., Mater. Chem., 7,549 ( 1982) 243. Tedeschi, R. J., Corrosion, 31, 130 (1975); Jovancicevic, V., Yang, B., and Bockris, J. O'M., Electrochim. Acta, 32, 1557 (1987) 244. Antropov, L. I., Protection of Metals, 13, 323 (1977) 245. Kolotyrkin, Ya. M. and Florianovich, G. M., Protection of Metals, 20, I 1 (1984) 246. Ekilik, V. V. and Grigoriev, V. P., Protection of Metals, 10, 303 (1974) 247. Antropov, L. I., Tarasevich, M. R. and Marinich, M.A.. Protection of Metals, 14, 588 (1978) 248. Mayne, J. E. 0. and Page, C. L., British Corrosion J., 9, 225 (1974); 10, 99 (1975) 249. Lumsden, J. B. and Szklarska-Smialowska, Z., Corrosion, 34, 169 (1978) 250. Bech Nielsen, G., Electrochim. Acta, 21, 627 (1976); 23, 425 (1978) 251. Geana, D., El Miligy, A. A. and Lorenz, w. J., Electrochim. Acta, 20, 273 (1975) 252. Ogura, K. and Majima, T., Electrochim. Acta, 23, 1361 (1978); 24, 325 (1979) 253. Fischer, M., Werkstoffe u. Korrosion, 29, 188 (1978) 254. Davies, D. H. and Burstein, G. T., Corrosion, 36, 416 (1980) 255. Reinhard, G. and Irmscher, C., Werkstoffe u. Korrosion, 28, 20 (1977) 256. Fischer, M., Z. Phys. Chem. (Leipzig). 260, 93 (1979) 257. Podobaev, N. I. and Lubenskii, A. P., Russ, J. Appl. Chem., 46, 2806 (1973) 258. Forker, W., Reinhard, G. and Rahner, D., Corrosion Science, 19, 11 (1979) 259. Kuznetsov, Yu. I., Rozenfeld. I. L., Kerbeleva, I. Ya., Balashova, N. N. and Solomko, N. A., Protection of Metals, 14, 282 (1978) 260. Stranick, M. A., Corrosion, 40, 296 (1984)
THE MECHANISM OF CORROSION PREVENTION BY INHIBITORS 261. 262. 263. 264. 265.
Thomas, J . G. N. and Davies, J . D., British Corrosion J . , 12, 108 (1977) Ogura. K. and Ohama, T., Corrosion, 40,47 (1984) Hunkeler, F. and Bohni, H., WerkstofJe u. Korrosion. 34, 68 (1983) Kuznetsov, Yu. I . , Protection of Metals, 20, 282 (1984) Ohsawa, M. and Sueteka, W . , Corros. Sci., 19, 709 (1979)
17 :65
17.4 Boiler and Feed-water Treatment
Introduction The explicit aims of boiler and feed-water treatment are to minimise corrosion, deposit formation, and carryover of boiler water solutes in steam. Corrosion control is sought primarily by adjustment of the pH and dissolved oxygen concentrations. Thus, the cathodic half-cell reactions of the two common corrosion processes are hindered. The pH is brought to a compromise value, usually just above 9 (at 25"C), so that the tendency for metal dissolution is at a practical minimum for both steel and copper alloys. Similarly, by the removal of dissolved oxygen, by a combination of mechanical and chemical means, the scope for the reduction of oxygen to hydroxyl is severely constrained. Deposit control is important because porous deposits, under the influence of heat flux, can induce the development of high concentrations of boiler water solutes far above their normally beneficial bulk values with correspondingly increased corrosion rates. This becomes an increasingly important feature with increase in boiler saturation temperature. In addition, deposits can cause overheating owing to loss of heat transfer. Finally, carryover of boiler water solutes, which can be either mechanical or chemical, can lead to consequential corrosion in the circuit, either on-load or off-load. Material so transported can result in corrosion reactions far from its point of origin, with costly penalties. It is therefore preferably dealt with by a policy of prevention rather than cure. All of these factors need to be taken into account when defining a tolerable boiler and feedwater regime for any given plant. This has been done, for example, in BS 2486 (BSI, 1968) and by various other bodies (see Section 5.5). With increasing boiler operating pressure, considerations of purity become even more important. High-pressure utility boilers, whilst in the main operating at a few percent make-up, are extremely sensitive to contaminant ingress, so that high-purity feed-water is essential. Above about 40 bar, complete demineralisation is therefore virtually mandatory, with the result that scaling due to hardness salts is impossible, barring inadvertent cooling water ingress due to, for example, condenser leakage, rotary air pump 17 :66
17 :67
BOILER AND FEED-WATER TREATMENT
suckback, or water treatment plant malfunction. However, carryover of solutes in steam becomes increasingly important. This can occur either mechanically (in its worst form as priming) or chemically. In the latter case, substances become distributed between steam and water in a ratio determined by the temperature. Material carried over and subsequently deposited may cause a number of problems elsewhere in the steam/water circuit. Assuming however, that appropriate water treatment has been arranged, the treated water, although unlikely to form scale, and with a greatly diminished tendency to carryover, nevertheless may still cause corrosion unless the quality is further adjusted. This is done by additions and/or removals of substances.
Fundamental Considerations Water Quality
The raw water available for any given installation will be of acceptable quality only rarely, and some degree of purification and adjustment will usually be needed. Following widely accepted usage we shall refer to the purification, i.e. the removal of undesirable constituents, as 'treatment' and the adjustment of quality or suitability, by, for example, alkalisation or deoxygenation, as 'conditioning'. Whilst a detailed consideration of methods of water treatment would be out of place here, nevertheless some brief discussion will be given as an aid to understanding the scope for corrosion control. Raw waters can be generally divided into four categories as follows: 1. Well water. This is rain water which has percolated through various
strata until it enters an underground aquifer. Well water usually contains dissolved calcium and magnesium salts, but is low in organic matter owing to natural filtration. 2. Upland surjace water. This is low in hardness salts having run over impervious rocks but will often be high in organic matter, Le. fulvic and humic acids. 3. Clean rivers. These originate from upland surface waters, but contain more organic matter and also silt. 4. Industrial rivers. These are essentially re-used waters and contain, in addition to those constituents originally present, sewage and industrial wastes. A typical analysis of each of these water categories in presented in Table 17.2. Table 17.2 Typical analysis of water categories Constituent
Water category Well
Alkalinity (mg kg-I CaCO,) Calcium (mg kg-' CaCO,) Magnesium (mg kg-' CaCO, Sodium (mg kg-I CaCO, Chloride (mg kg-' CaCO,) Sulphate (mg kg-' CaC03 Total organic carbon (mg kg "C)
200 90 120 20
10 20 0.2
Upland surface
Clear river
Industrial river
10
100
12 5 8 10 5 5
100 50 50 50 50 8
150 250 100 160
130 230 15
17 :68
BOILER A N D FEED-WATER TREATMENT
Treatment of waters for boiler feed For most boilers, then, raw water from any of the above sources will require treatment followed by appropriate conditioning. In simple, low pressure plants, precipitation of the hardness salts, e.g. by lime or lime/soda, will be adequate. For a wide range of low and intermediate pressure plant, base-exchange is applied. In this process, the potential scale-forming salts of calcium and magnesium are replaced by the equivalent sodium salts. This is achieved by passing the raw water through a bed-originally of a naturally occurring zeolite but now of a synthetic substitute-in which the exchange of unwanted for acceptable ions occurs. Periodic regeneration is necessary by passing brine through the bed. Whilst softened, i.e. rendered non-scaling, by this process, the water is corrosive owing to the presence of carbon dioxide, and in addition the solids content is undiminished. The corrosivity may be alleviated by conditioning, but the solids content becomes a constraining factor with increasing boiler pressure. Some means of lowering the solids content is therefore needed in many instances. Distillation was formerly extensively practised and is still viable in many situations. Increasingly, however, ion exchange is used. Ion exchange is analogous to base exchange and can be used for the removal of either cations or anions. Cations are replaced by hydrogen ions and anions by hydroxyl ions. Again, regeneration is required, this being most conveniently achieved in the UK with sulphuric acid for cation resins and sodium hydroxide for anion resins. For most modern high-pressure boiler plants ion exchange forms the basis of the water treatment plant. As many natural waters are coloured and turbid, and contain suspended solids (silt, clay etc) as well as organic matter, some pretreatment is necessary before ion-exchange as these impurities adversely affect resin performance and lead to lower water quality and higher operating costs. Pretreatment usually comprises coagulation/flocculation followed by settlement and filtration. Coagulation is often achieved by adding aluminium sulphate and, after subsequent flocculation etc, the suspended solids content of the water leaving the pretreatment plant should be less than 2 mg/kg. The water is then usually filtered through deep sand to provide an effluent of 60°C Boiling point
> 10%
> 90°C
Complete Partial (care necessary) Satisfactory Partial Complete Partial Complete Complete Complete Complete Complete Complete Complete Complete Complete Complete Partial Partial Complete Complete Complete Complete Complete Complete Complete Complete Complete Complete Complete
6 125°C 6 5ooc All All All 570
All All 100% 100% 100% 100% 95 q o 100% loo% 100%
100% 100% 100%
100% 100% 100%
Boiling point 10°C Boiling point Ambient Ambient Boiling point 170°C Boiling point Boiling point Boiling point Boiling point Boiling point Boiling point Boiling point 170°C Boiling point Boiling point Boiling point
18:B
CARBON
It is usual to protect carbon from oxidation at high temperature by the use of alternative gas atmospheres -these are generally hydrogen, nitrogen, argon or helium. The first two will react at temperatures above 1700°C to form methane and cyanogen, respectively. Carbon will react directly at high temperatures with many elements such as sulphur and iron. It also forms intercalation compounds in which a wide range of molecules enter the interlayer spacing of the graphite. This can lead to disruption of the material but also produces a whole new class of potentially useful materials. Carbon is inactive in blood and is not rejected from the human body. It is therefore increasingly used in artificial limbs, tendons and heart valves.
-
B. T. KELLY REFERENCES 1. Kelly. B. T., Physics of Graphite. Applied Science (1981) 2. McKee, D. W., Chemistry and Physics of Carbon, 16, 1 (1980)
BIBLIOGRAPHY Mantell, C. L. Carbon and Graphite Handbook. John Wiley Interscience, New York (1968) Ishikawa, T. and Nagaoki, T., Recent Carbon Technoiogy. JEC Press Inc. (1983) Thrower, P. A., (ed.), Chemistry and Physics of Carbon, Vols. 1-22 to date. Marcel Dekker, New York Carbon (An International Journal). Pergamon Press, Oxford England Kelly, B. T. Physics of Graphite. Applied Science (1981)
18.2 Glass and Glass-ceramics
General One of the most important properties of commercial glasses is their great resistance to corrosion; any chemical laboratory apparatus, any window or windscreen provides an excellent illustration. Windows remain virtually unchanged for centuries, resisting the influences of atmosphere and radiation. A vast range of products may be safely stored in glass for decades at ordinary temperatures, and the fact that glass can be used with alkaline, neutral and acid environments allows the same equipment to be used for a variety of processes. Glass is one of the engineer’s most useful and versatile materials. There are many types of glass to choose from to provide a wide range of physical, mechanical, electrical and optical properties for practically every type of environmental condition. The transparency of glass facilitates inspection of process operations and minimises the risk of failure due to unsuspected corrosion, while the hardness and smoothness contribute to easy cleaning. In recent years the development of glass-ceramics has further extended the range of glassy engineering materials. Glass-ceramics combine the formability of glasses with many of the advantageous properties of ceramics. They are finding increasing application by virtue of their strength and high chemical durability at elevated temperatures. The principal difficulty associated with the use of glass equipment is the fact that glass will break rather than deform on severe impact; thermal toughening or ion-exchange techniques may be used to mitigate this disadvantage in some applications. Glass is also more prone than metals to damage by thermal shock, although this difficulty can be largely avoided by the use of low-expansion glass formulations. Finally, the size of glassware which can readily be fabricated is sometimes below the needs of a particular process.
Commercial Glasses Glass Compositions
The term glass defines a family of materials that exhibit as wide a range of differences among themselves as exists among metals and alloys. The great 18:9
18: 10
GLASS AND GLASS-CERAMICS
variety of physical and chemical properties available arises from the possibility of including almost all the stable oxides, sulphides, halides, etc. throughout the periodic table in different glass formulations'. Many branches of the family, e.g. those borates, silicates and phosphates which are water soluble, are of little interest in the present context. It is, however, worth bearing in mind that glasses can be designed to combine particular physical properties with good chemical resistance. For most of the commercial glass families, silica sand is the main ingredient. However, greater melting economy and flexibility of properties are achieved with the addition of other oxides and modifiers. Depending upon the choice of these additional constituents, glasses are classified into groups, including fused silica, soda-lime, lead, aluminosilicate, borosilicate, etc. A cross-section of commercial glass compositions is given in Table 18.3, the glasses listed being: 1. Fused silica 2. Window glass
3. Container glass Fluorescent tubing Neutral glass Hard borosilicate Lead tubing 8. TV tube and screen 9. Textile glass fibre 10. Glass wool insulation 11. Superfine glass wool 4. 5. 6. 7.
Table 18.3 Typical glass composition, V'o
SOz
O3 B2 O3 MgO CaO BaO Na, 0 K2O Zr02 TiO, PbO Li20 Others
1
2
3
4
5
6
7
8
9
10
I1
100
72-7 1.1
72.8 1-7
71-4 2.2
71-5 5.5 10.0
80.3 2.8 12.3
57.2 1-0
67.5 4.8
4.6
14.5
0.8 15.0 1.7
4.0 8.5
0.1 12.0 7.2 6.9
63,6 2.9 5.0 3.2 7.3 2.6 14.5 0.6
57-1 4.6 11.8
10.5
54-2 14.3 8.3 4.5 17.7
3.9
3.8 8.4 13.1 0.5
0.2 3.0
8.0 1.2
4.0 0.4
0.6 0.1
0.4 14.2
0.8 3.8 7.5
29.0 0.4
0.5
0.4
0.6
0.2
0.3
0.5 0.3
0.3
0.3
0.4
Now. The designations 1-1 I represent the various types of glasses listed numerically in the text above this table, and they wd1
be referred to in this way in latcr tables in this section.
Fused silica is a general classification within which is a range of varieties and types with differences in purity, transmission and grade. This glass may be used up to 900°C in continuous service; it resists attack by a great many chemical reagents, rapid attack occurring only in hydrofluoric acid and concentrated alkali solutions.
GLASS AND GLASS-CERAMICS
18:11
The container glass is suitable for the storage of beverages, medicines, cosmetics, household products and a wide range of laboratory reagents. The tubing glass is suitable for general laboratory use and chemical apparatus construction, though neutral or hard borosilicate are preferred for more severe conditions, these representing the most resistant glasses available in bulk form. The neutral glasses are generally less resistant than the hard borosilicate type, but are more easily melted and shaped. They are formulated so that the pH of aqueous solutions is unaffected by contact with the glass, making it particularly suitable in pharmaceutical use for the storage of pH-sensitive drugs. Borosilicates, in terms of different types available, are the most versatile glasses produced. In general, the borosilicates are grouped into six types, viz. low expansion, low electrical loss, sealing, ultraviolet transmitting, laboratory apparatus and optical grade glasses. The example given, Le. hard borosilicate glasses, is used for ovenware, pipelines, sight glasses and laboratory ware, and combines low expansion and high chemical resistivity with chemical stability. They generally require high founding and fabrication temperatures compared with soft soda glasses. Glasses for electrical and electronic components are represented by the lead tubing and cathode-ray-tube screen and cone glasses. These glasses d o not operate under severe corrosion conditions, but surfaces must not leach excessive alkali under damp conditions or electrical breakdown can occur. The glass compositions are formulated to give the maximum electrical resistivity and moisture resistance compatible with other necessary properties. Glass fibres present particular problems in corrosive environments due to their very high surface/volume ratios. Glasses for electrical insulation are formulated from alkali-free aluminoborosilicate glasses (generally known as E-glass) and are frequently specified as containing less than 1% alkali (Na,O and K,O). This type of glass is also used extensively for the reinforcement of plastics where its high resistivity to moisture attack ensures a durable product. The glass wools are used for less demanding applications and generally contain some alkali. Superfine wool contains zirconia and titania to enhance the chemical resistance while retaining the properties necessary for economic fine-fibre formation. Physical Properties
Glass has been defined as ‘an inorganic product of fusion which has cooled to a rigid condition without crystallising’. The atomic structure of glasses is more closely related to liquids than to crystals. The properties of glasses are manifestations of this structure, being governed in particular by the random liquid-like disposition of the network-forming ions (commonly Si4+ and B3+), the presence of mobile, interstitial alkali ions and the ‘single-molecule’ nature of the lattice. The bonding within the atomic network is partly covalent, partly ionic; the network bonds are highly directional with a range of inter-bond angles, lengths and bond energies; the bonding electrons are restricted to particular energy levels within the bonds.
Table 18.4 Physical property data for some commercial glasses' c Units
Density
g/cm
1
2
3
2.20
2.49
2.46
5
6
2.42
2.24
4 2.49
7 3.03
8 2.62
9 2.58
10
11
2.57
2.54
PP c.
~~
Strain point
"C
987
520
490
495
518
515
Annealing point
"C
I 082
545
540
524
565
565
437
Littleton softening point
"C
I 594
735
720
705
631
Resistivity
Qm
Dielectric constant
at 1 kHz
l o i 5 1.1 x loi3 (20T) (2OOC)
-
-
780
820
9 - 6 x ioi4 (2OOC)
-
ioi5 3.2 x (20°C) (150T)
7.8
-
448
616
-
470
657
-
670
843
688
710
1.6 x io6 6.0 x ioi3 2 x 10-3 4.8 x 1 0 - ~ (350T) (ISS'C) (953°C) (862OC)
R
*z
V)
0
3.8
7.4
-
5.1
7.0
-
6-4
-
-
(50 Hz)
at 1 MHz
Tan 6
very small
Refractive index
0.03
-
0.008
-
0.02
0.0011
1.52
-
1.51
1.49
1.47
1.56
1.05
1.02
1.04
1.04
1.13
0.84
W m-2K-'
1.38
Thermal expansion
x 1O'K-I
5.4
Specific heat
J kg-I
775
O.ooo9
19.3
87
x 10-10Nm-2
modulus
-~~ 'Compositions are given in Table 18.3 ~
7.3
987 (200°C)
821
7.4
-
85.5 833
50
33
819
794
84
-
1.51
1.55
-
-
1.01
0.97
-
-
85.5
49
733 (23OC)
1%
(23°C)
~
Young's
-
(50 Hz)
(50 Hz)
1.48
Thermal conductivity
-
-
-
-
6.3
5.15
7.4
7.2
i?
V)
83
-
-
r
2w
-z
G:
GLASS AND GLASS-CERAMICS
la: 13
The network-modifying ions (commonly alkali and alkaline-earth ions) are ionically bound to the network although the field strength and diameter of the alkali ions allow them some mobility. The structural features are reflected in the characteristic properties of inorganic glasses and bring about a broad overall similarity in behaviour as summarised below. Values for the physical properties of the commercial glasses listed in Table 18.3 are given in Table 18.4.
Viscosity The random nature of the glass structure imparts a range of bond energies in the network, hence a characteristic feature of glasses is a continuous softening over a range of temperature, a continuous viscosity/ temperature curve and the absence of a true melting point. For convenience in comparing the viscosity behaviour of different glasses, arbitrary temperatures at which the glass has specific viscosities are often quoted2. Softening temperature The Littleton Softening Point3 is most commonly N s/m2. used. At this temperature the glass hs a viscosity of Transformation temperature, T, Tg corresponds to a viscosity from 10l2 to 1013 Ns/m2 depending on the definition and on the method of measurement. Annealing point and strain point An important range in practice is that Ns/m2 known as the annealing range. The annealing from 10" to point and the strain point are the temperatures at which the glass has a viscosity of Ns/m2 and 10'3'6Ns/m2 respectively. Within this range the glass is effectively a solid, but internal stresses can be relieved within a practical time scale. Rapid cooling of glass articles through the annealing range eventually results in permanent and sometimes catastrophic thermal stresses. However, it is possible to cool the glass relatively quickly from the lower end of the annealing range. Glass behaves as a Newtonian liquid at temperatures well above the glass transition. It is this behaviour which prevents the necking observed during gross deformation of metals and which allows glass to be formed into such a large number of useful configurations. Thermal shock resistance The ability of a glass article to withstand sudden changes of temperature depends primarily on its thermal expansion coefficient, its thickness and its design. For articles of identical shape a low expansion glass (such as a commercial borosilicate) will withstand appreciably greater temperature shocks than will glass of a higher expansion. Thermal shock-resistance testing is usually carried out by transferring the articles from a hot environment to a cold vessel containing water at a predetermined temperature4. In general, transitions from a hot to a cold environment are more likely to produce failure than those in the opposite direction since they tend to induce tensile stresses at the surface. Stress birefringence The presence of stress in glass articles may be monitored readily since stressed glass is birefringent. Standard methods exist for these measurements. Thermal expansion Glasses having coefficientsof linear thermal expansion
18: 14
GLASS AND GLASS-CERAMICS
of from 0 * 5 ~ 1 O - ~ / d eCg to over 10 x 10-6/degC are available. High expansion glass compositions do not generally have long-term chemical durability however. Glass-ceramics are remarkable for the very wide range of thermal expansion coefficients which can be observed. At one extreme, materials having negative coefficients are available while for other compositions very high positive coefficients can be obtained. Between these two extremes there exist glass-ceramics having thermal expansion coefficients practically equal to zero and others whose expansion coefficients are similar to those of ordinary glasses or ceramics or to those of certain metals and alloys. This range of expansion coefficients is allied with good chemical durability. Mechanical Properties
Characteristically, glasses are brittle solids which in practice break only under tension. The ionic and directional nature of the bonds and the identification of electrons with particular pairs of atoms preclude bond exchange. This, coupled with the random nature of the atomic lattice, i.e. the absence of close-packed planes, makes gross slip or plastic flow impossible.
Strength If flaws and stress concentrators, which emphasise the brittle nature of glass, can be avoided, then a glass article behaves as a single molecule in which the strength is governed by the very high interatomic bond strength. Glasses are therefore inherently very strong materials, theoretically capable of exhibiting a tensile strength of about 7 GNm-2. In practice however, surface flaws act as stress concentrators under tensile loading and commercial glasses in bulk form show a mean strength in tension of only about 40 MNm-’. The statistical variation of strength about this figure makes it desirable to allow a substantial safety margin and to design using a figure of about 7 MNm-’”). The strength of glass can be increased to about 200 MNm-2 by commercial toughening processes. The use of such glasses is not possible, however, at elevated temperatures since detoughening will occur. Commercial glass fibres display a strength of about 2GNm-’; this high figure is dependent upon surface protection, given usually by organic coatings. Removal of the coating will result in a marked decrease in strength. Elustic modulus Up to the fracture stress, glass behaves, for most practical purposes, as an elastic solid at ordinary temperatures. Most silicate-based commercial glasses display an elastic modulus of about 70 GNm-2, i.e. about 1/3 the value for steel. If stress is applied at temperatures near the annealing range, then delayed elastic effects will be observed and viscous flow may lead to permanent deformation. The brittle nature of glasses at normal temperatures makes them inappropriate for use in locations where severe impacts are likely to be encountered. In the design of pipelines or other equipment it is possible to use normal engineering assembly techniques provided that suitable gaskets or cushioning are provided at joints and supports and that care is taken in tightening bolts to avoid unequal or localised stresses.
18: 15
GLASS AND GLASS-CERAMICS
Chemical Properties
Technical glasses are now used so extensively and in such widely varying circumstances that it is necessary to be as accurate as possible in describing their chemical properties. The deterioration of individual glasses is dependent on composition, founding process and use and, unless the degradation processes are accelerated, may only be observable after very long periods of time. A typical figure for the corrosion rate of an ordinary soda-limesilica glass would be below 0.008 mm/y. The effect is accelerated when the exposure takes place at higher temperatures, e.g. in boiling water or in an autoclave. Table 18.5 compares the corrosion resistance of some commercial glasses. Table 18.5
The corrosion resistance of some commercial glasses ~
Glass
1
2
3
6
7
Water Acid Weathering
I 1 1
2 2 3
2 2 3
1-2 1-2 1-2
2-3 2-4 2-3
Key
I . Will virtually never show effects. May occasionally show effects. Will probably show erects. See Table 18.3 for compositions.
2. 3. Notes I . 2.
Table after Hauck. J . E., Murs. Engng.. 85, Aug. (1967).
One of the most commonly used measures of durability, i.e. the loss of sodium from the glass, is important to the pharmaceutical and chemical industries, but other changes such as loss of surface quality, are of equal importance for optical and window glasses. The properties of a wide range of technical glasses are well c a t a l o g ~ e d ~but - ~ , the data are often inadequate when considering a particular application and where possible nonstandard ‘whole article’ tests are advisable. In selecting a glass for chemical durability or weatherability regard must be paid to the temperature and concentration of the corrosive agent, length of exposure, the ratio of reagent volume to surface exposed and to the mechanical operating conditions. Guidelines on the durability of many commercial glasses in some attacking media are available from standard durability tests. Glass durability tests There are two types of durability tests for glassware, viz. ‘grain’ or ‘powder tests and ‘whole articles’ tests.
Grain tests In these tests, samples of glass, crushed and graded to a specified sieve size, are exposed under standard conditions of time and temperature to the attacking medium. The temperatures commonly used are 98OC (water bath) and 121°C (autoclave) and the attacking media are water, acid and alkali. The amounts of a particular glass constituent (usually soda or total alkali) removed from a standard weight of grains in a given time are determined. Standard grain tests have been established by various national standards bodies and by some pharmaceutical authorities. The most important of these
18: 16
GLASS AND GLASS-CERAMICS
standard tests are the American* and the German9. Several other continental standards are essentially based on the German. The present British standard" relates only to laboratory glassware. The German and American standards differ in a number of details and to try to establish an international uniformity the IS0 have issued recommended procedures". A new British standard in preparation will be based on these procedures. Careful comparison of results from different laboratories using a particular grain test has shown considerable divergence, and it appears that to obtain consistent results very close adherence to the details of the standard procedures regarding grain preparation, extracting media and analysis is necessary. Nevertheless, grain tests are extensively used, especially in USA and Germany where a further step has been taken of classifying glasses by their 'hydrolytic resistance,' a glass being placed in one of four classes according to the titre of the aqueous extract. The following examples describe the extraction of alkali by water from 2 g samples at 97*3OC'*for three glasses of different hydrolytic class. ~~~~~~~~~~~
Glass Class I Class I11 Class IV
Time (h) Titre (ml of 0 . 0 1 HCI) ~ Time (h) Titre (ml of 0 . 0 1 ~ HCI) Time (h) HCI) Titre (ml of 0 . 0 1 ~
1 0-12 0.5 0.65
2 0.20 1 0.96 1 3.0
3 0.29 2 1.39 2 4.4
4 0.40 4 2.0 4 6.3
6 0.54
Whole article tests Grain tests are open to the criticism that they do not necessarily reflect the behaviour of the finished product in service, hence various tests on complete glass articles have been developed. These are normally carried out under accelerated conditions, and on completion various relevant factors are determined, such as loss in weight, alkali or other constituents extracted, the weight of soluble and insoluble materials in the extract and an assessment of surface condition. The advent of the electron microscope as a standard tool has made the latter study much more objective. Whole article tests are particularly useful in the evaluation of window and optical glasses. Various tests have been proposed for window glass, but no standards exist. The usual procedure is to subject the glass to an accelerated humiditykemperature weathering cycle and to assess the surface conditions after a given period of treatment. The degree of haze formation has been suggested as a method of measuring surface damage, but generally visual comparison with a standard is used. Figure 18.1 illustrates the application of such a test to various optical gIasses. Many optical glasses are much less resistant to attack than are container and window glasses, and less severe tests are necessary. A commonly used method is to immerse specimens in either dilute nitric acid or standard acetate solution of pH 4 - 6 for specified periods at room temperature, then to examine the surfaces either visually or by interferometry. Glass fibres present a particular problem. The water resistance of the base glass can of course the measured by a grain test, but this is unlikely to be representative of the performance of the final product. Generally, purely empirical methods are used to test the glass fibres in situ in a composite
GLASS
18: 17
A N D GLASS-CERAMICS
-__--
__________ e
om----------------e----*
11E
0
LO
80
: 120
--e 1E
0 I.€
-
lmVE 2E 5€-
Days exposed-
-31OEE
Crown Crown --- 1E Dense barium crown ___---L E Borosilicate crown 11E Crown flint
Key
-6 E
Light flint
7 E Dense flint 2 E Extra dense flint 5 E Barium flint
6E -150 days, 3.2%; 7E -300 days, 1.65%; 2E-300 days, L-5%; 5E -156 days, 3.28%
Fig. 18.1 Comparison of optical glasses after exposure to an alkaline solution (after Simpson, H.E., Glass Tech., 37, 249 (1953)”)
material, e.g. the fibres are made up into rods or rings with the appropriate partly-polymerised plastic, the composites are then cured under specified conditions and the breaking stregth determined after various exposure to water or steam. A statement of standard tests for glass durability is given in Reference 14. Glass Durability Testing
Caution is needed in applying the results of general chemical tests for glass durability across a wide spectrum of glass properties. Glass fibre strengths, for example, are sensitive to the physical, as well as the chemical, nature of the environment and should only be assessed by the direct strength measurements in conditions which closely approximate the final application situation”. The effects of corrosion on other properties need similar direct assessment in many cases. However, in the absence of accepted standard tests the BS, DIN, IS0 tests for laboratory glassware are often used. At the present time, the British Standard BS 3473 ‘Methods of testing and classification of the chemical resistance of glass used in the production of laboratory glassware’ is being re-issued in six parts, of which the first five parts are identical to recently revised IS0 test procedures. There are also corresponding DIN tests in some cases which are very similar. The current situation is:
18: 18
GLASS AND GLASS-CERAMICS
BS 3473:part 1:1985 IS0 695/1984 DIN 523 22
Identical test procedures. Weight loss test using boiling alkali and glass pieces.
BS 3473:part 2: 1987 I S 0 719/1985
Grain test in water at 98°C Measures alkali extracted.
DIN 12111
BS 3473:part 3:1987 IS0 720/1985 No DIN equivalent
Grain test in water at 121OC (autoclave). Measures alkali extracted.
BS 3473:part 4:1983 I S 0 D P 4802/1982 DIN 52 329, 339
Interior surface of glass containers. Water at 121"C (autoclave). Measures alkali extracted.
BS 3473:part 5:1987 I S 0 1776/1986 DIN 12116
Resistance to 6 N HCI at 100°C using glass pieces. Measures alkali extracted.
BS 3473/6
Procedure for classifying glass articles according to chemical and thermal properties. Not yet published in its final form. May be adopted by ISO.
Mechanisms of Glass Corrosion
General corrosion properties The glass surface may react with a corrosive agent in one or a combination of the following waysI5: (a) By forming new compounds on the surface.
(b) By selectively losing material from a leached porous layer. (c) By continuous dissolution leaving a freshly exposed surface.
Under certain circumstances components may be leached out of the bulk of the glass to leave a new material.
The nature of the glass surface It is widely accepted that the composition of the glass surface is different t o that of the interior of the glass, but it is difficult to quantify the difference. Alkali loss during forming, grinding, polishing and surface treatments, affects the structure of the surface, but a more basic difference is brought about by the effect of the unbalanced force fields at the surface on the ions within the glass. Glass is composed of glass-forming cations (e.g. B 3 + , Si4+, P J + ) surrounded by polyhedra of oxygen ions in the form of triangles or tetrahedra. Two types of oxygen ions exist, viz. bridging and non-bridging. The former, bonded to two network-forming ions, link polyhedra, and the latter, bonded tb one network-forming ion only, carry an excess negative charge. To compensate for this, charged cations of low positive charge and large size (e.g. N a + , K+,C a 2 + )are located within the structure. Silicon may be substituted by other cations of large positive charge and small size which are collectively known as network-formers.
18: 19
GLASS AND GLASS-CERAMICS
The difference in size and field strength between ions is reflected in the polarisabilities of each ion, and their final position relative to the glass surface. Since the force field is unbalanced, ions of low polarisability will remain near the surface and ions of higher polarisability will move towards the interior of the glass. A strong feature of chemical reactions associated with the surface is the need to screen adequately (and not merely to neutralise) those cations which have strong electric fields. If the unbalanced force field is removed by the presence of materials, liquid, adsorbed vapour or solid in contact with the glass, there is sufficient mobility within the glass surface zone for it to revert to a more normal structure by the diffision of ions towards the surface. Corrosion mechanism^'^
There is no serious challenge to the view that the alkali or alkaline earth ions are removed from glass in water by an ion exchange process in which H + ions diffuse into the glass to preserve the electrical neutrality of the system. However, only under certain circumstances can the rate-controlling process be directly related to the diffusion of sodium in the glass. Most glasdcorrosive agent systems are treated as unique cases, since in addition to the concentration of the attacking agent, temperature, rate of flow, and reaction time contribute to what is observed. General chemical principles of electrophilic and nucleophilic types of general attack can be applied to glasses”. The first is considered as an attack on non-bridging oxygen atoms by reagents with an electron deficiency, and the second as an attack on bridging oxygens by reagents with an electron excess. For the most common series of corrosive agents, water, steam, acids, alkalis and salts, the hydrolytic processes peculiar to each determine the mechanism of attack. Thus, under the right circumstances, hydrolytic attack on the bridging oxygens* can occur in the following way:
I I -Si-O-si-I I OH(free ion)
+ -si
I
I OH6-
I O-siI
6-
I
+ -si-OH
I
I O---Si-I
. . .(18.1)
This is an irreversible reaction resulting in permanent damage to the glass network. The corrosion process is modified by the physical state of the surface. Grinding and polishing processes, in particular, leave the structure in a more open state and with a degree of roughness and residual stress; all can contribute to accelerated corrosion. The action of water and acids During attack the alkali and alkaline earth network-modifying ions are exchanged by H+or H,O+from acid solution. *It should be noted that the OH- is partly bonded to the network and has lost part of its charge. The oxygen has acquired a partial negative charge and is only partially bonded to the network. Thus the 6- indicates a transition state in which the charges on the OH and 0 lie between 0 and - 1.
18 :20
GLASS AND GLASS-CERAMICS
In some glasses the exchange process can go to completion causing only a small degree of network damage. In water, the exchange process proceeds at a very much slower rate relative to the acid conditions and some attack of the network is possible due to the presence of alkali ions from the glass moving into solution. This is most pronounced when glass is attacked by steam at high temperature and there is no mechanism for the removal of alkali from the hydrolysed zone. Acid solutions mitigate this form of attack by neutralising the alkali as it is formed. The attack of most glasses in water and acid is diffusion controlled and the thickness of the porous layer formed on the glass surface consequently depends on the square root of the time. There is ample evidence that the diffusion of alkali ions and basic oxides is thermally activated, suggesting that diffusion occurs either through small pores or through a compact body. The reacted zone is porous and can be further modified by attack and dissolution, if alkali is still present, or by further polymerisation. Consolidation of the structure generally requires thermal treatment. Attack by alkali solution, hydrofluoric acid and phosphoric acid A common feature of these corrosive agents is their ability to disrupt the network. Equation 18.1 shows the nature of the attack in alkaline solution where unlimited numbers of OH- ions are available. This process is not encumbered by the formation of porous layers and the amount of leached matter is linearly dependent on time. Consequently the extent of attack by strong alkali is usually far greater than either acid or water attack. Both acids form compounds of silicon as a result of attack on the network, silicon fluoride from hydrofluoric acid and silicyl phosphate from phosphoric acid. Chemical attack by other agents If the hydrogen ion concentration is high enough, the glass loses a substantial amount of weight by leaching, but these reactions are very dependent on the nature of the ions in solution. Certain salts, especially those of Zn, A1 and Be, if present as trace amounts, can have a beneficial effect by poisoning the process and preventing the occurrence of leaching. General Corrosion Behaviour
Important effects on glass durability in aqueous conditions, due to the interrelation between glass composition and environmental pH, have been reviewed recently by Doremus” and Adams”. To some extent, durability can be predicted on thermodynamic grounds. In general, additions of calcia and alumina to a basic alkali silicate glass confer water resistance, additions of zirconium, lanthanum, tin and chromium oxides improve alkali resistance and reducing the levels of boron, aluminium or lead oxides, in glasses where they are present, improves acid resistance. Predictions of the effects of added oxides on glass durability should, however, be treated with caution. Structural factors, such as occurrence of phase separation, coordination state, mixed alkali effect and/or kinetic effects resulting from the presence or absence of insoluble reaction product layers on the glass surface, can influence durability to a significant extent.
GLASS AND GLASS-CERAMICS
18:21
More generally, Valez et al.24have reviewed the corrosion behaviour of silicate and borate glasses in contact with alkali metals and molten salts, as well as in aqueous conditions. Another important aspect of glass corrosion behaviour which should be emphasised is the effect of applied stresses. As with most other brittle solids, glass is subject to stress-enhanced corrosion-often described as ‘static fatigue’. Under a continuously applied stress, and in the presence of normal environmental moisture (or other more aggressive corrosion conditions), cracks may grow from flaws on the glass surface and this can lead to delayed failure at stresses below the strength level which is measured in a short-term 26. As a consequence, common silicate glasses are usually loading expected to have a load-bearing capacity at one-quarter to one-third the short-term strength when continuously loaded over some 50 years in the normal atmosphere.
The Cleaning of Glass
There is no universally ideal technique for either cleaning glass or avoiding contamination of the surface. In the most severe circumstances of corrosion the only methods capable of restoring an acceptable surface finish consist of grinding and polishing or removing the contamination and corroded layers by strong etching agents such as hydrofluoric acid. Less severe conditions may respond to treatment by various detergent solutions or organic solvents, these being considerably aided by ultrasonic vibration. Manual washing or ultrasonic cleaning can be used to remove massive dirt accumulations. The vapour degreasing process, which uses isopropyl alcohol, has minimal corrosive action on the glass. To restore old stocks of corroded glass, treatment in hot 1% sodium hydroxide solution followed by rinsing in 5% hydrochloric acid and a final rinse in pure water at room temperature is recommended’8.
Mew Developments in Applications
Over recent years, a number of new applications of glasses have grown out of increased understanding and control of glass corrosion behaviour. Conventional silicate and borosilicate glasses are subject to severe corrosive attack in highly alkaline solutions (pH 12-13-5) such as those found in hydrating Portland cement. There is rapid and drastic loss of strength in fibres formed from glass compositions such as no. 2 or no. 9 of Table 18.3. Highly alkali-resistant glass fibres have been developed from silicate glass compositions containing about 16wtVo of zircona2” 28 and these have formed the basis for development of a range of glass-fibre reinforced cement (GRC) materials, analogous to glass-fibre reinforced plastics (GRP). In the GRP field, it has been shown that E-glass borosilicate fibres (composition no. 9, Table 18.3 are prone to strength loss and stress corrosion in acidic 30- this in turn led to the development of an acid-resistant environmentsZ9~ version of E-glass for use in such conditionsJo.
18:22
GLASS AND GLASS-CERAMICS
In the flatlcontainer glass fields a wide range of surface treatments now exist which modify the surface of the glass to confer corrosion or abrasion resistance, improve mechanical strength or optical properties. A number of treatments which claim to improve chemical or corrosion resistance are described in References 31 to 35. The slow rate of dissolution of, or leaching from, durable glasses has led to proposals for the vitrification of nuclear waste. Glasses based on the sodium borosilicate system appear to be favoured because of their ability to dissolve the waste, combined with good chemical durability. Intensive development has taken place over recent years” and a regular journal is devoted to this topic36. The above applications depend on the development of corrosion-resistant glasses. Equally interesting new developments depend on the controlled use of glass corrosion/dissolution. For example, glasses based on the network formers, P,O, and B203,have been formulated which even at a low pH react with attacking media by continuous dissolution at a constant rate leaving a freshly exposed surface. Heavy-metal ions incorporated in the glass are therefore released into the attacking medium at a controlled rate governed by the composition of the glass. The patent literature describes several uses of these types of glass, e.g. as biocides, as corrosion inhibitors and in animal husbandry3’-. Phosphate glasses, both in solid form or as a powder component in cements4’, have been used to provide the trace elements Cu, Co, Se to ruminant animals. More recently, polymer bonded soluble phosphate glass boluses have been developed to control breeding activity in sheep by releasing the hormone melatonin into the rumen4*, In a further application of ‘controlled release’ of glass components, glass ionomer cements have been used for dental cements, surgical splinting and foundry sands. These cements are based on the hardening reaction which occurs between a powdered ion leachable glass and aqueous solutions of homo-plus co-polymers of acrylic acidd3.
Glass-Ceramics Definition and Properties
Glass-ceramics are a family of materials that are polycrystalline in nature and are formed from the liquid or glassy state. A glass-ceramic article is made by the heat treatment of a vitreous body in two stages: 1. Nucleation. The glass is held at a temperature below its softening point
for a period of minutes or hours to allow nuclei to develop. 2. Crystallisation. The temperature of the nucleated glass is raised to just below the softening point when crystals form and grow around the nuclei. Ideally the product is a fine-grained ceramic containing interlocking crystals with sizes ranging from less than 10 nm in transparent glass-ceramics to several micrometres, with a residual, usually small, glass content. The behaviour of the material is largely determined by the choice of the cry-
GLASS AND GLASS-CERAMICS
18 :23
stalline phase; by suitable choice a range of useful properties has been obtained 19. As a class of materials, glass-ceramics have the following general characteristics: 1. Impervious, with moderate densities similar to those of glasses. 2. Rather high strengths for oxygen-rich solids, accompanied by some scatter of individual values. 3. Stiff, elastic, fully Hookean behaviour. 4. Considerable hardness and resistance to abrasion. 5 . Chemical stability and resistance to corrosion. 6. Resistance to medium-high temperatures (higher temperatures than most glasses but lower than the refractory oxides) and low thermal conductivity. 7. Resistance to the passage of electrical current.
Special characteristics can be developed in individual materials depending on the cations present and their arrangement relative to each other and to the oxygen anions. The most important of these characteristics is low, medium or high reversible thermal expansion. The properties of some commercially available glass-ceramics are summarised in Table 18.6. Chemical Durability of Glass-ceramics
Although the factors which govern the chemical stability of glasses are fairly well-known, there is little information concerning this aspect of glassceramics. While the chemical behaviour of a glass-ceramic is strongly influenced by the chemical composition of the parent glass, several different crystalline compounds together with a residual glass phase are likely to be present. The relative resistances of these phases to attack by water or other reagents will determine the chemical stability. In general, a glass which exhibits poor chemical stability is unlikely to give rise to a glass-ceramic of high stability. To this extent the factors which govern the stability of glassceramics can be equated to those which determine the chemical stability of glasses. In most cases, glass-ceramics possess good chemical stability and certainly compare favourably in this respect with other ceramic materials. Table 18.7 summarises makers’ data for chemical attack on commercially available materials. Certain types of glass-ceramic have good resistance to attack by corrosive chemical reagents. Low-expansion glass-ceramics derived from lithiumaluminosilicate glasses are only slightly inferior to borosilicate chemicallyresistant glass with regard to attack by strong acids and are somewhat more resistant to attack by alkaline solutions. Materials derived from magnesiumaluminosilicate glass compositions are slightly less resistant to attack by strong acids and alkalis than are chemically-resistant borosilicate glasses. Even at high temperatures these types of glass-ceramic retain resistance to attack by corrosive gases. For certain applications it is important that the glass-ceramic should be unaffected by contact with reducing gases at high temperatures. In such
N
v) m
N
v)
N
v) 'A
N
N
m
3
\ M
18 :25
GLASS AND GLASS-CERAMICS
Table 18.7 Chemical resistance data for some commercially available glass-ceramics Chemical resistance
Powder method Water solubility
A
Units
Acid solubility Alkali solubility
mg as Na,O per 1 mg sample loss in wt., Vo loss in wt., Vo
Surface method HZO (90°C x 24h) 5 % HCl(90" x 24 h) 5% NaOH (90' x 24 h)
mg/cm mg/cm2 mg/cm*
B
C
D
0.8
128
3.5
5
0.1-0.3 0*02-0*1
0.29
0.06 0.13
0-00 0.13 0.04
0.12
0.20 4.0
Key: A Heatron T B Nsaxram-15 C Corning code %50 D Pyroccram 9608 E Hcrcuvit
cases the composition must not include any oxides such as lead which are easily reduced to the metal. D. S . OLIVER B. A. PROCTOR
REFERENCES
Rawson, H., Inorganic Glass Forming Systems, Academic Press, London (1967) Viscosity-temperature Relations in Glass, International Commission on Glass (1970) Standard Method of Testf o r Softening Point of Glass, ASTM Designation C338-57 (1965) Thermal Shock Testing, BS 3517:1%2 5. Glass Pipelines and Fittings, BS 2598:1966 6. Volf, M. B., Technical Glasses, Pitman, London (1961) 7. Wessel, H., Silikattechnik, 18, 205-211 (1%7); ibid., 19, 6-10 (1968) 8. Standard Methods of Test f o r Resistance of Glass Containers to Chemical Attack, ASTM Method C 225-68 9. Testing of Glass: Determination of Resistance to Acid, DIN 12116:1955; Testing of Glass: Determination of Resistance to Caustic Liquors, DIN 12122:1955; Testing of Glass: Determination of Resistance to Water, DIN 52322: 1967 10. Method of Testing Chemical Resistance of Glass Used in the Production of Laboratory Glassware, BS 3413:1962 1 1 . Determination of resistance of Glass to Attack by Boiling Aqueous Solution of Mixed Alkali, I S 0 R695:1%8; Determination of Hydrolytic Resistance of Glass Grains at 98"C, IS0 R719:1968 12. Wiegel, E., Glas. Berichte, 29, 137 (1956) 13. Simpson, H. E., J. SOC.Glass Tech., 37, 249 (1953) 14. Fletcher, W. W., BGIRA Information Circ. No. 96 15. Holland, L., The Properties of Glass Surfuces, Chapman & Hall (1964) 16. Das, C. R. and Douglas, R. W., Phys. & Chem. of Glasses, 8, 178-184 (1967) 17. Budd, S., Phys. & Chem. of Glasses, 2, 111-114 (1%1) 18. Tichane, R. M., Amer. Cer. SOC.Bulletin, 42, 4 4 1 (1%3) 19. McMillan, P. W., Glass-ceramics, Academic Press, London (1964) 20. Doremus, R. H., Treatise on Materials Science & Technology, 17, 41-67 (1979) 21. Adams, P. B., J. Non-Crystalline Soli&, 67, 193-205 1984 22. Paul, A., J. Mat. Sci., 12, 2246-2268 (1977) 23. Newton, R. G. and Paul, A., Glass Technology, 21 No. 6, 307-309 (1980) 24. Velez, M. H., Uhlmann, D. R. and Tuller, H. L., 'Chemical Durability of Glasses', Revista Lotinoarnericano de Metallurgia et Moteriules, 11, No. 1 (1982) 1. 2. 3. 4.
18:26
GLASS AND GLASS-CERAMICS
25. Creyke, W. E. C., Sainsbury, 1. E. J. and Morrell, R., Design with Non-ductileMaterials’, Applied Science Publishers (1982)
26. Section V ‘Crack Growth and Fatigue’in ‘Sfrengfhoflnorganic Glass’, ed. Kurkjian, C. R., NATO Conference Series VI, Materials Science, Volume 1 I, Plenum Press, New York (1985) 27. Majumdar, A. J. and Ryder, J. F., Glass Technology, 9 No. 3, 78-84 June (1%8) 28. Proctor, B. A. and Yale, B., Phil. Trans. Roy. SOC.London, A294, 427-436 (1980) 29. Metcalfe, A. G., Gulden Mary, E. and Schmitz, G. K., Glass Technology, 12, 15-23, February (1971) 30. Torp, S. and Arvesen, R., 34th Annual Technical Conference, Reinforced Plastics/Cornposites Institute, SPI, 1979, Session 13-D 31. Yaschishin, I. N. and Sharagov, V. A,, USSR Patent 624,889 (1978) 32. Yaschishin, I. N., Stekloi Kerurnica, No. 8 , 6-7, (1974) 33. Desantis, U. J . and Synder, H. C., U S . Patent 3, 709,672 (1973) 34. Budd, S. M., U.S. Patent 1,244,832 (1971) 35. Wakai, N., Jap. Kokai Tokyo Koho 79. 40816 (1979) 36. Nuclear and Chemical Waste Management, Pergamon Press, Oxford. 37. U.K. Patent 2,116,424 A 38. U.K. Patent 2,133,182 A 39. U.K. Patent 2,037,735 A 40. U.S. Patent 4,098,610 41. U.K. Patent 2,123,690 A 42. Poulton, A. L., Symons, A. M., Kelly, M. 1. and Arendt, J., J. Reprod. Fert, 80.235-239 ( 1987) 43. McLean, J. W. and Wilson, A. D., Australian Dental Journal, 22, 31-36 (1977) 44. Cockram, D. R., Litherland, K. L., Proctor, B. A. and Yale, B., Glasstechische Berichre, LVI, 644-649.
18.3 Vitreous Silica
Introduction Vitreous silica, also referred to as quartz glass, fused quartz, or fused silica, is a material of considerable importance, possessing a unique combination of high softening temperature, excellent resistance to chemical attack, and high transparency. The general physical and mechanical characteristics are common to all glasses, and have been reviewed adequately in Section 18.2, the difference being that the high purity ( > 9 8 - 7 % S O 2 ) maximises the durability, fusion temperature and volume stability. The superiority of fused silica over conventional glasses is illustrated in Table 18.8. Table 18.8 Physical property data for used silica, crown glass and pyrex Fused Silica (Si02 ) Softening point ("C) Annealing point ("C) Resistivity (am) Dielectric constant at 1 kHz Thermal expansion X IO-'K-'
'Crown' glass (SO2-Ca-Na, 0)
Pyrex (Si02-B203-Na20)
1015
720 540 1 . 1 x ioL3
820 565 1015
3.8
7.4
5.1
5.4
8.8
3.3
1594 1082
Manufacturing Process The raw materials for the production of vitreous silica are either high purity rock crystal from which the transparent form is produced, or vein quartz (high grade glass sand) from which impurities have been removed by acid leaching. The persistence of liquid and gaseous inclusions is partly responsible for the translucency of the cheaper form of vitreous silica. The ground quartz is melted at around 2000°C by induction heating in a graphite crucible. Even at this temperature, the melt viscosity is so high (105-106 Pa s) that the trapped gas bubbles cannot be removed by ascent to the surface. Therefore gases from the intergranular spaces are removed by evacuation 18 :27
18 :28
VITREOUS SILICA
to high vacuum at the beginning of sintering (-1 400°C) to allow pores to be closed completely. At high temperature, the pressure cannot fall below the equilibrium vapour pressure, which is comparatively high as a result of partial reduction to SiO. Vitreous silica produced by this route contains small amounts of impurities such as Fe, Cr, A1 and Ca. To achieve metal ion impurities < IO-'% the synthetic hydrolysed silane process is used. Organic silica compounds or SiCI, are hydrolysed in a flame to produce fine molten droplets of S O 2 which is deposited on a cold base.
SiCl,
+ 2H20
4
SiO,
+ 4HC1
Contamination by hydroxyl groups is eliminated if oxygen is used instead of water in the reaction.
Structure and Physical Properties Polymorphism of Silica
Although vitreous silica is nominally a homogeneous isotropic amorphous material, and should normally remain so during its service life, it is in fact in a metastable condition. The tendency to revert to crystalline forms with attendant deterioration in mechanical durability places severe limitations on the range of applications. Figure 18.2 illustrates the polymorphic forms of silica, and the dimensional changes accompanying each transition.
200
400
600
800
1000
1200
1400
1600
Temperature ("C)
Fig. 18.2 Polymorphism and dimensional changes in silica as a function of temperature
VITREOUS SILICA
18 :29
Changes from one polymorphic form of crystalline SiO, to another, as from tridymite to quartz at temperatures below 870°C and to cristoballite above 1470”C, involve the breaking of very strong Si-0 bonds. High energies are required for these ‘reconstructive’ changes, and changes from one form to another require very long periods for completion. On the other hand, inversions from high to low temperature forms of quartz or cristoballite involve only changes in the angles between adjoining SiO, tetrahedra, and these ‘displacive’ transformations are accomplished almost instantaneously. The accompanying volume changes lead to disruption of ware containing significant amounts of quartz or cristoballite. Thermal Expansion
The coefficient of thermal expansion of vitreous silica is very small (5.4 x lo-’ over the range 0-1 OOOOC), about one-sixth that of porcelain. It is thus highly resistant to thermal shock. Heat Resistance
Being a glass, vitreous silica softens progressively as it approaches its melting point of 1713°C. The maximum recommended working temperature is 1050°C in an oxidising atmosphere, though it may be taken to 1 350°C for short periods. Surface devitrification to cristoballite occurs above this temperature. This causes some loss of transparency, but chemical and mechanical durability are unaffected provided the temperature does not fall below the P to a! inversion temperature (275”C),which would lead to the initiation of cracks. It should be noted that devitrification is accelerated by traces of alkali metal compounds, particularly potassium and lithium salts, sodium tungstate and ammonium fluoride. Devitrification is also enhanced by water vapour and oxygen, but inhibited by neutral or reducing atmospheres. Thermal Conductivity
The thermal conductivity of fused silica is low (1.38 W m-* K - ‘ ) . The transparent form passes infra-red radiation with little loss up to wavelengths of 3.5 pm. Electrical Characteristics
The insulating properties are excellent. At ordinary temperatures the resistivity of the translucent form is 10’’ Qm, and it is capable of withstanding high-frequency discharges at high voltages. See also Table 18.4 for data on other physical properties.
18:30
VITREOUS SILICA
Resistance to Chemical Attack Most glasses suffer chemical attack and deterioration due to ion exchange of sodium ions to form a highly alkaline solution which subsequently attacks the network. As network-modifying oxides are absent from fused silica, this mode of attack does not occur. Hence fused silica is highly resistant to most aqueous solutions, even aqua regia at elevated temperatures having no effect. Only those reagents that attack the network silica directly, such as strong alkalis and fluorides, are to be avoided. Boiling Water and Steam
There is negligible reaction with water and steam at moderate temperatures and pressures, as indicated by the free-energy change for the solution reaction: SiO,(quartz)
+ H,O(P) e H,SiO,
AGO = +29650 J mol-'
A solubility of < 6 ppm occurs, however, at temperatures in the range 400500°C and pressures of the order of 3*5MNrn-,. The solubility rises to 0-14% for the translucent form, and 0.035% for the transparent.
Fluorine, Hydrofluoric Acid and Alkaline Solutions
Silica is susceptible to attack by all three reagents, the rate of corrosion increasing with temperature and concentration. Hence 5% caustic soda solution can be contained in fused silica at room temperature, but attack becomes significant at pH values greater than 9, as shown in Fig. 18.3. The essential step in the dissolution reaction is the breaking of a siloxane bond Si-0-Si. This bond, although strong, is polar, and may be represented as (Si6+-06-). The excess positive charge associated with the
. 0
.E 60
-
E
N
. 0 v)
OI
40
-
N
v)
2 20 0
0
I
I
I
-
18:31
VITREOUS SILICA
silicon atom makes it susceptible to attack by nucleophilic reagents such as the OH- ion which attaches itself to the silicon, rupturing the network at that point as shown below:
I
I
-si-o-siI
+
I
I
si . . . o-si-
-
I
I
+ OH-
OH
(initial state)
(transition state)
I
I
I OH
I
+ - Si -0-Si-
(final state)
The attack of HF is thought to proceed by a similar mechanism in which there is simultaneous nucleophilic and electrophilic attack on the network silicon and oxygen atoms, respectively, according to:
-si-
I 0 -si - --* -Si-O.. I
+
H + F-
. Si-
i
I
i
I
H
F
+
I -Si-OH I
I F-Si-
I
It should be pointed out the H + ion alone is insufficiently powerful to affect disruption of the siloxane bond without the simultaneous action of the Fion. Consequently sulphuric and nitric acid d o not initiate attack, even at temperatures up to 1 OOO°C. Exceptions to the principle are hydrochloric and hydroiodic acid, which, although satisfying the requirements of simultaneous nucleophilic-electrophilic attack, exert a negligible degrading effect on silica. Sodium fluoride also attacks silica, as do sodium metaphosphate and sodium polyphosphate, and to a lesser extent sodium carbonate and sodium cyanide. Attack is particularly vigorous for fused alkalis, alkali halides and phosphates.
Basic Oxides
As an acidic oxide, SiO, is resistant to attack by other acidic oxides, but has a tendency towards fluxing by basic oxides. An indication of the likelihood of reaction can be obtained by reference to the appropriate binary phase equilibrium diagram. The lowest temperature for liquid formation in silicaoxide binary systems is shown below: Oxide eutectic("C)
Al,O, 1546
Be0
CaO
MgO
Tho,
TiO,
ZrO,
1670
1436
1543
1700
1540
1675
Metals
Silica is only decomposed by those metals which have a high affinity for oxygen as indicated by the Ellingham diagram (Fig. 18.4). On this basis, molten sodium should be compatible with silica:
18 :32
VITREOUS SlLlCA
r -200
-400
=.
!a?,
-800
c
a,
-1000 LL
-1 200
-1 400
0
200
400
600
800
1000
1200
1400
1600
Temperature ("C! Fig, 18.4 The standard free energy of formation of various oxides a5 a function of temperature
+
+
2Na(P) iSiO,(s) + Na,O(s) +Si(s) AGO = 73 kJ/g atom oxygen
+
The equilibrium levels of the reaction products are very small, but both can dissolve in liquid sodium, and sodium oxide can form compounds with silica. As a consequence, the reaction moves to the right, leading to further reduction of silica. Nevertheless, vitreous silica crucibles have been used sucessfully for containing molten antimony (S50°C), copper (1 210°C), gallium (1 100°C), germanium (1 100°C), lead (500OC) and tin (900°C). In accordance with the free energy diagram, silica is readily attacked by molten aluminium, lithium, magnesium and calcium.
Applications of Vitreous Silica The high thermal and electrical resistance of vitreous silica, and its imperviousness to chemical attack, make it suitable for a wide range of applications. These include chemical and physical laboratory ware, tubes and muffles for gas and electric furnaces (including vacuum furnaces), pyrometers, insulators for high-frequency and high-tension electrical work, mercury-vapour and hydrogen-discharge lamps, high-vacuum apparatus, plants (complete or partial) for chemical and related industries, equipment
VITREOUS SILICA
18:33
for the manufacture of pure chemicals, tubes, chimneys and radiants for the gas- and electric-heating industries, component material in refractory and ceramic mixtures, etc. It is used for pipes to carry hot gases and acids, acid distillation units, condensing coils, S-bend coolers, hydrochloric acid cooling and absorption systems, nitrating pots, and cascade basin concentrators for sulphuric acid. The inertness of vitreous silica to most acids is also utilised in the manufacture of electric immersion heaters and plate heaters for acidic liquors in chemical processes and electroplating baths. Vitreous silica wool is used for filtration of acidic liquids and filtering hot gases. The resistance of vitreous silica to water and steam at normal temperatures and pressures makes it applicable in the production of pure water for the manufacture of highly purified chemicals. Because of its thermal properties, it is used for the construction of muffles of oval cross-section used for the bright annealing of metal strip and wires. Special products include transparent vitreosil springs, which are ideal for continuous measurement in corrosive atmospheres, and quick-immersion thermocouple protection sheaths for rapid temperature measurement.
C. A. MAY BIBLIOGRAPHY Lay, L. A., The Corrosion Resistance of Technical Ceramics, NPL (1983) Paul, A., Chemistry of Glasses, Chapman and Hall, London (1982) Gulaev, V. M. ‘The Strength and Deformation Properties of Vitreous Silica’, Glass Ceramics, 30 No. 6, 279 (1973) Rawson, H., ‘Properties and Applications of Glass’, Glass Technology 3. Elsevier (1986) El-Shamy, T. M. M., and Douglas R . W . , Journal of the American Ceramic Society, 50, 1-7 ( 1%7)
18.4 Glass Linings and Coatings
While glass can provide many of the desirable features of an ideal inert material, fabrication difficulties prevent its use for large-diameter chemical process equipment, and mechanical considerations would in any case make it necessary to treat any such equipment with great care. The high chemical resistance, and the non-toxic, non-flavouring and thermal-resistance properties of glass, can however be combined with the mechanical strength of metals by covering metal surfaces exposed to corrosive media with a layer of suitable glass. It thus becomes feasible to produce large storage or transport tanks of over 10 OOO litre capacity, reaction vessels, valves, pipes, silos, smoke stacks, car exhausts etc. which have this serviceable combination of properties. The principal advantages of glass linings are the increased size and mechanical strength that are possible compared with all-glass equipment, and the flexibility of operation with different chemicals compared with allmetal equipment. The increased heat transmission, in comparison with glass equipment, can also be an advantage. The principal disadvantages are the loss of transparency and the potential vulnerability of the lining to mechanical damage unless sensible precautions are taken in handling, installation and service. A variety of metals can be protected in this way, including copper, gold, stainless steel, titanium and uranium, but by far the most extensive use of the technique is for steel equipment.
Glass Preparation There are a number of proprietary glass formulations for coating steel. In most cases the chemical constitution has not been disclosed but the more successful types are borosilicates containing more aluminium and alkali oxides than the typical heat-resisting borosilicates discussed in Section 18.2. In formulations, for ground-coating the steel, the glasses contain a much greater proportion of cobalt oxide than is found in ordinary glasses, to encourage the formation of a bond between the metal and the glass. The coefficient of linear expansion of a typical glass for this application is about 10 x 10-6/oC. The disparity between this figure and the higher expansion coefficient of the steel is quite deliberate and results in the development of 18 :34
GLASS LININGS AND COATINGS
18 :35
compressive stresses in the glass layer after processing. The stress which develops gives more resistance to thermal shocks or external stresses than if the expansion of the glass and steel were accurately matched. The second coat, or cover-coat, has a composition different from the first coat, or ground-coat, such that it has optimum durability. The raw materials required to produce the particular glass composition are intimately mixed and melted at a temperature near 1300°C. The melt is quenched with water, yielding the glass in a granular form known as frit, which is subsequently either ground to give dry glass powder or wetmilled with ball clay to form a creamy slip. It can be applied to the steel in either form, the particular method depending on the design and size of the structure.
Metal Preparation A low-carbon (less than 0.2% C) steel is usually employed in the manufacture of chemical storage or process vessels, water heaters, and the like; one writer refers more precisely to “a rimmed steel (ASTM A285), grade A or B flange quality” as being suitable, and states that with this base, defects such as fishscaling, blistering, crazing and poor adhesion are avoided. Cast iron is frequently employed in such equipment as valve bodies and pipe unions. It should be good quality, close-grained grey iron free from blow-holes, cavities and porosity. ‘Filling’ is not permissible. Conventional welding techniques are used in fabrication and assembly of the steel body, but special precautions must be taken to ensure that the weld metal is of similar composition to the base metal, and that moisture(hydrogen-) free electrodes are used. Lap-welded joints and riveted structures are not suitable for the subsequent enamelling operations; butt welding is much preferred. The completed assembly is examined in detail to check defects likely to impair serviceability, and is then normalised at 900°C to relieve stresses set up during fabrication. The heat treatment also has the advantage of burning-off any organic contamination. After cooling, the vessel is sand-blasted with a silica-type grit to remove oxide scale and to promote, by providing a roughened surface, mechanical keying of the glass ground-coat layer to be applied. The expensive preheating to burn off grease can be replaced by chemical descaling and degreasing I . Subsequently an alkaline cleaner, followed by a water wash and then by an acid pickle (HCI or H,SO,) and rinse prepares the steel for enamelling. A nickel can be used to further enhance adherence. Points which manufacturers note as being important in the design of these vessels are as follows: 1. The metal thickness should be as uniform as possible throughout,
avoiding heavy bosses, lugs or brackets. 2. The minimum radius recommended for all curved surfaces is 6 - 4 mm; sharp edges must be entirely avoided. On small radii there is a risk that shaling or other difficulties might arise. 3. The risk of distortion on firing is greater if the design calls for numerous apertures in the vessel or its cover.
18: 36
GLASS LININGS AND COATINGS
Lining Processes The ground coat, where used, is sprayed on in the form of a wet slip and after drying is fired at approximately 900°C. With open vessels or small cast-iron units, dry glass powder can be dusted directly on to the hot, fused surface of the ground coat; larger vessels such as storage tanks are allowed to cool, inspected and then sprayed uniformly with the wet slip. The coating is thoroughly dried and then fused by heating the vessel in a furnace at approximately 850°C. The vessel is cooled, and further coats of glass enamel may be applied as required in the same way. Thicker layers can thus be built up. Many cast iron items are preheated to 800°C to 1 O00"C and dusted with enamel powder. The coherent glass coating develops immediately and consolidates as the item is returned to the klin for a slow cool. It is clearly important that the vessels being treated should not distort under their own weight during furnace treatment, and it may be necessary to design the article with slightly greater thickness than usual to allow for the softening effect of high temperatures. Another way of avoiding the difficulty with closed vessels is to force inert gas into the tanks at controlled pressure. The integrity of the linings is tested electrically.
Properties of Glass Linings on Steel Mechanical Properties
With correctly formulated glass, the lining has the ability to withstand stresses up to the elastic limit of the steel without breaking. Impacts sufficiently severe to dent the steel will probably cause fractures in the lining. The hardness and abrasion resistance of the lining are similar to those of all-glass equipment and a smooth, easily cleaned surface is produced. The combined effect of mechanical keying of the glass to the metal and the chemical bond developed between the two results in a very high adhesive strength. Test figures indicate that the bond strength is of the order of 35-70 MN/m2.
Thermal Properties
The thermal shock resistance of glassed steel, i.e. the safe limit of temperature difference between the glass surface of the vessel and any charge introduced, varies according to the general operating temperature. This is because the desired compressive stress on the glass is reduced as the temperature increases. With a typical coating formulation on a vessel operating at 120"C, the recommended maximum thermal shock would be about 93"C, while at an operating temperature of 205°C the corresponding figure would be 55°C. Recent improved formulations have made it possible for vessels to withstand thermal shocks of some 30% greater than this. Plant can operate from -20°C to +300"C with the maximum operating temperature being determined by the corrosivity of the contact liquor.
18 :37
GLASS LININGS AND COATINGS
The heat-transfer coefficient for heating in glass-lined equipment is of the order of 340-455 W m-20C-I, but can be increased by agitation. For cooling, corresponding figures would be 200-285 W m-'"C-l. The numerical values depend on the thickness of the glass coating, increasing with decreasing glass thickness, and the figures quoted represent the behaviour of an average coating. Chemical Properties
The general pattern of chemical resistance of glass linings is very similar to that of all-glass equipment. Water absorption is negligible, resistance is very high to all acids except hydrofluoric and (at high temperatures and concentrations) phosphoric acid, attack by water is measurable only with difficulty, most organic liquids produce no measurable effect and strongly alkaline solutions are satisfactorily handled at near-ambient temperatures, but at higher temperatures there may be appreciable reaction. For most types of glassed chemical plant the enamel can be assessed by measuring the loss in weight per unit area when exposed to liquid. Conditions in service vary widely. Many cold liquors (milk and beverages) have little corrosive action and the chosen glass lining need not be very acid resistant. Alternatively, glass-lined reactors and stills might be exposed to high temperatures, under pressure (or vacuum) and the enamel then must be very resistant. A degree of resistance to alkali is always desirable for defence against caustic cleaning aids. Priest" has given the results of 15-day tests on a particular glassed steel, both for immersion and vapour-phase conditions, in a variety of media. Some of this information is extracted in Table 18.9 Table 18.9 Corrosion of glassed steel in boiling acid/
distilled water systems ~~
~
Test solution
~~
Corrosion rates (mmh)
Liquid phase Distilled H 2 0 50 p.p.m. HCI 500p.p.m. HCI 1 .O% HCI 10.0% HCI 14.0% HCI 18.0% HCI 50 p.p.m. H2S04 500 p.p.m. H2S04
25 years), durable (15-25 years); moderately durable (10-15 years); non-durable (5-10 years); and perishable, (< 5 years). Timber species also vary in the treatability of the heartwood’,’.’ with preservatives in an impregnation plant, and are classified into four categories according to the depth of penetration which can be achieved: permeable (complete penetration); moderately resistant (6- 18 mm lateral penetration); resistant (3-6 mm lateral penetration); and extremely resistant (no appreciable lateral and very little end-grain penetration). There are, however, timber species of low natural durability which are impermeable and therefore cannot be effectively preserved. Such timbers can only be successfully used in dry situations. Likewise where preservativetreated timber is to be used in a situation of the highest hazard, long service life is best achieved by using easily treated species impregnated with a suitable preservative. Timber species that are difficult to treat can be incised to improve preservative penetration. For outdoor applications including ground contact the most effective wood preservatives are and copper/chrome/arsenic (CCA)9*’o applied using a vacuum/pressure process. Creosote is used for railway sleepers, telephone and electricity transmission poles, agricultural buildings and fencing. CCA preservatives chemically react with the wood and cannot be leached out. The treated timber is mainly used in buildings and fencing and can be painted. Most building timbers out of ground contact, are impregnated by a double vacuum process using a clear solvent-type preservative incorporating a fungicide, an insecticide and a water repellant. It is the preferred treatment for profiled joinery timber as there is no distortion, and painting is permitted, as the solvent is volatile. An alternative treatment for building timbers out of ground contract, widely used in Australia and New Zealand, is impregnation with disodium octaborate using a diffusion process applied to freshly sawn green timber. In the maintenance of exterior timber, the surface application of creosote or a decorative solvent-type preservative at suitable intervals adds greatly to service life. For the remedial treatment of fungal decay and insect attack on timber in buildings, clear solvent-type preservatives applied by spraying all available surfaces are generally used, supplemented with wood preservative pastes to obtain deep penetration. Water-based emulsions are also used for woodworm control. There has been a general move towards the use of those
18: 100
WOOD
fungicides and insecticides which present a lower risk of environmental problems, especially in housing. Another category of treatment sometimes required is that against fire damage”. Surface coatings meet most requirements, but in some situations impregnation treatments of permeable timber species with fire retardant salts is specified. Impregnation solutions are usually based on ammonium phosphate and borax, and are commonly combined with a CCA preservation treatment. Salt-type treatments are only suitable for indoor use, but organo-phosphorus resins can also be used in exterior applications. With manufactured wood products, such as chipboard, fire-retardant salts can be incorporated in powder form. Unprotected timbers of large cross section survive structurally in fires better than unprotected metal, as an outer layer of charcoal protects the inner wood. Impermeable timbers have a good resistance to polluted atmospheres where acid fumes rapidly attack steel. Wood has given excellent service in the buildings of chemical works and railway stations. Permeable wood species and sapwood can suffer defibration problems caused by the sulphur dioxide of industrial atmospheres. Tile battens are particularly vulnerable. The heartwood of Douglas fir, pitch pine, larch, Scots pine/European redwood and many tropical hardwoods give good service in these conditions. Developments in glued laminated structures and panel products such as plywood and chipboard raises the question of the durability of adhesives ” as well as wood. Urea-formaldehyde adhesives are most commonly used for indoor components. For exterior use, resorcinol adhesives are used for assembly work, whilst phenolic, tannin and melamine/urea adhesives are used for manufactured wood products. Urea and casein adhesives can give good outdoor service if protected with well-maintained surface finishes. Assembly failures of adhesives caused by exudates from some timber species can be avoided by freshly sanding the surfaces before glue application.
Wood in Marine Conditions Of all the natural situations in which timber is used, marine conditions are considered the most severe, and only a few of the commonly available timber species can be relied on to give a very long service life. This elite of timber species includes oak, teak, mahogany, and pitch pine for boat building and greenheart for harbour works. These are all highly impermeable and naturally durable woods. Centuries of tradition in the successful use of these timbers impairs any judgement on the extent of the problems which would arise if other timbers were used for marine work. Until the voyages of discovery oak was very much relied on. Fungal decay in boats and marine structures is mostly associated with rainwater penetration and bad ventilation. Although it is generally thought the decay risk is less in saline conditions, this problem is largely controlled by the selection of timber species. Insect borer damage is uncommon in woods used for marine work, as again the selection of species and the avoidance of sapwood largely eliminates this risk. However, serious problems can occur in the form of marine borers1L’14 and chemical decay arising from attack on the wood by the products of metal corrosion”.
WOOD
18: 101
Shipworm (Teredo spp.), a mollusc which burrows into wood for protection, was a considerable problem in the days of wooden sailing ships, and can still be troublesome in wooden boats and harbour installations, especially in warmer climates. Gribble (Limnoriuspp.), a small wood-boring crustacean, also causes serious damage and operates in cooler waters. Dense, naturally durable tropical hardwoods, and permeable timber species impregnated with creosote or CCA preservative, give the best service life against these pests. Sheathing with other materials is also effective. Chemical decay of the wood around fastenings in boats and marine structures, commonly known as ‘nail sickness’ is caused by attack on the wood by the products of corrosion, i.e. alkali and iron compounds. Corrosion of fastenings operates on differences in oxygen availability” as in crack or crevice corrosion. Any part of a fastening which has access to oxygen and electolyte becomes cathodic, producing alkali from sea-water. In situations where this alkali cannot be lost by leaching or diffusion it chemically attacks the wood, gradually solubilising extractives, hemicelluloses and the lower molecular sizes of cellulose. Hardwoods, because of their much higher hemicellulose content are less resistant than softwoods, showing more shrinkage and distortion in damaged areas, but many hardwoods have the advantage of higher impermeability. Corrosion takes place on the embedded parts of fastenings where oxygen access is poor, Le. the anodic areas. The soluble iron salts migrate into the wood and gradually hydrolyse producing rust deposits and mineral acid. l 6 The latter causes softening and embrittlement of the wood by hydrolysis, converting long-chain cellulose into smaller molecular sizes. Corrosion of fastenings is normally prevented by the use of highly impermeable woods (traditional boat-building timbers) and totally enclosed fastenings to control oxygen and electrolyte access. High standards of workmanship are required for the same reason. Where parts of fastenings or fittings are exposed they should be protected with highly impermeable coatings. It has been demonstrated that further protection can be obtained so that the metalby the use of fastenings coated with suitable electrolyte contact is eliminated. The use of plastic-coated fastenings could widen the application of more permeable timbers for boat building and marine work. Pressure creosoting also makes a valuable contribution towards making permeable woods used for harbour works proof against seawater and oxygen penetration, thus reducing corrosion risks. Another problem associated with the marine situation, particularly in hot climates, is surface defibration or ‘furring’ of ship’s decking” and other timber surfaces. Disruption is due to the growth of salt crystals in the wood surface in the alternating cycles of wetting and drying, and is often assisted by ultraviolet radiation damage to the lignin-rich bond between fibres. Damage is superficial in the case of impermeable woods, but can be very destructive to permeable timbers. This problem is minimised by the use of impermeable timbers and the regular maintenance of protective surface coatings.
18 :102
WOOD
Corrosive Liquids: Wood in Chemical Service Wood in chemical applications l8 gives remarkably good service in the most severe conditions. Impermeable woods give the best results as any chemical degradation is confined to the surface, and the breakdown products are generally less harmful than metal corrosion products. Because of its permeability all sapwood must be removed. The low heat conductivity of wood is a property of considerable advantage for conserving heat and wood is commonly used for hotwells. Wood has a good resistance to a wide range of chemicals. The acid resistance is superior to that of most common metals. Iron begins to corrode at pH 5 , whereas attack on wood commences at pH 2, and even at lower values proceeds at a very low rate. Wood has excellent resistance to acetic acid which is particularly destructive to most common metals. In alkaline conditions wood has good resistance up to pH 11. Softwoods (pitch pine, Douglas fir, larch, and if available Southern cypress and kauri), because of their chemical composition (low hemicellulose, high lignin, and high cellulose content) are intrinsically more chemically resistant than hardwoods, especially for alkaline conditions where they are also less likely to produce coloured extracts (cf. hardwoods with tannins). Softwoods are generally of straighter grain than hardwoods which is an advantage in vat construction as it confers maximum strength and minimises the risk of distortion and permeability which could be introduced by sloping grain. (The longitudinal to transverse permeability ratio averages 1O:l for various wood species.) Hardwoods (oak, teak, iroko, idigbo and many other tropical hardwoods) give excellent service in acid conditions such as metal cleaning operations. Greater impermeability and higher densities can be found in hardwoods, and these properties largely offset their intrinsically lower chemical resistance and poorer straightness of grain. Wood is still a favoured material for constructing equipment used for handling chemical effluent. Although the use of wooden vats has declined with the advancement and wider use of plastics and corrosion-resistant metals, selected wood can still be the most economical material to use. Many old vats are still giving excellent service in industry. Vats give their longest service when used for one continuous operation, thus avoiding shrinkage and contamination troubles caused by changing usage. Vats should not be painted as any barrier on the outside will raise the moisture of the whole cross-section to a level promoting fungal decay in all but extremely durable timbers. Nitric acid, chlorine and sulphur dioxide are all destructive to wood, attacking the lignin component and causing surface defibration. To prevent or minimise chemical attack on wood surfaces, protective coatings of various waxes, bitumen, chlorinated rubber, polyvinyl chloride, phenolic and furfuryl resins can be applied. In recent years the plyvat has gained popularity. It is constructed of coldmoulded marine plywood and protected internally with suitable resin coatings. A variety of phenolic resin impregnated laminated wood products which have good chemical resistance (particularly against acids) are made by soaking beech veneers in solutions of phenolic resins, assembling and heat curing in plywood or moulded form.
WOOD
18: 103
Mildly Corrosive Liquids The chief advantage of wood for containers is that many common species are free from harmful contaminants. For this reason wood had widespread use in the food and beverage industries, but it has now suffered severe competition from corrosion-resistant metals, plastics and paper products. Oak had a very extensive use in tight cooperage in the brewing industry, and its use for barrels still survives in the maturing of whisky and brandy and in the wine industries. Wood is particularly useful where acetic acid is present as this acid is corrosive to most common metals. Wood is commonly used for the packing in large water-cooling towers associated with electricity generating stations. Tight control of mild pH values is essential to avoid chemical damage to the wood surfaces by oversoftening or overchlorination of the water 19. Biological degradation of wood by soft-rot fungi was first discovered in the softwood packing of cooling towers, and effective preservation has been achieved by impregnation with CCA preservative which cannot be leached out under proper operating conditions. With water-cooling towers which have outer wooden casings, defibration or ‘furring”’ of the outer surface sometimes occurs, caused by permeable wood allowing water evaporation, which leads to crystal growth of salts from the water causing surface disruption. Impermeable timbers avoid this problem.
Staining of Wood Impaired appearance of wood due to staining2’ by contaminating substances can also be considered as a type of degradation, and its avoidance is important in furniture, panelling and high-class joinery. The commonest staining trouble is ‘iron stain’- the blue-black stain caused by the interaction of soluble iron corrosion products and the natural tannins in wood. Hardwoods are generally more susceptible than softwoods. Steel wool should not be used for smoothing wood surfaces. Iron stains, if not too severe, can be removed with oxalic acid. Heavy contamination with soluble iron corrosion products usually results in migration and conversion to rust deposits in the wood. Tannin-containing woods can also suffer from dark-brown stains produced by alkali, usually from a concrete or a detergent source. Conversely, fair-faced concrete can be similarly stained by the run-off water from unprotected hardwood surfaces. Some woods, e.g. afzelia, ayan and idigbo, give yellow contaminating dyes with alkaline detergents and should not be used in washrooms and kitchens. Tannin-containing woods also darken with ammonia, a process which is usefully employed in darkening oak furniture by ‘fuming’. Ammonia stains can originate from animal glue, amino-type adhesives and concrete additive sources, particularly where damp conditions exist. Pink stains occur naturally in some acidic woods, e.g. oak and Douglas fir, but similar anthocyanidin stains can be produced in the wood of sycamore, maple, walnut, agba and sapele by acid-catalysed adhesives. A variety of enzymatic stains can be produced at mildly elevated temperatures in the
18: 104
WOOD
steaming and kiln-seasoning operations of various woods. These are notably pink stains in ash, reddish stains in beech, and brown stains in hemlock. Undesirable brown streaks or bands following the grain, sometimes occur naturally in oak, teak and afrormosia, but exposure to light reduces the differences, in some cases to normal. With most woods, new surfaces first of all darken when exposed to light, particularly with teak and afrormosia, where any masking of portions of the surfaces of furniture can give blemishes. After full colour development, strong sunshine bleaches most woods to a common brown colour.
Summary of Degradation Problems Associated with Wood Supedicial Degradation
Cofourchange Caused by exposure to light, the effect is greatest with hardwoods and is important in furniture, panelling and joinery; any masked areas become blemishes. Strong sunshine bleaches most woods to a common brown colour. ‘Weathered surfaces’ The main factors involved are: (a) Degradation of the lignin-rich bond between fibres by ultraviolet component of stong sunshine. (6) Surface checking caused by the stresses and movement in the wetting and drying cycles. (c) In marine and chemical applications, surface defibration caused by repeated crystal growth of salts in the wetting and drying cycles. ( d ) Biological degradation especially by soft rot fungi.
Weathering results in gradually receding wood surfaces. Chemical stains These occur in tannin-rich woods, especially hardwoods. (a) ‘Iron stains’, i.e. blue-black stains caused by the interaction of tannin with soluble iron corrosion products. Soluble iron corrosion products may also be converted to rust stains. (b) Dark-brown stains with alkali; usually from a concrete or detergent source. (c) Ammonia stains, produced in damp conditions from adhesive and cement additive sources. Chemical attack Damage is superficial in the case of impermeable timbers, but is deep with permeable wood species. (a) Damage by industrial atmospheres, e.g. defibration by sulphur dioxide. (b) Damage to cooling tower packing and casings due to, for example, oversoftening and overchlorination of water. (c) Damaging chemicals to wooden vats, e.g. delignifying (defibrating) agents, chlorine, nitric acid, sulphites and sulphur dioxide, some of which are usefully employed in pulping processes.
18 :105
WOOD
‘Nail sickness’ Nail sickness is chemical decay associated with corroded metals in marine situations. Chemical degradation of wood by the products of metal corrosion is brought about by bad workmanship or maintenance, or unsuitable (permeable) timber species, all of which permit electrolyte and oxygen access which promotes corrosion. Chemical decay of wood by alkali occurs in cathodic areas (metal exposed; oxygen present). Softening and embrittlement of wood occurs in anodic areas (metal embedded; oxygen absent) caused by mineral acid from hydrolysis of soluble iron corrosion products. Biological Degradation
Wood-rotting fungi These fungi are active in situations where the threshold value of 20% moisture content in wood is exceeded. (a) Wet-rot fungi: attack on all sapwood and the heartwood of nondurable species. (b) Dry rot (Serpuia lacrymans): very active in buildings in damp situations. (c) Soft-rot fungi: fungus not apparent; usually surface degradation in wet or moist conditions. Wood boring insects (a) (b) (c) (d)
Termites: very destructive; warm and hot climates. Common woodworm (Anobium punctatum). House longhorn (Hylotrupes bajuius): very destructive; softwoods. Death-watch beetle (Xestobium rufoviliosum): oak; associated with decay. (e) Wood-boring weevils (Pentarthrum and Euophryum): damp situations; associated with decay. (f)Powder-post beetle (Lyctus spp.): sapwood of ring-porous hardwoods.
Marine borers
(a) Shipworm (Teredo spp.): a mollusc; tropical waters. (b) Gribble (Limnoria spp.): a crustacean; cooler waters. L. PINION REFERENCES 1. Handbook of Hardwoods. SO 7 B.R.E., H.M.S.O., London (1975) 2. A Handbook of softwoods, SO 39 B.R.E., H.M.S.O., London (1983) 3. The Strength Properties of Timber, SO 38 B.R.E., H.M.S.O., London (1983) 4. The Movement of Timber, Tech. Note 38, B.R.E., H.M.S.O., London (1975)
5. Timber Drying Manual, 2nd edn., BR 76, B.R.E., H.M.S.O., London (1986) 6 . Nomenclature of CommerciaI Timbers, including Sources of Supply, BS. 881 and 589 British Standards Institution, London (1991) 7. The Natural Durability Classificntion of Timber. 1981. Tech. Note 40. BRE. 8. Timbers: their Natural Durability and Resistance to Preservative Treatment, Digest 296, B.R.E., H.M.S.O., London (1985) 9. TimberPrcservation;3rd edn., Timber Research and Development Association and British Wood Preserving Association (1986)
18: 106
WOOD
10. B WPA Manuol, British Wood Preserving Association (1986) 11. Timber Pests and their Control, 2nd edn., Timber Research and Development Association and British Wood Preserving Association (1984) 12. Flame Retordont Treatmenfs for Timber, Wood Information 2/3, Sheet 3, Timber Research and Development Association (1988) 13. Choice of Gluesfor Wood, Digest 175, B.R.E., H.M.S.O., London (1975) 14. Murine Borers ond Mefhods of Preserving Timber against their Attack, Tech. Note 59, B.R.E., H.M.S.O., London (1975) 15. ‘The Degradation of Wood by Metal Fastenings and Fittings’, Spring symposium 1970, 16. 17.
18. 19.
20.
University of Southampton, Deteriorofion of Materials in o Murine Environmenf, Timberlab Paper 27-1970, B.R.E. Pinion, L. C., Unpublished work, Building Research Establishment Pinion, L. C., The ‘Furring’of White Serayo Decking, Internal Report, Building Research Establishment Pinion, L. C., ‘Woodfor Chemical Service’, Chemical Processing, 172-173, Sept (1972) Ross, F. F. and Wood, M.J., ThePreservafionof Woodin WaferCooling Towers, BWPA Annual Convention Record, pp.171-197 (1957) Pinion, L. C., Chemical Sfoining in Wood, Information Sheet 2/73, B.R.E., H.M.S.O., London (1973)
18.10 The Corrosion of Metals by Wood
Wood can cause corrosion of metals by direct contact and, in confined spaces, also by the emission of corrosive vapour. With rare exceptions, all woods are acid, and the principal corroding agent in both types of attack is volatile acetic acid. Acetylated polysaccharides form part of the structure of wood, the acetyl radical constituting some 2-5% by weight of the dry wood. Hydrolysis to free acetic acid occurs in the presence of moisture at a rate varying from one species to another: a wood of lower acetyl content can liberate acetic acid much faster under given conditions than another wood of higher content',*. Small quantities of formic, propionic and butyric acids are also formed3, but their effects can be neglected in comparison with those of acetic acid. There is a broad, but only a broad, correlation between the corrosivity of a wood and its acidity. The chemistry of acetyl linkage in wood and of its hydrolysis has been examined in some detail4. Contact corrosion may be reduced by the presence of natural inhibitors, such as tannins, in the wood, and will be promoted by sulphates and chlorides in it*, especially if mineral preserving processes involving these ions have been applied.
Influence of Moisture The influence of moisture is fundamental, as it is with other forms of corrosion. Long-term contact tests' with ponderosa pine, some treated with zinc chloride, in atmospheres at 30, 65 and 95% r.h. showed that at 30 and 65% r.h. plain wire nails were not very severely corroded even in zinc chlorideimpregnated wood. At 95% r.h. plain wire nails were severely corroded, though galvanised nails were attacked only by impregnated wood. Brass and aluminium were also attacked to some extent at 95% r.h. Some concurrent outdoor tests at Madison, Wisconsin, showed that the outdoor climate there was somewhat more severe than a 65% r.h. laboratory test. *Woods contain from 0 . 2 to 4% of mineral ash. This consists largely of calcium, potassium and magnesium as carbonate, phosphate, silicate and sulphate. Aluminium, iron, sodium and chloride are also present. Sulphate contributes 1 to 10% by weight, usually 2 to 4%, and chloride 0.1 to 5%.
18 : 107
18: 108
THE CORROSION OF METALS BY woon
These are useful quantitative results, but they will cause little surprise, since the user of wood will expect metals in contact with damp wood to corrode. A degree of corrosion acceptable on the nails and fastenings on the outside of a packing case is not, however, acceptable on metal components inside, which the box is supposed to be protecting. Vapour corrosion is also governed by relative humidity and can occur whenever the internal humidity exceeds a critical value, as may happen for a few hours in the cold of the night even in quite good storage conditions. The critical humidity for corrosive attack has been reported6*' as 75%.
Less and More Corrosive Woods Table 18.19 list woods whose aggressiveness by vapour corrosion has been quoted in a survey', together with typical pH values of the aqueous extracts of these woods reported in another investigation'. Table 18.19 Relative corrosivity of woods by vapour corrosion
Wood
CIassi2cation in Defence Guide-3A
Oak Sweet chestnut Steamed European beech Birch Douglas fir Gahoon Teak Western red cedar Parana pine Spruce Elm African mahogany Walnut Iroko Ramin Obeche
Most corrosive Most corrosive Moderately corossive Moderately corrosive Moderately corrosive Moderately corrosive Moderately corrosive Moderately corrosive Least corrosive Least corrosive Least corrosive Least corrosive Least corrosive Least corrosive Least corrosive Least corrosive
'
Typical pH values 3.35, 3.45, 3.85, 3 . 9 3.4, 3.45. 3.65 3-85, 4.2 4.85, 5.05, 5 .3 5 3.45, 3.55, 4.15, 4 , 2 4.2, 4-45, 5.05, 5 , 2 4.65, 5-45 3.45 5 . 2 to 8.8 4.0, 4.45 6.45, 7-15 5 . 1 , 5.4, 5 - 5 5 , 6.65 4.4, 4.55, 4.85, 5.2 5.4, 6.2, 7.25 5.25, 5-35 4-15, 6.75
While certain reservations must be kept in view (i.e. there is not necessarily a correlation between pH and corrosivity, and different samples of the same species of wood show a wide scatter of pH values, which might well be even wider if differences in duration of seasoning were taken into account), the results of vapour corrosion tests nevertheless indicate a general correlation between quoted pH values and the corrosiveness of wood vapours. It may reasonably be concluded that a strongly acid wood, pH less than 4.0, is potentially dangerous, and a less acid wood, pH more than 5.0, is likely to be relatively safe. Heat treatments of wood are dangerous, for although existing acid vapours may be expelled, further vapours are formed by accelerated hydrolysis.
THE CORROSION OF METALS BY WOOD
18: 109
Volatile acid hardeners such as hydrochloric acid and formaldehyde (which oxidises to formic acid) present in glues in plywood contribute to vapour corrosion, as can varnishes and paints*. Wood preservatives appear not to affect emission of corrosive vapours from wood, suggesting that the hydrolysis of acetyl polysaccharides is chemical, not biochemical. Some copper-base preservatives can give enough leachable copper ions to cause galvanic corrosion of other metals, notably aluminium and steel.
Metals Affected The metals most susceptible to corrosion by wood are steel, zinc, cadmium, magnesium alloy and lead. The susceptibility of zinc and cadmium is no argument against the galvanising or cadmium plating of steel, since these coatings much reduce the rate of corrosion of steel by contact with wood or wood vapours, although they will not give the high degree of protection which they provide in open exposure to marine or tropical atmospheres. Aluminium is relatively resistant Io. So also are copper, brass, tin and stainless steel, but these metals should not be used as thin coatings on mild steel as they promote rusting at any points of breakdown. Bimetallic corrosion between two different metals (see also Section 1.7) embedded in damp wood, e.g. in the hull of a boat, can occur in two ways ' I . If the metals are joined by a metallic conductor, then the formation of the cell metal A/damp wood/metal B will result in accelerated corrosion of the metal which has the more negative potential in this electrolyte. Savory and Packman I ' point out that even if two metals are not connected by a metallic path, and project from wood into sea-water, then two opposing cells are set up, the first as described above and the second metal A/sea-water/metal B. These cells are unlikely to have the same potential so that a net potential will exist and one metal will corrode preferentially and the other will tend to be protected. However, although this situation may occur in practice it is difficult to see how the explanation given by Savory and Packman is tenable. In action, accelerated corrosion can occur on an individual metal, by the action of a concentration cell metal A/damp woodlsea-waterlmetal A. Iron salts from rusting steel, e.g. a nail, have a strongly deleterious effect on wood, causing 'charring' and complete loss of strength.
Practical Conclusions Contact corrosion Nails and fastenings in many non-durable wooden articles exposed to damp will outlive the useful lives of the articles, and their corrosion is of no great importance. Corrosion is, however, important in tile
18: 110
THE CORROSION OF METALS BY WOOD
and batten nails in roofs, fences and other more permanent structures. Unprotected steel should never be used. Galvanised steel is much better, and brass, copper, the more corrosion-resistant alloys of aluminium, and stainless steel, are likely to give even longer service. Vspour corrosion The best way to pack articles made of metals susceptible to vapour corrosion is in boxes made of metal or of those plastics which do not themselves emit corrosive vapours'. If wood cannot be avoided, then the less corrosive kinds should be chosen. Dryness, good ventilation and the inclusion of water-vapour barriers should be sought. Other obvious measures are the avoidance of susceptible metals and the use of protective treatments and paints.
Recent Developments Differential Oxygen Cell Cbrrosion
In addition to the basic corrosion mechanism of attack by acetic acid, it is well established I* that differential oxygen concentration cells are set up along metals embedded in wood. The gap between a nail and the wood into which it is embedded resembles the ideal crevice or deep, narrow pit. It is expected, therefore, that the cathodic reaction (oxygen reduction) should take place on the exposed head and that metal dissolution should occur on the shank in the wood. Alkaline areas around corroding nail heads were demonstrated by Pinion I' using phenolphthalein indicator. Pinion also measured the corrosion current which flowed between a bare steel rod and a similar rod embedded in wood both placed in aerated sea-water. This is an extreme case of the differential oxygen concentration cell; a less severe example is that of galvanised steel nails used to fasten wooden planks to oak frames. In the case of oak planks the oxygen content of the environment around the nail is uniformly low (due to the impermeable nature of oak) and no significant corrosion is observed in moist conditions. Relatively rapid corrosion can occur in a mixed wood construction. Beech is more permeable than oak, thereby facilitating the creation of a differential oxygen cell. The cathodic process generates hydroxyl ions which diffuse into the beech and the anodic process introduces ferrous ions into the oak. Wood Degradation
Two mechanisms of wood degradation have been identified which can be linked directly to metallic corrosion. Nail sickness is a term which has long been used t o describe the process by which soft and spongy areas of wood form around corroding fasteners. Cathodically generated hydroxyl ions attack the hemicelluloses, lignin and even the cellulose components of wood if present at sufficiently high concentrations. Because hardwoods contain higher proportions of hemicelluloses than softwoods, hardwoods can be severely degraded by alkali, but because they tend to be less per-
THE CORROSION OF METALS BY WOOD
18:111
meable than softwoods there is less tendency for oxygen concentration cells to be set up. Degradation of wood is also observed adjacent to anodic corrosion sites. Some metal ions, notably Fez+,catalytically decompose the cellulose components of wood. This significantly reduces the wood’s fastener-holding ability. Moisture Content and Corrosion
As outlined previously the moisture content of wood largely determines whether or not corrosion can occur. The moisture content of wood is reported in terms of weight percent, although it can be related to the relative humidity of the environment with which it is in equilibrium. A value of ea. 15% moisture content is generally taken to be the lower limit for corrosion of any metals to occur in commonly used softwoods 14. At moisture contents of 20% or above (equivalent to 80070, r.h. at 15’C) wood is at risk from rot. This risk together with the enhanced risk of fastener corrosion in damp wood was highlighted in a recent survey” of roofspaces fitted with varying qualities of between-joist thermal insulation. Treated Wood
There are two main reasons for treating wood: to provide flame retarding properties and to guard against rot. Flame retardants are usually inorganic salts such as phosphates, sulphates and borates, the cation being metal or ammonium. Bare steel screws and nails are not recommended for use in such flame-proofed wood; for this service more corrosion-resistant materials such as brass, silicon bronze or stainless steel are used. Galvanised, sheradised, zinc- and cadmium-plated fasteners have inadequate corrosion resistance. Some fire retarding chemicals now contain corrosion inhibitors. Preservative treatments used to protect wood against insect attack and fungus-related decay l6 are of two general types. Organic based treatments such as creosote and pentachlorophenol are applied in solution in organic solvents. They contribute little to the ionic conductivity of the wood and it is generally accepted that their presence in wood does not exacerbate corrosion problems. Indeed, creosote apparently acts as a corrosion inhibitor for many Inorganic preservatives are often based on copper salts or oxides, sometimes in combination with arsenic and chromium in the form of oxides or oxyanions. Due to the presence of uncombined by-product salts and leachable copper ions there is a serious corrosion problem for several metals especially steel, zinc and aluminium embedded in wet or recently treated wood’8’19.Untreated wood is also very corrosive in these circumstances, but fungal decay should preclude its use at moisture contents above ca. 20%. Since much of the wood in common use today for joinery, including external window frames and roof trusses, is redwood, such as Pinus sylvestris, preservative treatment is very necessary. However, whilst it is known that little corrosion of fasteners can take place in dry wood (< 15% moisture
18: 112
THE CORROSION OF METALS BY WOOD
content) and that serious corrosion can be expected in wet wood (both treated and untreated) there is no firm evidence concerning the corrosion of metal fasteners in wood of normal (15-22070) moisture content m. Research is in progress but this is a difficult corrosion environment to study. The electrical resistivity of wood in this moisture-content range is high, preventing the application of linear polarisation methods. The results of one research programme2' indicate that copperchromium-arsenic treatments are not corrosive towards galvanised steel and stainless steel in the moisture content range 15-22%.
Corrosion Prevention in Wood Wood is an intrinsically corrosive medium which becomes more corrosive in certain circumstances. In dry conditions, where the moisture content cannot exceed 15070, any metal can probably be used without any precautions being necessary. At normal moisture levels (ca. 15-22%) there is a definite, though illdefined, corrosion hazard in certain woods for commonly used metals such as steel, whether bare or protected by zinc in any form. The choice of material to be used for fasteners is difficult and depends on additional factors including possible wood treatments. In untreated, and solvent-borne preservative-treated redwoods, galvanised steel is commonly used. In flameproofed wood more corrosion-resistant metals, such as brass or even stainless steels, are necessary. The choice of metals for use in wood treated with inorganic preservatives, particularly those based on copper salts, is even more difficult because the necessary information does not exist. Galvanised nails and nailplates are still used but some users are looking to materials such as stainless steel. The long-term behaviour of stainless steels is not known; there may be a crevice corrosion or pitting problem with some grades. Wood at moisture contents in excess of 20% is liable to decay, and above ca. 22% it is quite corrosive in both the treated and untreated conditions. Wet wood can be found in leaky roofs, in ground contact and in many outdoor locations. Coated steel, stainless steel and, less satisfactorily, galvanised steel fasteners are used. This appears to be largely because of the high cost of more suitable materials. H.G. COLE M. J . SCHOFIELD REFERENCES 1 . Packman, D. F., HohJorschung, 14, 178 (1960) 2. Ami, P. C. Cochrane, G. C. and Gray, J. D., J. Appl. Chem., 15, 305 (1965) 3. Ibid., 15, 463 (1965) 4. Cochrane, G. C., Gray, J. D. and Ami, P. C., Eiochem. J., 113, 243 and 253 (1969) 5. Baechler, R. H., American Wood Preservers' Association, (1949)' 6. Schikorr, G., Werk. u. Korr., 12, 1 (1961) 7. Clarke, S. G . and Longhurst, E. E., J. Appl. Chem., 11, 435 (1961)
Reference 5 can be seen at the Library of the Timber Research and Development Association, Hughenden Valley, High Wycombe, Bucks.
THE CORROSION OF METALS BY WOOD
18: 113
8. Rance, V. E. and Cole, H. G., Corrosion of Metals by Vapoursfrom Organic Materials, H.M.S.O., London (1958) and D G 3 A . Defence Guidefor the Prevention of Corrosion
of Cadmium and Zinc Coatings by Vapoursfrom Organic Materials, H.M.S.O., London ( 1966) 9. Gray, V.R., J. Inst. Wood.Sci., 1, 58 (1958)* 10. Farmer, R. H. and Porter, F. C., Metallurgica, 68, 161 (1963) 11. Savory, J. G. and Packman, D. F.. DSIR Forest Products Bulletin No. 31 entitled Prevention of Decay of Wood in Boats, H.M.S.0 (1954) 12. Baker, A. J., ‘Corrosion of Metal in Wood Products’. In Durability of Building Materials and Components, ASTM STP 691. edited by Sereda, P. J. and Litvan. G. G., pp. 981-993, American Society for Testing and Materials, Philadelphia (1980)
13. Pinion, L. C., The Degradation of Wood by Metal Fastenings and Fittings, Timberlab Paper No, 27. Forest Products Research Laboratory, Princess Risborough (1970) 14. Ormstad, E., Corrosion of Metals in contact with Pressure Treated Wood, Report No. 47, Norwegian Institute of Wood Working and Wood Technology, Blindern, Norway: (1973) 15. Sanders, C. H., ‘Thermal Insulation and Condensation’, Building Research Esfablishmenf News, No. 55 (1982) 16. Preservative Treatments for External Softwood Joinery Timber, Technical Note No.24 (revised version), Building Research Establishment, Princess Risborough (1982) 17. Henningsson, B. and Jermer, J.. Studies on Corrosionof Metallic Obiects in Contact with Preservative-Treated Wood in the Open, Report No. 144, Swedish Wood Preservation Institute, Stockholm (1982) 18. Smith, R. S.. Johnson, E. L. and Cserjesi, A. J., Corrosion of Zinc-CoafedNails Used with Preservative-Treated WesternRed Cedar Shakes in Service. Document No. IRG/WP/ 3197. International Research Group on Wood Preservation, Stockholm (1982) 19. Whitney, R. S., Timber Fasteners and Potential Problems of Corrosion. Paper presented at N.Z. Wood Preservers Association Annual Conference, Blenheim, New Zealand. 1977, Building Research Association of New Zealand, Wellington (1977) 20. Bailey, G. and Schofield, M. J., Corrosion of Metal Fastenings in CCA-treated timber-the corrosion science assessment. J. Inst. Wood Sei., 14- 18 ( 1984) 21, Cross, J.N.‘Evaluation of metal fasteners performance in CCA-treated timber’, 5th International Conference on Durability of Building Materials and Components, Brighton, (1990)
* Reference 9 can be seen at the Library of the Timber Research and Development Association, Hughenden Valley, High Wycombe, Bucks.
19
CORROSION TESTING, MONITORING AND INSPECTION
19.1
Corrosion Testing and Determination of Corrosion Rates Test procedures Laboratory corrosion tests Electrochemical measurements Polarisation resistance Tests for bimetallic corrosion Accelerated tests-electrolyte tests Accelerated tests-simulated environments Intergranular attack of Cr-Ni-Fe alloys Crevice corrosion and pitting Impingement tests Corrosion fatigue Cavitation-erosion Fretting corrosion Corrosion testing in liquid metals and fused salts Tests in plant Atmospheric tests Tests in natural waters Field tests in soil Corrosion testing of organic coatings 19.1A Appendix-Chemical and Electrochemical Methods for the Removal of Corrosion Products 19.1B Appendix-Standards for Corrosion Testing 19: 1
19:3 19:5 19:18 19:30 19:37 19:44 19:46 19:48 19:57 19:71 19:75 19:77 19:80 19183 19:84 19:92 19:94 19:99 19.102 19:104
19:119 19.122
19:2
19.2
19.3 19.4
CORROSION TESTING AND CORROSION RATES
The Potentiostat and its Application to Corrosion Studies
19.133
Corrosion Monitoring and Inspection Inspection of Paints and Painting Operations
19.154 19.179
19.1 Corrosion Testing and Determination of Corrosion Rates*
Corrosion tests provide the basis for the practical control of corrosion and therefore deserve a more exhaustive discussion than limitations of space will permit. A detailed description of all the procedures and devices that have been employed in corrosion studies in many countries will not be attempted. Instead, attention will be directed principally to underlying principles and to comments on the significance and limitations of the results of the test methods that are considered. Further details may be obtained from the references and from the comprehensive works by Champion' and Ailor2. Tests may be classified conveniently under three headings. I. Laboratory tests, in which conditions can be precisely defined and controlled. 2. Field tests (tests in real environments), in which replicate test samples of metals or alloys -referred to as test coupons or specimens-are exposed to the actual environmental conditions expected in service, e.g. the atmosphere, the ground, the sea, etc. 3. Service tests, in which the test specimens-which may often take the form of manufactured components are exposed to the particular conditions in which they are to be used, e.g. in process streams of chemical plant.
-
Laboratory tests, although often necessarily conducted under conditions that are not met in service, nevertheless have a number of advantages over the other types of tests. Because conditions can be controlled at will it is possible to identify the separate effects of a number of factors on the corrosion behaviour. These factors include the type and condition of the metal surface, the environmental composition, temperature and pressure, movement of the specimen relative to the environment, time of exposure and SO on. Laboratory tests, at least in principle, also enable comparisons to be made under identical conditions of the relative corrosion behaviour of *Abbreviationsused in the text for specifications are as follows: BS. British Standard; ASTM. American Society for Testing and Materials Standard; NACE, National Association of Corrosion Engineers Standard; ISO, InternationalStandardsOrganisation Standard. Further details of English-language specifications relevant to corrosion testing are given in Appendix l9.1B.
19: 3
19:4
CORROSION TESTING AND CORROSION RATES
different metals and alloys and different protective schemes, e.g. coatings, environmental treatments, etc. In many cases attempts will be made to accelerate the test to produce results in a shorter period of time than might otherwise be possible in field or service tests. Such acceleration is usually achieved by intensifying one or more of the controlling factors. Tests might be conducted at a higher temperature, with more corrosive media, with activation of the corrosion process by electrochemical methods, etc., with the object of enhancing the aggressivity of the test conditions. While accelerated test procedures are often used, the results should always be treated with careful consideration. It is not unknown for a protective system to fail to meet the requirements of an accelerated test, although showing satisfactory performance in normal conditions of use. Nevertheless, a number of such tests, particularly for atmospheric corrosion where rates of corrosion in real conditions are often low, are accepted and correlations have been established with real conditions. Field tests do not have the uncertainties attached to accelerated laboratory tests since there is no attempt to adjust the controlling environmental conditions. The chief problem is obtaining reproducible conditions from one test to another. This is particularly the case with tests in the atmosphere. While broad classes of terrestrial atmospheres have long been recognised, e.g. tropical, rural, urban, marine, etc., difficulties remain that are associated with variations within these classes. In the 1980s steps were taken within IS0 to rationalise the situation by producing a standard on classification of atmospheres (ISO/DIS 9223: 1989). Other standards are available that provide guidance on the mounting and disposition of specimens for field tests (ISOIDIS 8565:1987) and for the statistical treatment of results where large numbers of specimens are used (see Reference 2 and ASTM G 16: 1984). Service tests will be used (1) where the operating conditions cannot be successfully reproduced in laboratory tests, (2) where the environment does not occur naturally, (3) where real components, as opposed to test specimens, need appraisal, and (4) to confirm laboratory and/or field tests. Often, all three types of test will be used sequentially. An example might be in the development of a coating to protect suspension cables for use on a bridge in a coastal region. The test programme could involve salt spray testing of candidate treatments in the laboratory, followed by field trials of the most successful materials at a site similar in aggressivity to the location of the final product and eventual testing at the site with loadings and positioning matching those of the end use. It would be expected that the number of candidate materials would decrease through this sequence of tests. Irrespectiveof the method of test or the purpose for which it is made there are certain practical features which require attention and which will be necessary to achieve good reproducibility (by one operator) and repeatability (by different operators).
CORROSION TESTING AND CORROSION RATES
19:5
Test Procedures Preparation of Surface
When the test is to be made to predict the performance of a material in a particular service, the ideal procedure would be to have the surface of the test-pieces duplicate the surface of the material as it would be used. Here, however, a complication is presented by the fact that materials in service are commonly used in several forms with different conditions of surface. Where the number of materials to be compared is large, it will usually be impractical to test all the conditions of surface treatment of possible interest. The best practical procedure, then, is to choose some condition of surface more or less arbitrarily selected to allow the materials to perform near the upper limits of their ability. If all the materials to be tested are treated in this way, and preferably with uniform surface treatment, the results of the test will indicate the relative abilities of the different materials to resist the test environment when in a satisfactory condition of surface treatment. Then, if it should be considered prudent or desirable to do so, the most promising materials can be subjected to further tests in a variety of surface conditions so that any surface sensitivity can be detected. These remarks apply as well to the treatment of the surfaces of specimens to be used in tests in corrosion research projects, except here selection of a particular method of surface preparation is required so as to achieve reproducibility of results from test to test and amongst different investigators. Methods of preparing specimens are described in ASTM Gl:1988 and IS0 7539-1:1987.
The final step in surface preparation should ordinarily be a cleaning and degreasing treatment to remove any dirt, oil or grease that might interfere with the inception or distribution of corrosion. The simplest test of a satisfactory surface condition in this respect is for the specimens to be free from ‘water break’ when rinsed with water after cleaning. As a final treatment for specimens to be weighed prior to exposure, a dip in a mixture of water and acetone or of alcohol and ether will facilitate quick drying and avoid water-deposited films. Specimens to be stored prior to weighing should be placed in a desiccator which, in best practice, should be sealed without grease . In addition to the preparation of the principal surfaces of the specimen it is essential to machine or grind any cut or sheared edges, since these could become sites of preferential attack. As a general rule, edge effects should be kept to a minimum by using specimens in which the ratio of surface area to edge area is large. With flat specimens a disc is best from this point of view, but other shapes may be more convenient and acceptable in many practical instances. When mass loss is to be used as a measure of corrosion, precision will be improved by providing a large ratio of exposed area to mass, and thin flat specimens or fine wires have obvious advantages. For accuracy of weighing, it is usually necessary to restrict the dimensions of specimens to what can be accommodated on the common analytical balances. It must be borne in mind that where attack occurs in the form of a very few pits or in crevices under supports, the extent of this localised
’
19:6
CORROSION TESTING AND CORROSION RATES
attack may be determined by the total area of the test-piece, as it establishes the area of passive metal acting as a cathode to the few anodic areas. Thus, larger specimens, or the much larger surfaces that will often be involved in field or service tests, may give rise to much more severe localised attack under nominally the same conditions of exposure. In certain tests it is sometimes desirable to eliminate any effects of a mechanically achieved surface condition by chemical treatment or pickling of the surface prior to test. This may be done in a pickling solution; alternatively, the test itself may be interrupted after sufficient corrosion has occurred to remove the original surface, the specimen then being cleaned and reweighed and the test started over again. Wesley4found it to be desirable to pickle off about 0.008mm from the surface of specimens in acid to improve the reproducibility of the tests. With materials like the stainless steels, which may be either active or passive in a test environment, it is common practice to produce a particular initial level of passivity or activity by some special chemical treatment prior to exposure. With stainless steels this objective may be subsidiary to eliminating surface contamination, such as iron from processing tools, by treatment in a nitric acid solution which might also be expected to achieve substantial passivity incidental to the cleaning action (ASTM A380: 1988). In studies of the behaviour of materials that may be either active or passive in the test environment, there would seem to be a real advantage in starting with specimens in an activated state to see if they will become passive, and to ascertain how fast they are corroded if they remain active. If passivity should be achieved after such an activated start, the material can be considered to be more reliable in the test environment than would be the case if by chance it managed to retain an originally induced passivity for all, or most of, the test period. It may also be valuable to know how fast the metal will be corroded by the test medium if activity should persist. A procedure for testing previously activated specimens applied in studies of titanium was described by Bayer and Kachik’. Renshaw and Ferree6 also employed prior activation in their studies of the passivation characteristics of stainless steels. In many cases there will be a need to test metal-coated specimens, e.g. galvanised steel, tin-plated copper, nickel-plated zinc, etc. It will then be necessary to test specimens in the completely coated condition and also with the coating damaged so that the basis metal is exposed. The latter condition will provide the conditions for galvanic action between the coating and the basis metal. With sheet specimens this condition is most readily achieved by leaving cut edges exposed to the test environment. There may also be a need to consider the performance of pre-corroded test specimens. Apart from the fact that these conditions frequently arise in service it is also important for two other reasons. First, the presence of corrosion products or other surface layers may affect the access of constituents of the environment to the underlying metal surface-where the corrosion process occurs -and, second, in the case of alloys some pre-corrosion may lead to compositional changes in the surface. These factors should be taken into account in the application of any test method.
CORROSION TESTING AND CORROSION RATES
39:7
Marking Spechens for IdentHcatrbn
The simplest way to identify a specimen is to mark it with letters or numbers applied by stamping with a stencil or number punch. There is, of course, always the danger that the identification marks will be obliterated by corrosion. To guard against this, the several specimens in a test should be identified further by a record of their positions relative to each other or to their supporting device. Before specimens are taken from test their identity should be established in this manner unless inspection has already shown that the identification marks have been preserved. Other means of identification can be used on specimens exposed to atmospheric corrosion. For example, where stamped letters cannot be expected to persist, identification may be provided by holes drilled in particular positions, or by notching the edges of specimens in particular places both in accordance with a template. Where severe corrosion is encountered, the identification by drilled holes is more permanent than that achieved by notching edges. Other means of identification sometimes used satisfactorily involve chemical etching of the surface (not to be generally recommended), or the formation of letters or numbers by means of a vibrating stylus. The former is advantageous in studies of stress-corrosion cracking in which stamped symbols could lead to regions of stress concentration. Number of Replicate Specimens
Practical considerations usually limit the number of replicate specimens of each kind that can be exposed for each period of test. At least two are recommended for obvious reasons, and if a larger number can be accommodated in the programme more valuable results can be secured -especially when it is desired to establish the reality of small differences in performance. For statistical analysis, five replicates are desirable. Accounts of statistical planning and analysis are given by F. H. Haynie in Reference 2 and in ASTM G16:1984.
In providing replicates for tests to be subjected to statistical analysis, it is necessary in the original sampling of the materials to be tested to ensure that normal variations in those qualities of the metals that might affect the results are represented in each set of samples. In order to secure information as to changes in corrosion rates with time, as in atmospheric exposure tests, it is necessary to expose sufficient specimens to allow sets to be taken from test after at least three time intervals. For preliminary tests where the number of test specimens that can be accommodated is limited, yet numerous materials are of possible interest, it is in order to expose single specimens. This may be more advantageous than limiting the compositions that can be investigated by exposing half the number of materials in duplicate. Probably the greatest advantage in exposing two specimens of a material instead of only one is in detecting gross errors, as in weighing, etc. rather than in any considerable improvement in the precision of the observations that may be made as to the relative behaviours of the metals tested.
19: 8
CORROSION TESTING AND CORROSION RATES
Test of Fusion Welds
In view of the widespread use of welded joints in equipment and structures exposed to corrosion, it is necessary to know whether such welded joints will demonstrate satisfactory resistance to attack. It is not necessary to include welded specimens of all materials in a preliminary study to discover which of them have satisfactory resistance to a particular environment. Weld tests can be postponed until the preliminary selection has been made, or, alternatively, those materials expected in advance to be most likely to be resistant can be exposed in the welded condition so as to expedite the final answer. There are several reasons for testing welded specimens. The first is to discover whether the weld itself will resist corrosion satisfactorily. A second purpose is to discover whether the heat effects associated with welding operations have been in any way detrimental to the corrosion resistance of the parent metal near the weld-as in the case of the so-called ‘weld decay’ of stainless steels (see Section 9.5). Since weld deposits may themselves be subject to a weld thermal cycle, it is necessary to include cross welds in the design of welded specimens for such corrosion tests. Further, a weld will generally constitute a stress concentration and, unless post-weld heat treated, will contain residual contraction stress. Thus, testing may be necessary to generate appropriate corrosion fatigue or stress-corrosion cracking data. The latter may follow ASTM standard G58:1985, ‘Practice for the Preparation of Stress-Corrosion Test Specimens for Weldments’, for example. A weld bead included in a test-piece is, to some extent, peculiar to itself and may not necessarily be representative of nominally similar welds to be made by other welders under other circumstances. To this extent, results of tests on welds must be subject to some qualification in interpretation, having in mind that what will be disclosed principally will be the overall ability of the composition of the weld metal to resist the corrosive environment. In some cases, entrapped flux, craters, fissures, folds, surface oxides etc. may introduce localised corrosion that may or may not occur with all welds of the type studied (see Section 9.5). The heat effects of welding are to an even greater extent peculiar to the particular test specimens used. They will be influenced by the welding process, by the skill of the welder, by the thickness of the metal welded, by the type of joint made, and by the geometry and mass of the surrounding structure in so far as they affect heating and cooling rates and areas over which these effects apply. Consequently, what happens to a particular welded testpiece has questionable general significance, especially when the result shows no apparent damage to a material known to be susceptible to welding heat effects in corrosive environments. It should not be assumed that high heat input during welding will represent the worst case. For example, with ferritic steels which are sensitive to hydrogen embrittlement stemming from environmental action, low welding heat input can be most detrimental because of the formation of hardened structures in the weld area. Moreover, the possible effects of multipass welding with attendant repeated thermal cycles must be recognised in design of a suitable test-piece.
CORROSION TESTING AND CORROSION RATES
19:9
With some materials, there are specific heat treatments that are known to reproduce the worst effects of the heat of welding. It is recommended, therefore, that in tests made to qualify a material for a particular service environment, in addition to the exposure of welded test specimens in order to observe effects of welding heat, specimens should be included that have been given a controlled abusive or sensitising heat treatment. As an illustration, austenitic stainless steels may be held at 650-700" for 0.5-1 h, followed by testing for susceptibility to intercystalline attack as in IS0 3651-1 or -2: 1976. If such sensitised specimens remain as free from accelerated corrosion as the welded specimens do, then it can'be concluded that no detrimental effects of the heat of welding need be anticipated in the environment covered by the test. However, if the sensitised specimens are corroded while the welded specimens are not, there will remain the possibility that, under some conditions of welding, difficulties due to the effects of the welding heat may be encountered, and appropriate action or the substitution of more reliable compositions will be required. Having in mind the effect of time in damage of this sort, it will be necessary to make a careful examination of the corroded specimens to detect the first signs of attack before it can be concluded that none has occurred. In assessing the significance of attack observed on drastically sensitised specimens, it is necessary to keep in mind that no similar sensitisation may result from good welding practice. Likewise, it should not be concluded that attack in a specific test environment will occur to a similar extent, or at all, in some quite different environment. The evaluation af heat treatments or the effectiveness of stabilisation by limiting carbon content of these stainless steels can be determined by subjecting specimens to the ASTM standardised acid copper sulphate test or boiling nitric acid test (ASTM A262:1986; see also Sections 9.5 and 1.3). Duration of Exposure
The duration of a particular test is likely to be determined by practical factors such as the need for some information within a particular limit of time, or the nature of the operation or process with which the test is concerned. Tests are rarely run too long; however, this can happen, particularly in laboratory tests where the nature of the corrosive environment may be changed drastically by the exhaustion of some important constituent initially present in small concentration, or by the accumulation of reaction products that may either stifle or accelerate further attack. In either case, the corrosivity of the environment may be altered considerably. Gross errors may result from the assumption that the results apply to the original conditions of the test rather than to some uncertain and continually changing conditions that may exist during the course of too extended a test period. Rates of corrosion rarely remain constant with time. More often than not, rates of attack tend to diminish as a result of the formation of adherent insoluble corrosion products or other protective films originating in the environment (Fig. 19.1). Therefore, extrapolation of results of tests that are too short is more likely to indicate a lower resistance to attack than will
19:10
CORROSION TESTING AND CORROSION RATES
actually be observed over a prolonged period of exposure. To this extent, such extrapolation may be considered as conservative. At the worst, it may lead to the use of a more resistant material or a heavier section than is actually needed, or to the exclusion from consideration of some materials that might be much better than the short-time test results would indicate.
20
~
i
I
u)
I
I
I
I
i
i
i
I
l
Fig. 19.1 Rate vs. time curve showing diminishing rate of attack (after Proc. A S . T.M., 51, 500, 1951)
Test should be of sufficiently long duration to permit demonstration of the possible protective nature of films, etc. Lengthy tests would not normally be required for materials that experience severe corrosion, although there are cases where this is not so. For example, lead exposed to sulphuric acid corrodes at an extremely high rate initially, while building up a protective film, then the rate decreases considerably and further corrosion is negligible. Short tests on such materials would indicate a high corrosion rate and be completely misleading. Short-term tests can also give misleading results on alloys such as stainless steels that form passive films. With borderline conditions, a prolonged test may be needed to permit breakdown of the passive film and subsequent more rapid attack. Consequently, tests run for long periods are considerably more realistic than those conducted for short periods. Where anticipated corrosion rates are moderate or low, ASTM G31:1972 (R1985) suggests that the following equation be used to estimate a suitable test duration: Test duration (h) = SO/(Corrosion rate in mm/y) For example, where the corrosion rate is 0.25 mm/y, the test should run for at least 200 h. Due to the relatively slow rate of the atmospheric corrosion process, it is recommended in ISO/DIS 8565:1987 that test exposures be on a schedule
CORROSION TESTING AND CORROSION RATES
19: 1 1
such as 1, 2, 5 , 10 and 20 years, depending on the corrosion resistance of the metal or coating being tested. In some cases, total exposure of less than 2 years may be suitable. It should be noted that, especially for short-term testing, the results may depend on the season of initiation of exposure. Therefore, it is recommended that exposures are commenced in the period of highest corrosivity (usually autumn). Tests in waters and soils should ordinarily be allowed to run for extended periods in excess of 3 years, with removals of specimens in groups after different time intervals. A desirable schedule for any extended test in a natural environment is one in which the interval between successive removals is doubled each time. For example, the first removal would be after 1 year, the second after 3 years, and the third after 7 years, and so on. On the other hand, test periods should not be significantly longer than the process or exposure time of the end-use requirement. The testing of inhibitors for use if pickling or cleaning treatments should be of a period commensurate with the practical requirement which may be for only a few minutes. In any event, the actual duration of a test must be reported along with the results, so that those who may wish to make predictions based on them will have an accurate idea of the extent to which they may undertake any extrapolation or interpolation. Heat Treatment
Many alloys are subject to drastic changes in their response to the effects
of corrosive media when they have undergone certain heat treatments. The principal effect of interest is a loss of corrosion resistance to some degree. This commonly takes the form of concentration of corrosion in particular regions or along certain paths as in the vicinity of grain boundaries -where phases formed by heat treatment are most likely to be concentrated. In other instances, and particularly in castings, homogenising heat treatments may improve corrosion resistance by eliminating ‘coring’ or major differences in composition from point to point in the original dendritic cast structure. Heat treatments that eliminate internal stresses are obviously helpful in connection with stress corrosion, but may induce structural changes that can affect corrosion in other forms. Heat treatments involving heating to a temperature high enough to take harmful phases into solution, followed by cooling (e.g. by quenching) at a rate fast enough to hold such phases in solution, may also be helpful in improving resistance to corrosion by avoiding attack that would otherwise be associated with a precipitated phase or compound. Obviously, some knowledge of the possible effects of such heat treatments is essential for a complete understanding of the corrosion behaviour of an alloy. Studies along this line should follow upon the initial selection of a material considered to be possibly useful for a particular service. Thus, it should be tested in the condition ,most likely to resist corrosion. Sometimes the obtaining of this condition may require annealing at a temperature sufficiently high to take any possibly harmful phases or compounds into solution followed by quenching to prevent them precipitating. Following this preliminary selection, it wouId be prudent to carry out additional
-
19: 12
CORROSION TESTING AND CORROSION RATES
corrosion tests on specimens that have been deliberately subjected to any possibly detrimental heat treatments to which the material may be subjected in processing, fabrication, or use. Heat treatment may also affect the extent and distribution of internal stresses. These may be eliminated by appropriate annealing treatments which can remove susceptibilityto stress-corrosion cracking. This must be explored in any studies of the performance of materials in environments where stresscorrosion cracking is a hazard. In particular cases, stress-relief annealing treatments may result in the appearance of new phases which, while eliminating the stress-corrosion effects, will induce another type of path of attack. This possibility must be kept in mind in assessing the overall benefits of heat treatments applied primarily for stress relief. In other instances, heat treatments involving quenching, tempering, or holding at some temperature to precipitate an age-hardening compound are employed to secure some desired level of hardness or other mechanical properties. It is obviously necessary to explore what effects such heat treatments may have on the corrosion resistance of the material in the condition, or conditions, of heat treatment in which it is to be used.
Stress Effects Techniques for studying effects of stress on corrosion are covered in some detail elsewhere in this work (see Section 8.10). So far as attention to stress effects in a general materials-selection programme is concerned, it is suggested that this should be a supplement to the initial selection of processing materials by exposing specimens in what approaches their best condition to resist corrosion, that is, free from stresses. Materials found to be worthy of further consideration in this way can be subjected to tests for stress effects. Where it is desired to discover whether severe internal stresses can be satisfactorily accepted, it will suffice to expose specimens in such a condition of stress. For example, a crucial test can be made by using a specimen in the form of a heavily cold-drawn tube in the as-drawn condition flattened on one end to introduce some additional multiaxial stresses. If such a severely coldworked specimen suffers no stress-corrosion cracking in a test, then the danger of this occurring on any structure of that metal in the environment represented by the test is extremely remote.
Appraisal of Damage
There are many ways of determining the extent or progress of corrosion. The choice may be determined either by convenience or on the basis of some special interest in a particular result of corrosion or in a particular stage of a corrosion process. Probably the most frequently made observation is the change in mass of a test-piece. This may take the form of a mass gain or a mass loss. Mass-gain determinations are most common in studies of the extent and rate of oxidation or scaling at elevated temperatures (see ASTM G54:1984).
CORROSION TESTING AND CORROSION RATES
19: 13
Very precise studies of this sort can be made by continuous observation of mass changes, as in the use of micro-balances, such as used and described by Gulbransen’. Such data have quantitative significance only when the exact composition (metal content) of the scale is known or can be determined and when there has been no loss of loose scale during or after the test. Fundamental studies of the initial stages of corrosion when films of a few monolayers are formed have made use of an ellipsometer to follow the increase of thickness of corrosion products without disturbing the specimen *. In most other cases, data on gains in mass due to the accumulation of corrosion products have little quantitative significancesince there is usually a question as to how much of the corroded metal is represented in the corrosion products that remain attached to the specimen at a particular time. There are also uncertainties as to the chemical composition of corrosion products, which may consist of mixtures of several compounds with varying amounts of combined or uncombined water, depending on the humidity of the atmosphere at the time. For these reasons, it is much better to determine the amount of metal removed by corrosion by weighing what is left after removal of all adherent corrosion products by some method that will not cause further attack in the process, or by making a proper correction for losses in the cleaning process. (Removal of corrosion products is dealt with in detail in Appendix 19.1A.) Subtracting this final mass from the original mass will give the loss in mass during the test. Since the extent of this loss in mass will be influenced by the area exposed, as well as by the duration of exposure, it is desirable, in order to facilitate comparisons between different tests and different specimens, to report the loss in mass in a unit which includes both area and time. A most commonly used unit of this sort is milligrams weight loss per square decimetre of exposed surface per day (24 h) (mdd). The unit gm-2 a- (gma) is sometimes used in atmospheric corrosion tests (see ISO/DIS 9226:1989) where ‘a’ represents ‘year’. It must be recognised that these units embody two assumptions that may not in fact be true. The first is that corrosion has occurred at a constant rate throughout the test period. This is rarely the case, since most rates of attack tend to diminish with time. But if the duration of the test and the actual loss in mass are also reported, the user of the data can take this into account. The second probable error in a mass loss/unit area unit is that it implies that corrosion has proceeded uniformly over the whole surface. These units, therefore, will give the wrong impression as to the probable depth of attack if corrosion has occurred at only a few spots on the surface of the specimen. Obviously, the mdd and gma units have limited significance when corrosion has taken the form of scattered pits or has been confined to the crevices where the specimen was supported. This should be covered by appended notes describing the nature and location of the corrosion represented and should be supplemented by data on the actual depths of the pitting or crevice attack. Here, again, the report should include data on the actual mass losses and duration of exposure. Expression of mass loss in terms of a percentage of the original mass of a test-piece is usually meaningless except for comparing specimens of the same size and shape, since it does not take into account the important relationship between surface and mass.
’
19: 14
CORROSION TESTING AND CORROSION RATES
As indicated, it is necessary to measure and report the depths of any pitting or other localised corrosion, such as in crevices, that may have occurred. It is also useful to provide information on the frequency of occurrence, distribution, and shape of pits, since these features are likely to have practical significance. Champion' has produced charts in which the number of pits/ unit area, the size of pits, the depth of pitting, cracking and general attack can each be rated by the numbers 1 to 7. Where the number of pits is very large, it is obviously impracticable to measure the depths of all of them. Consequently, the practice has developed of choosing 10 of the deepest pits and reporting their average depth and that of the deepest of them. All surfaces of the specimen should be examined in selecting the 10 deepest pits. There are several ways of measuring pit depths, but in all cases these measurements are facilitated if corrosion products are first removed (see Appendix 19. IA). If the pits are large enough, their depths may be measured directly with a pointed micrometer or with an indicating needle-point depth gauge. Otherwise, they may be measured optically with a microscope by focusing in turn on the surface of the specimen and, on the bottom of the pit using a calibrated wheel on the fine-focus adjustment rack for this focusing operation. In some instances the small dimensions or shapes of pits may require metallographic examination of a cross-section for a precise measurement of depth. Such metallographic examination may also be useful in detecting an association of pitting with a structural feature of the metal. Since it is often difficult to visualise the extent of attack in terms of depth from such mass-loss units as mdd, it is common practice to convert these mdd figures into others to indicate depth of penetration, Le. inches per year (ipy), mils or mm y-'. Such calculations suffer from the same defects as the mdd figuresin that they take into account neither changes in corrosion rates with time nor non-uniform distribution of corrosion. However, since such conversions are often made it is desirable for the initial reporter of the test results to make the calculations accurately and to report corrosion rates in both mdd and mm y-' or similar units. The basic formula for making such calculation is: mdd x
0-0365 ~
P
- mm y-'
where p is the density of the metal in g ~ m - Some ~ . values showing the relationship between mdd, and ipy and mm y - ' are given in Table 19.1. Losses in mass will also not disclose the extent of deterioration that may result from the distribution of a very small amount of attack concentrated along grain boundaries or in transgranular paths (as in some cases of stresscorrosion cracking). In such instances, an apparently trivial or even undetectable loss in mass may be associated with practically complete loss of the strength or ductility of the corroded metal. Where this may be suspected, or in any doubtful cases, the mass-loss determinations must be supplemented by other means of detecting this sort of damage, including simple bend tests followed by visual or metallographic examination to disclose surface cracking, quantitative tension tests, and direct metallographic examination of cross-sections. Changes in electrical resistance have been used as a measure
19:15
CORROSION TESTING AND CORROSION RATES
Tabk 19.1 Relationship between corrosion rate in mg dm-2 d - ’ (mdd) and penetration in iny-’ (ipy) and mmy-’
Density (g cm-?
Material ~
Aluminium 2 s Ambrac (Cu-6.5Si) Brass (admiralty) Brass (red) Brass (yellow) Bronze, phosphor (5% Sn) Bronze (silicon) Bronze. cast (85-5-5-5) Cast iron copper Cu-30Ni Hastelloy A Hastelloy B Hastelloy C Inconel 600 Iron-silicon alloy Lead (chemical) Monel Nickel Nickel silver (I8Vo Ni) Ni-resist Silver Stainless steel Type 304 Stainless steel Type 430 Steel (mild) Tin Zinc
Penetration equivalenf to a corrosion rate of 1 mdd ipy x 10’
mm y-l x 10’
~~~
2.72 8.86 8-54 8.75 8.47 8-86 8-54 8.70 7-20 8-92 8.95 8-80 9-24 8.94 8.42 7.0 11.35 8.84 8.89 8.75 7-48 10-50
7.92 7-61 7-86 7-29 7-15
0.528 0.162 0.168 0.164 0.170 0.162 0.168 0.165 0.200 0.161
0-161 0.163 0.155 0.161 0.171 0,205 0.127 0.163 0.162 0.164 0.192 0.137 0.181 0.189 0.183 0.197 0.201
1-342 0.412 0.427 0.416 0.432 0-412 0-427 0-419 0-508
0.409 0.409 0.414 0.394 0.409 0.434 0.521 0.323 0.414 0.412 0.417 0.488 0-348 0-462 0.480 0-465 0 -500 0.510
of intergranular attackg. Because of the nature of such resistance determinationslOs”they have been more useful for comparing specimens of a particular kind and size than as a basis for quantitative expression of rates of attack. The characteristic mode of corrosion of some alloys may be the formation as a corrosion product of a redeposited layer of one of the alloy constituents, as in the case of the brasses that dezincify, or of a residue of one of the components, as in the case of the graphitic corrosion of cast iron. Particularly in the case of the dezincified brass, the adherent copper is not likely to be removed with the other corrosion products, and therefore the mass-loss determination will not disclose the total amount of brass that has been corroded. This is especially important because the copper layer has very little strength and ductility and the extent of weakening of the alloy will not be indicated by the mass loss. In these cases, also, the mass-loss determinations must be supplemented by, or replaced by, mechanical tests or metallographic examination, or both, to reveal the true extent of damage by corrosion. Difficulties in obtaining accurate mass losses of heavily graphitised specimens have been reported 12. Whenever changes in mechanical properties, such as performance in tension tests, fatigue tests, and impact tests, are to be used as a measure of
19: 16
CORROSION TESTING AND CORROSION RATES
corrosion damage, it is obviously necessary to provide test data on the relevant properties of the uncorroded metal. When tests extend over long periods during which the alloys being tested may be subject to changes in mechanical properties due to ageing effects, entirely aside from corrosion, it will be necessary to provide sets of specimens that may be subjected to similar ageing in a non-corrosive environment so that by direct comparison with corroded specimens of the same age the changes due to corrosion can be separated from those due to ageing. Preferably the control specimens should be stored so that they will be subjected to the same thermal experience as the specimens undergoing corrosion. This is usually very difficult to accomplish while maintaining the control specimens completely protected from corrosion. In calculating the strength properties of the corroded specimens and comparing them with those of the uncorroded control specimens after appropriate mechanical tests, it will be necessary to take into account the actual area of the cross-section of the corroded metal and report results on this basis instead of, or as well as, on the basis of the original cross-section prior to exposure such as would be represented by the uncorroded control specimens. In view of possible or probable variations in mechanical properties among different specimens of the same metal cut from different sheets or other pieces, or even from different sections of the same sheet or piece, it is necessary to pay careful attention to the initial sampling of stock to be used for control, as well as exposure, specimens. An interesting case in which several of these considerations were involved was provided by the long-time atmospheric exposure tests of non-ferrous metals carried out by Subcommittee VI of ASTM Committee B-3 on Corrosion of Non-Ferrous Metals and alloy^'^ in which changes in tensile properties were used as one of the means of measuring the extent of corrosion. Tests carried out for particular purposes may make use of other special means to measure the progress of corrosion. For example, changes in the reflectivity of polished have been used as a sensitive means of following changes in the very early stages of corrosion in laboratory studies. A similar technique has been applied on a practical scale in connection with the direct evaluation of the relative merits of different alloys as used for mirrors in searchlights exposed to corrosive natural atmospheres. KrugerI6, while at the then U.S. National Bureau of Standards, used an ellipsometer to follow the growth of very thin corrosion-product films (oxides) during the initial stages of corrosion. This requires knowledge of the composition of the oxide and its refractive index. An outline of modern physical techniques for studying the nature and kinetics of the growth of oxide films and scales is given in Table 1.6, Section 1.2. In some cases, the principal interest is in the possibility of undesired contamination or other alteration of an environment rather than in the rate of destruction of the metals being tested. Here, in addition to paying attention to the usual factors that influence rates of corrosion, it is necessary also to consider the ratio of the area of the test specimen to the volume or mass of test solution, and the time of contact. All of these factors may be quite different in a test from what would obtain in a practical case, and any dis-
CORROSION TESTING AND CORROSION RATES
19: 17
tortions of the test in these ways must be taken into account in planning the test and in interpreting the results. In cases such as this, the possible contamination of the solution by corrosion products may be estimated from the loss in mass of the test specimen. This, however, does not make any distinction between soluble and insoluble corrosion products, which may have different effects and which can be studied best by chemical analysis of the test solution and the materials filtered from it. Similarly, chemical analysis may be required to detect any other changes in the composition of the test solution that may be of interest. Particularly in theoretical studies of corrosion processes, it has been useful to measure the progress of corrosion in terms of the rate or extent of consumption of oxygen in the corrosion reactions. This technique has been very useful in following the progress of wet corrosion or of oxidation in its initial stages ”. Somewhat along the same lines is the measurement of the volume of hydrogen generated as corrosion proceeds ’** 19. This technique has been used not only in theoretical studies, but as a means of comparing some corrosion-resisting characteristics of different lots of steel which seem to affect their behaviour when used as a base metal for tin cans”22. The polarograph has been found to be a very useful tool for following the progress of corrosion, especially in its early stages, by measuring minute changes in the composition of the solution, as in the consumption of some constituent, such as oxygen, or by the accumulation of metal salts or other reaction products, such as hydrogen peroxidez3. An electrical resistance methods which directly measures loss of metal from a probe installed in the corrosive system under study is described in Section 19.3. It is reported that corrosion equivalent to a thickness loss of as little as 2.5 x lO-’cm can be dete~ted”*~’ . This technique is most useful as a means of monitoring steps taken to reduce corrosion, e.g. by inhibitors, or to detect changes in the corrosivity of process streams. Electrical methods of determining corrosion rates are considered subsequently. Temperature effects may also be used in test methods and notably for assessing the effects of inhibitors in acid solutions. The technique is based on that first proposed by MyliusZ6which records the temperature-time behaviour associated with the exothermic reaction resulting from the initial contact of a metal with a corrosive acid solution. The effectiveness of inhibitors may then be determined from their effects on the temperaturetime behaviour*’. Removal of Corrosion Products
An ideal method for removing corrosion products would be one that would remove them completely without causing any further corrosion or other deterioration of a test specimen in the process. Procedures that achieve this ideal or approach it very closely have been developed for many of the common alloys. Steels, for example, have been cleaned in such a manner that the loss due to cleaning is about 0.01%.
19: 18
CORROSION TESTING AND CORROSION RATES
There are numerous satisfactory methods of cleaning corroded specimens, but whatever the method its effect in removing base metal should be determined for each materia128-30 taking into account possible differences between the behaviour of ‘as-new’ and corroded base metal (see Appendix 19.1A). The various methods may be classified as follows: 1. Mechanical treatment.
(a) Scrubbing with bristle brush. (b) Scraping. (c) Wire brushing. ( d ) Grit, shot sand blasting. 2. Chemical treatments. (a) Organic solvents. (6) Chemical reagents. 3. Electrolytic treatments as cathode in the following. (a) Sulphuric acid usually inhibited. (b) Citric acid. (c) Potassium cyanide. ( d ) Caustic soda. Further details of removing corrosion products are given in Appendix 19.1A.
Laboratory Corrosion Tests Total-immersion Tests
The total-immersion corrosion test is most adaptable to rigorous control of the important factors that influence results. This control may be achieved in different ways and it is unnecessary and undesirable to seek a standardised method or apparatus for universal use. All that is required is a recognition of what is essential, as covered, for example, by the ASTM procedure G3 1: 1972 (R1985). This represents a code of minimum requirements without insisting on the use of any particular kind of apparatus or specifying the exact conditions of aeration, temperature or velocity to be used. Since different metals respond differently to effects of aeration, temperature and velocity, the setting up of standard test conditions in terms of these factors would be inappropriate. Depending on the environment, such standardised testing conditions would favour maximum corrosion of some materials and minimum corrosion of others and thus lead to gross errors in indicating any general order of merit applicable under conditions differing from those of a standardised test. In some instances it may be possible, though it is usually very difficult, to undertake laboratory corrosion tests under conditions that will be the same as those encountered in some practical application, and thus to secure some directly applicable data. More often, the conditions of service are so variable or so difficult to appraise accurately and duplicate in the laboratory that it is impractical and probably unwise to attempt to do so. A better procedure is to examine the individual effects of the several controlling factors by varying them one at a timeso as to provide a picture of their influence on the
CORROSION TESTING AND CORROSION RATES
19: 19
behaviour of the materials of interest in the corrosive medium being investigated. This information will be helpful in deciding whether the conditions of a particular use are favourable or unfavourable to the materials being considered. It will also serve as a guide to account for behaviour in service and to suggest changes in the operating conditions that may be expected to reduce corrosion of a material being used. In many cases, and particularly in aqueous solution, the most important controlling factors will be solution composition, temperature, aeration and velocity.
Solution composition When designing tests to determine the effects of solution composition on corrosion it is important to understand the nature of the controlling process. In the case of most metals and alloys the ratedetermining step will be the rate of supply of cathodic depolariser to the metal surface. This is particularly true in neutral solutions where corrosion will often be under oxygen diffusion control. Thus, tests in stagnant (unstirred, quiescent) conditions may be inappropriate since the effects of solution composition will be insignificant compared with the oxygen diffusion effect. In stagnant conditions corrosion rates of mild steel in, for example, sodium chloride, sodium sulphate and other salt solutions will be effectively the same over a range of concentrations and equal to that in pure water*. The effects of anion type and concentration begin to be shown only with movement of the solution, i.e. when oxygen access to the metal surface is facilitated to the point where it may be no longer rate controlling. Specific effects of anions in stagnant solutions will, however, be found when the anion has oxidising properties as in the case of nitrate. Care must therefore be taken in designing tests to study the effects of solution composition since different results will be obtained depending on the degree of aeration and/or movement of the solution. Variations in solution composition throughout a test should be monitored and, if appropriate, corrected. Variations may occur as a result of reactions of one or more of the constituents of the solution with the test specimen, the atmosphere or the test vessel. Thus, it is important that the composition of the testing solution is what it is supposed to be. Carefully made-up solutions of pure chemicals may not act in the same way as nominally similar solutions encountered in practice, which may, and usually do, contain other compounds or impurities that may have major effects on corrosion. This applies particularly to ‘artificial’ sea-water, which is usually less corrosive than natural sea-water. This subject is discussed in detail in a Special Technical Publication of ASTM 31, and tests with natural, transported and artificial sea-water have been described3’. Suspected impurities may be added to the pure solutions in appropriate concentrations or, better still, the testing solutions may be taken directly from plant processes whenever this is practical. It should also be pointed out than in exploring the effects of the concentration of a particular acid or other chemical on its corrosivity, it is necessary to cover the full possible variation of concentrations thoroughly, since it *An effect apparently first noticed by Heyn, E. and Bauer, O., Mitt. K. Mater. Pruf Amt BerlDahlem., 28, 62 (1910)
19 :20
CORROSION TESTING AND CORROSION RATES
frequently happens that particular ranges of concentration are especially corrosive to some metals. This extends to the highest degrees of concentration where sometimesthe complete elimination of water may increase corrosion a great deal-as in the case of aluminium in acetic acid. On the other hand, the presence of a trace of water may make other chemicals much more corrosive -as in the case of bromine and other halogens. It should be noted, also, that exposing a specimen to a solution of some chemical while it is being concentrated by evaporation practically to dryness will not suffice to explore the effects of the complete range of concentration, simply because the period in which any particular concentration range exists is not likely to be long enough to permit any especially corrosive effects to be detected in the overall result. Temperature control Of the factors mentioned, temperature is probably
the easiest to control; this can be accomplished by means of a thermostat or by operating at the boiling point of the testing solution with an appropriate reflux condenser to maintain the solution at a constant concentration. Control to 1°C is not hard to accomplish. The need for temperature cycling should be taken into account when designing or conducting tests. The nature of the test vessel should be considered for tests in aqueous solutions at temperatures above about 60°C since soluble constituents of the test vessel material can inhibit or accelerate the corrosion process. An inhibiting effect of soluble species from glass, notably silica, on the behaviour of steel in hot water has been shown”. Pure quartz or polymeric materials are often more appropriate for test vessel construction.
*
Aeration Control of aeration is more difficult. Aeration here means the amount of oxygen supplied either as such or, more commonly, in air. In some situations, it may not require a large amount of air bubbled through a solution to accommodate even a modest rate of corrosion of a small testpiece. Figure 19.2 shows the relationship between the rate of supply of air used for aeration and the rate of corrosion of Monel alloy in 5% sulphuric acid. To facilitate rapid solution of oxygen from air bubbles it is desirable to make these as small as possible, e.g. by having the air enter through a porous thimble or sintered glass disc. Much less satisfactory results are secured by simply letting air escape into the solution from a tube drawn to a fine tip. It is also undesirable to permit air bubbles to impinge directly on the testpieces. This can be avoided by placing the aerator inside a chimney. When it is desired to study effects of various degrees of aeration, it is better to do this by varying the oxygen content of the saturating gas (e.g. by using controlled mixtures of oxygen and nitrogen) introduced at a constant and adequate rate than by attempting to vary the rate of admission of a gas (such as air) of constant composition. This extends as well to zero aeration, which can best be accomplished by saturating the test solution with de-oxygenated nitrogen or other inert gas. It is unwise to assume that, because no air is purposely added, oxygen has been excluded from a test solution in a vessel open to the air. Such a practice provides a low oxygen availability that is not sufficiently under control to ensure reproducible results ’.
CORROSION TESTING AND CORROSION RATES ...
...
,
19: 21
T
I
I
II
1 I 280 I 160 200 2 Air (rnl/rnin)
320
Fig. 19.2 Effect of rate of supply of air used for aeration on corrosion of Monel alloy in 5% sulphuric acid
Velocity The precise control of velocity and the study of effects of velocity on corrosion are extremely difficult, especially when high velocities are involved. A major problem is to prevent, or to take into account properly, the tendency of a liquid to follow the motion of a specimen moved through it, e.g. by rotation at high velocity. This can be controlled to some extent by proper baffling, but uncertainties as to the true velocity remain-as they do also when the test liquid is made to pass at some calculated velocity over a stationary test-piece or through a test-piece in the form of a tube or pipe”. Velocity effects can be achieved either by having the test-piece move through a presumably stationary liquid or by having a moving liquid come into contact with a stationary test-piece. Occasionally tests may involve both types of exposure. Details of test procedures are given in NACE TM 0270-70 Method of Conducting Controlled Velocity Laboratory Corrosion Tests. The achievement of zero velocity in a test set-up is about as difficult as the accurate control of some high velocity. It is a common mistake to assume that by not making any attempt to move either the specimen or the testing liquid, the relative velocity between them will be zero. This neglects such effects as convection currents and the agitation due to the effects of corrosion products streaming under the influence of gravity. The most common difficulty arising from this situation is that these uncontrolled effects in tests made under presumably quiet or stagnant conditions make it very difficult to secure reproducible results from test to test. Therefore, even when there is no practical interest in the effects of any appreciable velocity, it is desirable to provide for some controlled movement of either the specimens or the solution at some velocity such as 7.5 cm s-’, readily achieved with a vertical circular-path machine. Equipment of this type in which the specimens are moved in a vertical circular path with all portions of the surface of a specimen moving at the same speed has been used where such moderate test velocities are required.
19:22
CORROSION TESTING AND CORROSION RATES
Statistical analysis of data from tests with an apparatus by Wesley4 has demonstrated satisfactory reproducibility of results not only among specimens in a particular test, but also from test to test undertaken at different times. Where effects of much higher velocities are to be studied, various devices have been used to move test-pieces through the testing solution at high velocity. One procedure is to use test specimens in the form of discs which can be rotated at the desired speed while either wholly or partly immersed in the testing solution, and Freeman and Tracy described a device of this sort in a contribution to the ASTM Symposium on Corrosion Testing Proceduresz8. With their apparatus the specimen discs were mounted on horizontal shafts and were partially immersed in the testing solution. A similar method of test was used at the International Nickel Company’s Corrosion Laboratory at North Carolina. The specimen discs are mounted on insulated vertical spindles and submerged in sea-water, which is supplied continuously to the tank in which the specimens are immersed. The rnaximum peripheral speed of the spinning disc is about 760cms-’, and the characteristic pattern of attack is shown in Fig. 19.3a. Studies of variation of depth of attack with velocity indicate that at low velocities (up to about 450 cm s-’) alloys such as Admiralty brass, Cu-1ONi and cupro-nickel alloys containing iron maintain their protective film with a consequent small and similar depth of attack for the different alloys. At higher velocities the rate increases due to breakdown of the film. Tests of this sort indicate a sort of critical velocity for each material that marks the boundary between the maintenance and loss of protective films. These apparently ‘critical’velocities must be considered as relative and only applicable to the conditions of test in which they are measured. Because of the complex effects associated with the differences in velocity from point to point on such rotating specimens, the apparent ‘critical velocity’ obtained in a given test may be quite different from what might be indicated by another test in which the same velocity is achieved in some other way-as by moving the liquid past a stationary specimen at a uniform velocity from point to point. The apparent ‘critical velocity’ indicated by this latter method of test will likely be higher for many materials than that shown by the spinning disc test. Thus, the establishment of critical velocities by a particular method of test will afford only qualitative data regarding the relative abilities of a number of materials to resist the destructive effects of high velocity. Furthermore, the critical velocity at which severe attack commences has been found to depend on the diameter of the disc so that no quantitative significance can be attached to it. This restriction extends as well to tests with iron discs, where attack is concentrated at the centre of the disc rather than at the periphery, irrespective of its diameter.* Somewhat similar tests may be made by attaching specimens to discs that can be rotated at some desired velocity in the testing medium. A machine of this sort that is used extensively in studying corrosion of metals by sea* Small variations in solution composition may also affect the value of any critical velocity. In laboratory tests using recirculating artificial sea-water the presence of dissolved copper from copper alloy test-pieces has been shown to affect the value of the critical velocity for such materiaIs35.
(a)
,
Fig. 19.3
CORROSION TESTING AND CORROSION RATES
(0) Distribution of
19: 23
corrosionon surfaces of rotating disc specimens and (b)assembly of specimens attached to rotating discs
19 :24
CORROSION TESTING AND CORROSION RATES
water at high velocity was developed by the staff of the US Naval Engineering Experiment Station at Annapolis, Maryland36. A typical assembly of discs and specimens is shown in Fig. 19.3b. The action of the rotating discs with their attached specimens causes violent agitation of the liquid in the tank. Depending on the height of liquid above the specimens, as determined by the location of the overflow pipe, there may be considerable whipping of air bubbles into the liquid or none at all, as desired. The heat of agitation causes the temperature to rise. This may be controlled readily by adjusting the amount of fresh cold liquid, for example sea-water, allowed to pass into the tank and out through the overflow pipe. It is not difficult to hold the temperature within 1-2OC of the desired value. The use of rotating discs to carry test specimens has been extended to studies of protective coatings in what are considered to be ‘accelerated’tests of such coatings for service underwater”. Velocity effects involving a high differential in velocity between adjacent areas are achieved simply by exposing a test specimen to the action of a submerged jet. This sort of test has been very popular and very useful in studying impingement attack or erosion of condenser tube alloys. It was introduced originally by Bengough and May” and later modifications were described subsequently by May and S t a ~ p o o l e The ~ ~ . appearances of typical specimens from this test are shown in Fig. 19.4. In this test the dimensions of test specimens should be standardised, since the depth of attack has been found to be influenced by the extent of the immersed area of the specimen that is outside the impingement zone. Along the same general lines is an apparatus employed by Brownsdon and Bannisterw in which a stream of air at high velocity is directed against the surface of a submerged test specimen (see also p. 19:75). A straightforward way to study velocity effects is to force the testing liquid through tubular specimens, which may be arranged to form model piping systems for studying the peculiar corrosion that may result from severe turbulence effects downstream of valves, reducers, branch connections, elbows, and other fittings. In such systems the rates of flow can be measured by suitable orifice meters and regulated by control valves. A somewhat similar technique applied to condenser-tube alloys is to test them as installed in model tube-bundle assemblies4’. Butler et al. have described a laboratory test rig for studying the effects of flowing water on steel pipework”. Other methods involve holding specimens in suitable fixtures so that they form the walls of channels through which the test solution can be passed at controlled rates of flow. Such devices have been used at the Harbor Island Test Station in North Carolina primarily for studying the electrode potential and polarisation characteristics of metals and alloys, but they are also suitable for observing effects of velocity on corrosion. This is illustrated in Fig. 19.5 in which the specimen and Pt electrode are of the same size and are placed parallel to one another in the holder. When required potentials are measured by inserting a capillary through the hole in the Pt; it is then removed to avoid shielding effect. Effects of velocity are sometimes aggravated by the presence of abrasive solids in suspension, which increases deterioration by straight mechanical
19:25
CORROSION TESTING AND CORROSION RATES
Sample
no.
Alloy
Mass loss affer I OOO h (g)
Penefrafion of impingemenf pit (% of cross-section)
Appearance
A
Arsenical 70/30 brass
0.262
40
Deep pit opposite jet, surrounding area darkly stained
B
Arsenical Admiralty brass
0-422
44
Deep pit opposite jet
C
Arsenical aluminium brass
0-010
I
Very slightly attacked opposite jet
D
Cu lONi
0-026
1 *5-2-0 indicates susceptibility; for Type 304 Streicher considers a rate > 0.76 mm/y to indicate susceptibility, but Brown considers a higher figure to be acceptable (see Table 19.4) Accumulation of corrosion products does not stimulate attack so that several specimens may be tested in the same solution, but additional Fe,(SO,), may have to be added (or the solution changed) if there is considerable attack on severely sensitised specimens, as is indicated by a colour change of the solution from brown to dark green. The redox potential of the solution is that of the Fe3+/Fe2+equilibrium and lies within the range 0.80-0.85 V (vs. S.H.E.). The high weight loss of susceptible alloys is due to undermining and grain dislodgement at the sensitised zones, which occurs at about twice the rate of that in the Huey test. Another difference is that whereas in the Huey test corrosion products
CORROSION TESTING AND CORROSION RATES
19:65
[Cr(VI)] increase the rate by raising the potential of the alloy into the transpassive region, the converse applies in the acid (Fe,(SO,), test, since reduction of Fe3+to Fe2+during the test will result in a decrease in the redox potential and the whole sample will corrode with hydrogen evolution. According to Cowan and Tedmon189the test can selectively attack some types of a-phase. Those of Types 322 and 347 are readily attacked, whereas the molybdenum-bearing a-phase of Type 326 is unattacked. The test will also show Hastelloys and Inconels to be susceptible to intergranular attack when there are either chromium- (or molybdenum-) depleted grain boundaries or grain-boundary a-phase present. Ferritic (200 series) and austenoferritic stainless steels can also be tested for chromium-depletion sensitisation in this reagent, but whether a-phases formed in these alloys affect the test has not been established. In conclusion it must be emphasised again that all the tests used are accelerated tests and only provide information on susceptibility to intergranular attack under the precise test conditions prevailing. They are quality control tests that may be used to demonstrate either that heat treatment has been carried out adequately or that a steel will withstand the test for a certain sensitising heat treatment. Electrolytic Oxalic Acid Etching Test
This test, which was developed by Streicher is used as a preliminary screening test to be used in conjunction with the more tedious testing procedures such as the boiling HNO, test. The specimens are polished (3/0grit paper) and then anodically polarised for 1 5 min at 1 A cm-, at room temperature in a solution prepared by dissolving 100 g of H,C,O, *2H20 in 900ml of distilled water. The surface is then examined at about x 500 magnification and the structure is classified as ‘step’, ‘ditch’ or ‘dual’ (both ‘step’ and ‘ditch’). If the surface shows a ‘step’ structure it is immune to intergranular attack and no further testing is necessary; if the structure is ‘ditch’, further testing by the Huey test or some other chemical test is necessary; if ‘dual’ further testing may be necessary. Thus the test, by identifying ‘Ditch’ structure
’step structure
Fig. 19.17 ‘Ditch’ and ‘step’ structures (after Streicher
19 :66
CORROSION TESTING AND CORROSION RATES
structures that are immune to intergranular attack, eliminates unnecessary testing, although where a ‘ditch’ (or possibly a ‘dual’ structure) is obtained, final confirmation by the Huey test is essential. Figure 19.17 shows diagrammatically the ‘ditch’ and ‘step’ structures, and Fig. 19.18 photomicrographs of these structures and a ‘dual’ structureI9*.
Fig. 19.18 Micrographs of (a) a ‘step’ structure, (b) a ‘ditch’ structure and (c) a ‘dual’ structure (after Streicher ‘53 and Cowan and Tedmon
The test operates at a potential above 2.00 V (vs. S.H.E.), and the ‘ditch’ structure obtained with sensitised alloys must be due, therefore, to the high rate of dissolution of the sensitised areas as compared with the matrix. The ‘step’ structure is due to the different rates of dissolution of different crystal planes, and the ‘dual’ structure is obtained when chromium carbides are present at grain boundaries, but not as a continuous network. Electrochemical Tests
The difficulties associated with the ASTM Recommended Methods for Detecting Susceptibility to Intergranular Attack in Austenitic Stainless
CORROSION TESTING AND CORROSION RATES
19:67
Steels’ (A262-1986)are that the methods are destructive and qualitative in nature. Early attempts to develop quantitative, non-destructive electro’ ~ chemical techniques to detect sensitisation by Clerbois et ~ 1 . employed potentiostatic techniques and it was observed that sensitised 18-8 stainless steel when anodically polarised potentiostatically in 1 -0 mol dm-3 H2S04 gave rise to a secondary active peak in the range 0-14-0-24V (vs. S.H.E.) that was not present in the curve for the annealed alloys. This observation was criticised by France and GreenezW,who consider that the active peak is due to the dissolution of Ni that had accumulated at the surface during active dissolution at lower potentials. Clerbois, et a1.Iw also noted that if a sensitised sample is held at 0 - 14V in 1 -0mol dm-3 H,SO, for 24 h and then bent around a mandrel, it fissures and cracks, and it can be seen from Fig. 19.15 that at this potential the chromium-depleted grain boundary will corrode actively, whereas the matrix will be passive. The potentiostatic test using cracking to detect suceptibility is thus analogous to the acid-copper sulphate test. France and Greene2” proposed that it should be possible to predict service performance by potentiostatic studies of steels in the environments encountered in practice coupled with metallographic examination of the surfaces. They argued that many environments do not selectively attack the grain boundaries of sensitised stainless steels so that the use of costly preventative measures is unnecessary. Since the intergranular attack of austenitic stainless steels occurs only in limited potential regions it should be possible to predict service performance providing these regions are precisely characterised. In their studies, specimens of different sensitised steels were held at various constant potentials in different concentrations of the acid under study at various temperatures and the surfaces were then examined metallographically for intergranular attack. Data obtained in this way enabled E-concentration of acid diagrams to be produced showing the zones of general corrosion, fine intergranular corrosion and coarse intergranular corrosion for a given sensitised stainless steel in a given acid at various constant temperatures (Fig. 19.19). Streicher201, however, considered this approach to be unsound and pointed out that the short duration of the potentiostatic studies carried out by France and Greene cannot be used to predict long-term behaviour in ser203 was well sumvice. The prolonged dialogue between these workerszo2* marised in the review article by Cowan and Tedmon who concluded that these particular potentiostatic tests cannot be regarded as accelerated tests for service environments and that predicting future industrial service for periods longer than the test is not advisable. EPR Test
The electrochemical potentiokinetic reactivation (EPR) test was proposed by Cihal et and developed by Novak and ~ t h e r s ~ ~as ~ -a’fast, ~ ’ quantitative and non-destructive technique for establishing the degree of sensitisation of austenitic stainless steels. The test is accomplished by a potentiodynamic sweep from the passive
19: 68
CORROSION TESTING AND CORROSION RATES 1.6C
_1
1LO
Sensitised cast CF-8 at 10°C
1.20
1~00
-w: 0.80
f
0.60
2 1
G 0.10 0.20
General corrosion
/\
' ,
\'
i
0.00
-0.20 -040
0
5 Normality of H2S04
Fig. 19.19 Intergranular corrosion plot for a sensitised cast CF-8 stainless steel (0-08Vo C max., 8-1 1% Ni, 18-21% Cr) in H2 SO, at 40°C as a function of potential and concentration of acid (after France and Greene2T
to the active regions of electrochemical potential (a process referred to as reactivation) for a given alloy in a specific electrolyte, during which the amount of current resulting from the corrosion of the chromium-depleted regions surrounding the precipitated chromium carbide particles is measured. In a sensitised microstructure, the bulk of these particles is located at the grain boundaries and are particularly susceptible to corrosion in oxidising acids. Proposed national and international standards on EPR testing specify 0.5 M H,SO, + 0.01 M KSCN at 30°C as the EPR test environment for sensitised austenitic stainless steels. Three different forms of EPR test can be employed, designated as the single loop, double loop and reactivation ratio methods in Figure 19.20. Single Loop EPR Test The single loop method requires the sample to be polished to a 1 fim finish then passivated at +200 mV (S.C.E.) for 2 min following which the potential is decreased at 1 -67 mV s-' until the corrosion potential of approximately -400 mV (S.C.E.) is reached. The reactivation process results in the preferential breakdown of the passive film in the
19 :69
CORROSION TESTING AND CORROSION RATES
(2 rnin]
+200 mV
Non-sensitised
Passive
m
c
0
a
Ecorr.
rnV, 2 min)
( * -400
-
(a)
Log current
+300 rnV
4
2
Anodic scan
-d
.-+m.
C
Reverse scar!
Q) +A
0
a
Ecorr. (- -400
rnV)
Scan rate = 6 Vlh Log current
I
4
'a
+200mV I
m
c.
0 (L
I (- -400
rnV, 2 rnin)
I
Log current
4
(C)
Fig. 19.20 Schematics of reactivation polarisation curves. (a) Single loop EPR test method, (b) double loop EPR test method and (c) reactivation ratio EPR test method
chromium-depleted grain boundaries of sensitised material and an increase in the current through the cell. The area under the E us. log Z curve (Fig. 19.20~)is proportional to the electric charge, Q, measured during the reactivation process. On non-sensitised materials, the current density during the reactivation step is very low because the passive film remains essentially intact. A measure of the degree of sensitisation is obtained by calculating the normalised charge, Pa,where:
19 :70
CORROSION TESTING AND CORROSION RATES
Pa(C cm-') = Q / A where Q = integrated charge during the reactivation scan, and A = grain boundary area ( 5 - 1 x 10-3exp0-35G, where G is the ASTM grain size at l00X magnification). Pitting caused by the dissolution of non-metallic inclusions can increase the Pavalue. Consequently, the microstructures of specimens with a high Pa value must be examined to identify the source of the elevated value. In general, Pa values below 0.10 are characteristic of unsensitised microstructures, while sensitisation is indicated if Pa exceeds 0-4. Single loop tests are sensitive to mild degrees of sensitisation but do not readily distinguish between medium and severely sensitised materials.
Double Loop EPR Test Details of this procedure are given in Japanese Industrial Standard JIS G 0580 (1986). The sample is ground to a 100 grit finish then placed in the test solution for about 2 min to establish the rest potential (about -400 mV (S.C.E.) for AISI Types 304 and 304L stainless steel). The sequence of polarisation steps is shown in Figure 19.206. The surface is first polarised anodically from the corrosion potential to +300 mV (S.C.E.) at a rate of 1 *67mV s-'. As soon as this potential is reached, the scanning direction is reversed and the potential is decreased at the same rate to the corrosion potential. The ratio of the maximum current in the reactivation loop, I , , to that in the larger anodic loop, Za, is used as a measure of the degree of sensitisation. Reactivation Ratio EPR Test (Fig. 19.20~) This is a simpler and more rapid method than the single or double loop tests, and depends on the fact that the value of Z, determined during the anodic scan of a double loop test (which produces general dissolution without intergranular attack on sensitised material) is essentially the same for all AISI Type 304 and 304L steels. The specimen is ground to a 100 grit finish then, after 2 min at the corrosion potential (about -400 mV (S.C.E.)), it is conditioned by a 2 min treatment at -230 mV (S.C.E.) in order to eliminate the need for polishing prior to the reactivation procedure. Passivation is then accomplished at +200 mV (S.C.E.) for 2 min after which the specimen is reactivated by scanning back to the corrosion potential at 1 -67 mV/s. During this reactivation scan, the maximum current, I,, is measured and is divided by the surface area as an indication of the degree of sensitisation. EPR Tests for Ferritic Stainless Steels
Leezo8has demonstrated that in slightly modified forms the single loop EPR test can be used to quantify the degree of sensitisation in ferritic stainless steels. For AISI Types 430, 430Ti, 430Nb and 446 stainless steels, the test consists of passivating the specimen in deaerated 3 N H,SO, solution at 30°C at +400mV (S.C.E.) for IOmin, followed by reactivation at a scan rate of 250mV min-'. The EPR test for AISI Type 434 stainless steel requires a reactivating scan rate of 150 mV min-' (the other test conditions remaining unchanged). For AISI Type 444 stainless steel, the test is
CORROSION TESTING AND CORROSION RATES
19:71
conducted in deaerated 5 N H,SO, solution at 30°C and involves passivation at +400 mV (S.C.E.) for two min followed by reactivation at a scan rate of 100mV min-I.
Crevice Corrosion and Pitting Crevice corrosion and pitting have been dealt with in some detail in Section 1.6, and it is not appropriate here to discuss the nature of the phenomena nor the methods that have been used to determine the mechanisms of these forms of localised attack. However, it should be noted that many of the methods of testing follow directly from the concepts that have been discussed in Section 1.6, and in particular the potentials Eb (the critical pitting potential) and Ep (the protection potential) have been investigated by a number of workers as possible criteria for the resistance of metals and alloys to pitting and crevice corrosion in service. It should also be noted that since crevice corrosion and pitting have similar mechanisms, and since the presence of a crevice is conducive to pitting of alloys that have a propensity, to this form of attack, it is appropriate to consider them under the same heading. In general, the tests may be classified as follows: 1. Laboratory tests in which the specimen is immersed in a solution conducive to pitting such as an acidified FeCl, solution (redox potential above the critical pitting potential Eb). 2. Laboratory tests in which the specimen is anodically polarised in a chloride-containing solution to evaluate Eband Ep. 3. Field tests in which the specimen (with or without a crevice) is exposed to the environment that it will encounter in service. As far as tests for crevice corrosion are concerned all that is required is a geometrical configuration that simulates a crevice, which may be achieved in a variety of ways using either the metal itself or the metal and a non-metallic material. Figure 1.49 (Section 1.6) shows the testing arrangement used by Streicher- to study the crevice corrosion of Cr-Ni-Fe alloys, in which two plastics cylinders are held on the two opposite faces of a sheet metal specimen by two rubber bands, thus providing three different types of crevice in duplicate. A simple method of testing for crevice corrosion produced by contact with different materials is to use a horizontal strip of the metal under study and place on its upper surface at intervals small piles of sand, small piles of sludge, pieces of gasket material, rubber, etc. More precise crevices can be produced by bolting together two discs of the metal, which are machined on the facing surfaces so that there is a flat central portion followed by a taper to the periphery of the disc, the flat central portion providing a very fine crevice and the tapered portion a coarser onez1o. Figure 19.21 shows the types of crevices used by Wilde'" for studying crevice corrosion and pitting of Cr-Ni-Fe alloys in the laboratory and in the field. Types 1 and 5 were used for anodic polarisation studies in nitrogensaturated 1 mol dm-3 NaCl and in aerated 3 . 5 mass% NaCI, respectively, and it can be seen that attachment to the conducting lead is by means of a Stern-Makrides pressure gasket; Types 3 and 4 were used for field tests
19:72
CORROSION TESTING AND CORROSION RATES
in sea-water for periods up to 4f years; Type 2 was used for laboratory studies in which the specimens were immersed in acidified FeCl, (108 g I-’ FeC1,.6H20 with the pH adjusted to 0.9 with HCl).
e--
Sheet
Teflon
gasket
Nylon ’bolt
sample
Nylon nut \ bolt
U
bP 2
Type 1 Nvlon
Type 3
Plote
ylindrical specimen
Bolt
Neoprene O-ring
Type 5
Fig. 19.21 Various types of crevices used for investigating crevice corrosion of Cr-Ni-Fe alloys (after Wilde’”)
The value of electrochemicalevaluation of the critical pitting potential as a rapid method of determining pitting propensity is controversial. France and Greene’” studied the pitting of a ferritic steel (Type 430) using a controlled potential test in 1 mol dm-3 NaCl and a conventional immersion test in oxygen-saturated 1 mol dmd3NaClY but found that at the same potential (-0.17 to 0.09V us. S.C.E.) the corrosion rates were 390 and 5 - 2 mm/y, respectively. Similar studies were carried out on Zr using 0.5 mol dm-3 H,SO, + 1 mol d1n-~NaC1for the controlled potential test and 0-5 mol dm-3 H,S04 FeCl, -6H,O for the immersion test, and again the former gave a much higher corrosion rate than the latter. France and Greene conclude that these two types of test give rise to significantly different results under identical test conditions. To explain the results obtained with the ferritic stainless steel they pointed out that during the controlled potential test the anodic reaction occurs at the metal’s surface whereas the interdependent cathodic reaction takes place at the counter-electrode. Under these circumstances the M 2 +ions produced anodically result in increased migration of C1- to maintain electroneutrality, and this in turn results in a higher concentration of C1- at the metal/solution interface with consequent increase in the rate of pitting. A similar situation does not arise during the immersion test where the anodic and cathodic sites are in close proximity and charge balance is maintained without migration of C1- from the bulk solution (see Fig. 19.39, Section 19.2). Potentiostatic t e ~ t s ~ ’ ~have - ~ ’ ’been used and Wilde and Williams * I 3 in potentiokinetic studies of the critical breakdown potential of stainless steels (Types 430 and 304) in 1 -0mol dm-3 NaCI, showed that the nature of the gas used to purge the solution has a pronounced effect on the value of Eb
+
19: 73
CORROSION TESTING AND CORROSION RATES
Table 19.5 Variation in Eb(V) for stainless steels in 1.0 mol dm-3 NaCl at 25°C with nature of dissolved gas (Eb vs. S.C.E.)* ~~
~
Gas
Type 430 stainless steel
Type 304 stainless steel
Hydrogen Nitrogen Argon Oxygen
-0.185 -0.130 -0.100
-0.050 -0.020 +0-050 +0*065
-0.035
Data after Wilde and WilliamsL'
(Table 19.5). In particular, they have established that the presence of dissolved O2enhances passivity thus causing E,, to become more positive, and consider that this explains the failure of France and Greene to obtain accord between controlled potential tests in hydrogen-saturated chloride solutions and immersion tests in oxygenated chloride solutions at the same potentials. Wilde and william^^'^ have used the redox system Fe(CN)i- /Fe(CN):in 0.1 mol dm-3 for their immersion tests, which for Type 403 stainless steel gives a corrosion potential of -0. lOOV (vs S.C.E.); selection of this system was based on the premise that being large anions they would be less likely than dissolved 0, to be involved in the adsorption processes that stabilise the passive state. Pitting occurred within 60 s, and equivalent tests on the same alloy conducted potentiostatically at - 0 - 1 V (vs. S.C.E.) in hydrogen-saturated 1 * O mol dm-3 NaCl gave similar results. They conclude that these two tests give comparable results, but that extreme caution must be used in utilising E,, as an index of pitting, since its value is dependent upon environmental variables and in particular the nature of the dissolved gas in the corrodent. Wilde and have also shown that the critical pitting potential can be used to predict the behaviour of alloys exposed for long periods to sea-water or to industrial chemical environments. In a subsequent paper Wilde*" pointed out that although Eb is qualitatively related to resistance of a material to breakdown of passivity and pit initiation, it is of questionable value in predicting performance when crevices are present. Wilde found that although the Fe 30Cr-3Mo alloy appeared to indicate total immunity to breakdown when tested anodically in 1 mol dm-' NaCI and in the freely corroding condition in 10% FeCl,, it pitted within the crevice when an artificial crevice was present. Exposure in sea-water for a 16 month period showed that AIS1 Types 304 and 316 stainless steels and the Fe30Cr3Mo alloy all pitted to the same extent when a crevice was present, although the former two alloys are considered to be less resistant to pitting than the Fe-30Cr-3Mo alloy. Pourbaix, et ai. have defined the protection potential E, (see Section 1.6) as the potential below which no pits can initiate and preexisting pits cannot propagate, since they are passive at that potential. However, Wilde using cyclic potentiodynamic sweeps at varying sweep rates has established that E, is not a unique parameter and that it varies in a semi-logarithmic manner with the extent of localised attack produced during the anodic polarisation, Le. E,-log(extent of pit propagation) is linear. Thus at a sweep rate of 1OV h-', E, was found to be -0.290V (vs. S.C.E.) whilst it fell to a more negative value of
19 :74
CORROSION TESTING AND CORROSION RATES
-0.410V at the slower sweep rate of 1 V h-'. This was explained by Wilde as being due to the chemical changes that occur in the growing pit by hydrolysis of corrosion products and by the increased migration of Cl- ions. Since Epis a variable that depends upon experimental procedures it cannot be used on its own as a criterion for protection against the propagation of pre-existing pits or crevices in an engineering structure. Wilde considers that a more useful parameter appears to be the 'difference potential' (&-Ep), which is used as a rough measure of the hysteresis loop area produced during the cyclic determination of Eb and Ep. The area of the hysteresis loop obtained in a potentiodynamic sweep using a specimen with an artificial crevice provides a measure of the resistance to crevice corrosion in service, Le. the greater the area the lower the resistance. Figure 19.22 shows the linear relationship between the 'difference potential' and the mass losses of various stainless alloys containing an artificial crevice that have been exposed to seawater for 4+ years.
-, E
I
$
Eb
determined at 0-600V h
.Ebdetermined after reversing sweep at 2 mAlcm2 3.5 mass% sodium chloride, aerated, 25OC Sample area = 1220cm*
(410)
The above considerations show that although considerable advances have been made in developing laboratory controlled potential tests for evaluating crevice corrosion and pitting, the results must be interpreted with caution. Guidance on crevice corrosion testing of iron-base and nickel-base stainless alloys in sea-water and other chloridecontaining aqueous environments is given in ASTM G78: 1989, while ASTM G61: 1986 provides a standard test method for conducting cyclic potentiodynamic polarisation measurements for localised corrosion susceptibility (Le. pitting and crevice corrosion) of iron-, nickel-, and cobalt-based alloys. Guidance on the selection of procedures for the identification and examination of pitting corrosion to determine the extent of its effect is available in G46:1976 (R1986).
CORROSION TESTING AND CORROSION RATES
19:75
Impingement Tests/Erosion Corrosion
The method most commonly used for testing condenser materials is the BNFMRA May jet impingement testz1’ in which small sections of tube, abraded to a standard finish, are immersed in sea-water and subjected to an underwater jet of sea-water containing air bubbles. However, at high velocities cavitation can occur in the water box in this test. An alternative design has been described to overcome this’”. Resistance to impingement attack is also assessed by the Brownsdon and Bannister test 2’9 in which a stream of air bubbles is directed onto the surface of the test specimens immersed in sea-water or sodium chloride solution. Special tests for resistance to corrosion under localised heat transfer conditions (hot-spot corrosion) have been described by Breckon and Gilbert”’ and by Bem and Campbell but temperature effects are usually ignored when comparing condenser tube materials. Campbell’’2 points out that in evaluating condenser tube materials a test apparatus is required that will include all the principal hazards likely to be encountered in service and should thus cater for the following conditions: impingement, slow moving water, heat transfer and shielded areas. Furthermore, the internal surfaces should not be abraded, as in the jet impingement test, but should be tested in the ‘as-manufactured’condition, particularly in view of the deleterious effect of carbon films produced during manufacture (see Sections 1.6 and 4.2). LaQue has pointed out the importance of specimen area in impingement The general arrangement of the apparatus is shown in Fig. 19.23. It accommodates 10 vertical 200 mm lengths of condenser tube spaced equally around a 125mm diameter circle. Water enters the bottom of each tube through an inlet nozzle (PartNo. 6 in Fig. 19.23) which fits inside the tube and also locates it. The nozzle has a 5 mm diameter blind hole up the centre connecting with a 2.4mm diameter hole, set at 45” to the vertical, through which the water emerges at a velocity of 10 m s-’ to impinge on the wall of the tube. The water then rises up through the tube at a mean velocity of 0.1 m s-’ (in a 22-24 mm dia. condenser tube) and leaves through an outlet nozzle (Part No. 1) fitted into the top end of the tube. Half the length of each outlet nozzle has a 2” taper on the outside to provide a reproducible annular crevice between it and the inside of the condenser tube. Neoprene ‘0’-rings (Part No. 3) provide seals between the tube and the top and bottom nozzles, and the tubes are held in place by a common clamping plate (Part No. 2) at, the top. The 10 inlet nozzles are fed with water through a distributor (Parts Nos. 7, 8 and 15) of the design used in the May jet impingement apparatus, which ensures equal distribution of water between them. The distributor and nozzles are all of non-metallic materials. The part of each test piece between 40 and 65 mm from the top is fine-machined externally to fit a semi-circular notch in a 15 mm thick brass heater block (Part No. 4), the tubes being held in contact with the block by a circumferential clip to ensure efficient and equal heat transfer between the block and each tube. The diameter of the inlet and outlet nozzles and that of the semicircular notches in the heater block are made to suit the size of condenser tube to be tested.
’”,
19:76
CORROSION TESTING AND CORROSION RATES Water outlet to waste m
-
A
-.__
@----Annular crevice
e \
\
I'
!A
Y
_-I
A
Condenser
tube
I
'!
i
I
I
I
I
%----
L'S.
1,.*_I
U'..+,"...
L.0.
Upy'U.'.."1
.".
"-..LL..l.....e
....
*. .,""."...." ".
L..U.
thit
condenser-tube materials are subjected to in service (after Campbell 222)
The common heater block shown in Fig. 19.23 can itself be subject to corrosion leading to different heat transfer conditions for different tubes, and in some later versions of the apparatus individual short heating jackets are used for each tube, which are heated with oil from either a steam-heated or electrically heated heat exchanger. This modification not only avoids corrosion problems but also obviates the necessity to machine a length of the outside of each tube to fit the semi-circular notches in the single heater block. The oil flow is adjusted to give an oil temperature of 95°C at each outlet.
CORROSION TESTING AND CORROSION RATES
19 :77
The test usually lasts 8 weeks, after which the tubes are sectioned longitudinally and their interiors inspected for accumulated deposits. Loose deposits are then removed by washing in water and the internal surfaces are examined for impingement attack, pitting and blistering or flaking of the corrosion-product film, using a low-power binocular microscope. After cleaning the section in 10% HzS04.the depth of impingement attack, pitting or other localised corrosion, is determined. Observations and measurements are recorded for each of the following five areas of the section: (a) impingement area opposite the inlet nozzle where water velocity and turbulence are greatest, (b) the slow-moving cold water area from the impingement area upwards to the heated area, (c) the heated area including the heat-transfer area itself and the warm water area above and ( d ) the two annular crevices formed between the tapered portions of the cold-water inlet and warm-water outlet nozzles and the tube wall. The Campbell apparatus is cheap to construct and easy to use, and can be installed on site to assist the selection of condenser or heat-exchanger tube materials, or to monitor changes in the corrosivity of the cooling water. The information that it provides on the various forms of attack is more comprehensive than that of any other existing apparatus for corrosion testing condenser tubes, and it is therefore particularly suitable also for assesbing new materials or the effect of surface conditions arising from changes in manufacture. Impingement and erosion-corrosion forms of attack will usually be intensified by the presence of solid particles in the fluid. Variations of the jet test have been proposed to take this effect into Test equipment for the study of erosion-corrosion by liquids with sand content, as met in formation waters in oil and gas production, has been described by Kohley and HeitzZz5.
Corrosion Fatigue The simultaneous action of alternating stresses and corrosion usually has a greater effect than when either is operating separately, and in this respect corrosion fatigue is analogous to stress-corrosion cracking. The important factors in corrosion fatigue (see Section 8.6) include the following: 1. environmental conditions; 2. magnitude of the alternating stress;
3. magnitude of mean stress. 4. frequency of reversal of the stress. 5 . load-versus-time waveform;
6. characteristics of the metal.
Depending on the intended purpose, corrosion fatigue tests can be conducted on smooth, notched or pre-cracked specimens as well as on components and parts joined by welding. Because of the time-dependent nature of corrosion processes, it is essential that the mechanical variables employed during corrosion fatigue testing, including cyclic frequency and load-versustime waveform, as well as the chemical and electrochemical conditions, are relevant to the intended application. For example, it is unlikely that data generated in a laboratory test at a frequency of 10 Hz would be applicable
19:78
CORROSION TESTING AND CORROSION RATES
for predicting corrosion fatigue behaviour in a structure which is cycled at 0 - 1 Hz. Laboratory corrosion fatigue tests can be classified as either cycles to failure (crack initiation) or crack propagation testszz6.Cycles to failure tests employ plain or notched specimens to provide data on the intrinsic corrosion fatigue crack initiation behaviour of a metal or alloy. Crack propagation tests use pre-cracked specimens to provide information on the threshold conditions for the propagation of pre-existing defects by corrosion fatigue and on the rates of corrosion fatigue crack growth. It is often difficult to conduct laboratory tests in which both the environmental and stressing conditions approximate to those encountered in service. This applies particularly to the corrosive conditions, since it is necessary to find a means of applying cyclic stresses that will also permit maintenance around the stressed areas of a corrosive environment in which the factors that influence the initiation and growth of corrosion fatigue cracks may be controlled. Among these factors are electrolyte species and concentration, temperature, pressure, pH, flow rate, dissolved oxygen content and potential (free corrosion potential or applied). For tests on plain or notched specimens, a simple approach can be to use a conventional Wohler rotating cantilever beam modified so as to permit the specimen to be brought into contact with the corrodent. This may be achieved by surrounding the specimen with a cell through which the corrosive solution is circulated or by applying it by a padz27,wick2’* or drip feedU9. Four-point loading or push-pull machines can be used in a similar way, and have the advantage over the Wohler machine when testing plain specimens that the length of the test-piece between the two points of loading is subjected to an approximately uniform stress. Rawdon 230 used flat specimens that were subjected to repeated flexure while they were being immersed periodically in the corrosive solution. Kenyonz3’used a rotating wire specimen in the form of a loop, the upper part of which was attached to the motor whilst the lower part of the loop passed through the corrodent, and a somewhat similar device was developed by Haigh-Robertson and used in several ~ t u d i e s * ~Gough ~ ~ * ~ and ~. used this machine in their studies, the corrodent being Sopwithz34-235 applied as a spray. Figure 19.24 shows a slow fatigue machine236that has been developed to study the performance of welded butt and fillet joints for steels used in the construction of North Sea oil drilling-rigs; the bending stress and frequency have been selected to simulate the forces produced by the wave motion. The specimens, 1500 x 100 x 12-5 mm with the weld 25 cm from the base, are clamped at the lower end and the stress is applied as a variable bending moment at the upper end by rams. The rams, which are attached to a sliding frame, are activated by a pneumatic cylinder that can be automatically programmed for stroke and frequency and the stress level is monitored by strain gauges. The maximum amplitude of stress is 150 mm, the stress range is up to 300 M N m-’ and the frequency can be varied from 1 cycle/2 s to 1 cycle/ 20 s. The corrodent is artificial sea-water, and provision is made for studying the effect of cathodic protection by means of Zn anodes. H ~ p p n e r ’pointed ~~ out that until the early 1970s. most investigators conducted fatigue tests utilising rotating bending, flat plate bending or
CORROSION TESTING AND CORROSION RATES
19 :79
(b) Fig. 19.24 (0) Rig for a laboratory study of the corrosion fatigue of welded joints in sea-water and (b)view of test-pieces showing welded joint (after Jarman et ~ 1 . ~ ~ 9
19:80
CORROSION TESTING AND CORROSION RATES
torsion-type loading configurations, which have the disadvantage that tests at positive or negative mean-stress values are difficult to achieve. In addition, the rotating bending and flat-plate-bending tests create complex stress states upon crack initiation, e.g. a shifting neutral axis. For these reasons, axial load fatigue machines, as recommended by the ASTM Committee E9, are preferred. The results obtained from the tests described above are presented in the form of the conventional S-N curve, where S is the stress and N the number of cycles to cause fracture. Curves of this type are obtained for the metal in air and for the metal in the corrodent, and comparison provides information on the effect of the corrosive environment on the fatigue life. Hoeppner points out that even though the S-N curve for either notched or unnotched specimens may be useful for certain applications it cannot always be employed to evaluate the effect of the environment on the fatigue life. This is because in some materials the inherent metallurgical and fabrication discontinuities, which may be undetectable by non-destructive testing will be so large that the only factor of engineering significancewill be the rate of propagation of a crack from the initial defect, i.e. the fatigue-crack propagation rate may play the dominant rdle in the useful life of the component. For this reason it is important to conduct fatigue crack growth tests on pre-cracked specimens, and the data are then presented in the form of curves showing crack growth rate, da/dN, vs. stress intensity factor range, AK. The NACE publication Corrosion FatigueZ3’gives a comprehensive account of all aspects of the subject, and in this work a review of the application of fracture mechanics for studying the phenomenon has been presented by McEvily and WeiZ3*,whilst K i t a g a ~ a ’has ~ ~given a detailed account of crack propagation in unnotched steel specimens. This work should be consulted for details of testing and interpretation of results. Special requirements for fatigue testing in aqueous environments are addressed in the Annexe to ASTM E647:1986a ‘Standard Test Method for Measurement of Fatigue Crack Growth Rates’.
Cavitation-erosion The phenomenon and mechanisms of cavitation-erosion have been considered in Section 8.8 and here it is only necessary to consider laboratory test methods that have been designed to simulate conditions that prevail in practice and which may be used to evaluate the performance of materials. In considering these tests it should be remembered that the phenomenon of cavitation-erosion is often accompained by corrosion effects and that a synergistic effect may operate between the mechanically and chemically induced forms of attack. In fact the term cavitation-erosion-corrosion may often be more applicable in describing the requirements of a test procedure. The subject has recently been discussed by Wood et ’ follows: The methods used have been classified by Lichtman, et ~ 7 1 . ’ ~as 1. High-velocity flow. (a) Venturi tubes. (b) Rotating discs. (c) Ducts containing specimens in throat sections.
CORROSION TESTING AND CORROSION RATES
19:81
2. High-frequency vibratory devices. (a) Magnetostriction devices. (b) Piezoelectric devices. 3. Impinging jet. (a) Rotating specimens pass through continuous, stationary jets or droplets. (b) Stationary specimens exposed to high-speed jet or droplet impact. All tests are designed to provide high erosion rates on small specimens so that the test can be conducted in a reasonable time, and although vibra-
tory and high-velocity jet methods may not simulate flow conditions they give rise to high-intensity erosion and can be used, therefore, for screening materials. The essential component of many high-velocity flow rigs is a Venturi-type section in which cavitation occurs in the low-pressure high-velocity region created by the Venturi throat. Typical of this type is the double-weir arrangement used by S c h r ~ t e r ~ but ~ ~since , this technique requires very large volumes of water it is not readily adaptable to laboratory use. H ~ b b s and '~~ others have used a uniform-area rectangular-cross-section duct in which a cylinder of small diameter is inserted; cavitation occurs in the wake of the cylinder, which may be used as the test specimen or the specimen may be set in the side wall of the duct near the cylinder. The cavitation intensity will be dependent on the configuration of the test section and the velocity, pressure, temperature, viscosity, surface tension, corrosivity, gas content and density of the liquid. Devices in which cavitation is achieved by vibrating a test specimen at high frequencies are often used. The original apparatus was developed by Gaines 244, and was adapted for cavitation-erosion studies by Hunsaker and Peters as described in the paper by it has been used also by Beeching%, R h e i n g a n ~ ~and ~ ' Leith, et al."*. In this method cavitation is produced by attaching the specimen to the vibrating source or by means of a partially immersed probe vibrating axially at a high velocity and low amplitude and placed close to the test specimen. Although originally magnetostriction oscillators were usedN these have now been largely superseded by piezoelectric oscillators, which are more efficient. The apparatus consists basically of a conventional ultrasonic generator, a piezoelectric transducer and a resonating horn or probe, and the majority of tests are carried out at a frequency of 20 kHz. Originally the test specimen was fastened to the end of the ultrasonic probe, and this is still specified in ASTM D2939-7 1 which describes a method of testing aluminium in antifreeze solution. However, this arrangement also subjects the test-piece to high alternating stresses as a result of the high accelerations associated with vibration at ultrasonic frequencies, which may be overcome by using a stationary test-piece and locating it immediately below a dummy tip placed on the end of the ultrasonic probe. Vibratory test apparatuses are relatively cheap to build and run, and have low power consumption, while flow rigs are bulky, expensive to build and run, and have high power consumptions but have the advantage that they simulate more closely practical conditions of hydrodynamic cavitation. On the other hand, the damage rate is higher in the vibratory tests than in the
19: 82
CORROSION TESTING AND CORROSION RATES
flow test, although whether this is advantageous depends on the objectives of the test. A further criticism of the vibratory test is that the mechanical component is over-emphasised in relation to the effect produced by corrosion. For this reason P l e ~ s e t "uses ~ a technique in which cavitation is intermittent with short bursts of vibration followed by longer static periods, which significantly increases the erosion rate of materials with poor corrosion resistance but has little effect on materials with good corrosion resistance. Tests of this type have distinguished readily between materials having the same hardness but different resistances to corrosion, and between corrosive and non-corrosive solutions. Figure 19.25 shows an apparatus for studying cavitation-corrosion using the magnetostriction principle for vibration. A nickel tube is made the core of a magnetic field tuned to the natural frequency of the tube assembly, and since nickel changes its length as it is magnetised and demagnetised it will vibrate with the frequency of the magnetising current. The specimen under test vibrates with the nickel tube, and a commonly used frequency is 6 500 Hz with an amplitude of 0-008-0.009 cm. Damage is increased by the amplitude of vibration, and the more resistant the material the greater the amplitude to achieve substantial attack. Increase in temperature decreases damage by increasing the vapour pressure within the cavitation bubbles, thus reducing the force of their collapse, but in opposition to this effect is the increased damage resulting from the lower solubility of gases which cushion the collapse of the cavitation bubbles. Consequently, under many circumstances damage reaches a maximum at a test temperature of about 46-52°C.
Nickel tube vibrator
4" 704A
-
IIU v -I
bubbles
1 Fig. 19.25
I
a.c
Polarisinq current
2kVdc Plate
voltage
Vibratory cavitation-erosion test using magnetostriction
CORROSION TESTING AND CORROSION RATES
19:83
Assessment of cavitation-erosion is based on mass loss and the results are expressed as curves showing cumulative mass (or volume) loss vs. the time of the test. Eisenberg, et QI.~” have expressed the cumulative mass loss plot on the basis of the rate vs. time curve as follows: 1. Incubation zone (little or no mass loss). 2. Accumulation zone (increasing rate to a maximum). 3. Attenuation zone (decreasing loss rate to a steady-state value). 4. Steady-state zone (loss rate at a constant value).
It has been proposed that evaluation of the resistance of materials, or the study of experimental variables, should be based on the results obtained for the attenuation zone. Other methods of assessment have been proposed by Hobbs2”, and by Plesset and Devine252. Examples of various vibratory test procedures for studying cavitation: erosion of metals in inhibited engine coolants have been given in an ASTM Special Technical P ~ b l i c a t i o n ~ ~ ~ - ~ ~ ~ .
Fretting Corrosion The deterioration of surfaces that occurs when parts supposedly tightly fitted together nevertheless move slightly relative to each other in some sort of cycle under load is called fretting corrosion (see Section 8.7). With ferrous materials the characteristic corrosion product is a finely divided cocoacoloured oxide. The general state of knowledge of the subject was reviewed in a symposium on fretting corrosion held by the ASTM in 1952256and more recently by W a t e r h o u ~ e ~ ~ ~ ~ ~ ’ ~ . Several techniques for reproducing fretting corrosion have been used. All involve some means for controlling contact pressure, and for achieving and measuring small-amplitude cyclic motion or slip between the contacting surfaces; some control of the environment, particularly moisture which has a considerable effect on the extent of damage, is also desirable. Finkzs9 used an Amsler wear machine. Another early series of tests on fretting corrosion arose from a study of the bottom bearings of electricity meters by Shotter 260. Tomlinson, Thorpe and Gough26’adapted a Haigh alternatingstress machine by which annular specimens were pressed together under load while being subjected to vibration to achieve the required slip. These investigators also used apparatus in which a specimen having a spherical surface was moved cyclically through a small amplitude while in contact under load with a plane surface. A further modification involved an upper specimen machined to provide an annulus which was oscillated under load in contact with a lower plane specimen. A similar technique was used by Wright262.The area of damage was measured optically and the maximum depth of damage was calculated by carefully lapping the lower surface and determining the change in mass. In addition, the amount of oxidised debris was determined chemically. Uhlig, Tierney and McClellanZa measured fretting damage by mass loss of recessed 25 - 4 mm diameter steel cylinders subjected to radial oscillating motion. The specimens were loaded pneumatically, frequency was varied,
19: 84
CORROSION TESTING AND CORROSION RATES
and slip was adjusted up to 0.020 mm. Mass loss was determined after debris had been removed by pickling the specimens in inhibited acid. McDowel1261used a set-up which took advantage of the elastic modulus of one of the test materials to provide a definite deflection subject to control. A rotating-beam fatigue-testing machine was used to produce an alternating compressive and tensile deflection on the surface of the rotating specimen. A sliding specimen slipped back and forth on the rotating specimen as the outer fibres were strained alternately in tension and compression in proportion to the extent of deflection of the rotating specimen. Horger 2’6 undertook rotating-beam fatigue tests of press-fitted assemblies using specimens as large as 305 mm diameter shafts. W a r l ~ w - D a v i e sused ~ ~ ~a technique in which specimens were subjected to fretting corrosion and then tested in fatigue to show the effect of fretting damage in lowering resistance to fatigue. Herbeck and Strohecker 256 used machines designed particularly for comparing the merits of lubricants in preventing fretting corrosion of antifriction bearings. One provided for both oscillating conditions and combination radial and thrust loads to simulate service. Another was concerned primarily with thrust bearings and correlated satisfactorily with the radial load tester. An interesting approach involved microscopic observation of fretting corrosion; a glass slide mounted on the stage of a microscope was used for the bearing surface which pressed against a spherical specimen being vibrated by a solenoid 266. Other testing machines and techniques have been described by Gray and Jenny267Villemeur268,Wright269,270, Barwell and Wright”’, Field and Waters 272, and Waterhouse 273.
Corrosion Testing in Liquid Metals and Fused Salts* Liquid metals have high heat capacities and heat transfer coefficients, and these and other properties make them attractive as coolants for hightemperature nuclear reactors and as heat-transfer and working fluids in power-generation systems that operate in conjunction with nuclear reactors. However, austenitic cladding and ferritic structural steels can suffer rapid corrosion when exposed to liquid metals at high temperatures (e.g. in liquid sodium at temperatures above 600°C or in liquid Pb-17atVo-Li eutectic alloy at temperatures above 500°C). Similar corrosion processes affect numerous solidlliquid metal systems, including molybdenum in liquid sodium or lithium, stainless steel in liquid aluminium, platinum in liquid sodium and carbon steel in liquid zinc. Corrosion by liquid metals is usually controlled by diffusion processes in the solid and liquid phases and, unlike aqueous corrosion, does not generally involve galvanic effects, and, even where electrochemical phenomena are known to occur, it has not, in general, been demonstrated that they have been responsible for a significant portion of the corrosion observed274.In ‘See
also section 2.10.
CORROSION TESTING AND CORROSION RATES
19:85
fused salts, there is evidence that electrochemical factors are i n v ~ l v e d ~ ~ ~ ~ ~ ~ ~ . Nevertheless, the corrosion process in relation to liquid metals and fused salts may conveniently be considered under one of the following processes which do not directly include electrochemical factors: 1. chemical reaction; 2. simple solution; 3. mass transfer; 4. impurity reactions. Several of the above processes may be involved in a single corrosion reaction, but for simplicity they will be treated separately.
Chemical reaction This involves the formation of distinct compounds by reaction between the solid metal and the fused metal or salt. If such compounds form an adherent, continuous layer at the interface they tend to inhibit continuation of the reaction. If, however, they are non-adherent or soluble in the molten phase, no protection will be offered. In some instances, the compounds form in the matrix of the alloy, for example as grain-boundary intermetallic compound, and result in harmful liquid metal embrittlement (LME) although no corrosion loss can be observed. Simple solution The liquid phase may simply dissolve the solid metal or the liquid may go into solid solution with the metal to form a new phase. In some instances, only a single constituent of an alloy will dissolve in the liquid phase; in this case, a network of voids extending into the metal will result, with obvious deleterious effects. Mass transfer This phenomenon manifests itself as the physical transport of a metal from one portion of the system to another, and may occur when there is an alloy compositional difference or a temperature gradient between parts of the unit joined by the flowing liquid phase. An exceedingly small solubility of the metal component or corrosion product in the molten metal or salt appears sufficient to permit mass transfer to proceed at a fairly rapid pace.
Impurity reactions Small amounts of impurities in the liquid phase or on the surface of the solid metal may result in the initiation of attack or in increased severity of attack by one of the mechanisms just outlined. In general, it is fair to state that one of the major difficulties in interpreting, and consequently in establishing definitive tests of, corrosion phenomena in fused metal or salt environments is the large influence of very small, and therefore not easily controlled, variations in solubility, impurity concentration, temperature gradient, et^.^^^. For example, the solubility of iron in liquid mercury is of the order of 5 x at 649"C, and static tests show iron and steel to be practically unaltered by exposure to mercury. Nevertheless, in mercury boiler service, severe operating difficulties were encountered owing to the mass transfer of iron from the hot to the cold portions of the unit. Another minute variation was found substantially to alleviate the problem: the presence of lOppm of titanium in the mercury reduced the rate of attack to an inappreciable value at 650°C; as little as 1 ppm of titanium was similarly effective at 454°C 278. In the case of the alkali metals, impurities such as oxygen and carbon can have a significant effect on the corrosion of steel and refractory metals.
19:86
CORROSION TESTING AND CORROSION RATES
Borgstedt and Frees2" have shown that for the corrosion of both stabilised and unstabilised austenitic stainless steels in flowing liquid sodium at 700°C there is an almost linear dependence of the corrosion constant, k, on the oxygen content of the sodium, as follows: log k = - 5.6637
+ 0.919 log [ O ]
. . .(19.17)
where k is in mg cm-2h-' and [O] is in ppm). Barker et a1.'" have demonstrated that oxygen exerts a similarly deleterious effect on the corrosion of AIS1 Type 316 austenitic stainless steel in liquid Pb-17Li eutectic by increasing the depth of the ferritic corrosion layer and the extent of chromium depletion within the layer. The effect of carbon on the corrosion of stainless steels in liquid sodium depends upon the test conditions and the composition of the steels279. Stabilised stainless steels tend to pick up carbon from sodium, leading to a degree of carburisation which corresponds to the carbon activity in the liquid metal. Conversely, unstabilised stainless steels suffer slight decarburisation when exposed to very pure sodium. The decarburisation may promote corrosion in the surface region of the material2" and, under creep rupture conditions, can lead to cavity formation at the grain boundaries and decreased strength. Testing
As in all corrosion testing, the procedure which most nearly duplicates the conditions anticipated in service will provide the most satisfactory and useful information for those aspects of corrosion under consideration here. In fact, in view of the extraordinary sensitivity of fused metal and salt corrosion phenomena to minute variations in operating conditions and purity of components, as already discussed, failure to reproduce these conditions with considerable accuracy may well make any test results completely unrealistic and worthless. In all of the following, then, it should be understood, if not explicitly stated, that all extraneous matter must be carefully excluded from the system and that only materials closely simulating those to be employed in service (including prior history and surface preparation of the metals) should be used. Other factors affecting the corrosion in liquid metals and fused salts include the heat flux of the corroding surface, the volume of liquid to the surface area of the solid, the heat flux of the corroding surface and the liquid flow rate. If, however, screening tests to establish the compatibility of a relatively large number of metals with a given molten metal or salt are to be run, it is often useful to commence with static.tests even though the ultimate application involves a dynamic system. This is desirable because static tests are comparatively simple to conduct and interpret, and considerably more economical to operate, and because experience has shown that a metal which fails a static test is not likely to survive the more severe dynamic testzg2.Static tests have been used by Grabner to investigate the compatibility of metals and alloys in liquid Pb-Li eutectic at temperatures up to 650°C. Static Tests Ideally, a static test would consist of immersing a test sample
in the liquid medium held in an inert container under isothermal conditions.
CORROSION TESTING AND CORROSION RATES
19:87
Tests in mercury, for example, may be contained in glass at temperatures of several hundred degrees2". Unfortunately, at the higher temperatures and with the aggressive metals and salts of interest there are few readily available inert container materials, and results will often vary according to the nature of the container. The most satisfactory solution is to make the container of the same material as the test sample, or even in some cases to let it be the sample. Klueh used small capsules for determining the effect of oxygen on the compatibility of Nb and Ta with sodiumtB5and potassiumts6. For the Nb-K tests the Nb specimen was approximately 2.5 x 1 - 4 x 0.1 cm and was contained in a Nb capsule surrounded by another capsule of welded Type 304 stainless steel. It was demonstrated that the oxygen concentration, added as K20, markedly increased the solubility of the Nb in the molten K. DiStefano2" studied the interaction of Type 316 stainless steel with Nb (or Nb-1Zr) in Na and Na-K by exposing tensile specimens of Nb (Nb-1Zr) to the liquid metal in a stainless steel container. Carbon and nitrogen from the stainless steel were transferred to the Nb resulting in carbide-nitride at the surface and diffusion of nitrogen into the metal, thus producing an increase in tensile strength and a decrease in ductility. Close control of temperature is also essential if reproducible results are to be obtained, because of differential solubility as a function of temperature. For example, the corrosion rate for Cu-Bi at 500 f 5-0"Cis several times its rate at 500 f 0.50C2".
Refluxing capsules In systems where a liquid metal is used as the working fluid, the liquid is converted to vapour in one part of the system whilst the converse takes place in another, and the effect of a boiling-condensing metal on the container materials is most readily studied in a refluxing capsule. DiStefano and D e VanB9 used a system in which the lower part of the capsule was surrounded by a heating coil whilst the upper part was water cooled. Specimens were inserted in the upper part of the capsule and thus exposed to the condensing vapour, the rate of condensation being controlled by the water flow rate. When close control of purity is essential it may be necessary to assemble the test specimens in a dry box under an inert atmosphere and to weld the containers shut under inert gas or vacuum before placing on test. With some environments even the small amount of oxygen and moisture adsorbed on the component surfaces will significantly affect the test results. In one laboratory this problem was eliminated by maintaining within the dry box a container of molten sodium at 25OoCtB2-a rather cumbersome procedure, but one which emphasises again the importance of purity. Static test results may be evaluated by measurement of change of mass or section thickness, but metallographic and X-ray examination to determine the nature and extent of attack are of greater value because difficulty can be encountered in removing adherent layers of solidified corrodent from the surface of the specimen on completion of the exposure, particularly where irregular attack has occurred. Changes in the corrodent, ascertained by chemical analysis, are often of considerable value also. In view of the low solubility of many construction materials in liquid metals and salts, changes in mass or section thickness should be evaluated cautiously. A limited volume of liquid metal could become saturated early in the test and the reaction would thus be stifled when only a small corrosion loss
19:88
CORROSION TESTING AND CORROSION RATES
has occurred, whereas with a larger volume the reaction would continue to destruction288*2w.
Dynamic Tests Various tests have been devised to study the effects of dynamic conditions and one of the simplest tests is to use a closed capsule that contains a sample at each end and is partially filled with the liquid metal or saltz9’.A temperature gradient is maintained over the length of the tube, and the capsule is rocked slowly so that the liquid metal passes from one end to the other. After the test, the extent of mass transfer is determined from the two specimens placed at each end of the capsule. Tests of this type are useful to establish whether thermal-gradient mass transfer (or concentration-gradient mass transfer if dissimilar metals are incorporated in the system) will occur, but although the method is useful for screening purposes, the dynamic nature of the heating and cooling cycles prevents a rigorous analysis of mass transfer in terms of time and temperature. High velocity effects can also be studied in spin tests using cylindrical specimens of the solid metal and rotating them at high velocities in an isothermal-metal bath. Although strictly speaking only a single alloy should be tested at a time, it is generally satisfactory to include a variety of alloys since the velocity effects become manifest at considerably shorter times than does mass transfer. KassnerZ9’used a rotating disc, for which the hydrodynamic conditions are well defined, to study the dissolution kinetics of Type 304 stainless steel in liquid Bi-Sn eutectic. He established a temperature and velocity dependence of the dissolution rate that was consistent with liquid diffusion control with a transition to reaction control at 860°C when the speed of the disc was increased. The rotating disc technique has also been used to investigate the corrosion stability of both alloy and stainless steels in molten iron sulphide and a copper/65% calcium melt at 1220°C293.The dissolution rate of the steels tested was two orders of magnitude higher in the molten sulphide than in the metal melt.
Loop Tests Loop test installations vary widely in size and complexity, but they may be divided into two major categories: (a) thermal-convection loops; and (b) forced-convection loops. In both types, the liquid medium flows through a continuous loop or harp mounted vertically, one leg being heated whilst the other is cooled to maintain a constant temperature across the system. In the former type, flow is induced by thermal convection, and the flow rate is dependent on the relative heights of the heated and cooled sections, on the temperature gradient and on the physical properties of the liquid. The principle of the thermal convective loop is illustrated in Fig. 19.26. This method was used by De Van and Sessionszwto study mass transfer of niobium-based alloys in flowing lithium, and by De Van and J a n ~ e n ’to~ determine ~ the transport rates of nitrogen and carbon between vanadium alloys and stainless steels in liquid sodium. The thermal-convection loops are limited to flow velocities up to about 6 cm s-’. Where higher velocities are required, the liquid must be pumped, either mechanically or electromagnetically; the latter is usually preferred as it avoids the problem of leakage at the pump seal. Basically, these forcedconvection systems295-z98 consist of (a) a hot leg, where the liquid metal is
CORROSION TESTING AND CORROSION RATES
19 :89
Specimen port and sampling port
Corrosion specimen hot zone
Drain valve
Fig. 19.26 Loop test for studying the corrosion produced by molten metals or salts
heated to the maximum temperature, (b) an economiser or regenerative heat exchanger and (c) a cold leg, where the liquid is cooled to its minimum temperature. The economiser consists of concentric tubes with the hotter liquid flowing through the inner tube whilst the cooler liquid flows in the opposite direction through the annulus between the two tubes, thus minimising power requirements. The material under test may be used for constructing all parts of the loop, and the loop is then destructively examined after a given period of test. However, this is costly and it is now usual practice to use the loop as a permanent testing facility and to test specimens that are generally placed in the hot leg. Assessment of corrosion is based on changes in weight, dimensions, composition, mechanical properties and microstructure. The final stage in a testing programme is the design, construction and testing of loops that simulate the type of system for which data are required. Because sodium, which is liquid between about 100°C and 881"C, has excellent properties as a heat-transfer medium, with a viscosity comparable with that of water and superior heat conductivityzw, much attention has been paid to liquid sodium corrosion testing of metal and alloys. Indeed, ASTM have issued a Standard Practice which can be used for determination
19:W
CORROSION TESTING AND CORROSION RATES
of the corrosion of ferrous alloys, austenitic stainless steels, high nickel alloys and refractory metals in pumped flowing sodium (ASTM G68:1980). This includes guidance on the monitoring and control of impurity levels in liquid sodium. The oxygen content of the liquid sodium can be measured continuously by an electrochemical oxygen meterJm. Similar electrochemical sensors have been used to monitor the carbon content of liquid sodium30' and the oxygen content of liquid Li-17Bi eutectic302.The purity of the liquid metal can be maintained by means of a cold trap through which a small part of the flow is continuously bypassed, the purity level being determined by the temperature of the trap. The ASTM Standard Practice gives the following relationship between the cold trap temperature and oxygen content of the liquid sodium: log,,C(ppm oxygen) = 7.0058 - 2820/T (K)
. . .(19.18)
and recommends that the oxygen level of liquid sodium be lowered to 2-85 ppm or less, corresponding to a cold trap temperature of about 150°C.
Borgstedt and Freesm found that a cold trap operating at 125°C further reduced the oxygen content of liquid sodium to 1-2ppm and acted as a sink for carbon, reducing the level of this element to about 0.01 ppm. The maintenance of low impurity levels in the liquid metal is facilitated if the inert cover gas in the expansion chamber is of high purity (e.g. 299-996'70 argon).
Evaluation of loop-test results Although the thermal loop test approximates to the conditions which obtain in a dynamic heat-transfer system, in evaluating the results it is necessary to be aware of those aspects in which the test differs from the full-scale unit, as otherwise unwarranted confidence may be placed in the data. Assuming that adequate attention has been paid to the purity and condition of components, etc., the following factors will, according to ASTM G68: 1980, influence the observed corrosion behaviour: 1. liquid metal temperature; 2. degree of non-isothermality of the liquid metal system; 3. liquid metal flow rate; 4. heat flux at the corroding surface; 5 . surface-area/volume ratio of solid metal/liquid metal; 6. relative sizes of dissimilar metal surface areas exposed to the liquid metal at the various system temperatures.
The relation between corrosion, and maximum temperature and temperature gradient is obvious, since solubility varies as a function of temperature. If the results are to be useful, these factors should match those anticipated in service. Erratic temperature cycling should be avoided as this can also be modify the corrosion behaviour. The effect of surface-to-volume ratio will be more pronounced in thermal-convection than in pump loops. It can readily be seen that if a relatively small volume of liquid passes through a given isothermal segment of loop per unit time it will become saturated quickly and the corrosion rate will appear lower than would be the case if a substantially larger volume of liquid were passing at the same velocity. In a pumped loop, the velocity can be maintained sufficiently high to prevent the attaining of equilibrium between the solid and liquid phases, and the
CORROSION TESTING AND CORROSION RATES
19:91
rate of dissolution of the solid will be the controlling step. The flow velocity, or Reynolds number, will affect this step too, in that increased velocity will decrease the stagnant or lamellar layer adjacent to the tube wall and decrease the diffusion path that particles must negotiate to enter the rapidly moving stream2**.The turbulence of the flow may also be modified by the manner in which test specimens are inserted in the loop, and this should also be considered carefully in designing a test unit. The corrosion rates of the materials of construction are always of importance, but it has been found that, whereas the uniform removal of metal from the hot leg may not impair the load-carrying ability of the container, the deposition of metal in the cold leg can cause the cessation of flow, and the measure of the suitability of an alloy is often the time, under given conditions, that it takes for plugging to occur. Again, the flow velocity and the cross-sectional area are of primary importance in relating test results to operating conditions. The ultimate test, short of constructing a full-scale unit, is t o build a smallscale system in which each item to be incorporated in the final device is represented. Such programmes are too specialised to warrant discussion here, and are fully described in the l i t e r a t ~ r e ~ ’ ~ - ’ ~ . Liquid-Metal Embrittlement
Metals have sometimes been observed to crack almost instantaneously when wetted by certain molten metals and subjected to plastic strain at temperatures far below those at which the diffusion-ruled processes involved in liquid-metal corrosion attain significance3”. The fracture appears to be more brittle than in the absence of the liquid metal, leading to decreased elongation and reduction of area values and, in severe cases, brittle intergranular fracture. Like other forms of environmental cracking, liquid-metal embrittlement is highly specific according to alloy and environment. For example, molten zinc can cause liquid-metal embrittlement of stainless steel if the oxide film is damaged, and because of this molten zinc from associated galvanised parts poses the greatest hazard in welding stainless steel equipment 308. Other well known examples of liquid-metal embrittlement include the effects of solder on copper alloys and carbon steels and those of mercury on aluminium and nickel alloys. It is generally accepted that most cases of liquid-metal embrittlement arise from the effects of chemisorption of liquidmetal atoms and the consequent reduction of the tensile strength of interatomic bonds at the crack tip since rates of crack growth (up t o lOcm s-’) are usually rapid compared with rates of diffusion of embrittling atoms ahead of cracks or dissolution of the solid in the liquid metal’09. However, there are a few cases where diffusion of embrittling atoms ahead of cracks or selective dissolution of a particular phase of an alloy can produce degradation of materials in liquid-metal environments 310. Prerequisites for liquid-metal embrittlement are that a solid metal should be subjected to tensile plastic strain while wetted by a liquid metal in which it has low solubility. It has been suggested that such embrittlement may be a general phenomenon occurring under appropriate conditions and to varying degrees between all solid-metal Aiquid-metal couples and that a
19:92
CORROSION TESTING AND CORROSION RATES
single mechanism may be responsible for all liquid metal embrittlement failures3". The occurrence and severity of the embrittlement are governed by: 1. the particular solid-metal/liquid-metal combination; 2. the temperature; 3. the strain rate; 4. the initial mechanical and metallurgical state of the solid metal.
The most commonly used method for assessing liquid-metal embrittlement is by tensile deformation at a slow strain rate. During testing, the specimen should be immersed in the liquid metal in a sealed autoclave to avoid contamination by atmospheric gases3". Electrochemical probes similar to those employed in liquid metal corrosion testing can be used to monitor the purity of the liquid metal. Susceptibility to liquid-metal embrittlement can be assessed in terms of the uniform elongation, reduction in area and fracture appearance of the specimen relative to that determined under similar testing conditions in an inert environment at the same temperature. Where information on crack-propagation behaviour is required, use can be made of pre-cracked specimens. These can be tested under static or cyclic loading conditions to determine threshold stress intensity factors and crack growth rates 'I2.
Tests in Plant Although laboratory tests (NACE TMO 169-76, and Reference 313) are obviously of value in selecting materials they cannot simulate conditions that occur in practice, and although an initial sorting may be made on the basis of these tests ultimate selection must be based on tests in the plant. This is particularly important where the process streams may contain small concentrations of unknown corrosive species whose influence cannot be assessed by laboratory trials. Testing is also important for monitoring various phenomena such as embrittlement, hydrogen uptake, corrosion rates, etc. which are considered in Section 19.3. Corrosion Racks
Exposure of coupons or specimens to the process stream cannot be achieved satisfactorily unless they are rigidly supported in a rack, although in some cases it may be possible to simply hang them in by means of a wire. Methods of exposure coupons are described in ASTM Method G4:1984. In the birdcage rack, disc specimens are mounted on a central rod, and are insulated from each other and from the rod by insulating spacers and an insulating tube, respectively. P.t.f.e. has been found to be suitable for this purpose in aggressive media, particularly at high temperatures. Plates at the end of the rack act as bumpers to prevent the specimens touching the side walls, and the assembly is constructed from a corrosion resistant material such as Monel. Advantages of this method are (1) electrical insulation avoids galvanic effects and (2) the method of holding the specimen at
19:93
CORROSION TESTING AND CORROSION RATES
the centre avoids losses due to corrosion around the point of support. The disadvantages are (1) specimens are not subjected to either heating or cooling effects and thus will not disclose ‘hot-wall’ effects and may also escape corrosive condensates when the specimens are in a vapour stream above the dew point, and (2) the corrosivity of the environment may be affected by the presence of corrosion products of the construction material of the racks or by corrosion products of adjacent specimens. A further disadvantage is that because of its size and shape it must be inserted into the process stream when the plant is out of service. Special devices are required for mounting specimens within pipelines so that they will be subjected to velocity effects. The insert rack is designed for easy installation and removal through an unused nozzle. The supporting rods (one for each specimen) are welded to a single support plate that is of a width that enables it to be introduced through the nozzle. However, this too cannot be inserted unless the equipment is out of service, although its introduction does not require removal of gas. A slip-in rack is described by Dillon, et u I . ” ~ (Fig. 19.27) that is designed to be inserted and removed during the operation of the plant through a full-port gate valve attached to a nozzle of suitable diameter (3 8-5 * 1 cm). It consists of a short length of pipe flanged at one end to match the gate valve and having a backing-gland arrangement at the other. The coupons are mounted on a rod of small diameter welded to a long heavier rod. The valve is opened and the support rod is pushed through the packing gland so that the specimen is introduced into the process stream. The specimens are removed by withdrawing the rod until they are again within the pipe section, the gate valve is closed and the rack removed from the valve. Access fittings are available (e.g. Cosasco* access fitting) that enable specimens to be introduced into plant that is operating at high pressures, but can also be used for ambient pressures (see Section 19.3). In some instances it is possible to secure valuable information by substituting experimental materials for parts of the operating equipment, a practice that is used most frequently with condenser tubes, evaporators or other heat exchangers or sections of piping systems. The prediction of materials performance in plant conditions using modelling and corrosion test methods has been discussed by Strutt and Nichols3”.
-
Specimens
-
A convenient size for a circular coupon is 3 8 cm dia., a thickness of 0.32 cm and a central hole of 1- 1 cm. Although inherent in the philosophy of corrosion testing, the use of coupons with surfaces that simulate those in service has been found to be unsatisfactory owing to irreproducibility, and the standard procedure normally adopted is to abrade down to 120-grit. ASTM Method G4: 1984 gives details of preparation of specimens, evaluation of replicate exposures and the application of statistical methods. *Grant Oil Tool Company.
19:94
CORROSION TESTING AND CORROSION RATES
Fig. 19.27 Slipin corrosion test rack (after Dillon et a/.314)
Atmospheric Tests More or less standardised techniques have been developed for the exposure of specimens to atmospheric weathering. ASTM G50:1976 (R1984) and I S 0 8565: 1992 provide guidance on conducting atmospheric corrosion tests on metals, alloys and metallic coatings. Procedures for recording data from atmospheric corrosion tests on metallic-coated steel specimens are given in ASTM G33:1988. The usual practice in the USA3I6 is t o mount bare specimens on racks that slope 30" from the horizontal and painted specimens on racks that slope 45" from the horizontal. The usual orientation is to have the specimens face south. In coastal exposures it is not uncommon to have the specimens face the ocean. Steel specimens exposed vertically have been found t o corrode about 25% more than similar specimens exposed at the 30" angle3". Vertical exposure was used in the large-scale tests of non-ferrous metals undertaken by Subcommittee VI of ASTM Committee B-3318.Vertical exposure is also favoured by Hudson3". A typical test installation uses a frame to support racks on which the specimens are mounted by means of porcelain or plastics insulators. The insulators may be spaced to take specimens varying in size from 10.1 x 13.4cm to 10.1 x 32cm and even larger specimens may be used for certain tests. Special types of exposure have been devised to take into account important effects of partial shelter and accumulation of pools of water, as in the case of the specimen and method of support used by Pilling and Wesley3" to compare steels for roofing. C o p s ~ n has ~ ~ 'described in considerable detail the several factors that require attention in studying atmospheric corrosion, particularly of steels. Several sizes and shapes of specimens have been used in addition to the common ones already mentioned. In the long-time test of bare and zinccoated steels undertaken by ASTM Committee A-5 on Corrosion of Iron and Steel, full-size sheets were used3". This Committee has also exposed specimens in the form of hardware323and wire and fencing324. The extent of deterioration may be measured by one or more of the following methods: visual examination, change in weight, change in tensile properties. Visual inspection was depended upon primarily in the A-5 tests
CORROSION TESTING AND CORROSION RATES
19 :95
of steel sheets'". Here, visible perforation more than 6mm from an edge was the criterion of failure. This leaves much to be desired for close comparisons because of the frequency with which perforations may be obscured by heavy coats of rust 'I7. Other shortcomings of the use of time to visible perforation as the criterion of corrosion resistance are as follows. 1. The removal of rust films or other corrosion products to facilitate inspection for perforation prior to termination of the exposure will change the natural performance of the material, and is therefore not tolerable. 2. The recording of a perforation establishes only the time to failure and provides no idea of the progress of corrosion up to the point of failure. 3. The time to perforation may be influenced considerably by the random occurrenceof pits that happen to meet after starting from opposite sides of a sheet. This chance meeting of pits may be determined only to a slight extent by the composition of the material and, therefore, will interfere with observations of the effects of composition. Where changes in appearance are of paramount interest, as in the case of metallic and organic coatings on steel or other metals, visual examination is most desirable. To facilitate ratings on such a basis, photographic standards have been employed, as, for example, in tests on chromium-plated steel undertaken by ASTM Committee B-8 on Electrodeposited Metallic C ~ a t i n g s ' ~These ~ . ratings are supplemented by a shorthand description of the nature of the deterioration observed. Similarly, photographic standards are recommended for rating organic coatings with respect to different modes of deterioration in ASTM
D6101985. The most precise measurements of corrosion resistance require the use of specimens that can be weighed accurately after careful removal of corrosion products by the techniques described earlier. A sufficient number of specimens should be exposed initially to permit their withdrawal from test in appropriate groups, for example 3 to 5 duplicates after at least three time intervals. For long-time tests, a suitable schedule would call for removals after 1, 2, 5 , 10 and 20 years. It is good practice to determine depths of pitting as well as mass loss. As is the case with other types of corrosion testing, mass-loss determinations may fail to indicate the actual damage suffered by specimens that are attacked intergranularly or in such a manner as dezincification. In such cases, mechanical tests will be required as discussed already in the section on evaluation techniques. It is desirable for reporting of atmospheric corrosion tests to include a precise description of the climatic conditions that prevailed at the test site during the test so that the weather factors can be tied in with the results of exposure. Progress towards this aim has been made recently with the development of international standards which provide guidance on evaluating the corrosivity of atmospheric environments. Atmospheric corrosivity can be expressed in terms of environmental factors, the most important of which have to do with contaminants of the atmosphere and the time that the specimens are actually wetted by condensed moisture. In the case of organic coatings, the interacting effects of sunlight and moisture, and their
19:96
CORROSION TESTING AND CORROSION RATES
sequence, complicate this problem even more326. Methods of measuring pollution (deposition rates of sulphur compounds and chlorides) are provided in ISO/DIS 9225: 1989. ISO/DIS 9223: 1989 defines five different categories of atmospheric corrosivity based on time of wetness and pollution. An alternative approach is to express atmospheric corrosivity in terms of the corrosion rate of standard materials, including carbon steel, weathering steel, zinc, copper and brass. Methods of determining the corrosion rates for this purpose are given in ISO/DIS 9226:1989, while ISO/DIS 9224: 1989 provides a classification of atmospheric corrosivity based on both the average and the steady-state corrosion rates of the standard metals. Sereda”’ has described a method of determining time of wetness in which a strip of platinum foil (0.8 x 7cm) is mounted on a zinc panel (10.1 x 13.4cm) on both the skyward and groundward face. Condensed moisture from dew, or rain or snow, results in a galvanic cell whose potential is monitored on a recorder, thus giving the time of wetness. Guttman and Sereda found that if the SO, content remained essentially constant the corrosion rate of zinc was related to time of wetness; furthermore, the dew detector registered the presence of moisture on the panel when the relative humidity ranged between 82 and 89%, thus providing a means of estimating from long-term weather data, such as temperature and relative humidity, the time a specimen is likely to be wet.
’”
Atmospheric Galvanic Tests Studies of galvanic corrosion in the atmosphere are experimentally simpler than those conducted in solution in the laboratory. The environment is taken as it comes and the relatively high electrical resistance of the rain and moisture films that serve as electrolytes restricts the distance through which the galvanic action can extend, and thus limits the relative area effects that complicate galvanic corrosion in solutions of high conductivity. Standard test methods for assessing galvanic corrosion caused by the atmosphere are given in ASTM G104:1989 and in IS0 7441:1984 (see also Section 1.7). Because of the limited proportion of the areas of a couple that actually participates in the galvanic action, it is difficult to make quantitative measurements that separate the galvanic action from the total effects of exposure. Thus many of the observations are likely to be qualitative ones, and often no more than what can be determined by visual inspection or measurements of changes in strength, etc. as a result of any localised galvanic action. An idea of the distribution of galvanic corrosion in the atmosphere is provided by the location of the corrosion of magnesium exposed in intimate contact with steel in the assembly shown in Fig. 19.28 after exposure in the salt atmosphere 25 m from the ocean at Kure Beach, North Carolina, for 9 years. Except where ledges or crevices may serve to trap unusual amounts of electrolyte, it may be assumed that, even with the most incompatible metals, simple galvanic effects will not extend more than about 4-5 mm from the line of contact of the metals in the couple.
CORROSION TESTING AND CORROSION RATES
h
19 :97
- #
Fig. 19.28 Distribution of galvanic effects around contact of a magnesium casting and a steel core
The extent of galvanic action in atmospheric exposure may also be restricted by the development of corrosion products of high electrical resistance between the contacting surfaces - this is especially likely to occur if one of the metals in the couple is an iron or steel that will rust. In long-time tests such possible interruptions in the galvanic circuit should be checked by resistance measurements from time to time so as to determine the actual periods in which galvanic effects could operate. The test assembly used originally by Subcommittee VI11 of ASTM Committee B-3 in its comprehensive studies of atmospheric galvanic corrosion 329 had the disadvantage that it depended on paint coatings to confine corrosion to the surfaces in actual contact with each other. In interpreting the results, it was frequently difficult to decide how much corrosion was due to galvanic action and how much to a variable amount of normal corrosion through failure of the paint system. These difficulties were overcome in a design developed by Subcommittee VI11 of ASTM Committee B-3330(Fig. 19.29). In this assembly each of the two middle specimens has a specimen of the other metal each side of it and only these middle specimens are considered in appraising the results. A fairly direct way of observing galvanic effects, which also permits changes in mechanical properties to be measured, involves the preparation of a composite specimen formed by attaching a strip, or strips, of one metal to a panel of another one. Tensile test specimens that include the areas of galvanic action can be cut from these panels after exposure, as shown in Fig. 19.30.
19:98
CORROSION TESTING AND CORROSfON RATES
Fig. 19.29 Atmospheric galvanic couple test assembly
KEY ( I ) Bakelite washer 19-0 x 3.2 mm. (2) Metal B disc 30 x 1 e 6 mm. (3) Metal B disc 36.6 x 1 e6 mm. (4) Bakelite washer 35.5 x 3 . 2 rnm. (5) Stainfess steel lock washer. (6) Stainfess steel bolt 4 . 8 x 38. I mm. (7) Stainless steel washer 15.9 mm 0.d. (8) Metal A disc 25.4 x 1 - 6mm. (9)MetalAdisc35.5 x 1.6mm.(10) 1 1 . 1 mmBakelitebushing,5.2rnmi.d. x 7 . 9 m m o . d . (11) Stainless steel washer 15.9 mm 0.d. (12) Galvanised angle support
A modification of the specimen shown in Fig. 19.30 may be made simply by lapping a panel of one material over a panel of another one. The greatest effects may be observed when such panels are exposed with the laps facing up so as t o favour retention of corrosive liquids along the line of contact. To permit observations of secondary effects of corrosion products, or exhaustion of corrosive constituents, the relative positions of the dissimilar metals should be changed from top to bottom in duplicate test assemblies. Where the practical interest is in possible galvanic effects of fastenings, it is simple to make up specimens to include such couple assemblies as illustrated in Fig. 19.31. A type of assembly calculated to favour maximum galvanic action was developed by the Bell Telephone Laboratories and is illustrated in Fig. 19.32. Here, the less noble metal is in the form of a wire wound in the grooves of a threaded specimen of the metal believed to be more noble. Good electrical contact is achieved by means of set screws covered with a protective coating. This assembly favours accumulation of corrosive liquids around the wire in the thread grooves. Corrosive damage is also favoured by the high ratio of surface to mass in the wire specimens. To determine whether a protective metallic coating will retard or accelerate corrosion of a basis metal, and to what distance either effect will extend, specimens in which strips of various widths are left bare or made bare have been used by Subcommittee 11 of ASTM Committee B-8330.The extent of corrosion in and near the bare strips as compared with that on a
19:W
CORROSION TESTING AND CORROSION RATES
101 6 m m
*-
1
1 58 rnm
125 4 mn
.--+
158 mm thick metal strips
,28,57mm /,
-7.93mm
t-
/
5
c
87 rnrn
$15
Metal f3-d
\
'Metal
',,
8 or C
', \$, Clearance holes for L 76 mm bolt, use NO 8
drill
250 8 mrn
I
/ 1.58 mm thick metal strips , '
..-I
5 15 07 mrn
t
A ,/I' r,
L 76 m m diaheter bolt with nut 12 7 mm long
125 Lrnr
----
Metal A
! Fig. 19.30 Plate and fastening type galvanic couple test specimen
completely bare or completely coated specimen will provide a measure of the extent of galvanic action and the distance through which the effect is able to extend from the edges of the bare strips.
Tests in Natural Waters The corrosion testing of metals in natural waters is most usually conducted in field or service tests since the conditions of flow are important and often rate-determining. Testing will be concerned with mains water (potable water), river-water and sea-water or combinations of these as in estaurine conditions. Test specimens of various geometries will be used, e.g. in the
19: 100
CORROSION TESTING AND CORROSION RATES
Fig. 19.3 I
Specimen for studying galvanic corrosion resulting from fasteners
Anodic element 0 813 rnrn wire
9 52mm-16 USS Thread
38f159rnm ___-
Fig. 19.32 Bolt and wire type atmospheric galvanic couple test specimen
form of wires, plates, tubes, etc., and certain general precautions should be followed. 1. The specimens should be mounted so that they are insulated from their supporting racks and from each other. Such insulation can be achieved by the use of fastening assemblies, such as illustrated in Fig. 19.33. Occasional difficulties have been encountered with this sort of assembly for tests of copper and high-copper alloys because of deposition of copper from corrosion products along the surfaces of the insulating tubes which provided a
CORROSION TESTING AND CORROSION RATES
19: 101
metallic bridge between the specimens and the rack and introduced undesired galvanic effects. The required insulation and support can be provided by use of porcelain or plastics knob insulators in much the same manner as used on atmospheric test racks. A modified design has the advantage of offering less resistance to the flow of water and is less likely to serve as a form of screen to catch debris floating on, or suspended in, the water. Additional details of rack design may be found in the section on sea-water tests in The Corrosion Handbook (Uhlig, Selective Bibliography).
Specimen Monel bolt and washer Bakelite tube and washers Support
Fig. 19.33 Scheme for insulating specimens from metal test racks
2. In the case of corrosion tests in the sea or in other large volumes of water, Le. as opposed to tests in waters flowing within pipes, all the specimens to be compared should be suspended at the same depth or should pass through the same range of depths. Isolated specimens exposed at different depths will not be corroded in the same way as continuous specimens that extend through the total range of depth to be studied. This is especially the case with specimens exposed to sea-water from above high tide to below low tide. Where the behaviour of structures, such as piling, that pass through these zones is to be investigated, the test specimens must be continuous and large enough to extend through the total range in order to take into account differential aeration and other possible concentration cells that may have such a tremendous effect on the results secured’”. For example, in sea-water exposure, isolated specimens of steel exposed in the tidal zone have corroded 10 times as fast as portions of continuous specimens of the same steel in the same zone that extended also below low-tide level’”. 3. The specimens should be oriented so that their flat surfaces are parallel to the direction of water flow and so that one specimen will neither shield an adjacent specimen from effects of water velocity nor create any considerable extra turbulence upstream of it. 4. In tests in sea-water where accumulations of marine organisms are likely, specimens exposed parallel to each other should be spaced far enough apart to ensure that the space between specimens will not become completely
19: 102
CORROSION TESTING AND CORROSION RATES
clogged by fouling organisms. A minimum spacing of 100 mm is suggested. 5 . Wooden racks used in sea-water tests are likely to be subject to severe damage by marine borers. The wood used, therefore, must be treated with an effective preservative, for example creosote applied under pressure, if the test is to extend for several years. Organic copper compound preservatives may suffice for shorter tests, for example 2 or 3 years. Since the leaching of such preservatives may have some effects on corrosion, metal racks fitted with porcelain or plastics insulators have an advantage over wooden racks. 6. Where constant depth of immersion is desired in spite of tidal action, it is necessary to support the test racks from a float or raft. Recommended methods for assessing the corrosivity of waters, including flowing potable waters, are described in ASTM D2688:1983. Three procedures are described in which test specimens in the form of wires, sheets or tubes are placed in pipes, tanks or other equipment. The test assembly for the first of these consists of three helical wire coils mounted in series on, and electrically insulated from, a supporting frame. The assembly must be installed so that flow is not disturbed and turbulence and high velocities, e.g. of more than 1 -53ms-’, are avoided. A minimum test period of 30 days is recommended. Procedures for the other specimen forms are given in the standard. An extensive study of the corrosion of metals in tropical environments has been carried out by Southwell, et ai.”*. Tests have included atmospheric exposure, and exposure in sea-water under mean tide and fully immersed conditions for a range of ferrous and non-ferrous metals and alloys. The Marine Corrosion Working Party of the European Federation of Corrosion has published valuable advice on corrosion testing in ~ervice’~’.
Field Tests in Soil The precautions generally applicable to the preparation, exposure, cleaning and assessment of metal test specimens in tests in other environments will also apply in the case of field tests in the soil, but there will be additional precautions because of the nature of this environment. Whereas in the case of aqueous, particularly sea-water, and atmospheric environments the physical and chemical characteristics will be reasonably constant over distances covering individual test sites, this will not necessarily be the case in soils, which will almost inevitably be of a less homogeneous nature. The principal factors responsible for the corrosive nature of soils are the presence of bacteria, the chemistry (pH and salt content), the redox potential, electrical resistance, stray currents and the formation of concentration cells. Several of these factors are interrelated. These considerations will significantly affect the location of test specimens in field testing. It is clearly important to ensure that the conditions of exposure are accurately known so that the corrosion test results may be interpreted with respect to the end-use requirements. Two civil engineering operations require particular attention when soil corrosion tests in the field are required. These are (1) the use of reinforced earth structures in which the corrosion conditions will differ from those at
CORROSION TESTING AND CORROSlON RATES
19: 103
the site from which the soil has been taken and which may take some time to come to equilibrium in the new site, and (2) the use of reclaimed or contaminated land where unusual corrosive agents may be present in irregular distribution. In both these situations considerable thought should be given to the corrosion test procedures. Soil burial tests are popular despite the precautions that are needed. It is also important that a sufficient number of specimens are exposed so that statistical treatment of the results may be applied to compensate for some of the inevitable variations in the exposure conditions. Certain precautions originally set out in 1937334 are still valid, and are as follows: 1. A sufficient number of specimens to yield a reliable coverage should be included. 2. The test site should be typical of the type of soil to be investigated. 3. The depth of burial should be that which will be occupied by the structure of interest. Specimens to be compared should be buried at the same depth. Ideally, tests for structures, such as piling, that wi11 extend through several horizons would require the use of test specimens long enough to extend to the same depth. 4. Specimens should be separate so that they will not affect the corrosion of each other. A minimum spacing of two diameters was proposed. 5 . Cylindrical specimens should be laid horizontally. 6. Sheet or plate specimens should be placed on edge. 7. The ends of pipe specimens should be closed to prevent internal corrosion. 8. Sufficient specimens should be provided to allow withdrawals after several time intervals so as to permit observations of changes in corrosion rates with time. 9. A portion of the original surface should be protected so as to provide a datum line for the measurement of pit depths. 10. In applying results of tests on small specimens to estimating corrosion, particularly by pitting on large structures, the effect of the increased area in increasing the depth of pitting must be taken into account 335. Other Tests
Other tests to determine bacterial-notably sulphate reducing-activity, soil resistivity, pH, redox potential, etc., will provide valuable data to supplement the results obtained with test specimens. A useful account of some of these was given in Reference 336 and they are also discussed in Sections 2.6 and 10.7. A scheme for assessment of corrosivity of soils based on some of the above parameters has been given by Tiller3”. A number of standards exist for the determination of some of these parameters. BS 1377:Part 3:1990 refers to methods of tests for soils for civil engineering purposes, and Part 9 refers to these and corrosivity tests in situ. It is significant that the standard draws attention to the fact that the results of the tests that are described should be interpreted by a specialist. ASTM tests for pH and resistivity of soil used for corrosion testing are covered by G51: 1977(R1984) and G57:1978:(R1984), respectively.
19: 104
CORROSION TESTING AND CORROSION RATES
Corrosion Testing of Organic Coatings Programmes to evaluate the corrosion protection by organic coatings on metals are intended to establish relationships between coating properties and performance. Such knowledge is essential to the most effective use of organic coating systems in corrosion control. Depending on the detail with which such studies are performed, light may be shed on the mechanism of coating deterioration as well. If valid and useful relationships are to be established, it is essential that the factors affecting performance be recognised and form part of the test record. Since the performance is determined by interactions between the coating, the substrate and the surrounding environment against which protection is sought, significant factors and their interrelationships will vary with the nature of the service. Care in designing and conducting the test in no way reduces the need for discrimination on the part of the person using the test data in the selection of a coating for a particular purpose. Test environments must reflect the deteriorating influences of the service for which they are applicable. A coating system cannot reliably be selected for service in a chemical plant on the basis of performance determined in a rural atmosphere. Thus, both the proper conduct of the testing programme and the valid use of the data depend on an understanding of the nature of organic coatings and of the forces through which they are degraded. Beheviour of Organic Coatings
An organic coating provides corrosion protection through the interposition of a continuous, adherent, high-resistance film between the metal surface and its environment (see Section 14.3). In principle, its function is the mechanical exclusion of the environment from the metal surface. It seldom, if ever, succeeds practically in achieving this since all continuous organic films are permeable to some degree to moisture and many coatings either have occasional physical defects or acquire them in service. Surface conversion treatments, such as phosphate and chromate dips, are used to supplement the physical protective properties of coatings, as are chemically inhibiting primers and wash primers. When such treatments are used, they must be included in the record as constituting a part of the coating system. Paints are considered in detail in Chapter 14, paint failures being discussed in Section 14.4. Critical parts of the test programme are the preparation of test specimens, the selection of the exposure conditions (both in laboratory and field tests) and the selection of significant coating properties to be evaluated as a measure of deterioration with time. Preparation
Specimens will normally be flat panels, large enough to avoid any effects caused by nearby edges of the specimen. Edges and backs are usually coated
CORROSION TESTING AND CORROSION RATES
19: 105
unless the effecc c -incoated edges is an intended test variable. Panels may include the structural features of plates, channels, welds, sharp edges, pits or depressions, depending on the service for which the data are to be applicable. The composition of the basis metal has been found to influence the performance of organic finishes in many cases. Thus, composition is a significant test variable and must be considered in comparing test data. It is particularly important that surface roughness and cleanness, which greatly affect adhesion, should be carefully controlled and that the procedures used to achieve them be a part of the test record. A high degree of cleanliness is normally sought. If, however, the data are to be applicable to the painting of outdoor structures, a certain amount of outdoor weathering becomes a part of the specimen preparation prior to coating. Specimens again will be of the basis metal appropriate for the related service application. The thickness of a coating plays an important part in determining its physical characteristics. Uniformity of thickness among specimens therefore is necessary, particularly when coating deterioration is to be assessed by changes in such properties. For the preparation of reproducible specimens, methods of applying coatings in uniform thicknesses are available, as are methods for accurately measuring film thickness. Exposure Conditions
In considering exposure tests, whether in the form of laboratory, field or service tests it is important to consider the purpose of the test and the relevance of the data to the anti-corrosion function of the coating. Thus, in the case of paint coatings, factors such as gloss deterioration, chalking and colour retention are of considerable importance in some industries, for example the automotive industry, but perhaps of minor importance in the painting of structural steelwork. These assessment factors can nevertheless be of significance since they may be the precursors of corrosion of the basis metal.
Laboratory tests Laboratory tests are often conducted with the purpose of providing an accelerated test procedure and if intelligently used, Le. with proper respect for their limitations, are of value in determining the probable order of durability and hence, by implication, corrosion protection of a group of paints. They can also be of value in assessing the quality of a range of similar compositions where there is already some knowledge of the performance of the general composition. Although continuing attention is given to the correlation of accelerated tests with field trials and service performance, caution must always be exercised in attempting to predict the type of failure likely to occur under conditions of natural exposure. Certainly an approach based on 'the rougher the treatment the better the test' cannot be justified. Three main classes of laboratory test can be identified and may be conveniently classified under the headings of (1) electrochemical (2) coating adherence and (3) exposure cabinets -including weatherometers.
19: 106
CORROSION TESIINC AND CORROSION RATES
Electrochemical tests This group includes the various electrochemical tests that have been proposed and used over the last fifty or so years. These tests include a number of techniques ranging from the measurement of potential-time curves, electrical resistance and capacitance to the more complex a.c. impedance methods. The various methods have been reviewed by Walter”*. As the complexity of the technique increases, i.e. in the above order, the data that are produced will provide more types of information for the metal-paint system. Thus, the impedance techniques can provide information on the water uptake, barrier action, damaged area and delamination of the coating as well as the corrosion rate and corroded area of the metal. However, it must be emphasised that the more comprehensive the technique the greater the difficulties that will arise in interpretation and in reproducibility. In fact, there is a school of thought that holds that d.c. methods are as reliable as ax. methods. Adherence tests This group of techniques involves the testing of the metalto-paint adherence. These techniques are covered by descriptions such as p r ~ h e s i o n blister340, ~~~, pull off (BS 3900:Part E10: 1979(1989)) and crosscut (BS 3900:Part E6: 1976(1989)). Detailed descriptions of these techniques will be found in the appropriate references. Exposure cabinets This group of laboratory tests include the so-called exposure cabinets, salt spray and weatherometer tests in which the paintcoated panels are subjected to various cycles of wetting and exposure to ultra-violet light to simulate atmospheric conditions of exposure. BS 3900: Part F3:1971 (1986) describes a weatherometer consisting of a 1 . 2 m dia. drum that rotates at 1 rev/20 min, and has facilities for spraying the panels (100 x 150 mm) periodically during a 24 h cycle and exposing them to ultraviolet light by means of an enclosed carbon arc. Spraying with distilled water is effected by means of an atomiser and fan using the following 24 h cycle: 4 h off, 2 h on, 10 h off, 2 h on and 1 h off; the final 1 h is used for checking the arc. The test is continued for 7 days at the end of which the panels are examined visually for change of colour, loss of gloss and blistering, and for checking, cracking and chalking by means of a lens ( x 25). An appraisal of artificial weathering methods was given in a report by Hoey and H i p ~ o o d ’ who ~ ’ described the effectiveness of various weatherometer tests such as are described in BS 3900:Part F3 and ASTM E42:1964 (now ASTM G23: 1989). Although these tests simulate atmospheric exposure it is not possible t o obtain a direct correlation owing to variation in outdoor exposure conditions from place to place, but they serve a very useful purpose in providing a preliminary sorting of paints that can then be tested in the field. Field and Plant Tests Field exposure of test panels offers the benefit of a high degree of control over surface preparation and application. Moreover, through standardised exposure conditions, broader comparisons between both paint systems and locations are possible. More importantly, since replicates may be removed and laboratory tested periodically, changes in p:operties can be followed in considerable detail. At least four replicates should be examined for each exposure period to minimise the effects of atypical specimens.
CORROSION TESTING AND CORROSION RATES
19: 107
The exposure site is selected according to the service for which the data are to be applicable. For atmospheric service, such factors as marine and industrial contaminants, sunlight, dew and sand abrasion, must be considered. Atmospheric specimens are normally mounted at 45", facing south. This has been shown to provide about a 2:l acceleration of failure compared with a vertical exposure. Whether this or other standardised positions are used, the details of the exposure are an important part of the test record. The degree of deterioration experienced over a given test period varies with climatic conditions. Since these differ significantly from one season to another, a standard specimen, the performance of which is well known, should be included with each exposure to increase the validity of coating comparisons. For other environments, such as in sea-water or in chemical plants, exposure conditions that most nearly duplicate those of the related service and are at the same time reproducible, are used. Impingement by water or water carrying entrained solids, thermal effects and physical abuse are among the factors to be considered. Coating Evaluation
The performance of organic finishes on test is evaluated by visual observation and by physical tests made upon coated specimens that have been exposed for various periods of time to natural or accelerated weathering conditions. Electrical tests are sometimes used on immersed specimens. Inspection of test panels at the test sites consists of visual observations of blistering (see ASTM D714:1987) and the appearance of rust (see ASTM D610:1985). For these, photographs showing various degrees of degradation which serve as observational standards greatly reduce variations between observers. Results of consecutive observations entered on charts provide visual records of the trend of these features with time. These more serious evidences of degradation, however, are preceded by invisible physical alterations within the coatings which can be detected readily and quantitatively by suitable physical tests of replicate specimens removed from tests at periodic intervals. The use of such tests to reveal incipient coating changes is, in a sense, a means of accelerating the test programme without distortion of the test environment. This approach is especially dependent on uniformity of properties among replicates, hence on reproducible application techniques. Moreover, since many coating properties are highly sensitive to changes in temperature and relative humidity, equilibrium of the specimens during testing is necessary. Testing conditions are commonly 25°C and 50% r.h. Physical tests appropriate for this type of evaluation are not necessarily limited to those properties which the coating may be called upon to display in service. A coating that shows a decrease in distensibility from 20 to 10% is still quite capable of withstanding the expansion and contraction of the substrate in atmospheric temperature cycling, yet such a coating can be expected to fail in service earlier than one which shows no decrease. Thus,
19: 108
CORROSION TESTING AND CORROSION RATES
the properties of value are those that have been established as reliable indices of deterioration. Besides providing early comparative data on coating performance, physical tests in dealing with intrinsic coating properties provide much-needed quantitative information on the relationships between the several factors affecting the ageing of organic films. The tests cited below are those which have been shown to indicate reliably significant changes in the condition of coatings on tests. Distensibility This property is very sensitive to chemical changes within the coating. Its measurement thus shows the beginning of normal ageing or of deterioration through reaction with the environment. Distensibility is generally determined by bending the best panels over a conical mandrel of known radius and calculating the 070 elongation at first rupture. Abrasion tests In these tests the end point is normally taken as the amount of abrasion required to penetrate the coating. The results thus reflect the strength of the coating, its cohesion, and in some cases its adhesion to the basis metal as well as resistance to abrasion. Hardness Coating hardness is related to the method of measurement. Results reflect the resistance t o scratching as well as to indentation. Impact tests Such tests reveal the resistance of coatings to deformation and destruction by concentrated sudden stresses. They thus throw considerable light on the integrity of the metal-coating bond. Changes in adhesion through chemical reaction at the paint/metal interface will be reflected in the impact-test values. The above tests for characterising coating properties necessarily continue to involve a certain amount of empiricism. The intelligent use of these tests, however, has shown that wide variations of physical and electrochemical characteristics of coatings as a function of composition may be obtained, and further, that significant changes in these characteristics, that can be measured before the usual evidence of failure appears, occur upon natural and accelerated ageing.
Test Methods for Corrosion Inhibitors Immersed conditions
Since corrosion inhibitors are used in a wide range of applications, no universal test method exists. Recognised methods tend to relate to a product or process in which the inhibitor forms a part rather than to the inhibitor per se. Thus, tests exist for inhibited coolants, cooling waters, cutting oils, pickling liquids, etc. The considerations applicable to corrosion test methods also apply to tests for inhibited products. The metals and alloys used, their surface preparation, the temperature, flow rate, composition of the test medium, the presence of heat transfer, and so on, must all be relevant to the proposed use of the inhibited product. As with other test methods there are those tests
19: 109
CORROSION TESTING AND CORROSION RATES
that have been developed in particular laboratories for the development of inhibitors for particular purposes and those that have acquired national or international recognition by appropriate standards-writing bodies. The three types of test procedure discussed in this chapter may often be identified in testing of inhibitors or inhibited products. The testing of inhibited engine coolants provides a suitable example. Cooling water
1
Air
1
t
t
k P.t.f.e. (or metal) end cap 1
E E 7
N
dia
F
it.f.e. end cap (a)
(b)
14 mm dia (C)
Fig. 19.34 Experimental arrangement for corrosion tests without heat transfer. (u) Test vessel and specimen, (b) specimen assembly for single metal or bimetal specimens, and (c) assembly for brass/solder/brass specimens
Laboratory tests used in the development of inhibitors can be of various types and are often associated with a particular laboratory. Thus, in one case simple test specimens, either alone or as bimetallic couples, are immersed in inhibited solutions in a relatively simple apparatus, as illustrated in Fig. 19.34. Sometimes the test may involve heat transfer, and a simple test arrangement is shown in Fig. 19.35. Tests of these types have been described in the l i t e r a t ~ r e ~ " .However, ~~~. national standards also exist for this type of test approach. BSI and ASTM documents describe laboratory test procedures and in some cases provide recommended pass or fail criteria (BS 5 117:Part 2:Section 2.2: 1985; BS 6580:1985; ASTM D1384: 1987). Laboratory testing may involve a recirculating rig test in which the intention is to assess the performance of an inhibited coolant in the simulated flow conditions of an engine cooling system. Although test procedures have been developed (BS 5 177:Part 2:Section 2.3: 1985; ASTM D2570: 1985), problems of reproducibility and repeatability exist, and it is difficult to quote numerical pass or fail criteria.
19: 110
CORROSION TESTING AND CORROSION RATES a.c. supply
E E 0
z
85 mm Fig. 19.35
Apparatus for testing inhibitors with metal-to-coolant heat transfer
These laboratory tests may be followed by engine dynamometer tests (BS5117:Section 2.4: 1985(1989)) and finally by road tests in working vehicles (BS51 17:Part 2:Section 251985 (1989)), thus completing the sequence of laboratory, field, service testing. The problems that have been experienced in the recirculating rig test are indicative of those often met in performance testing. Attempts to reproduce the service conditions in a laboratory test inevitably involve attempting to reproduce each of the controlling conditions that exist in the real situation. Variations, which may be relatively small, in these simulations can lead to significant differences in test results. There is therefore much to be said for keeping test conditions as simple as possible rather than attempting to reproduce accurately the conditions in practice. A balance between reproducibility and realism has to be struck. An example of a relatively simple, but effective, test method is that developed for inhibited mineral oils in the presence of water (BS 2000:Part 135:1983). Typically, in these tests, a mixture of the inhibited oil and distilled water or sea-water in specified proportions is stirred at 60°C in a beaker containing a steel specimen for a test period of 24 h, followed by a visual inspection for rusting. Similar test procedures exist for inhibited fuel products. The testing of inhibitors for use in oil and gas production, transport and processing normally involves two-phase oil-water fluids with, sometimes, a
19: 111
CORROSION TESTING AND CORROSION RATES
solid phase, e.g. entrained sand particles. Tests are usually of the dynamic variety with continuous movement so that test specimens contact all phases present. A well known laboratory test procedure is the so-called wheel test in which bottles of about 200cm3 volume containing weighed test specimens and a two-phase fluid saturated with an appropriate gas (CO, or H,S) rotate inside a temperature-controlled chamber 344, For many applications in these technologies data are required for high-temperature high-pressure conditions, and the use of autoclaves then becomes e~sential’~’.Two reviews on test procedures for corrosion inhibitors have been published3469347. Vapwr Phase Conditions
The testing of vapour phase inhibitors, usually referred to as volatile corrosion inhibitors, is essentially a matter of placing a test specimen in the vapour space of a closed vessel containing an aggressive atmosphere frequently water vapour, perhaps with SO2present-and a quantity of the inhibitor. Variations on the basic technique include provision for circulation of the vapour, the use of paper impregnated with inhibitor, provision for temperature cycling, etc. In the early 1950s, Wachter etaf.348, in the USA, described a humidity cabinet test in which metal specimens were supported inside inverted glass tubes containing a slip of inhibitor-impregnated paper in the lower end. The test was conducted at 3 7 ~ 7 ° Cand 100Vor.h. In the UK, Stroud and Vernon”’ described two types of test: (1) with a single test specimen suspended from a cork in the neck of a 250cm3 conical flask containing 25cm3 of water with 5 massolo of the inhibitor which was held at 35°C during the day and at room temperature overnight, and (2) with specimens suspended in the upper part of glass tubes containing water and inhibitor with the lower part of the tubes immersed in a thermostatted bath so that condensation occurred in the upper part of the tubes. Other forms of test using a climatic cabinet with tropical or industrial atmospheres have also been described350.
-
Acknowledgements Acknowledgements are made to Dr. R. Francis, Dr. T. G. Gooch, Prof. J. S. Llewelyn Leach, Dr. T. C. Lindley, Mr. G. 0. Lloyd, Dr. D. Mills, Mr. T. E. Such, Mr. A. K. Tiller and Mr. P. J. Trant for their helpful comments in connection with the updating of this section which is based on the original by Dr. L. L. Shreir and Dr. F. L. LaQue. P. McINTYRE A. D. MERCER REFERENCES
1. Champion, F. A.. Corrosion Testing Procedures, 2nd edn., Chapman and Hall, London (19W 2. Ailor, W. H. (4. Handbook ). on Corrosion Ttsting and Evaluation, John Wiley (1971) 3. Gilroy, D. and Mayne. J. E. 0.. Corrm. Sci.. 5 , 55 (1965)
19: 112
CORROSION TESTING AND CORROSION RATES
4. Wesley, W. A., Proc. Amer. SOC. Tesf. Mafer.,43, 649 (1943) 5. Bayer. R. 0. and Kachik, E. A., Corrosion, 5. 308 (1949) 6. Renshaw, W. G. and Ferree, J. A., Corrosion, 7, 353 (1951) 7. Gulbransen, E. A., Min. and Metall.. N.Y., 25, 172 (1944) 8. Kruger, J., J. Electrochem. SOC., 106, 847 (1959); see also ‘Ellipsometry in Corrosion Testing’ by Kruger, J. and Hayfield, P. C. S. in Reference 2 9. Rutherford, J. J. 8. and Aborn, R . H., Trans. Amer. Insf. Min. fMeral1.) Engrs., 100, 293 (1932) IO. Burns, R. M. and Campbell, W. E., Trans. Electrochem. SOC.,55, 271 (1929) 11. Hudson, J. C., Proc. Phys. SOC. Lond., 40, 107 (1928) 12. Mercer, A. D., Butler, G. and Warren, G. M., Br. Corros. J., 12, 122 (1977) 13. Finkeldey, W. H., Proc. Amer. SOC. Tesf. Mafer.,32, 226 (1932) 14. Kenworthy, L. and Waldram, J. M., J. Inst. Met., 55, 247 (1934) 15. Tronstad, L., Trans. Faraday SOC., 29, 502 (1933) 16. Kruger, J.. J. Electrochem. SOC., 106, 847 (1959) and 108, 504 (1961); see also Kruger, J., ‘Recent Developments in Ellipsometry’, Symp. Proc., University of Nebraska, Bashara, N. M., Buckman, A. B. and Hall, A. C. (eds.), published in SurfaceScience, 16 (1969) 17. Bengough, G . D., Stuart., J. M. and Lee, A. R., Proc. Roy. SOC. A , 116, 425; A , 121, 89 (1928) 18. Shipley, J. W., McHaffie, I. R. and Clare, N. D., Industr. Engng. Chem., 17, 381 (1925) 19. Bloom, M. C. and Krulfeld. M.. J. Elecfrochem. Sm., 104, 264 (1957) 20. Vaurio, V. W., Clark, B. S. and Lueck, R. H., Indusfr. Engng. Chem. (Anal.), IO, 368 (1938) 21. Hudson, R. M. and Stragland, G. L., Corrosion, 15, 13% (1959) 22. Willey, A. R., Krickl, J. L. and Hartwell, R. R., Corrosion, 12, 433t (1956) 23. Burns, R. M., J. Appl. Phys., 8, 398 (1937) 24. Dravnieks, A. and Cataldi, H. A,, Corrosion, 10, 224 (1954) 25. Marsh, G. A. and Schaschl, E., Corrosion, 14, 155t (1958) 26. Mylius, F., Z . Merallkd., 14, 233 (1922) 27. El-Kot, A. M. and AI-Suhybani, Br. Corros. J., 22, 29 (1987) 28. Amer. Soc. Test. Mater. Spec. Tech. Publ. No. 32 (1937) 29. Knapp, B . B., in The Corrosion Handbook, (ed. Uhlig. H. H.), Wiley, New York; Chapman and Hall, London, 1077 (1948) 30. Teeple, H. O., Amer. SOC.Test. Mater. Spec. Tech. Publ. No. 175, 89 (1956) 31. Amer. SOC.Test. Mater. Spec. Tech. Publ. No. 970 (1988) 32. Gallagher, P., Malpas, R. E. and Shone, E. B., Br. Corros, J . , 23, 229 (1988) 33. Mercer, A. D. and Brook, G. M.. La Tribune de Cebedeau, 417-418. 299 (1978) 34. Romeo, A. J., Skrinde, R. T. and Eliassen, R., Proc. Amer. SOC. Civ. Engrs., 84, No. SA4 (1958) 35. Cheung, W.K. and Thomas, J. G. N. in The Used ofSynfheficEnvironmenfsfor Corrosion Testing, ASTM STP 970, 190 (1988) 36. LaQue, F. L. and Stewart, W. C., M i f a u x e f Corros., 23, 147 (1948) 37. Vernon, W. H. J., J. SOC. Chem. Ind., Lnd., 66, 137 (1947); Corrosion, 4, 141 (1948) 38. Bengough, G. D. and May, R., J. Insl. Mef.. 32, 81 (1924) 39. May, R. and Stacpoole, R. W. de Vere, J. Insf. Met., 77, 331 (1950) 40. Brownsdon, H. W. and Bannister, L. C., J. Inst. Met., 49. 123 (1932) 41. Freeman, J. R. Jr. and Tracy, A. W., Corrosion, 5 , 245 (1949) 42. Butler, G. and Ison, H. C. K., J . Appl. Chem., 10. 80 (1960) 43. Fontana, M. G.. Indusfr. Engng. Chem., 39, 87A (1947) 44. Wagner, H. A., Decker, J. M. and Marsh, J. C., Trans. Amer. SOC. Mech, Engrs., 69, 389 (1947) 45. Trembler, H. A,, Wesley, W. A. and LaQue, F. L., Indusfr. Engng. Chem., 24, 339 (1932) 46. Brennert, S., J . Iron S f . Insf., 135, lOlP (1937) 47. Smith, H. A., Mefal Progr., 33, 596 (1938) 48. Hanawalt, J. D.,Nelson, C. E. and Peloubet, J. A., Trans. Amer. Insf. Min. fMefal1.) Engrs., 147, 273 (1942) 49. Benedicks, C., Trans. Amer. Insf. Min. (Mefall.)Engrs., 71, 597 (1925) 50. McAdams, D. J., Heat Transmission, McGraw-Hill, New York, 3rd edn., 370 (1954) S I . Groves, N. D. and Eisenbrown, C. M.. Meful Progr., 75, 78 (1959)
CORROSION TESTING AND CORROSION RATES 52. 53. 54. 55. 56. 57. 58. 59. 60. 61.
19: 113
Fisher, A. 0. and Whitney, F. L. (Jr.), Corrosion, 15, 257t (1959) Fisher, A. O., Corrosion. 17, 21% (1961) Gleekman, L. W. and Swandby, R. K., Corrosion, 17, 144t (1961) Groves, N. D.. Eisenbrown, C. M. and Scharfstein, L. R., Corrosion, 17, 173t (1961) Hart, R. J., Reference 2, p. 367 LaQue, F. L., ‘Electrochemistry and Corrosion (Research and Tests)’, Acheson Memorial Address, J. Elecfrochem. Soc., 116, 73C (1969) Stern, M., and Geary, A. L., J. Elecfrochem. Soc.. 104, 56 (1957) Stern, M., Corrosion, 14, 44Ot (1958) Pourbaix, M., Lectures on Electrochemical Corrosion Plenum Press (1973); see also references to potential-pH diagrams given in Section I .4 Armstrong, R. D., Henthorne, M. and Thirsk, H. R., J. Elecfroanal. Chem., 35, 119
(1972) 62. Epelboin, 1. Keddam, M. and Takenouti, H., J. App. Elecfrochem.. 2, 71 (1972) 63. Sathyaharayana, S.. Elecfroanal. Chem. and Interfacial Electrochem., 50, 41 I (1974) 64. Macdonald, D. D., Corrosion, 46, 229 (1990) 65. Kendig, M. W. and Mansfeld, F., Proc. Fall Meefing, Detroit, Electrochem. Soc., 82-2, 105 (1982) 66. Ferreira, M. G. S. and Dawson, J. L., Passivity of Metals and Semiconducfors,(ed. M. Froment) 359 (1980) 67. Ferreira, M. G . S. and Dawson, J. L., J. Elecfrochem.Soc., 132, 760 (1983) 68. Keddam, M., Oltra, R., Colson, J. C. and Desestret, A,, Corros. Sci., 23, 441 (1983) 69. Isaacs, H. S . and Kendig, M. W., Corrosion, 36, 269 (1980) 70. Park, J. K. and Macdonald, D. D.. Corros. Sci., 23. 293 (1983) 71. Epelboin. I., Gabrielli, C., Keddam, M. and Takenouti. H., 2. fhysik. Chem., 98. 215 (1975) 72. Oltra, R. and Keddam, M., Corros. Sci., 28, 1 (1988) 73. Sato, N., J. Elecfrochem. SOC., 123, 1197 (1976) 74. Shibata, T. and Takeyama. T., Corrosion, 33, 243 (1977) 75. Williams. D. E., ‘The Analysis of Current and Potential Fluctuations in Corroding Systems’, Proc. Elecfrochemical Corrosion Testing. Ferrara, 10-14 September 1985 DECHEMA, (1986) 76. Williams, D. E., Wescott, C. and Fleischmann, M., J. Elecfrochem. SOC., 132, 1796 ( 1985) 77. Williams. D. E., Wescott, C. and Fleischmann, M., J. Elecfrochem. SOC., 132, 1804 (1985) 78. Gabrielli, C., Huet, F., Keddam, M. and Oltra, R., Corrosion, 46, 266 (1990) 79. Ives, D. J. G. and Janz, G. J., Reference Elecfrodes, Academic Press (1961) 80. Compton, K. G., Materials Research and Sfandards, 10, 13 (1970) 81. Covington. A. K., Elecfrochemistry. Vol. I., The Chemical Society, London, 56 (1970) 82. Meites, L. and Moros. S. A., Analyf. Chem., 31, 25 (1959) 83. European Federation of Corrosion Publication No. 4, A WP report Guidelines on Elecfrochemical Corrosion Measurements, The Inst. of Metals, London (1990) 84. Berzins, T. and Delahay, P., J . Am. Chem. SOC., 77, 6448 (1955) 85. Pouli, D., Huff, J. R. and Pearson. J. C., Anal. Chem., 38, 382 (1966) 86. Kooijman, D. J. and Sluyters, J. H., Electrochim. A c t a , 11, 1147 (1966) 87. Bewick, J., Elecfrochim. Acta., 13. 825 (1968) 88. Piontelli, R. and Bianchi, G., h o c . 2nd. Meefing C.I.T.C.E., Milan (1951) 89. Piontelli, R., Proc. 4fh Meefing C.I.T.C.E.,London and Cambridge (1952) 90. von Fraunhofer J. A. and Banks, C . A., Potenfiosfafandifs Applications, Butterworths, London (1972) 91. Stern, M. and Makrides, A. C., J. Electrochem. Soc.. 107, 782 (1960) 92. Greene, N. D., France, W. D. and Wilde, B. E.,Corrosion, 21, 275 (1965) 93. Creme, N. D., Acello, S. J. and Greif, A. J., J . Elecfrochem. SOC., 109, 1001 (1962) 94. Cleary, H. J. and Greene, N. D., Elecfrochim. Acfa., 10, 1107 (1965) 95. France, W. D. (Jr.), J. Elecfrochem. Soc., 114, 818 (1967) %. Wilde, B. E., Corrosion, 23, 331 (1%7) 97. Smith, L. W. and Pingel, V. J., J. Electrochem. Soc., 98, 48 (1951) 98. Budd, M. K. and Booth, F. F., Metallurgia, 66, 245 (1962) 99. Cleary, H. J., Corrosion, 24, 159 (1968) 100. Doig, P. and Edington, J. W . , Br. Corros. J . , 9, 88 (1974)
19: 114
CORROSION TESTING AND CORROSION RATES
101. Davis, J. A., Proc. ConJ on Localised Corrosion, Williamsburg, NACE, 168 (1971) 102. May. R., J. fnst. Metals, 40, 141 (1928) 103. Hines, J., private communication 104. Hoar, T. P. and Hines. J. G., J. Iron Sreel fnsr., 182, No. 124. 156 (1965) 105. Hoar, T. P. and West, J. M., Roc. Roy. Soc.,A268, 304 (1962) 106. Horst, R. L. (Jr.), Hollingsworth. E. H. and King, W., Corrosion. 25. 199 (1%9) 107. Pearson. J. M., Trans. Electrochem. Soc..81 485 (1942) 108. Schwerdtfeger, W. J., Corrosion, 19, 171 (1963) 109. Schwerdtfeger, W. J. and Manuele. R. J.. Corrosion, 19, 59t (1963) 110. Skold, R. V. and Larson, T. E., Corrosion, 13, 139t (1957) 111. Stern, M. and Weisert. E. D.. froc. Amer. SOC. Test. Mater., 59, 1280 (1959) 112. Barnarrt, S., Corrosion, 27, 467, (1971) 113. Barnarrt, S.. Corros. Sci., 9. 145 (1969) 114. Leroy, R. L., Corrosion, 29. 272 (1973) 115. Oldham, K. B. and Mansfeld, F., Corrosion. 27, 434 (1971) 116. Oldham, K. B. and Mansfeld. F.. Corns. Sci.. 13. 811 (1973) 117. Hickling, J., Ph.D. Thesis, University of Cambridge (1974) 118. Hoar, T. P., Corros. Sci., 7 , 455 (1967) 119. Mansfeld, F., Corrosion, 29, 397 (1973) 120. Mansfeld. F., Corrosion, 30. 92 (1974) 121. Mansfeld, F., J. Electrochem. Soc., 118, 545 (1971) 122. Mansfeld, F., J. Electrochem. Soc., 120, 515 (1973) 123. Makrides. A. C..Corrosion. 25,455 (1969) 124. Legault, R. A. and Walker, M. S . , Corrosion, 19, 222 (1963) 125. Walker, M. S. and France, W. D.. Mat. Protect.. 8. 47 (1969) 126. Jones, D.A. and Greene, N. B., Corrosion, 25, 367 (1969) 127. Wilde, B. E., Corrosion, 23. 379 (1967) 128. Jones, D.A., Corros. Sci.. 8, 19 (1968) 129. Bureau, M.,9th Fatipec Congress, 79 (1968) 130. Mikhailovskii, Y. N.,Leonev, V. V. and Tomashov, N. D.. Korroz.yaMefallov iSplavov Sbornik 2, Metaflurgizdat, Moscow (1965) 131. Butler, T. J. and Carter, P. R., Elecfrochem. Tech.. 1, 22 (1963) 132. Walpole, J. F., Bull. Inacol., 23, 22 (1972) 133. Bird, D. W.,Bull. fnacol.. 22, 149 (1971) 134. Rowlands, J. C. and Bentley. M. N., Brit. Corros. J . , 7, 42 (1972) 135. Makar, D. R. and Francis, H. T.. J. Elecfrochem. SOC., 102, 669 (1955) 136. Wesley, W. A., Trans. Electrochem. Soc., 73, 539 (1938) 137. Copson, H.R., Industr. Engng. Chem., 37,721 (1945) 138. Rowe. L. C..J. Mater., 5. 323 (1970) 139. Denison, I. A., J. Res. Nut. Bur. Stand., 17, 363 (1936) 140. Ewing. S. P., Amer. Gas. Ass. Mon., 14,356 (1932) 141. Schwerdtfeger, W. J., J. Res. Naf. Bur. Stand., 50, 329 (1953) 142. Schwerdtfeger, W. J., J. Res. Nut. Bur. Stund.. 52, 265 (1954) 143. Schwerdtfeger, W. J., J. Res. Naf. Bur. Sfand., 58, 145 (1957) 144. Schwerdtfeger, W. J., J. Res. Nut. Bur. Stand., 65C, 271 (1961) 145. Logan, R. H.. Ewing, S. P. and Denison, 1. A., Amer. Soc. Test. Mater. Spec. Tech. Publ. No. 32, 95 (1937) 146. Jones, D.A. and Lowe, T. A., J. Materials. 4, 600 (1969) 147. Fuller, T. S., Proc. Amer. SOC. Test. Mater., 27, 281 (1924) 148. Lathrop, E. C., Proc. Amer. SOC. Test. Mater., 24, 281 (1924) 149. Todt, F., Z.Elektrochem., 34, 586 (1928) 150. Todt, F.,Z.Ver. Dtsch. Zuckerind., 79, 680 (1929) 151. Streicher, M. A., ASTM Bull. No. 188. 35, Feb. (1953) 152. Streicher, M. A., ASTM Bull. No. 195,63, Jan. (1954) 153. Streicher. M.A., J. Electrochem. Soc.. 106, I61 (1959) 154. Streicher, M. A., Corrosion, 19,272t (1963) 155. Saur. R. L. and Basco, R. P . , Plating, 53, 33 (1%6) 156. Saw, R. L. and Basco, R. P., flaring, 53, 320 (1966) 157. Saur, R. L. and Basco, R. P., Plating, 53, 981 (1966) 158. Saur, R. L., Pluting, 54, 393 (1966) 159. Englehart, E.T. and George, D. J.. Mater. Prof., 3 No. 11, 25 (1964)
CORROSION TESTING AND CORROSION RATES
19: 115
160. Capp. J. A., Proc. Amer. SOC. Tesf. Mafer., 14,474 (1914) 161. LaQue, F. L., Proc. Amer. Soc. Test. Mater.. 51, 495 (1951) 162. LaQue, F. L., Mater. and Merh.. 35. No. 2, 77 (1952) 163. Sample, C. H.,Bull. Amer. Soc. Test. Mater.. No. 123, 19 (1943) 164. May, T. P. and Alexander, A. L.. Proc. Arner. Soc. Tesf. Mater.. 50, 1131 (1950) 165. Darsey. V. M. and Cavanaugh, W. R.. Proc. Amer. Soc. Test. Mafer., 48, 153 (1948) 166. Nixon. C. F.. Mon. Ren. Amer. Elecfmpl. Soc., 32, I105 (1945) 167. Pinner, W. L., Plating, 44,763 (1957) 168. Nixon, C. F., Thomas, J. D. and Hardesty, D. W., 46fh Ann. Tech. Proc. Amer. Elecfropl. SOC., 159 (1959) 169. Thomas, J. D., Hardesty, D. W. and Nixon, C. F., 47th Ann. Tech. Proc. Amer. Electropl. Soc., 90 ( 1960) 170. LaQue, F. L., 46fh Ann. Tech. Proc. Amer. Elecfropl. SOC., 141 (1959) 171. Bigge, D. M., 46fh Ann. Tech. Proc. Amer. Elecfropl. SOC., 149 (1959) 172. Edwards, J., 46fh Ann. Tech. Proc. Amer. Elecfropl. SOC., 154 (1959) 173. Edwards, J., Trans. Inst. Metal Finishing. 35. 55 (1958) 174. Kesternich. W.. Stahl u. Eisen. 71. 587 (1951) 175. Dix, E. H. (Jr.) and Bowman, J. J.. Amer. Sac. Test. Mater. Spec. Tech. Publ. No. 32, 57 (1937) 176. Carter, V. E. (ed.), Corrosion Tesfingfor Metal Finishing, Butterworths. London (1982) 177. Swinden, T. and Stevenson, W. W., J. Iron SI. Insf., 142, 165P (1940) 178. Lloyd, T. E., J. Mefals, N.Y.,188. 1092 (1950) 179. Evans, U. R. and Britton, S. C., J. Iron SI. Insf., Spec. Rept. No. 1, 139 (1931) 180. Dennis, J. K. and Such, T. E., Trans. Insf. Metal Finishing., 40, 60 (1963) 181. Dennis, J. K. and Such, T. E., Nickel and Chromium Plating, 2nd edn., Butterworths. London (1986) 182. Chandler, K. A. and Kilcullen, M. B., Brif. Corros. J., 5, 1 (1970) 183. Bromley, A. F.. Kilcullen, M. B. and Stanners, J. F., 5th European Congress of Corrosion. Paris, Sept. (1973) 184. Pourbaix. M.,CEBELCOR RT 160,Aug. (1%9) 185. Legault. R. A., Mori, S. and Leckie. H. P.. Corrosion, 26, 121 (1970) 186. Legault. R. A.. Mori. S. and Leckie. H. P.. Corrosion, 29, 169 (1973) 187. Okada, H.. Hosio, U. and Naito, H., Corrosion, 26, 429 (1970) 188. Brown, M. H., Corrosion, 30, I (1974) 189. Cowan, R. L. and Tedmon, C. S. (Jr.), ‘Intergranular Corrosion of Iron-NickelChromium Alloys’, in Advances in Corrosion Science and Technology, Vol. 3, (eds. M. G.Fontana and R. W. Staehle). Plenum Press (1973) 190. Huey, W. R., Trans. Amer. SOC.Steel Treat., 18, 1126 (1930) 191. Henthorne, M., Corrosion, 30, 39 (1974) 192. Strauss, B., Schottky, H, and Hinnuber, J.. 2. Anorg. Allgem. Chem., 188, 309 (1930) 193. Rocha. H. J., in discussion of paper by Brauns, E. and Pier, G., Srahl u. Eisen, 75, 579 (1955) 194. Scharfstein, L. R. and Eisenbrown, C. M.,ASTM STP No. 369, 253 (1%3) 195. Tedmon. C. S. and (Jr.), Vermilyea. D. A. and Rosolowski, J. H.. J. E/ecfrochem. SOC., 118, 192 (1971) 1%. Ebling, H. and Scheil, M. A.. ASTM Special Tech. Publ. No. 93, 121 (1949) 197. Warren, D., ASTM Bulletin No. 230, 45 May (1958) 198. Streicher, M. A,, ASTM Bulletin No. 188,35 (1953) 199. Clerbois. L.. Clerbois. F. and Massart. J.. Electrochem. Acfa.. 1, 70 (1959) 200: France, W. D. and Greene, N. D., Cofros: Sci. 8, 9 (1968) 201. Streicher, M. A., Corros. Sci., 9, 55 (1969) 202. France, W.D. and Greene, N. D., Corros. Sci., 10, 379 (1970) 203. Streicher, M. A., Corros. Sci., 11, 275 (1971) 204. Cihal, V.. Desestret, A., Froment. M and Wagner, G. H.. Proc. Conf. European Federation on Corrosion, Paris, France. 249 (1973) 205. Novak, P.. Stefec. R. and Franz. F., Corrosion. 31, 344 (1975) 206. Kolotyrkin, Ya. M.. Zashch. Mer.. 11. 699 (1975) 207. Clark, W. L., Cowan, R. L. and Walker, W. L., Compararive Methods for Measuring Degree of Sensirisofion in Sfainless Steel, ASTM STP 656, ASTM. Philadelphia, 99 (1978) 208. Lee, J. B., Corrosion, 42, 106 (1986)
19: 116
CORROSION TESTING A N D CORROSION RATES
209. Streicher, M . A., Corrosion, 30, 77 (1974) 210. Fontana, M. G. and Greene, N. D., Corrosion Engineering, McGraw-Hill (1967) 211. Wilde, B. E., Corrosion. 28, 283 (1972) 212. France, W. D. and Greene, N. D., Corrosion, 26, 1 (1970) 213. Wilde, B. E. and Williams, E., J . Electrochem. SOC.,117, 775 (1970) 214. Wilde. B. E. and Greene, N. D., Corrosion, 25, 300 (1969) 215. Henry, W. D. and Wilde, B. E., Corrosion, 25, 515 (1969) 216. Wilde, B. E. and Williams, E., J. Elecfrochem. SOC.. 118. 1058 (1971) 217. May, R. and Stacpoole, R. W. de V., J. Inst. Met., 77, 331 (1950) 218. Efird, K. D., Corrosion, 33, 347 (1977) 219. Brownsdon, H. W. and Bannister, L. C., J. Inst. Me/., 49, 123 (1932) 220. Breckon, C. and Gilbert, P. T., 1st Int. Congress on Met. Corrosion, Butterworths, London, 624 (1962) 221. Bern, R. S. and Campbell, H. S., ibid., 630 (1962) 222. Campbell, H. S.,MP577, BNFMRA, Feb. (1973) 223. LaQue, F. L., Marine Corrosion, J. Wiley and Sons, Inc.. 62 (1975) 224. Grant, A. A. and Phillips, L., The Application of Advanced Materials Technology in Fluid Engineering, 1. Mech. E., London, Feb. (1990) 225. Kohley, T. and Heitz, E., The Use of Synthetic Environments for Corrosion Testing, ASTM STP970, eds. P. E. Francis and T. S. Lee, 235-245 (1988) 226. Sprowls. D. O., ASM Metals Handbook, Vol. 13, 9th Edn., 291-302 (1987) 227. Haigh, 8. P., J. Inst. Metals, 18, 55 (1917) 228. Huddle, A. U. and Evans, U. R., J. Iron and Steel Inst., 149, 109P (1944) 229. Inglis, N. and Lake, G. F., Trans. Faraday SOC., 17, 803 (1931) 230. Rawdon, H. S., Proc. Amer. SOC. Test. Mater., 19, 314 (1929) 231. Kenyon, J. N., ibid., 40, 705 (1940) 232. Could, A. J. and Evans, U. R., Iron and Steel Inst. Spec. Report No. 24., 325 (1939) 233. Evans, U. R. and Simnad, M. T., Proc. Roy. SOC.,A188, 372 (1947) 234. Gough. H. J. and Sopwith, D. G., J. Iron Sfeel Inst., 127, 301 (1933) 235. Cough, H. J. and Sopwith, D. G., Engineering, 136, 75 (1933) 236. Jarman, R. A., Smith, S and Williams, R. A., Br. Corros. J., 13. 195 (1978) 237. Hoeppner, D. W., Corrosion Fatigue, NACE-2, University of Connecticut, 3 (1972) 238. McEvily, A. J. and Wei, R. P., ibid., 381 (1972) 239. Kitigawa, H., ibid., 521 (1972) 240. Wood, R. J. K. and Fry, S . A., J. Fluids Eng., 111, 271 (1989) 241. Lichtrnan, J. Z., Kallas, D. H. and Rufola, A., Handbook on Corrosion Testing and Evaluation, ed. W. H. Ailor, John Wiley, 453 (1971) 242. Schroter, H., 2. Ver. Dtsch. Ing., 78, 349 (1934) 243. Hobbs, J. M.,Proc. Cavirarion Forum. ASME, 1 (1966) 244. Gaines, N., Physics, 3, 209 (1932) 245. Kerr, S. L., Trans. Amer. SOC.Mech. Engrs., 59, 373 (1937) 246. Beeching, R., Trans. Insfn. Engrs. Shipb. Scot., 90, 203 (1946) 247. Rheingans, W. J., in Engineering Approach to Surface Damage, (eds. C. Lipson and L. V. Colwell), University of Michigan, 249 (1958) 248. Leith, W. C. and Thompson, A. L., Trans Amer. Soc. Mech. Engrs., J. Basic Engng., 82, Ser. D., 795 (1960) 249. Plesset, M . S . , Trans. ASME Series D, J. Basic. Engng.. 85, 360 (1963) 250. Eisenberg, P., Preiser, H. S. and Thiruyengadam, A., Trans. SNAME, 73, 241 (1965) 251. Hobbs, J. M., ASTM STP 408, 159 (1967) 252. Plesset, M. S. and Devine, R. D., J. Basic Eng., Trans. ASME, 692, Dec. (1966) 253. Schulmeister, R. and Speckhardt, H., ASTM STP 705, 81 (1980) 254. Hudgens, R. D., Carver, D. P.. Hercamp. R. D. and 1-auterback, J.. ibid. 233 (1980) 255. Chance, R. L., ibid., 270 (1980) 256. ASTM STP 144 (1953) 257. Waterhouse, R. B., Fretting Corrosion, Pergamon Press (1972) 258. Waterhouse, R. B., Proc. 10th International Conference on Metallic Corrosion, Madras, India, 7-11 November 1987, 5, 63, Oxford and IBH Publishing Co., New Delhi (1987) 259. Fink, M., Trans. Amer, SOC. Steel Treat., 18, 1026 (1930) 260. Shotter, G. F., J. Inst. Elec. Engrs., 75, 755 (1934) 261. Tomlinson, G. A., Thorpe, P. L. and Gough, H. J., J. Insf. Mech. Engrs.. 141, 233 (1939)
CORROSION TESTING AND CORROSION RATES 262. 263. 264. 265. 266. 267. 268. 269. ?70.
'I. - '2. '73. 3:4. 275. 276. 277. 278.
279. 280. 281. 282. 283. 284. 285. 286. 287. 288. 289. 290. 291. 292. 293. 294. 295. 296.
297. 298. 299. 300. 301. 302. 303. 304. 305. 306. 307.
19: 117
Wright, K . H. R., Proc. Instn. Mech. Engrs., lB, 556 (1952-53) Uhlig, H. H., Tierney, W. D. and McClellan, A., ASTM STP 144, 71 (1953) McDowell, J. R., ibid., 24 (1953) Warlow-Davies, E. J., J. Inst. Mech. Engrs., 146, 32 (1941) Godfrey, D., Tech. Note 2039, Natl. Advisory Comm. Aeronaut. (1950) Gray, A. C. and Jenny, R. W., S.A.E.J.. 52, 5 1 1 (1944) de Villerneur, Y., Metaux. Paris, 34, 413 (1959) Wright, K. H. R.. Proc. Instn. Mech. Engrs.. lB, 556 (1952-53) Wright, K. H. R., ibid., 181, Pt. 30, 256 (1966-67) Barwell, F. T. and Wright, K. H. R., J. Res. Erif. Cast Iron. Ass., 7, 190 (1958) Field, J. E. and Waters, D. M.. N.E.L. Rep. No. 275 (1967) Waterhouse, R. B., J.I.S.I.. 197, 301 (1961) Epstein, L. F., Proc. Int. Conf. PeacefulUsesofAtomicEnergy,New York,9,311 (1956) Bakish, R. and Kern, F., Corrosion, 9. 533t (1960) Edeleanu, C and Gibson, J. G.. J. Inst. Met., 88, 321 (1960) Brasunas, A de S., Corrosion, 9. 78 (1953) Miller, E. C., in Liquid Metals Handbook (ed.-in-chief, R. N. Lyon), Atomic Energy Comm. and Dept. of the Navy, Washington DC, June, 144 (1952) Borgstedt, H. U. and Frees, G., Proc. 10th In!. Cong. on Metallic Corrosion, Madras, India, 7-11 Nov. 1987, 3, 1843 (1988) Barker, M. G., Coen, V., Kolbe, H, Lees, J. A., Orecchia, L and Sample, T., J. Nucl. Mater., 155-157B, 732 (1988) Borgstedt, H. U. and Frees, G., Werkst. Korros., 41, I (1990) Vreeland, D. C., Hoffman, E. E. and Manly, W. D., Nucleonics, 11, 36 (1953) Grabner. H., Feurstein, H. and Oschinski, J., J. Nucl. Mater., 155-157B. 702 (1988) Strachan, J. F. and Harris, N. L., J. Inst. Met., 85, 17 (1956-57) Klueh, R. L., in Proc. Int. Conf. Sodium Technol. Large Fast Reactor Design, Nov. 7-9 1968; ANL-7520, PI. I , 171, Argonne National Laboratory Klueh, R. L., Corrosion, 25, 416 (1969) DiStefano, J. R., ORNL-4028, Oak Ridge Laboratory (1966) Manly, W. D., Corrosion, 12, 336t (1956) DiStefano, J. R. and De Van, J. H., Nuclear Appl. Tech., 8, 29 (1970) Koenig, R. F. and Vandenberg, S. R., Metal. Prog., 61, 71 (1952) Hoffman, E. E., Corrosion of Moteriols by Lithium at Elevated Temperatures, ORNL-2924, Oak Ridge National Laboratory (1960) Kassner, T. F., AIME Met. SOC.Trons., 239. 1643 (1967) Shibanova, L. N.. Vostryakov, A. A. and Lepinskikh, B. M., Zashch. Met., 22, 124 ( 1986) De Van, J. H. and Sessions, C. E., Nucl. Appl.. 3, 102 (1967) De Van, J. H. and Jansen. D. H., Fuels and Materials Development Program Quart. Progr. Rept., Sept. 30 (1968); ORNL-4350. Oak Ridge National Laboratory, p 91 Bonilla. C. F., in Reactor Handbook, Vol. IV (ed. S. McLain), Interscience. New York, 107 (1964) Romana, A. J., Fleitman, A. H. and Klamut, C. J., Proc. AEC-NASA Liquid Metals Inform. Meeting, CONF-65041I (1965) Fuller, L. C. and MacPherson, R. E., ORNL-TM-2595, Oak Ridge National Laboratory (1967) Borgstedt, H. U. and Frees, G., Werkst. Korros., 38, 732 (1987) RDT Standard C8-5T, Electrochemical Oxygen Meter f o r Service in Liquid Sodium, RDT Standards Office, Oak Ridge National Laboratory, Tennessee Pillai. S. R. and Mathews, C. K., J. Nucl. Mater., 137, 107 (1986) Schutter, F. de, Dekeyser, Ja, Tas, H. and Burbure, S. de, J . Nucl. Mater., 155,744 (1988) Roy, P., Wozaldo, G. P. and Comprelli, F. A., in Proc. Int. ConJ Sodium Technolog. LargeFast ReactorDesign, Nov. 7-9 (1968); ANL-7520, Pt. 1, Argonne National Laboratory, 131 Roy, P. and Gebhardt, M. F., GEAP-13548, General Electric Company (1969) Hoffman, E. E. and Harrison, R. W.,in Metallurgy and Technology of Refractory Meral Alloys, Plenum Press, New York, 251 (1969) Harrison, R. W., GESP-258. General Electric Company (1969) Borgstedt, H. U. and Grundrnan. M.,Preprints of Eurocorr '87 conference, Karlsruhe, Germany, 6-10 April 1987. DECHEMA, Frankfurt, 141 (1987)
19: 118
CORROSION TESTING AND CORROSION RATES
308. Dillon, C. P., Mater. Perform., 29, (1 I), 54 (1990) 309. Lynch, S. P., Proc. 2ndInt. Conf. on Environmental Degradation of Engineering Materials, Blacksburg, Virginia, Sept. 31-23 1981, Virginia Polytechnic Institute, 229 (1981) 310. Old, C. F., Metal Science, 14, 433 (1980) 31 1. Preece, C. M., Proc. Int. Conf. on Stress Corrosion Cracking and Hydrogen Embrittlemen1 of Iron Base Alloys, Unieux-Firminy. France, June 12-16 1973, NACE, Houston, 625 (1977) 312. Kapp, J. A., Duquette, D and Kamdar, M. H., J. Eng. Mater. Techno/.. 108, 37 (1986) 313. Thompson, D. H., in Ref. 2 314. Dillon, C. P., Krisher, A. S. and Wissenburg, H., Ref. 2, 599 (1971) 315. Strutt, J. E. and Nichols, J. R. (eds.), Plant Corrosion: Prediction of Materials Performance, Ellis Horwood, Chichester (1987) 316. Rawdon, H. S., ASTM STP 32, 36 (1937) 317. LaQue, F. L.. Proc. Amer. Soc. Test. Mater., 51. 495 (1951) 318. Finkeldey, W. H., Proc. Amer. Soc. Test. Muter., 32, 226 (1932) 319. Hudson, J. C.. J . Iron SI. h i . , 148, 161P (1943) 320. Pilling. N. B. and Wesley, W. A., Proc. Amer. Soc. Test. Mater., 40, 643 (1940) 321. Copson, H.R., Proc. Amer. SOC. Test. Mater., 48, 591 (1948) 322. Gibboney, J. H., Proc. Amer. SOC. Test. Mater.. 19, 181 (1919) 323. Mendizza, A., Proc. Amer. SOC.Test. Mater., 50, I14 (1950) 324. Passano, R. F., Proc. Amer. SOC. Test. Mater., 34, 159 (1934) 325. Pinner. W. L.. Proc. Amer. SOC. Test. Mater.. 53. 256 (1953) 326. Wirshing, R. J. and McMaster, W. D., Paint Varn. Prod., 41, 13 (1951) 327. Sereda, P. J., Bull. Amer. Soc. Test. Mater., No. 228, 53 (1958) 328. Guttman, H. and Sereda. P. J.. ASTM STP 435 (1968) 329. Gorman, L. J . , Proc. Amer. SOC. Test. Mater., 39,247 (1939) 330. Pray, H. A., Proc. Amer. Soc. Test. Mater., 44. 280 (1944) 331. Humble, H. A.. Corrosion, 5, 292 (1949) 332. Southwell, C. R., NRL Reports, Naval Research Laboratory, Washington, D.C. 333. IJsseling, F. P., Brit. Corros. J., 24, 55 (1989) 334. Logan, R. H., Ewing, S. P. and Denison, 1. A., ASTM STP 32, 95 (1937) 335. Scott, G. N., Proc. Amer. Petrol. Inst., 95 (1937) 336. Escalante, E. (ed.), Underground Corrosion, ASTM STP 741 (1981) 337. Tiller, A. K., Biocorrosion in Civil Engineering, Cranfield Institute of Technology (1990) 338. Walter, G. W., Corros. Sci., 26, 681 (1986) 339. Timmins, F. D., J. Oil Col. Chem. Assoc.. 62, 131 (1979) 340. Ali Elbasir. Scantlebury, J. D. and Callow, L. M., ibid., 67, 282 (1985) 341. Hoey, C. E. and Hipwood, H. A., J . OilCol. Chem. Assoc., 57, 151 (1974) 342. Butler, G., Mercer, A. D. and Warren, G. M., Eurocorr 77, 6th Eur. Congr. Metall. Corros., SOC.Chem. Ind., London, 349-355 (1977) 343. Mercer, A. D.. ‘Laboratory Research in the Development and Testing of Inhibited Coolants in Boiling Heat Transfer Conditions’, in Engine Coolant Testing: State of the Art 1979, ASTM STP 705, W. H. Ailor. ed., 53-80 (1980) 344. NACE Publication ID 182, Wheel Test Method for Evaluation of Film Persistent Inhibitors for Oilfield Applications, Mater. Perform., 21, (12), 45 (1982) 345. Schmitt, G. and Bruckhoff, W., Proc. 5th European Conference on Corrosion Inhibiiors (5 SEIC), University of Ferrara, 323 (1980) 346. Mercer, A. D., Br. Corros. J., 20, 61 (1985) 347. Mercer, A. D., Proc. 6th European Symposium on Corrosion Inhibitors, (6 SEIC), Ann. Univ. Ferrara, N.S., Sez. V, Suppl. N8 (1985) 348. Wachter, A,, Skei, T. and Stillman, N., Corrosion, 7 , 284 (1951) 349. Stroud, E. G . and Vernon, W. H. J., J. Appl. Chem., 2, 178 (1982) 350. Levin, S. 2.. Gintzberg, S. A., Dinner, S. M. and Kuchinsky, V. N., 2nd Ferrara Conference on Corrosion Inhibitors (2SEIC) 1965, University of Ferrara, 765 (1966)
19.1A Appendix -Chemical and Electrochemical Methods for the Removal of Corrosion Products
This appendix provides information on chemical and electrochemical treatments which have been recommended for the removal of corrosion products. In using these methods the following points need to be borne in mind: 1. The duration of chemical or electrochemical treatment should be kept to the minimum necessary to remove the corrosion product. Loosely adherent material should be removed beforehand by suitable mechanical means, e.g. scrubbing. 2. The combined action of chemical (or electrochemical) treatment and scrubbing is often more effective than either method alone. It is frequently advantageous to alternate short periods of immersion with scrubbing to remove any corrosion product that has become loosened by the action of the chemical reagent. 3. The rate of attack of the chemical reagent on sound metal should be determined on a separate uncorroded sample of the material being cleaned, and if necessary a correction should be applied to the loss in weight of the corroded specimen. However, where a metal, and particularly an alloy, is heavily corroded- thus exposing a different surface structure from that of an uncorroded surface-it will be necessary to check the reliability of the cleaning method [Mercer, A. D., Butler, G. and Warren, G. M.,Br. Corms. J . , 12, (2), 122-126 (1977)l. A procedure for obtaining more accurate weight loss data in these circumstances has been described [ISO/DIS 8407.2:1989]. 4. The possibility of redeposition of metal from the dissolved corrosion product or, if electrochemical treatment is employed, from the anode material should always be kept in mind. If there is reason to believe this has occurred during removal of the corrosion product, further treatment to remove the redeposited metal will be necessary before the weight loss due to corrosion is measured.
19: 119
19: 120
APPENDIX-THE
REMOVAL OF CORROSION PRODUCTS
Procedures for Removing Corrosion Products 'The removal of corrosion products from metal specimens is described in Reference 1 and in ASTM RGI: 1988 and ISO/DIS 8407.2:1989 and certain of these procedures are described below. Electrolytic Catholic Cleaning
After scrubbing to remove loosely attached corrosion products, cathodically polarise in hot dilute sulphuric acid under the following conditions: Electrolyte-sulphuric acid (5% wt.%) plus an inhibitor (0.5 kg m - 3 ) such as diorthotolyl thiourea, quinoline ethiodide or 0-naphthol quinoline. The temperature should be 75"C, the cathode current density 2000 Am-* and the time of cathodic polarisation 3min. The anode should be carbon or lead. If lead anodes are used, lead may deposit on the specimens and cause an error in the weight loss. If the specimen is resistant to nitric acid the lead may be removed by a flash dip in 1:1 nitric acid. Except for this possible source of error, lead is preferred as an anode, as it gives more efficient corrosion product removal. After the electrolytic treatment, scrub the specimen with a brush, rinse thoroughly and dry. Electric treatment may result in the redeposition of a metal, such as copper, from reducible corrosion products, and thus decrease the apparent weight loss. Chemical Cleaning
Copper and nickel alloys Dip for 1-3 min in 1:l HCl or 1:lO H2S0, at room temperature. Scrub lightly with bristle brush under running water, using fine scouring powder if needed. Aluminium alloys
Dip for 5-10min in an aqueous solution containing 2 wt.% chromic acid (CrO,) plus 5 vol.% orthophosphoric acid (H,PO,, 8 5 % ) maintained at 80°C. Ultrasonic agitaion will facilitate this procedure. Rinse in water to remove the acid, brush very lightly with a soft bristle brush to remove any loose film, and rinse again. If film remains, immerse for 1 min in concentrated nitric acid and repeat previous steps. Nitric acid may be used alone if there are no deposits. (See comments on this method when used for corroded specimens in the paper by Mercer, A. D., Butler, G . and Warren, CV.M., Br. Corros. J . , 12, 122 (1977).) Tin alloys Dip for 10 min in boiling trisodium phosphate solution (15%). Scrub lightly with bristle brush under running water, and dry.
Lead alloys Preferably use the electrolytic cleaning procedure just described. Alternatively, immerse for 5 min in boiling 1To acetic acid. Rinse in water to remove the acid and brush very gently with a soft bristle brush to remove any loosened matter.
APPENDIX-THE REMOVAL OF CORROSION PRODUCTS
19: 121
Alternatively, immerse for 5 min in hot 5% ammonium acetate solution, rinse and scrub lightly. This removes PbO and PbSO,.
Zinc Immerse the specimens in warm (60-80°C) 10% NH,C1 for several minutes. Then rinse in water and scrub with a soft brush. Then immerse the specimens for 15-20 s in a boiling solution containing 5% chromic acid and 1% silver nitrate. Rinse in hot water and dry. Note: in making up the chromic acid solution it is advisable to dissolve the silver nitrate separately and add it to the boiling chromic acid to prevent excessive crystallisation of the silver chromate. The chromic acid must be free from sulphate to avoid attack on the zinc. Immerse each specimen for 15 s in a 6% solution of hydriodic acid at room temperature to remove the remaining corrosion products. Immediately after immersion in the acid bath, wash the samples first in tap water and then in absolute methanol, and dry in air. This procedure removes a little of the zinc and a correction may be necessary. Magnesium alloys Dip for approximately 1 min in boiling 15% chromic acid to which has been added with agitation 1% silver chromate solution.
Iron and steel Preferably use the electrolytic cleaning procedure, or else immerse in Clark’s solution (hydrochloric acid 100 parts, antimonious oxide 2 parts, stannous chloride 5 parts) for up to 25 min. The solution may be cold but it should be vigorously stirred. Remove scales formed under oxidising conditions on steel in 15 vol.% concentrated phosphoric acid containing 0.15 vol.% of an organic inhibitor at room temperature. Stainless steels Clean stainless steels in 20% nitric acid at 60°C for 20 min. In place of chemical cleaning, use a brass scraper or brass bristle brush or both, followed by scrubbing with a wet bristle brush and fine scouring powder. Other methods of cleaning iron and steel include immersion in molten sodium hydride and cathodic treatment in molten caustic soda. These methods may be hazardous to personnel, and should not be carried out by the uninitiated, or without professional supervision.
General Note Whatever cleaning method is used, the possibility of removal of solid metal is present. This will result in error in the determination of the corrosion rate. One or more cleaned and weighed specimens should be recleaned by the same method and reweighed. Loss due to this second treatment may be used as a correction to that indicated by the first weighing. F. L. LaQUE
19. I B Appendix -Standards for Corrosion Testing*
British Standards
BS 1224:1970 BS 16151987
BS 1872:1984 BS 2000:Part 135:1983 BS 2011: Part 2.1 Ka: 1982 Kb:1987 Kc: 1977 Kd: 1977
BS 3597:1984 BS 3745:1970 (1988) BS 3900: Part E6:1974 (1 989) Part E10:1979 (1989) Part F3:1971 (1986) Part F4:1968 (1985) Part F6:1976 (1 984)
Specification for electroplated coatings of nickel and chromium Method of specifying anodic oxidation coatings on aluminium and its alloys Specification for electroplated coatings of tin Rust-preventing characteristics of steamturbine oil in the presence of water Environmental testing Tests Test Ka. Salt mist Test Kb. Salt mist, cyclic (sodium chloride solution) Test Kc. Sulphur dioxide test for contacts and connections Test Kd. Hydrogen sulphide test for contacts and connections Specification for electroplated coatings of 65/35 tinhickel alloy Method for the evaluation of results of accelerated corrosion tests on metallic coatings Methods of test for paints Cross-cut test Pull-off test for adhesion Resistance to artificial weathering (enclosed carbon arc) and Addendum No. 1 Resistance to continuous salt spray Notes for guidance on the conduct of natural weathering test
Corrosion standards, including test methods, in use in Europe, including national, ISO, ASTM. NACE and CEN documents up to the year 1990 are described in the conference proceedings ‘Corrosion Standards: European and International Developments’, P. McIntyre and A. D. Mercer, eds, The Institute of Metals, London (1991).
19: 122
APPENDIX-STANDARDS FOR CORROSION TESTING
Part F8: 1976 (1986)
19: 123
Determination of resistance to humid atmospheres containing sulphur dioxide Part F9: 1982 (1985) Determination of resistance to humidity (continuous condensation) Part F12: 1985 Determination of resistance to cathodic disbonding of coatings for use in marine environments Part F13: 1986 Filiform corrosion test on steel Part G5 :1976 (1984) Determination of resistance to liquids BS 4292: Method for specifying electroplated coatings of gold and gold alloys Part 1:1989 Gold and gold alloys for engineering purposes BS 5117: Testing corrosion inhibiting, engine coolant concentrate (‘antifreeze’) Part 2: Methods of test for corrosion inhibition performance Section 2.1:1985 (1989) General procedures Section 2.2:1985 (1989) Glassware tests Section 2.3:1985 (1989) Recirculating rig test Section 2.4: 1985 (1989) Static engine test Section 2.5:1985 (1989) Field test BS 5466: Methods for corrosion testing of metallic coatings Part 1:1977 (1988) Neutral salt spray test (NSS test) Acetic acid salt spray test (ASS test) Part 2:1977 (1988) Copper-accelerated acetic acid salt spray test Part 3:1977 (1988) (CASS test) Part 4: 1979 Thioacetamide test (TAA test) Corrodkote test (CORR test) Part 5: 1979 Rating the results of corrosion tests on elecPart 6: 1982 troplated coatings cathodic to the substrate Guidance on stationary outdoor exposure Part 7: 1982 corrosion tests Sulphur dioxide test with general condensaPart 8: 1986 tion of moisture Saline droplets corrosion test (SD test) Part 9: 1986 Test methods for determining electrolytic BS 5735:1979 corrosion with electrical insulating materials Method for determination of resistance to BS 5903: 1980 (1 987) intergranular corrosion of austenitic stainless steels: copper sulphate-sulphuric acid method (Moneypenny Strauss test) Specification for electroplated coatings of BS 6137:1982 tin/lead alloys Specification for corrosion inhibiting, engine BS 6580:1985 coolant concentrate (‘antifreeze’) Method for the determination of bimetallic BS 6682~1986 corrosion in outdoor exposure corrosion tests
19: 124
APPENDIX -STANDARDS FOR CORROSION TESTING
BS 6918:1990 BS 6980: Part 1:1988 Part 2: 1990 Part 3: 1990 Part 4: 1990 Part 5: 1990 Part 6: 1990 Part 7: 1990
Glossary of terms for corrosion of metals and alloys Stress corrosion testing Guide to testing procedures Method for the preparation and use of bentbeam specimens Method for the preparation and use of U-bend specimens Method for the preparation and use of uniaxially loaded tension specimens Method for the preparation and use of C-ring specimens Method for the preparation and use of precracked specimens Method for slow strain rate testing
ASTM Standards A 262: 986 A 708: 979
A 763:1986
B 117:1990 B 154:1987 B 368:1985 B 380:1985 B 457:1967 (1980)
B 537:1970 (1980) B 735:1984 C 876:1987
D 130:1988 D 610:1985 D 714:1987
Practices for detecting susceptibilityto intergranular attack in austenitic stainless steels Recommended practice for detection of susceptibility to intergranular corrosion in severely sensitised austenitic stainless steel (intent to withdraw) Practice for detecting susceptibility to intergranular attack in ferritic stainless steels Method of salt spray (fog) testing Method for mercurous nitrate test for copper and copper alloys Method for copper-accelerated acetic acidsalt spray (fog) testing (CASS test) Methods for corrosion testing of decorative chromium electroplating by the Corrodkote procedure Method for measurement of impedance of anodic coatings on aluminium Recommended practice for rating of electroplated panets subjected to atmospheric exposure Test method for porosity in gold platings on metal substrates by gas exposures Test method for half-cell potentials of uncoated reinforcing steel in concrete Method for detection of copper corrosion from petroleum products by the copper strip tarnish test Method for evaluating degree of rusting on painted steel surfaces Method for evaluating degree of blistering of paints
APPENDIX
D 807:1982
D 849:1988
D 1014:1983 (1988) D 1280:1989 D 1374:1989
D 1384: 1987 D 1611:1981 (1986) D 1654: 1979a (1984)
D 1743:1987
D 1748:1983 D 1838:1984 D 2059: 1987
D 2247: 1987 D 2251:1985 D 2570:1985 D 2688:1983
D 2758:1986 D 2776:1979
D 2803:1982 (1987) D 2809: 1989 D 2847: 1985
D 2933: 1974 (1986) D 2943 :1986
- STANDARDS
FOR CORROSION TESTING
19: 125
Method of assessing the tendency of industrial boiler waters t o cause embrittlement (USBM embrittlement detector method) Test method for copper corrosion of industrial aromatic hydrocarbons Method for conducting exterior exposure tests of paints on steel Method for total immersion corrosion test for soak tank metal cleaners Method for aerated total immersion corrosion test for metal cleaners Method for corrosion test for engine coolants in glassware Test method for corrosion produced by leather in contact with metal Method for evaluation of painted or coated specimens subjected to corrosive environments Test method for corrosion preventive properties of lubricating greases Test method for rust protection by metal preservatives in the humidity cabinet Test method for copper strip corrosion by liquefied petroleum (LP) gases Test method for resistance of zippers to salt spray (fog) Practice for testing water resistance of coatings in 100% relative humidity Test method for metal corrosion by halogenated organic solvents and their admixtures Method for simulated service corrosion testing of engine coolants Test method for corrosivity of water in the absence of heat transfer (weight loss methods) Method of testing engine coolants by engine dynamometer Test methods for corrosivity of water in the absence of heat transfer (electrical methods) Test method for filiform corrosion resistance of organic coatings Test method for cavitation erosion-corrosion characteristics of aluminium pumps with engine coolants Practice for testing engine coolants in car and light truck service Test method for corrosion resistance of coated steel specimens (cyclic method) Method of aluminium scratch test for 1,1,1-trichloroethane
19:126
APPENDIX- STANDARDS FOR CORROSION TESTING
D 3262:1982 D 3310:1974 (1983) D 3842:1986 D 3843:1980
D 3911:1980 D 3912:1980 (1985) D 3929:1980 (1984) D 43401989 D 4350:1986 D 4627:1986 D 4798:1988 E 647:1988a
F 363:1979 (1985) F 482:1984
F 483:1977 F 981:1987
F1110:1988 G 1:1988
G 2:1988 G 2M:1988
G 3:1989
Test methods for corrosivity of solvent systems for removing water-formed deposits Recommended practice for determining corrosivity of adhesive materials Guide to the selection of test methods for coatings used in light-water nuclear power plants Practice for quality assurance for protective coatings applied to nuclear facilities Method for evaluating coatings used in lightwater nuclear power plants at simulated loss of coolant accident (LOCA) conditions Test method for chemical resistance of coatings used in light-water nuclear power plants Practice for evaluating the stress cracking of plastics by adhesives using the bent-beam method Test method for corrosion of cast aluminium alloys in engine coolants under heat-transfer conditions Test method for corrosivity index of plastics and fillers Test method for iron chip corrosion test for water soluble metalworking fluids Test method for accelerated weathering test conditions and procedures for bituminous materials (xenon-arc method) Test method for measurement of fatigue crack growth rates Method for corrosion testing of enveloped gaskets Method for total immersion corrosion test for tank-type aircraft maintenance chemicals Method for total immersion corrosion test for aircraft maintenance chemicals Practice for assessment of compatibility of bio-materials (non-porous) for surgical implants with respect to effect of materials in muscle and bone Test method for sandwich corrosion test Recommended practice for preparing, cleaning, and evaluating corrosion test specimens Practice for aqueous corrosion testing of samples of zirconium and zirconium alloys Test method for corrosion testing of products of zirconium, hafnium and their alloys in water at 633 K or in steam at 673 K [metric] Recommended practice for conventions applicable to electrochemical measurements in corrosion testing
APPENDIX-STANDARDS FOR CORROSION TESTING
G 4:1984 G 5:1987 G 7:1989
G 11:1983 G 15:1989a G 16:1971 (1984)
G 23:1989 G 28:1985 G 30:1979 (1984) G 31:1972 (1985)
G 32:1985 G 33:1988 G 34:1986 G 35:1988
G 36:1987 G 37:1985 G 38:1973 (1984) G 39:1979 (1984) G 41:1985
G 44:1988
19: 127
Guide for conducting corrosion coupon tests in plant equipment Standard reference test method for making potentiostatic and potentiodynamic anodic polarisation measurements Recommended practice for atmospheric environmental exposure testing of nonmetallic materials Test method for effects of outdoor weathering on pipeline coatings Definitions of terms relating to corrosion and corrosion testing Recommended practice for applying statistics to analysis of corrosion data Practice for operating light- and waterexposure apparatus (carbon-arc Type) for exposure of nonmetallic materials Method for detecting susceptibility to intergranular attack in wrought nickel-rich, chromium-bearing alloys Practice for making and using U-bend stress corrosion test specimens Recommended practice for laboratory immersion corrosion testing of metals Method for vibratory cavitation erosion test Practice for recording data from atmospheric corrosion tests of metallic-coated steel specimens Test method for exfoliation corrosion susceptibility in 2XXX and 7XXX series aluminium alloys (EXCO test) Practice for determining the susceptibility of stainless steels and related nickel-chromium-iron alloys to stress corrosion cracking in polythionic acids Recommended practice for performing stress corrosion cracking tests in a boiling magnesium sulphate solution Test method for use of Mattsson’s solution of pH 7 . 2 to evaluate the stress corrosion cracking susceptibility of copper-zinc alloys Practices for making and using C-ring stress corrosion cracking test specimens Practice for preparation and use of bentbeam stress corrosion specimens Test method for determining cracking susceptibility of metals exposed under stress to a hot salt environment Practice for alternate immersion stress corrosion testing in 3 5% sodium chloride solution
-
19: 128
APPENDIX-STANDARDS FOR CORROSION TESTING
G 46: 1976 (1986) G 47:1979 (1984) G 48: 1976 1980)
G 49: 1985 G 50: 1976 1984) G 51:1977 (1984)
G 52:1988 G 57:1978 (1984)
G 58:1985 G 59:1978 (1984) G 60:1986 G 61: 986
G 64: 985
G 66: 1986
G 67:1986
G 68: 1980
G 69:1981 G 71:1981 (1986) G 73:1982 (1987) G 78:1989
Recommended practice for examination and evaluation of pitting corrosion Test method for determining susceptibility t o stress corrosion cracking of high-strength aluminium alloy products Test method for pitting and crevice corrosion resistance of stainless steels and related alloys by the use of ferric chloride solution Recommended practice for preparation and use of direct tension stress corrosion test specimens Recommended practice for conducting atmospheric stress corrosion tests on metals Test method for pH of soil for use in corrosion testing Practice for exposing and evaluating metals and alloys in surface seawater Method for field measurement of soil resistivity using the Wenner four-electrode method Practice for the preparation of stress corrosion test specimen for weldments Practice for conducting potentiodynamic polarisation resistance measurements Method for conducting cyclic humidity tests Test method for conducting cyclic potentiodynamic polarisation measurements for localised corrosion susceptibility of iron-, nickel-, or cobalt-based alloys Classification of resistance to stress corrosion cracking of high-strength aluminium alloys Method for visual assessment of exfoliation corrosion susceptibility of 5XXX series aluminium alloys (Asset test) Test method for determining the susceptibility t o intergranular corrosion of 5XXX series aluminium alloys by mass loss after exposure to nitric acid (NAMLT test) Practice for liquid sodium corrosion testing of metals and alloys Practice for measurement of corrosion potentials of aluminium alloys Practice for conducting and evaluating galvanic corrosion tests in electrolytes Practice for liquid impingement erosion testing Guide for crevice corrosion testing of iron base and nickel base stainless steels in seawater and other chloride-containing aqueous environments
APPENDIX -STANDARDS FOR CORROSION TESTING
G 82:1983 G 84:1989 G 85:1984 G 87:1984 G 90:1985
G 91:1986
G 92:1986 G 100:1989 G 101:1989 G 102:1989
G 103:1989
G 104:1989
19:129
Guide for development and use of a galvanic series for predicting galvanic corrosion performance Practice for measurement of time-of-wetness on surfaces exposed to wetting conditions as in atmospheric corrosion testing Practice for modified salt spray (fog) testing Practice for conducting moist SO, tests Practice for performing accelerated outdoor weathering of nonmetallic materials using concentrated natural sunlight Test Method for monitoring atmospheric SO2using sulfation plate technique Practice for characteristics of atmospheric test sites Method for conducting cyclic galvanostaircase polarisation Guide for estimating the atmospheric corrosion resistance of low-alloy steels Practice for calculation of corrosion rates and related information from electrochemical measurements Method for performing a stress-corrosion cracking test of low copper containing AlZn-Mg alloys in boiling 6% sodium chloride solution Test method for assessing galvanic corrosion caused by the atmosphere
International Standards IS0 2160:1985 IS0 2810:1974
I S 0 3651-111976
IS0 3651-2~1976
IS0 3768:1976 IS0 3769:1976 I S 0 3770:1976 I S 0 4536:1985 I S 0 4538:1978
Petroleum products-corrosiveness to copper. Copper strip test Notes for guidance on the conduct of natural weathering tests Austenitic stainless steels-determination of resistance to intergranular corrosion. Part 1: Corrosion test in nitric acid medium by measurement of loss of mass (Huey test) Austenitic stainless steels-determination of resistance to intergranular corrosion. Part 2: Corrosion test in a sulphuric acidlcopper sulphate medium in the presence of copper turnings (Moneypenny Strauss test) Neutral salt spray test Acetic acid salt spray test Copper accelerated acetic acid salt spray test Saline droplet test Thioacetamide corrosion test
19: 130
APPENDIX-STANDARDS FOR CORROSION TESTING
IS0 4539:1980
IS0 4540: 1980 IS0 4541:1978 IS0 4542: 1981
IS0 4543:1981 I S 0 4623:1984 IS0 6251:1982 I S 0 6314:1980 IS0 6315:1980 IS0 6505:1984
IS0 6509:1981 IS0 6988: 1985 IS0 7 120: 1987
IS0 7384: 1986 IS0 7441:1989
IS0 7539-1 :1987 IS0 7539-2~1989
IS0 7539-3:1989 SO 7539-4: 1989
IS0 7539-5~1989
IS0 7539-6: 1990 I S 0 7539-7~1989 IS0/8044: 1986
Electrodeposited chromium coatings. Electrolytic corrosion testing Coatings cathodic to the substrate-rating of electroplated test specimens subjected to corrosion tests Corrodkote corrosion test General rules for stationary outdoor exposure corrosion tests General rules for corrosion tests applicable to storage conditions Filiform corrosion test on steel Liquefied petroleum gases-corrosiveness to copper. Copper strip test Road vehicles-brake linings-resistance to water, saline solution, oil and brake fluidtest procedure Road vehicles-brake linings-seizure to ferrous mating surfaces due to corrosion-test procedure Rubber vulcanised-determination of adhesion to, and corrosion of, metals Corrosion of metals and alloys; determination of dezincification resistance of brass Sulphur dioxide test with general condensation of moisture Petroleum products and lubricants-petroleum oils and other fluids-determination of rust preventing characteristics in the presence of water Corrosion tests in artificial atmosphere; general requirements Corrosion of metals and alloys; determination of bimetallic corrosion in outdoor exposure corrosion tests Corrosion of metals and alloys; stress corrosion testing. Part 1: General guidance on testing procedures Corrosion of metals and alloys; stress corrosion testing. Part 2: Bent-beam specimens Corrosion of metals and alloys. Part 3: Ubend specimens Corrosion of metals and alloys. Part 4: Uniaxially loaded tensile specimens Corrosion of metals and alloys. Part 5: Cring specimens Corrosion of metals and alloys. Part 6: Precracked specimens Corrosion of metals and alloys. Part 7: Slow strain rate stress corrosion tests Terms for corrosion of metals and alloys
APPENDIX -STANDARDS FOR CORROSION TESTING
ISO/DIS 8407.2: 1989 IS0 8565:1992 ISO/DIS 9223:1990 ISO/DIS 9224: 1990 ISO/DIS 9225:1990 ISO/DIS 9226: 1990
ISO/DIS 9227.2: 1989 ISO/DIS 9400: 1989 ISO/DIS 10062:1990 IEC 68-2-42: 1982 IEC 68-2-43 :1976 IEC 68-2-46: 1982 IEC 68-2-60: 1980
19: 131
Metals and alloys; removal of corrosion products from corrosion test specimens Metals and alloys; atmospheric corrosion testing; general requirements for field tests Corrosion of metals and alloys. Classification of corrosivity of atmospheres Corrosion of metals and alloys. Guiding values for the corrosivity categories of atmospheres Corrosion of metals and alloys. Corrosivity of atmospheres. Methods of measurement of pollution Corrosion of metals and alloys. Methods for the determination of corrosion rate of standard specimens for the evaluation of corrosivity Corrosion tests in artificial atmospheres; salt spray tests Nickel-based alloys-Determination of resisance to intergranular corrosion Corrosion tests in artificial amospheres at very low concentrations of polluting gas(es) Test K,-sulphur dioxide test for contacts and connectors Test &-hydrogen sulphide test for contacts and connections Guidance to test Kd Test K,-corrosion tests in artificial atmosphere at very low concentrations of polluting gases
NACE Standards NACE TM 0169-76 NACE TM 0170-70 NACE TM 0171-71 NACE TM 0172-86 NACE TM 0174-74 NACE TM 0177-86 NACE TM 0184-88
Laboratory corrosion testing of metals for the process industries Visual standard for surfaces of new steel airblast cleaned with sand abrasive Autoclave corrosion testing of metals in high temperature water Antirust properties of cargoes in petroleum product pipelines Laboratory methods for the evaluation of protective coatings used as lining materials in immersion service Testing of metals for resistance to sulphide stress cracking at ambient temperatures Accelerated test procedures for screening atmospheric surface coating systems for offshore platforms and equipment
19: 132
APPENDIX- STANDARDS FOR CORROSION TESTING
NACE TM 0270-72 NACE TM 0274-74 NACE TM 0275-89 NACE TM 0286-88
Method of conducting controlled velocity laboratory corrosion tests Dynamic corrosion testing of metals in high temperature water Performance testing of steel and reinforced plastic sucher rods by the mixed string alternate rod method Cooling water test units incorporating heat transfer surfaces P. McINTYRE A. D. MERCER
19.2 The Potentiostat and its Application to Corrosion Studies
The potentiostatic technique discussed here involves the polarisation of a metal electrode at a series of predetermined constant potentials. Potentiostats have been used in analytical chemistry for some time'; Hickling2was the first to describe a mechanically controlled instrument and Roberts3 was the first to describe an electronically controlled instrument. Greene4 has discussed manual instruments and basic instrument requirements. The determination of polarisation curves of metals by means of constant potential devices has contributed greatly to the knowledge of corrosion processes and passivity. In addition to the use of the potentiostat in studying a variety of mechanisms involved in corrosion and passivity, it has been applied to alloy development, since it is an important tool in the accelerated testing of corrosion resistance. Dissolution under controlled potentials can also be a precise method for metallographic etching or in studies of the selective corrosion of various phases. The technique can be used for establishing optimum conditions of anodic and cathodic protection. Two of the more recent papers have touched on limitations in its applications, and differences between potentiostatic tests and exposure to chemical solutions6. In this section an attempt is made to give a more detailed introduction to experimental procedures, as well as to some of the ideas where the use of the potentiostat has helped in the understanding of corrosion processes.
Experimental Apparatus Instruments very suitable for corrosion work are readily available, with several different models produced commercially. Although most, if not all, of the available potentiostats are properly designed, it should be kept in mind that corrosion studies require the instrument to have a low internal resistance and to react quickly to changes of potential of the working electrode. A basic circuit is shown schematically in Fig. 19.3qa). The specimen C., or working electrode W.E. is the metal under study, the auxiliary electrode A.E. is usually platinum and R.E. is the reference electrode, for instance a saturated calomel electrode. The desired potential difference between the specimen and the reference electrode is set with the backing circuit B. Any 19:133
19: 134
THE POTENTIOST'AT AND CORROSION STUDIES
0
rJ Amplifier
controller
(C)
Fig. 19.36 Basic circuit for a potentiostat. (a) Basic circuit for a potentiostat and electrochemical cell. (b) Equivalent circuit. (c) Circuit of a basic potentiostat. A.E. is the auxiliary electrode, R.E. the referenceelectrodeand W.E.the working electrode (band care from Potenriostat and its Appkcations by J. A. von Fraunhofer and C. H. Banks, Butterworths (1972))
unbalance between the electrode potential and the backing potential produces an error signal at the input of the amplifier-controller circuit. The latter rapidly adjusts the cell current between the specimen and the auxiliary electrode until the error signal is reduced to zero. The electrical characteristics of the cell and electrode will comprise both capacitative and resistive components, but for simplicity the former may be neglected and the system can be represented by resistances in series (Fig. 19.366 and c). The resistance R, simulates the effective series resistance of the auxiliary electrode A.E. and cell solution, whilst the potential developed across R, by the flow of current between the working electrode W.E. and A.E. simulates the controlled potential W.E. with respect to R.E. Figure 19.36~shows a basic circuit of a potentiostat in which the difference between the desired potential ( Vz)and the actual potential of the working
THE POTENTIOSTAT AND CORROSION STUDIES
19: 135
electrode (V,)is amplified by a high gain differential pre-amplifier. The output is an error signal A V, = A ( V2- V,),where A is the gain of the amplifier, which is arranged to control the power amplifier in such a way that the potential of A.E. is continuously adjusted to minimise V,. If the loop gain is high, V, can be made to approach zero very closely, the limit being determined by the electrical noise in the system. The potential of W.E. with respect to R.E. is thus held constant at the desired potential, V,. When potential setting is varied manually during the determination of polarisation, each change can be made after a constant time interval or when the rate of current change reaches a predetermined low level. A number of instruments for programmed potential changes have been introduced, permitting a variety of continuous sweeps or stepwise traverses over a desired range of potential. This, together with a suitable electrometer, recorder, noise filter (when necessary) and logarithmic converter, provide an automated procedure for plotting E-log i curves. An ASTM recommended practice (A Standard Reference Method for Making Potentiostatic and Potentiodynamic Anodic Polarisation Measurements, G5:1972) has been issued. It provides a means of checking experimental technique and instrumentation using a specimen from a single heat of AIS1 Type 430 stainless steel, which is available from ASTM*. Scanning Rate
The time factor in stepwise potentiostatic or potentiodynamic polarisation experiments is very important, because large differences can be caused by changes in the scanning rate. Since the steady state depends on the particular system and conditions of exposure, no set rule exists for the magnitude or frequency of potential changes. Chatfield et aLh have studied the Ni/H,SO, system and have shown how E,,,,. becomes more passive with increase in sweep rate. In order for the potentiostatic technique to provide an accelerated test, whether for general or localised corrosion, it is obvious that an accelerating factor is needed. Merely duplicating service conditions by substituting the potentiostat for chemical potential control does not necessarily shorten the required testing time. When employing an accelerating factor, such as higher temperature, change in chemistry of the environment,' or a greater driving force (potential), care should be taken to ensure that the mechanism of the reaction@)under examination is not altered. iR Corrections, Probe Positioning, Specimen Masking and Mounting
There is no difference between galvanostatic and potentiostatic polarisation experiments regarding the iR potential drop between the specimen and the tip of the probe used for measuring the electrochemical potential. In either case corrections should be made for accuracy. These could be quite large if the current density is high and/or the conductivity of the electrolyte is low. *American Society for Testing and Materials, Headquarters, 1916 Race Street. Philadelphia, Pa., USA.
19: 136
THE POTENTIOSTAT AND CORROSION STUDIES
The position of the probe relative to the test specimen surface can cause differences in potential readings’. Adequate specimen masking is one of the major problems in corrosion testing. Crevices with non-uniform current distribution and in which changes in the chemistry of the electrolyte can take place rapidly, are particularly undesirable’. In potentiostatic work, pitting o r crevice attack is frequently found near the masking interface (or under it when seepage and undermining occur). Unless this factor is specifically being investigated, it should be avoided. The use of a partially immersed specimen can eliminate the need for masking, although it leaves the water line to be dealt with. A method for mounting a cylindrical specimen has been described 9- lo, which avoids crevices when the holder is properly tightened (see Section 19.7). Specimens can also be rotated during the test, if desired, when this method is used.
Applications Studies of Passivity
The potentiostat is particularly useful in determining the behaviour of metals that show active-passive transition. Knowledge of the nature of passivity and the probable mechanisms involved has accumulated more rapidly since the introduction of the potentiostatic technique. Perhaps of more importance for the subject at hand are the practical implications of this method. We now have a tool which allows an ‘operational’ definition of passivity and a means of determining the tendency of metals to become passive and resist corrosion under various conditions. The use of the potentiostatic method has helped to show that the process of self-passivation is practically identical to that which occurs when the metal is made anodically passive by the application of an external current ”-”. The polarisation curve usually observed is shown schematically in Fig. 19.37~.Without the use of a potentiostat, the active portion of the curve AB would make a sudden transition to the curve DE, e.g. along curve AFE or AFD, and observation of the part of the curve BCDE during anodic polarisation was not common until the potentiostat was used. The current-potential relationship ABCDE, as obtained potentiostatically, has allowed a study of the passive phenomena in greater detail and the operational definition of the passive state with greater preciseness. Bonhoeffer, Vetter and many others have made extensive potentiostatic studies of iron which indicate that the metal has a thin film, composed of one Similar or more oxides of iron, on its surface when in the passive studies have been made with stainless steel, nickel, chromium and other metai~~~~~. Since the corrosion potential of a metal in a particular environment is a mixed potential -where the total anodic current is equal to the total cathodic current -the potentiostatic curve obtained by external polarisation will be influenced by the position of the local cathodic current curve. (Edeleanu” and Mueller” have discussed the details which must be considered in the analysis and interpretation of the curves.) For this reason, residual oxygen in the test solution can cause a departure from the usual curve; in such a
19: 137
THE POTENTIOSTAT AND CORROSION STUDIES
t
t
Positive
Positive
I
Transpassive region
I \
E
I I
\ ‘
Passive region
I
E
H
Gakdown G potential
EP
i
lo)
I
i,
I
lbl
Fig. 19.37 Schematic polarisation curves from anodic potentiostatic polarisation
case, a ‘negative loop’ B’ corresponding with cathodic reduction of dissolved oxygen occurs after passing the critical point, with the normal passive region and low positive currents only being resumed at Ec2,as shown in Fig. 19.38. Other ions can also interfere with the currents observed. If they are oxidising, then they will have much the same effect as dissolved oxygen. However, some of them may increase the observed current to values that suggest corrosion rates much higher than are actually taking place. The
Fig. 19.38 Schematic polarisation from potentiostatic polarisation. B‘ shows the ‘negative loop’ and represents the cathodic reduction of dissolved oxygen. The dashed curves in the diagram are cathodic currents and are frequently drawn on the left-hand side of the E axis
19: 138
THE POTENTIOSTAT AND CORROSION STUDIES
polarisation curve for titanium29 indicated current levels in the passive region that did not agree with the (lower) corrosion rates determined gravimetrically, and this was found to be due to the presence of Ti3+ions in solution. In the case of iron in neutral water3’, the passive anodic current densities were found to be proportional to the concentration of Fez+. Indig 3’ observed similar current increases in stagnant high-temperature water tests due to the formation of Fe,O, from Fez+;he largely eliminated this by changing to a flowing electrolyte system, and achieved correlation with actual corrosion rates. The critical current and primary passivation potential Eppwill not appear on an anodic polarisation curve when the steady-state potential already is higher than Epp. In such a case the potentiostat is unable to provide direct data for constructing the full polarisation curve. If that portion of the curve below the steady-state potential is desired, then the potential has to be held constant at several points in this range and corrosion currents calculated from corrosion rates as determined from solution analyses and/or weight losses. The potentiostat is very useful for determining the effects of composition and heat treatment on the corrosion resistance of alloys. Sometimes it is possible to understand what appear to be discrepancies in practice. Edeleanu2’ used the method for determining the resistance of stainless steels to acids. The potentiostatic curves showed that the current in the transpassive region (at a very high potential) increased with the chromium content, while the current in the passive region (at lower potentials) decreased. This explained the behaviour of several steels in service, where steels of higher chromium content showed poor resistance to corrosion in environments of high redox potentials -nitric acid plus chromic acid mixtures-but greater resistance in nitric acid. Edeleanu also discussed potentiostatic curves which showed the beneficial effects of nickel, copper and molybdenum on the corrosion resistance of stainless steel in sulphuric acid. This paper should be consulted for an excellent discussion on the use of these techniques for determining the effects of alloy composition on corrosion resistance. Cihal, eta/.32presented early data on the effects of chromium, nickel, molybdenum, titanium, niobium and silicon on the passive behaviour of stainless steel. Pitting
Some stainless steels and aluminium alloys are examples of metals that show pitting corrosion when exposed to aqueous solutions containing halide ions, although the phenomenon is not confined to these alloys. Various factors influence the onset of pitting, one of which is the interfacial potential; pits are thought to form only at potentials more positive than a certain critical value. This can be demonstrated by electrochemical measurements using potentiostatic techniques3342,and Fig. 19.37b which is similar to that shown by K~lotyrkin’s~~, represents a typical curve. In the absence of aggressive (pitting) ions, ABCDE represents the usual polarisation behaviour, where DE is the region of transpassivity. However, when conditions suitable for
THE POTENTIOSTAT AND CORROSION STUDIES
19: 139
pitting prevail, the curve ABCGH is typically found, and the breakout (or breakthrough) G H from the passive region is accompanied by pitting. The first point of departure from the passive region has been referred to as the ‘critical pitting potential’, but it should be noted that its value is time dependent and that it varies with the rate of potential change, i.e. if the potential sweep rate or stepping rate is too rapid, a more positive ‘critical’ value is obtained. In general, the shorter the time the more positive is the ‘critical pitting potential’, and in order to be certain that a reliable estimate of the pitting potential has been made, it is necessary to hold the potential at a constant value just below the critical point for a suitably long time (in practice, several days*), in order to demonstrate retention of passivity and the absence of pits in the specimen surface (see also Section 1.6). The correlation between the redox potential of a system and the occurrence of pitting attack was established some 30 years ago”. Also, the use of passivity breakdown as a screening method for alloy resistance was described over the years by several workers, for example, B r e n n e ~ ? ~Mahla ~, and NielsenQ6,and Pourbaix4’. For a metal in a given solution, it may appear that the electrochemical potential, regardless of its origin, will be the only determinant of whether or not pitting will take place. While this is generally expected, France and Greene6 suggested that a potentiostatically controlled corrosion test could be more severe than a conventional one (chemically controlled potential). Their reason is related to local chemistry changes required to preserve charge neutrality; during anodic polarisation migration of C1ions occurs to balance the excess positive charge produced by the Fez+ions and this results in an increase of C1- ions at the metal/solution interface and a consequent increase in pitting propensity. A similar movement of anions in, for example, a solution of FeCI, ,does not occur (see also Table 21.33). Figure 19.39 is taken from the work of France and Greene to illustrate the movement of C1- ions across a boundary line. Correspondence between electrochemical tests and field exposures has not always been found. Therefore, it is difficult to interpret potentiostatic data in terms of service performance. Disagreement between results can be caused by factors that are not immediately obvious. For example, gases such as hydrogen, argon or nitrogen are generally used to remove oxygen from solution before proceeding with a potentiostatic test. At first sight, it might appear immaterial which gas is used, but Wilde and Williams4’ found differences in the breakout (critical pitting) potential of stainless steels, depending on the selection of gas. More details of other factors that affect the critical pitting potential have been discussed by Uhlig and his c o - w ~ r k e r s They ~ ~ ~ ~indicated ~. that for stainless steel the critical pitting potential decreased with increasing concentration of chloride ion. At a fixed chloride level, passivating ions in solution, such as sulphate and nitrate, etc., cause the pitting potential to become more positive; at a sufficient concentration these ions totally inhibited pitting, as shown in Fig. 19.40 for S O 2 and (210,. Lizlovs and Bonds1 reported a molar ratio of 5:l (SO&:CI-) for inhibiting pitting in ferritic stainless steels. A plot of critical potential vs. *Even after several days of no localised attack, the question could legitimately be asked whether the test was long enough to establish a true value, below which pitting would never occur.
19: 140
THE POTENTIOSTAT AND CORROSION STUDIES
t
1
'1' I I
Na+CI-
I
Na*CI-
Na*CI-
j
Na'Ci
I
I
I
Fig. 19.39 Schematic representation of reactions during (u) controlled potential and (b) conventional corrosion tests in acidic chloride solutions. In (a)charge balance must be maintained by migration of CI- ions, since the cathodic reaction occurs elsewhere at the counter-electrode. In (b)the anodic and cathodic sites are in close proximity, and charge balance is maintained without migration of C1- ions from the bulk solution (after France and Greene6)
z
L
0001
0005001 Activity of
005 01 05 1.0 SO:- or CIO;
Fig. 19.40 Activity of SO$- or CIO, required to inhibit pitting as a function of CI- activity; 25°C (after Leckie and Uhlig44
ratio of S:- :Br- is reproduced in Fig. 19.41 from the work of Kolot ~ r k i n Lowering ~~. the temperature causes the critical potential to become more positive49.Changes of pH in the acid range did not affect the pitting potential appreciably, but in the alkaline region it increased markedly with pH, as shown in Fig. 19.42.
THE POTENTIOSTAT AND CORROSION STUDIES
1.0
-
; 3 g
i
19: 141
\
1.2
\
I
1.4-
1 I I
i 160 Y -0
L
-
1.8 L
5
0
10
IS
20
Ratio R of concentrution of
25
SO:- Br-
Fig. 19.41 Dependence of breakdown potential of Fe-Cr alloys (containing 13% Cr, in 0 - 1 mol dm-' HBr K, SO., solution) on the ratio of sulphate and bromide concentration in
+
solution (after Kolotyrkin43)
2
,
1
I
4
6
8 1 0 1 2
1
5
PH Fig. 19.42 Effect of pH on critical potential for pitting in 0 . I mol dm-' NaCi; 25°C (after Leckie and Uhlig44
The fact that scanning speed can affect polarisation behaviour has already been mentioned. In the case of stainless steel a plot of critical potential Eb vs. rate shows how Eb becomes more positive with potential change rate (Fig. 19.43)52.When a specimen was held at a fixed passive potential while aggressive ions (C1-) were added to determine the concentration required
19:142
Scan rate(mV/h)
Fig. 19.43 Effect of potential scan rate on the value of E,, for Type 304 stainless steel in 0-1 mol dm-’ NaCl (after LeckieS2)
for loss of passivity (pitting), then a similar time effect was observed53;a higher apparent resistance, Le. greater apparent C1- ion tolerance, was found when making more rapid additions. Numerous references regarding alloying additions have been published, and the reader should look up specific effects for relevant alloying systems. Potential-pt i Di8gr8ms
Pourbaix has contributed substantially to the science of corrosion through plotting thermodynamic data of systems as a function of electrode potential and pH. Numerous publications of his have appeared in the literature, as well as an atlas of potential-pH diagrams. Reference 54 exemplifies the usefulness of potentiostatic polarisation curves in the experimental plotting of various domains, such as protection, pitting, general corrosion and passivity, in these diagrams. This particular procedure, which has been dealt with in some detail in Section 1.6, is a very powerful tool which is now availabie for studying corrosion. De-all0 ying
The selective net loss of a component such as zinc, aluminium or nickel from copper-base alloys sometimes occurs when these alloys corrode. Early studies of the phenomenon were done by simple immersion. More recently, however, the potential-pH dependence of de-alloying has been examined ”, and it appears that this approach can provide a much more detailed understanding of the mechanism. Future experimental work is expected t o include potentiostatic and potentiodynamic techniques to a much greater extent.
THE F'OTENTIOSTAT AND CORROSION STUDIES
19:143
Selective Etching
Dissolution kinetics are influenced by pH, potential and the ions present in the test solution, and this forms the basis of selective metallographicetchThe potentiostat is ing techniques that have been used for some time32*56.57. often used to hold the potential of a multi-phase alloy constant at a level
y+ u+a'+/3
/v\
-4
-3
,A -2
-1
0
Potential, E ( V ) Fig. 19.44 Typical current ws. potential curves for alloys of various phase combinations (after Jones and Hume-Rothery58)
19: 144
THE POTENTIOSTAT AND CORROSION STUDIES
suitable for attack on a specific phase while the rest of the surface remains passive. A scan over a potential range can be done as a preliminary step to ascertain whether the electrochemical properties of various phases differ sufficiently in the chosen electrolyte for selective etching or phase extraction. One such procedure was reported by Jones and Hume-Rothery5*for austenitic steels alloyed with aluminium, and Fig. 19.44 is reproduced from their work. Another example is the behaviour of Fe-Fe3C as a function of potential, pH and anion59.In this latter work, the conditions were defined under which four different modes of attack took place, i.e. general-, matrix-, interface- and carbide attack, and the data were interpreted in terms of thermodynamics, kinetics and the influence of complex formation. These potentiostatic experiments were coupled with detailed electron transmission microscopy and the comprehensive nature of the results demonstrate the effectiveness of such a combined approach. The extraction of precipitates for further examination is also possible by the same techniques. Conditions would have to be chosen to give matrix or interface attack. Grain-boundiwy Corrosion
Intergranular corrosion is encountered in many metal systems, often associated with the presence of precipitates at grain boundaries. In the case of stainless steels, one widely accepted theory states that the precipitation of chromium carbides leads to a chromium-denuded zone which undergoes rapid corrosion. Potentiostatic methods, being capable of detecting differences in corrosion and passivation behaviour of various parts of a heterogeneous surface, have been applied to the electrochemical determination of grain boundary corrosion~o’63.61,66,68 Cihal and Prazakm determined the resistance of 1 8 0 stainless steel to this type of corrosion. They claimed that the technique could be used on steels which are difficult to test by other methods, including steels of low carbon content, and steels in which stabilising elements are present. By means of potentiostatic curves and light etching at constant potential they confirmed that the extent of intergranular corrosion depended upon the amount of precipitated chromium carbide. Corradi and Gasperini6’ claimed that the potentiostatic method was more effective and simpler than the Strauss test for determining intergranular corrosion of stainless steels, and suggested that the method may lend itself for use on finished equipment in service as a ‘non-destructive’ test. tabulated potentials which may be used for the selective Cihal, et etching of the various phases in several stainless steels. Bergholtz6’ suggested a potential of + 0.160 V vs. S.H.E. for grain-boundary etching of stainless steel, while D e s e ~ t r e t favoured ~,~~ various levels of potential, depending on its chromium content. He concluded that potentiostatic etching was more sensitive for determining susceptibilityto intergranular corrosion than chemical tests in boiling nitric acid or acidified copper sulphate. Budd and Booth6’ found the potentiostatic test best for investigating the intergranular and layer corrosion of aluminium alloys.
THE POTENTlOSTAT AND CORROSION STUDIES
19: 145
Not all test methods are necessarily accelerated by the use of a potentiostat. France and Greene6%used a potentiostat to hold sensitised 1818 stainless steels at various constant potentials in 1 N H,SO, in order to determine the range of potentials at which intergranular attack occurred (see Section 19.1, Fig. 19.19). However, this method of testing for sensitivity has been criticised by S t r e i ~ h e rwho ~ ~ ~points , out that the duration of the potentiostatic test is too short, and that alloys found to be immune during this test will suffer intergranular attack when the duration of exposure is more prolonged. Streicher's work69 indicates how useful the potentiostat has been in studying intergranular corrosion. Ideally, future data would be expanded to provide Pourbaix-type diagrams that also contain kinetic information showing various rates of attack within the general domain of intergranular corrosion. (Similar data for cases other than intergranular attack would be equally valuable.)
Stress-cornsion Cracking /s.c.c.1 S.C.C. has received a share of the potentiostatic approach to corrosion. Barnartt and van Rooyen7' reported that potentiostatically controlled corrosion in a potential range 50-100 mV above the corrosion potential provided an accelerated test for the S.C.C. of stainless steels. The elevation of the potential by means of a potentiostat eliminated the incubation period, and also increased the density of cracks. Booth and Tucker7' used potentiostatic methods in the S.C.C. of AI-Mg alloys. Hoar and his co-workers72,73 at first used galvanostatic equipment in their investigations into the 'mechano-chemical' dissolution of metal during plastic deformation; subsequently, the potentiostatic rather than galvanostatic control of potential was reported to give better results, and it enabled them to show that high corrosion rates were possible without appreciable elevation of the driving potential. This mechano-chemical theory has recently been refined in work reported for copper-base alloy^^^*^^. In the latter case, the potential dependence of the reactions leading to cracking has been analysed very carefully as a function of pH76 Staehle, et al.77have considered several aspects of S.C.C. from the electrochemical standpoint, including a feature of a recently suggested cracking mechanism which relates to the amount of corrosion that takes place each time that a slip step emerges at a surface. They recorded the shape of the current rise and decay curve that accompanied instantaneous straining (impact load), while the potential was controlled with a potentiostat. The number of coulombs of charge indicated the magnitude of metal dissolution. They believe that S.C.C. would be likely in a metal that showed the right amount of corrosion per slip step event. Another contribution of the potentiostatic technique to S.C.C. studies has been the report7' that cracking prevails essentially at two potential levels for metals showing an active-passive transition. These potentials are located near the top and bottom of the passive region. Along the same lines, Uhlig and his co-workers have determined critical ranges of potential for S . C . C . ~ ~ ~although ~, their theoretical interpretation differs from that of the other references cited.
19:146
THE POTENTIOSTAT AND CORROSION STUDIES
High-temperature Water
Pressurised water nuclear reactors require metals that will have a high degree of corrosion resistance to pure water at around 300°C. Laboratory testing of materials for this application have included potentiostatic polarisation experiments designed to clarify the active-passive behaviour of alloys as well as to establish corrosion rates. Since pressure vessels are used for this work, it is necessary to provide sealed insulated leads through the autoclave head83. Care should be taken to avoid short circuits; for instance, an insulated specimen, being common with the ground point of some potentiostats, can become electrically reconnected to the autoclave if the latter is not separated from ground by using an isolation transformer. Not all reference electrodes are suitable for use at high temperatures, and in addition, they may cause contamination if placed directly in pure water. A liquid junction consisting of a pressure-reducing tube with a wet string (or plug) has been employed81*84; this enabled locating a reference electrode at room temperature and low pressure outside the autoclave, but compensation for contact and other potentials was difficult. Platinum has been used directly in the high-temperature e l e c t r ~ l y t e ~and ~ * it~ ~functioned , as a hydrogen electrode as long as hydrogen was present. Most recently, a canned electrode has been described”, in which a silver-silver chloride reference is used inside a small container suspended in the hot water. Tiny pinholes and a long diffusion distance permitted a continuous electrolytepath while avoiding contamination of the test medium with chloride. A review article on techniques for electrochemical measurements in pressurised water has been written by Jones and Masterson88, which describes many of the experimental ramifications involved. The low conductivity of high-purity water makes it difficult to study electrode processes potentiostatically, since too high an electrical resistance in the circuit can affect the proper functioning of a potentiostat, and it can also introduce large iR errors. The increase in conductivity of water with temperature has been measured89and iR-corrected polarisation data have been obtaineda6in hot water that originally had very low conductivity at room temperature. Other r e ~ u l t s ” * ~ in ~ ’high-temperature ~’~~~~’ water are all for tests where the conductivity was deliberately increased through the addition of electrolytes. The interpretation of the polarisation curves requires care. Wilde” determined corrosion rates by linear polarisation in pure degassed water. Indig and Grootgo, however, found that electrochemical methods were unsuccessful in measuring the corrosion rate of Ni-Cr-Fe alloy 600in the presence of hydrogen, owing to the kinetic ease of the redox reaction on the metal surface. In another paper, the same authors” also indicated that linear polarisation generally gave corrosion rates that were not accurate, as a result of competing half-reactions, under conditions that were different from those used by Wilde. In a test with stainless steel they found that the removal of hydrogen could reduce the problems. Corrosion products are another source of error in the potentiostatic determination of polarisation curves in high-temperature water. In stagnant tests Fez+ could be converted to FeZ0?’, causing a false anodic current
THE POTENTIOSTAT AND CORROSION STUDIES
19: 147
reading in the passive region. This effect was eliminated by using a flowing electrolyte. S.C.C. has been examined as a function of potentials2 in high-temperature water with chlorides present and an increased susceptibility of stainless alloys to intergranular attack was found as the potential was increased. Additional work9' reported that no intergranular cracking was observed in tests of short duration. Hydrogen Penneation N2
Saturated calomel electrode
Saturated calomel electrode
pre -electrolysis cell Fig. 19.45
Apparatus for studying the permeation of hydrogen through thin metal foils
Reference elect
Cathodically polarised side of test membrane d side of the electrode uxiliary electrode
Auxiliary electrodes
Electrical timer and coupled relo
Fig. 19.46 Electrical circuit of cell and pre-electrolysis vessel (after Devanathan and Stachew~ki~~)
19: 148
THE POTENTIOSTAT AND CORROSION STUDIES
A very sensitive technique using the potentiostat was developed at the Unifor studying the permeation of hydrogen through versity of Penn~ylvania~~ thin metal foils. Such studies have and will continue to contribute important information in areas where hydrogen embrittlement is a problem. The technique involves the use of a double cell coupled by a thin metal membrane. Hydrogen is generated on the input side of the membrane which is maintained at a cathodic potential. Upon diffusion through the membrane, the hydrogen is electrochemically oxidised at the exit surface by an anodic potential that must be maintained constant by a potentiostat, and the anodic current provides a direct measure of the hydrogen flux. The technique is capable of detecting fluxes of 3 x 10-’cm3 of H, per second and can provide information on diffusivity, permeability, solubility and the interaction of hydrogen with metallic lattices. The apparatus is shown in Figs. 19.45 and 19.4693 and further details of the underlying theory are given in Section 20.1.
Molten Salts
The potentiostatic technique has been used in the investigation of the behaviour of metals in molten salts. In principle, the experimental method is the same as the one for aqueous media. Results are also capable of interpretation in the same way as those in aqueous solutions, and typical activepassive behaviour as well as anodic and cathodic Tafel lines have been observed. Reference 94 also contains several references to earlier work. These authors state that ’the potentiodynamic method, so successful in evaluating corrosion-resistant materials for aqueous systems, appears to be quite suitable also in selecting materials to be employed in molten salts’. In addition to the plotting of individual polarisation curves, it is possible to construct stability diagrams for molten salt systems resembling the well-known Pourbaix diagrams. The main difference is that the oxygen anion potential PO’-replaces the pH function, since the former is more important in molten salts. It has been established that salts can deposit or form on metals during gasmetal reactions. Molten layers could then develop at high operating temperatures. Consequently, the laboratory testing of corrosion resistance in molten salts could yield valuable results for evaluating resistance to some high-temperature gaseous environments.
Inhibitors
There are many published papers dealing with the electrochemical investigation of the effects of inhibitors and surfactants on corrosion processes, using the potentiostatgs*%.Adsorption of organic and inorganic ions on metal surfaces is found to be important, since it is related to their positive or negative charges as well as the potential of the metal surface. Some details regarding the use of polarisation techniques for examining specific effects on anode and cathode kinetics are described in References 95 and 96;the reader will find that numerous other papers are also available.
THE POTENTIOSTAT AND CORROSION STUDIES
19: 149
Rotating Disc-ring Electrodes
Frumkin and N e k r a ~ o vintroduced ~~ a rotating disc-ring electrode suitable for the detection of intermediates in corrosion reactions, and its theory was considered by Ivanov and Levich9*.In this method, a disc electrode (specimen) can be corroded under controlled conditions, and a metal ring around it is held potentiostatically at a predetermined potential E, in order to measure the rate at which an ionic species arrives at its surface; this is proportional to the current flowing in the potentiostat circuit. E, is varied to appropriate values for particular ions of interest. Figures 19.47 and 19.48 are taken from the work of Pickering and Wagnerw, who applied the technique in their study of the de-alloying phenomenon. Modifications used by other workers include the ‘split-ring’ technique.
Gbss
Plastic
Au(or C U )
\---cu-Au
(or
fl
Cu-Zn)
= 0,26cm
f z = 0.29cm r3= 0 4 L cm
--I k-r1 Fig. 19.47 Disc-ring electrode assembly
19: 150
THE POTENTIOSTAT AND CORROSION STUDIES
Cell
Fig. 19.48 Circuit used for the ionisation and redeposition experiments
Anodic Protection
The potentiostat has supplied an experimental tool for the study of anodic protection. The elucidation of passive behaviour made possible by potentiostatic anode polarisation curves allowed investigators to determine the conditions necessary for maintaining a metal in a stable passive condition by provision of a suitable environment, addition of cathodic alloying elements loo,lo', and/or maintenance of the required potential by means of external anodic p~larisation~'*~~.'". Edeleanu '02.'03 made use of potentiostatic curves to determine the optimum conditions for the protection of stainless steel in sulphuric acid. A pilot plant was then used to determine the practicability of anodic protection at a constant potential. He pointed out several factors necessary for proper control and indicated the spectacular results obtained. Stern, et ai.'' obtained potentiostatic polarisation curves for titanium alloys in various solutions of sulphuric acid and showed that the mixed potentials of titanium-noble metal alloys are more positive than the critical potential for the passivity of titanium. This explains the basis for the beneficial effects of small amounts of noble metals on the corrosion resistance of titanium in reducing-type acids. Hoar's review of the work on the effect of noble metals on including anodic protection should also be consulted'@'. The use of potentiostatic curves has also facilitated the study of the r61e of oxidising agents and inhibitors in corrosion processes. Stern lo discussed the r81e of passivating-type inhibitors and used potentiostatic curves to explain their action. Posey lo6 used the potentiostatic technique for determining the reduction of cupric ion on stainless steel. Both of these references
THE POTENTIOSTAT AND CORROSION STUDIES
19: 151
should be consulted for an extensive bibliography on these subjects. The reader will also find many subsequent papers dealing with this subject, and they generally confirm the principles that were set out in the aforementioned work.
Fast Electrode Reactions
Cahan, Nagy and Genshaw"' examine design criteria for an electrochemical measuring system to be used for potentiostatic transient investigation of fast electrode reactions. They emphasise the importance of co-design of the experimental cell and electronics. Accurate control of potential, stability, frequency response and uniform current distribution required the following: low resistance of the cell and reference electrode; small stray capacitances; small working electrode area; small solution resistance between specimen and point at which potential is measured; and a symmetrical electrode arrangement. Their design appears to have eliminated the need for the usual Luggin capillary probe. D. van ROOYEN
REFERENCES Lingane, J. J., Electroanalytical Chemistry, Interscience, N.Y., 2nd edn (1958) Hickling, A., Trans. Faraday Soc., 38. 27 (1942) Roberts, M. H., J. Appl. Phys., 5, 351 (1954) Greene, N. D., Corrosion, 15, 369t (1959) 5. France, W. D., Mats. Res. and Stds., 21, August (1969) 6. France, W. D. and Greene, N. D., Corrosion, 26, I (1970) 6a. Chatfield, C. J. and Shreir, L. L., Electrochim. Acta., 12, 563 (1972) 7. Mears, D. C. and Rothwell, G. P., J. Electrochem. Soc., 115, 36 (1968) 8. Greene, N. D., France, W. D. and Wilde, B. E., Corrosion, 21, 275 (1965) 9. ASTM Recommended Procedure, Committee G5 (1969) 10. Stern, M. and Makrides, A. C., J. Electrochem. Soc., 107, 728 (1960) 11. Tomashov, N. D., Adv. Chem., Moscow, 24, 453 (1955) 12. Heumann, T. and Rosener, W., Z. Elektrochem., 57.17 (1953) 13. Batrakov, V. P., C. P . Acad. Sci. U.R.S.S., 99, 797 (1954) 14. Edeleanu, C., Metallurgia, Manchr., 50, 113 (1954) 15. Stern, M., J. Electrochem. SOC., 105, 638 (1958) 16. Bonhoeffer, K. F. and Gerischer, H., Z. Elektrochern., 52, 149 (1948) 17. Weil, K. G. and Bonhoeffer, K. F., Z. Physik. Chem., 4, 175 (1955) 18. Franck, U. F. and Weil K. G., Z. Elektrochem., 56, 814 (1952) 19. Vetter, K. J., Z. Elektrochem., 55,274(1951)56, lM(1952); 58,230(1954); 59,67 (1955) 20. Kolotyrkin, Y. M., Z. Electrochem., 62, 664 (1958) 21. Frankenthal, R. P., J. Electrochem. SOC., 114, 542 (1967); 116, 680 (1969); Pickering, H. W. and Frankenthal, R. P., J . Electrochem. Soc., 112, 761 (1965) 22. Prazak, M.,Prazak, V. and Cihal, V., Z. Elektrochem., 62, 739 (1958) 23. Schwabe, K. and Dietz, G., Z . Elektrochem., 62, 751 (1958) 24. Okamoto, G., Kobayashi, H., Nagayama, M. and Sato, N., Z. Elektrochem., 62, 775 (1958) 25. Weidinger, H. and Lange, E., Z. Elektrochem., 64, 468 (1960) 27. Edeleanu, C., J. Iron St. Inst., 188, 122 (1958) 28. Mueller, W. A., Canad. J. Chem., 38, 576 (1960) 29. Stern, M. and Wissenberg, H., J. Electrochem. Soc., 106, 755 (1959) 30. Nagayama, M. and Cohen, M., J. Electrochem. SOC.,110, 670 (1963) 31. Indig, M. E., Ph.D. Thesis, RPI, Troy, N.Y. (1971) 1. 2. 3. 4.
19: 152 32. 33. 34. 35. 36. 37.
THE POTENTIOSTAT AND CORROSION STUDIES
Cihal, V. and Prazak, M., JfSf, 193, 360 (1959) Franck, U. F., Werkstofle u. Korrosion, 9, 504 (1958); 11, 401 (1960) Weil, K. G. and Manzel, D.. Z. Elektrochem.. 63. 669 (1959) Engell, H.J. and Stolica. N. D., Z. Phys. Chem.. 20, 113 (1959) Stoffils, H. and Schwenk. W., Werkstofle u. Korrosion, 12, 493 (1961) Kolotyrkin. Ja. M., Progress in Chemistry (USSR),21,322 (1962); J. Electrochem. Soc., 108,209 (]%I)
38. Kolotyrkin, Ja. M. and Gilman. V. A., Proc. Acad. Sci. U.S.S.R.,137, 642 (1961) 39. Vanleugenhaghe, C., Klimzack-Mathieiu, I., Meunier, J. and Pourbaix, M., International Conf. Corr. Reactor Materials. Salzburg (1962) 40. Popova, T. I. and Kabaniov. B. N., J. Physic. Chem. (U.S.S.R.).35, 12% (1961) 41. Acello, S. J. and Greene, N. D.. Corrosion, 18. 286t-290t. Aug. (1962) 42. Trumpler, G. and Keller, R., Helv. Chim. Acta.. 44, 1785 (1961) 43. Kolotyrkin, Ja. M., Corrosion. 19. 261t (1963) 44. Uhlig, H. H., Trans. Hur. Met. Min. Met. Eng., 140, 422 (1940) 45. Brennert, S., J. Iron Steel Inst., 135, 101 (1937) 46. Mahla, E. M. and Nielsen, N. A., Trans. Electrochem. SOC.. 89,167 (1946) 47. Pourbaix, M., Klimzack-Mathieiu, L., Mertens, Ch., Meunier, J . , Vanleugenhaghe, CI., deMunk, L., Laureys, L., Neelemans, L. and Warzee, M., Corrosion Sci., 3, 239 (1963) 48. Wilde, B. E. and Williams, E., J. Electrochem. Soc., 116, 1539 (1969) 49. Leckie, H. P. and Uhlig, H. H., J. Electrochem. SOC., 113, 1262 (1966) 50. Horvath, J . and Uhlig, H. H., J. Electrochem. Soc., 115, 791 (1%8) 51. Lizlovs, E. A. and Bond, A. P., J. Electrochem. Soc., 116 No. 5 , 574 (1969) 52. Leckie, H.P., J. Electrochem. SOC., 117, 1152 (1970) 53. Jackson, R. P. and van Rooyen, D., Corrosion, 27, 203 (1971) 54. Pourbaix, M., Corrosion, 26, 431 (1970) 5 5 . Verink, E. D. and Parrish, P. A., Corrosion, 26, 214 (1970) 56. Edeleanu, C., JISI, 185, 482 (1957) 57. Greene, N. D., Rudaw, P. S. and Lee, L., Corrosion Science, 6 , 371 (1966) 58. Jones, J . D. and Hume-Rothery, W., JfSI, 204, 1 (1966) 59. Cron, C. J., Payer, J . H. and Staehle, R. W., Corrosion, 27, l(1971) 60. Cihal, V. and Prazak, M., Hum. Lisf., 11,225 (1956); Corrosion, 16, 53Ot (translation) ( 1960) 61. Corradi, B. and Gasperini. R.. Mefallurg. /fa/., 52. 249 (1960) 62. Cihal, V. and Prazak, M. J., J. Iron St. Inst.. 193. 360 (1959) 63. Voeltzel, J. and Plateau, J., Compf. Rend.. 254, 1791 (1962) 64. Desestret, A., Mem. Sci. Rev. Met., 59. 553 (1%2) 65. Desestret, A.. Scuola Azione. 13, 165 (1964) 66. Schueller, H. J. et a/., Arch. Eisenhuttenw., 33, 853 (1962) 67. Bergholtz, G., (ZIS) Mitt.. 4, I134 (1%2) 68. Budd, M. K. and Booth, F. F., see Ref. 95. New York. 44 (1%3) 690. France, W. D. and Greene. N. D., Corrosion Science, 8, 9 (1968) 696. Streicher, M. A., Corrosion Science, 11. 275 (1971) 70. Barnartt, S, and van Rooyen. D.. J. Elecfrochem. Soc., 108, 222 (1961) 71. Booth, F. F. and Tucker, G. E. G., Corrosion, 21, 173 (1965) 72. Hoar, T. P. and West, J. M., Proc. Roy. Soc., A268, 304 (1962) 73. Hoar, T. P. and Scully, J. C., J. Electrochem. Soc., 111.348 (1964). Seealso Ref. 95, New York, 184 (1963) 74. Hoar, T. P. etal., Curr. Science, 11, 231 and 241 (1971) 75. Hoar, T. P., N A T O Conference on The Theory ofStress Corrosion in Alloys, Ericeira, Portugal, to be published 76. Hoar, T. P. and Rothwell, G . P., Electrochim. Acta., 15,1037 (1970) 77. Staehle, R. W., Royuela, J. J., Raredon, T. L., Serrate, E., Morin, C. R. and Farrar, R. V., Corrosion, 26, 451 (1970) 78. Staehle, R. W., NATOConferenceon The Theoryof StressCorrosion in Alloys, Ericeira. Portugal, to be published 79. Uhlig, H.H.and Cook, E. W., J. Electrochem. Soc., 116, 173 (1969) 80. Lee, H. H.and Uhlig, H. H., J. Electrochem. Sac., 117, 18 (1970) 81. Bacarella, A. L. and Sutton, A. L., J. Electrochem. Soc., 112, 5 4 6 (1965) 82. Hiibner, W.. de Pourbaix, M. and Ostberg, G., Papers at 4th Int. Cong. Met. Corr., Amsterdam (1969)
THE POTENTIOSTAT AND CORROSION STUDIES
19: 153
83. Taylor, A. H.and Cocks. F. H., Br. Corrosion J., 4,287 (1969) 84. Cowan, R. L.,Ph.D. Thesis, Ohio State University (1969) 85. Indig, M.E. and Groot, C., Corrosion, 25. 455 (1%9) 86. Wilde, B. E., Corrosion, 23, 331 and 379 (1967) 87. Indig. M. E. and Vermilyea. D. A.. Corrosion. 27. 312 (1971) 88. Jones, D. de G. and Masterson. H.G., Advances in Corrosion Science and Technology Vol. I (Ed. M. G. Fontana and R. W.Stacehle), Plenum Press (1970) 89. Wilde, B. E., Elecrrochim. Acta., 12,737 (1%7) 90. Indig, M.E. and Groot, C., Corrosion, 26, 171 (1970) 91. Hiibner. W., Johansson, B. and de Pourbaix. M.,Paper presented at Annual NACE Conference. Chicago, March (1971) 92. Devanathan, M. A. V. and Stachewski, Z., Proc. Roy. Soc., A270, 90 (1962) 93. Devanathan, M.A. V. and Stachewski, Z . , Journal of Electrochemical SOC., 111, 619 (1964)
94. Baudo, G., Tamba, A. and Bombara, G., Corrosion, 26, 293 (1970) 95. International Congress on Metallic Corrosion (first -London. 1961;second- New York, 1963; third-Moscow, 1 W . fourth- Amsterdam, 1969) 96. Symposium on Corrosion Inhibitors, Ferrara. Italy (1%1 and 1965) 97. Frumkin,A. N.andNckrasov, L. I., Dokl. Akud. Nauk. S.S.S.R., 126,115(1959);Proc. Acad. Sei. U.S.S.R., Phys. Chem. Sect., 126, 385 (1959) 98. Ivanov, Yu. B. and Levich, V. G., Dokl. Akad. Nauk. S.S.S.R., 126, 1029 (1959);Proc. Acad. Sci. U.S.S.R., Phys. Chem. Sect., 126. 505 (1959);Levich. V. G.,Physiochemical Hydrodynamics. Prcntice-Hall. Inc.. Englewood Cliffs. N.J.. 329 (1962) 99. Pickering, H.W.and Wagner, C., J. Elecrrochem. Soc., 114, 702 (1967) 100. Tomashov, N. D.,Adv. Chem.. Moscow, 24, 453 (1955) 101. Tomashov, N. D., Corrosion, 14, 22% (1958) 102. Edeleanu, C.. Nature, Lond.. 173, 739 (1954) 103. Edeleanu, C.. Metallurgia, Manchr., 50, 113 (1954) 104. Hoar, T. P., Plutinum Merals Rev., 4, 59 (1960) 105. Stern, M.,J. Electrochem. Soc., 105,638 (1958) 106. Posey, F.A., Cartledge, G. H.and Yake, R. P., J. Electrochem. SOC., 106,582 (1959) 107. Cahan, B. D., Nagy, Z. and Genshaw, M. A.. J. Elecrrochem. Soc., 119, 64 (1972)
19.3 Corrosion Monitoring and Inspection
Introduction The previous edition of Corrosion indicated an increased emphasis in monitoring of internal corrosion in high capital-cost process plant in the decade prior to 1976’. A number of reasons for this were given; for example the requirement for process plant to operate for longer periods between scheduled shut-downs, avoidance of unscheduled stoppages, increased management efficiency and reduction in incidents resulting in hazard or injury to both plant personnel and the general public. These requirements particularly applied to oil and gas production as well as production of chemicals and petrochemicals. Since 1976, all these factors have intensified, and reactions to political e.g. the OPEC crisis, which led to considerable expansion in the European offshore sectors, which would otherwise have evolved more slowly. Also, during this time a number of events have led to greater awareness of the benefits to be obtained from on-line inspection to gain, in particular, information on the corrosion conditions of the interior of plant, pipework and pipelines. An example was the destructive explosion at the chemical plant at Flixborough (UK). Although the cause was not directly attributable to corrosion per se, the event catalysed mandatory requirements for overall inspection relating to the operation of process plant. In addition, industry in general has been undergoing a massive reduction in the workforce at all levels of responsibility-inspection departments were not exempted. It was inevitable therefore, for management to exercise interest in automatic systems for both corrosion monitoring and inspection. The trend to automatic systems was considerably assisted by developments in computer technology especially the introduction of the microprocessor. Exploratory work regarding computer storage of corrosion monitoring data was reported in the last edition’. Since that time rapid progress has been made in computer involvement in both monitoring and inspection techniques. Details will be found under the respective techniques. ‘Closing the loop’ where a computer can control and provide a remedy for up to 80% of corrosion ‘alerts’ arising from on-line monitoring is already operating in a number of oil refineries. 19: 154
CORROSION MONITORING AND INSPECTION
19: 155
A factor which previously limited installation of automatic corrosion monitoring systems was the cost of cabling between sensors and control room instrumentation-this was particularly relevant to the electrical resistance (ER) systems. Developments to overcome this have included transmitter units at the probe location providing the standard 4-20mA output (allowing use of standard cable) for onward transmission to data systems or the use of radio linkage which has been successfully used for other processplant instrumentation. Industry in general requires user-friendly equipment for industrial monitoring without the need for expertise for operation and data interpretation. Equipment and techniques that do require corrosion expertise will therefore, be limited to a ‘service’, Le. specialist, company providing both equipment and personnel on a contract basis. In addition to the general range of monitoring equipment, a number of instruments have been developed for detection and measurement of corrosion for specific requirements. Examples are the ‘wheeled-trolley’using eddy currents for assessment of galvanised coating deterioration of high-tension transmission cables (Fig. 19.49) an electrochemical monitor for high-tension electricity cable pylons, a device for atmospheric monitoring (Fig. 19.50) and the use of thin-layer activation (TLA) for a subsea assembly used in oil production.
Fig. 19.49 Overhead line corrosion detector (courtesy Cormon Ltd.)
Developments in electrochemical methods since 1976 for measurement of corrosion have been rapid. Research and development has produced several new techniques, e.g. a.c. impedance and electrochemical noise. These methods require corrosion expertise for both operation and interpretation. Industry generally prefers instrumentation that can be operated by process
19: 156
CORROSION MONITORING AND INSPECTION
Fig. 19.50 Environmental corrosion monitor (courtesy Cormon Ltd.)
workers, so it is unlikely that such instrumentation systems will be sold ‘over the counter’, but will be supplied as a service to industry, as are the many advanced non-destructive testing (NDT) systems. A considerable catalyst to the corrosion monitoring market has been expansion in the production of oil and gas, not only in the usual oil areas (US and the Middle East), but also the offshore developments in Europe. In addition to the usual uncertainty of the onset or progress of internal corrosion in the operation of plant, the oil industry has to face the considerable problem concerning prediction of ‘field corrosivity’ and the possibility of the producing field becoming corrosive or more corrosive as depletion progresses. These factors have considerable influence on the installation of corrosion monitoring as oil and gas production is the major user of such equipment. It should be noted that there are still many deficiencies in the science and technology of corrosion monitoring, mainly in the areas of localised corrosion (pitting) and the inability to monitor at inaccessible sites such as ‘downhole’ (oil and gas wells) and subsea installations (satellite wells and pipelines). Other problems (and successes) experienced by users of equipment are outlined in surveys (1981 and 1984) which present a broad spectrum of industry applications*. The industries surveyed are shown in Table 19.6 and summaries of the findings of these surveys are shown in Tables 19.7 and 19.8. It is interesting to note that some of the problems highlighted in the 1981 survey had been considerably improved by the time of the 1984 survey. Plant designers are well supplied with corrosion data by materials manufacturers. These data are based on both experience and laboratory studies, but the information is usually based on specific parameters such as concentration of chemical or temperature. Edeleanu has emphasised this problem
CORROSION MONITORING AND INSPECTION Table 19.6
19: 157
Response rate by industrial sectors in the 1981 corrosion monitoring survey
Response Oil and gas production Oil treating Chemical/petrochemical Industrial boiler plants (excluding CEGB) Industrial cooling water systems Gas distribution Electricity generation Pipelines Mining Process plant contractors Miscellaneous
8/22 6/20 9/11 (5/50)' 2/4 2/10 0/5 3/7 o/ I 2/5 1/5 5/20
Total
38 responses total
36% 30% 82% 50% 20%
-
43%
-
40% 20% 25% 35%
'Five additional respanses from a blind survey of 50 companies. Table 19.7
Summary of the findings of the 1981 corrosion monitoring survey
Varied response Widespread concern with the significance of the results and their interpretation Ruggedness and reliability problems with equipment Concern with intrinsic safety of equipment Where corrosion monitoring is successful, several techniques are almost always in use Widespread interest in new techniques Universal use of NDT (especially ultrasonics and radiography Table 19.8
Summary of the findings of the 1984 corrosion monitoring survey
Increased awareness of the need for corrosion monitoring requirements to be considered at the plant design stage Expanded use of corrosion monitoring in many companies Greater appreciation of the qualitative nature of corrosion monitoring data and the need to learn by experience Improved reliability of equipment, but much still to be done Maintenance costs often too high Desire to go more automatic and on-line Awareness that system installation is expensive Continuing interest in new techniques
with regard to stainless steel corrosion in sulphuric acid. Although the designer may select the correct choice of steel, it is certain that many times in the life of the plant, changes in concentration and/or temperature (as well as other parameters) will occur with a consequent increase in the corrosion rate above that anticipated. Considerable corrosion monitoring is carried out utilising invasive methods, i.e. where the corrosion sensor is required to penetrate the pipe or vessel wall. Avoidance of penetration using non-invasive' methods (thin layer activation, ultrasonics, radiography and magnetic fingerprinting) is receiving considerable developmental attention. No one method for corrosion inspection is sufficient in itself and it is extremely dangerous to rely on data provided by one method only. A study is required of all methods available and the most suitable then chosen -
19: 158
CORROSION MONITORING AND INSPECTION
usually two or three methods are necessary. These are then used and additional methods can be called upon to supplement data if excessive corrosion is experienced or requires verification. This principle, outlined in the previous edition of Corrosion', was confirmed in the results of the surveys described above, Le. an integrated system for corrosion monitoring and inspection achieved most of the objectives required'. Since 1976, two conferences have been held in London solely addressed to corrosion monitoring (see Bibliography). In addition, the main annual conferences devoted to corrosion in the US and UK, sponsor sessions in corrosion monitoring and the conference proceedings should be consulted for additional information. Also the NACE* and I Corr-t Technical Committees have produced Guidance and Recommended Practices relating to various aspects of corrosion monitoring (see Bibliography; also details in text). Several publications and composite articles describing general aspects of the subject have been published since 19764-9. Several economic benefits arising from the use of corrosion monitoring have been published. An example of the savings that can be achieved relates to an offshore oil-field in the Norwegian sector". Carbon dioxide corrosion problems necessitated a chemical inhibitor programme combined with the installation of automatic corrosion monitoring (cost $1 2 million). The number of workovers (repairs to an oil-well), was reduced from twelve (1981) to one (1983). Each workover cost an average of $2 million. This confirms the savings resulting from corrosion monitoring indicated previously
-
'.
Variables Affecting Corrosion Monitoring The shortcomings of plant testing are considerable. Variables that affect the rate and type of corrosion are chemical composition, temperature, pressure, trace compounds or contaminants, velocity, presence of insoluble metal compounds, presence of insoluble materials (either as abrasives or deposits), crevices, stress (both magnitude and type are important), interface effects, phase changes (vaporising or condensing), chemical composition of the metallurgical condition of the metal and galvanic effects". No single corrosion test can include all of these variables and the corrosion data obtained should only be considered in this context. Mechanical phenomena, localised corrosion and stress-corrosion cracking are some of the factors that cannot be assessed with accuracy. Some sensor systems are now available which can make provision for thermal conditions.
Selection of Inspection Points The selection of inspection points is of paramount importance, and factors
to be considered have been outlined by Abramchuk ''as follows:
1. Abrupt changes in direction of flow such as elbows, tees, return bends
and changes in pipe size which create turbulence or changes in velocity. *National Association of Corrosion Engineers (USA). t Institute of Corrosion (UK).
CORROSION MONITORING AND INSPECTION
19: 159
2. Presence of ‘dead-ends’, loops, crevices, obstructions or other conditions which may produce turbulent flow causing erosion or stagnant flow which will allow debris or corrosive media to accumulate and set up corrosion cells. 3. Junctions of dissimilar metals which might promote severe galvanic (bimetallic) action. 4. Stressed areas such as those at welds, rivets, threads or areas that undergo cyclic temperature or pressure changes.
Selection of inspection points therefore, should be based on a thorough knowledge of process conditions, materials of construction, geometry of the system, external factors and historical records. Some of these factors may not be present, for example, when new plant is commissioned. There is greater knowledge now available concerning hydromechanics of fluid flows which should be considered when installation is planned. Oil and gas streams require additional consideration regarding the phase to be monitored, Le. water or oil stream, which in turn, will depend on the objectives of the monitoring, Le. metal loss and/or optimisation of corrosion inhibitor addition. Factors to be considered are: (a) Top, middle or bottom line monitoring?
(b) Flow regimes existing in pipeline or vessel? (c) Presence of different phases?
Access, Fittings ASTM G4 (latest revision) gives guidance for conducting plant corrosion tests, and in particular, for various methods for mounting specimens (coupons) in process plant. This standard evolved from ASTM* and NACE Technical Committees. Monitoring devices can be installed via standard gate-valves but these are only suitable for low pressures. Since the last edition of Corrosion’, there has been a trend to mechanical insertion/retrieval devices for coupons and sensors. These devices can be operated at pressure (whilst plant is operating), and utilise access fittings (Fig. 19.51). The equipment can now be obtained from several manufacturers in USA and Europe. These systems are safer to operate compared with probes utilising pressure glands fitted to plant via gate-valves. Consideration should be given to adequate space surrounding an access fitting thus allowing easy operation of the retrieval tool. Access fittings can be used for a variety of sensing devices (coupons and electronic probes), and can also be used for corrosion inhibitor injection using a suitable ‘quill’ injection nozzle. The exposure of sensors in a by-pass stream (which can be valved off), is an alternative way of collecting monitoring data although correlation is required between the main-stream and the by-pass The use of a side-stream taken either side of a choke in the main-stream can provide a useful monitoring point. Traps where product streams can be condensed can offer alternative sampling systems. *American Society for Testing Materials.
19: 160
CORROSION MONITORING AND INSPECTION
HOW THE PLUG IS REMOVED AND INSTALLED UNDER PRESSURE 1. Access Fitting in service,
3. Retriever is attached lo plug.
externals removed. 2. Service Valve installed, retriever goes in.
4. Retriever extended.
1.
2.
plug past
gate, valve closed
5. Plug removed.
3.
4.
5.
Fig. 19.5 1 Access fitting and retrieval tool (courtesy Rohrback-Cosasco Systems Ltd.)
Methods Available Coupons These are usually strip, flush discs or cylindrical rods mounted in suitable racks (inserted and retrieved at shutdown), or installed in the plant using high-pressure access systems Coupons are available from several manufacturers in a variety of materials and surface finishes and are supplied pre-weighed. Comprehensive guidance for the preparation and installation of coupons is given by ASTM Method G4 (latest revision) and NACE Standard RP0775 (latest revision). Coupons can be heat-treated to represent plant material. It is important to anneal any stress arising from cold-work such as stamping or guillotining. Coupons can be welded for assessment of weld-corrosion and used to assess the possibility of crevice corrosion (using plastics or rubber bands). Photographic records provide valuable data as to the condition of coupons following exposure. Penetration of metal due to corrosion is calculated from the weight loss assessed gravimetrically.
CORROSION MONITORING AND INSPECTION
19: 161
Post-exposure techniques are well documented for a variety of metals (ASTM and NACE). A disadvantage of coupon techniques is that the response to severe corrosion that may occur for short periods of time is not detected because the response measured is only an average for the period of exposure, although coupons can be withdrawn at intervals of time provided a sufficient number are placed at the start of exposure. The important aspect of coupons is that data obtained from their use can be used as base-line and therefore can be used to correlate corrosion data from other methods. Electrical Resistance Electrical resistance (ER), is the oldest electronic method for measuring corrosion following development for industrial moniGuidance for using ER probes is toring in the US in the late 1940~'*~-*. given in NACE publication 3DI70. Most of the development was initiated by the oil industry. Instrumentation is now available from around six manufacturers in both Europe and the USA compared with only two in 1976. A wire element (made of the metal of interest) is mounted in a suitable casing and exposed to the corrosive medium, which can be either liquid or gas. The element decreases in thickness due to corrosion and, as most corrosion products have greater electrical resistance than the metals from which
brrosive nvironment
Probe
)Cable or radio link
1
1
lnstrument
Fig. 19.52 Electrical resistance probe-circuit
19: 162
CORROSION MONITORING AND INSPECTION
they were formed, an accurate measurement of the increase in resistance can be equated with metal lost (Fig. 19.52). Temperature compensation for resistance changes is provided by reference elements mounted in the stem of the probe and protected from the corrosive environment. A variety of elements in different geometric forms, e.g. wire, tube and strip, corresponding to the commercial metals and alloys used in the process plant, are available. Also a variety of casings or housings are available depending on temperature and pressure requirements (Figs 19.53 and 19.54).
- - - - c e + L - --..l905 mrn full
,
*
r-,---J
port or 25.40mrn reduced port gate or ball valw
(customer furnished)
t
-___
0 dim
Fig. 19.53 Electrical resistance probe with screw-in fitting (courtesy Rohrback-Cosasco Systems Ltd.)
12 70 mrn npt mounting
1 12 70mrn k S + d length k 5 I
O
m
m
dI
Fig. 19.54 Electrical resistance probe with retractable fitting (courtesy Rohrback-Cosasco Systems Ltd.)
Probes require insertion and removal from plant whilst the plant is operating and various methods for this are available (see section Access Fittings). Consideration should be given to the compatibility with the corrosive environment of the probe casing materials as well as seals used in construction. A complete range of instrumentation is available from portable units to automatic systems utilising many probes. Transmitter units are available which can be located at the probe and transmit ER data into the 4-20 mA standard instrument signal. Radio linkage from transmitter to control room or nearby offshore platform is available commercially. A satellite link has been used to monitor offshore platform ER probes at the onshore base" in a Norwegian oilfield. The data obtained from ER probes, and those provided by test coupons, are similar in giving an integrated or average rate, but the former has the advantage that the data are obtained whilst the plant is operating. The time periods can be decided by the frequency of measurement, and periods of
CORROSION MONITORING AND INSPECTION
19: 163
changing corrosion rate can be detected and measured whilst the plant is on-stream. It is usual to plot the probe reading against time, the slope of the line giving the corrosion rate at any particular time. Pitting of the wire element increases the slope of the graph as the instrumentation cannot discriminate between general and localised attack. Pitting should be suspected if an increase in slope does occur and no changes have occurred in plant process conditions that would increase the general corrosion rate. Inspection of the element (another advantage in using retractable probes) can usually confirm whether pitting is occurring or not. A ‘flush’ strip element is available for use in pipelines and obviates the necessity for withdrawal of probes prior to use of inspection vehicles. This element design more closely represents the inner wall of a pipe or pipeline. The ER system has been used successfully in a range of industries for process plant monitoring. As ER can be applied in any liquid or gaseous environment the areas of application are considerable. However, there is a problem with ER if a conductive corrosion product is produced as is the case with sour crude oil or gas due to the deposition of iron sulphide. Specific applications (apart from process plant monitoring) reported are:
(a) atmospheric monitoring (b) assessing effectiveness of cathodic protection (connecting an ER probe to the structure being cathodically protected); (c) automobile body corrosion; ( d ) reinforcement corrosion (concrete); (e) marine piling; (f)aircraft An ER probe specifically designed for assessing the effectiveness of cathodic protection is shown in Fig. 19.55. The elements for this probe can be machined from the actual pipeline. Electrochemical Techniques
An overall assessment and guidance to electrochemical techniques has been published 1 3 . Direct Current
Potential measurement This technique* has provided valuable information as to the condition of ‘passive/active’ materials, particularly in the chemical industry ’.3-5*8. Although quantitative weight loss measurements are not obtained, measurements can be on-line and more importantly, can be monitored using the actual plant material (in situ) as a sensor. Choice of an appropriate reference electrode remains an enigma for ‘noncorrosion-aware’ personnel-although commercially available polarisationresistance probes can be adapted. An interesting aspect concerns localised corrosion in that, for some materials, localised corrosion only occurs within characteristic potential ranges. * Potential measurementsconcerning cathodic and anodic protection are excluded here.
19:164
Fig. 19.55
CORROSION MONITORING AND INSPECTION
Electrical resistance probe for assessment of cathodic protection (courtesy SSL Ltd.)
Potentiodynamicpolarisation The characteristics of passive/active conditions for metals can be readily defined using this techniquei4. Details for laboratory application can be found in ASTM Standard G5 (latest revision). Application in plant is easily performed as portable equipment (potentiostat) is available from several manufacturers, with some models incorporating built-in computer facilities. Tafel plots The linear part of the anodic or cathodic polarisation logcurrent and potential plot is extrapolated to intersect the corrosion potential linei4. Low corrosion rates can be measured relatively quickly. Note that resultant oxide films may be of different composition from those occurring in practice owing to the several decades of current applied which may not relate to actual plant practice. Portable apparatus including computing facilities is commercially available for plant testing. Polarisation resistance This technique, sometimes referred to as ‘linear polarisation resistance’ (LPR), has been applied widely in industrial monitoring because of its ability to react instantaneously to a corrosion situation . The limitation of the technique arises or change in corrosion rate i*4*5*8*i3-’7 from the necessity to have a defined electrolyte as the corrosive (the author has seen an LPR probe installed in a dry gas-line in an oil refinery). The method applies a small potential (usually 10-30mV) to a test elecThe resultant current trode on either side of the corrosion potential (Emrr).
CORROSION MONITORING AND INSPECTION
19: 165
is plotted against potential and the slope is used to calculate the corrosion rate. Aspects and errors arising from the use of two- or three-electrode probes (Figs 19.56 and 19.57) have been discussed Recently there have been indications that modifications (by chemical treatment) to the probe dielectric (insulating the electrodes from one another) enable measurements to be made in less conducting solutions”. The process is thought to render the dielectric more conducting, minimising the IR drop which can be responsible for errors in using LPR. Cyclic polarisation This type of measurement is similar to potentiodynamic anodic polarisation with the difference that, following an anodic polarisation plot, the test specimen is subjected to a cathodic stimulus, Le. a reverse scani4. Any hysteresis, Le. deviation from the anodic plot, can
\
TWO
‘ 0 0 Fig. 19.56 Two- and three-electrode polarisation probes (courtesy Rohrback-Cosasco Systems Ltd.)
Standard probe body from cadmium-plated mild steet (other metals available on special order)
Each probe furnished with one set (3) M-620 S mild steel replaceable electrodes and gaskets
-
Moulded Viton A surface
n 9 mm
r--
bM I L electrical f ittii’9
38 1 mm-&F’robe assemblyd Bottom v i e w normal thread 47.6mm for M-620SMS engagement electrodes (dimension will vary for other materials)
Fig. 19.57 Polarisation
probe-three-electrode in screw-in fitting (courtesy Petrolite Corporation)
19: 166
CORROSION MONITORING AND INSPECTION
indicate a tendency for localised corrosion such as pitting. This method is widely used in the laboratory mainly for specifying corrosion properties of new alloys. With availability of portable potentiostats (plus computer data and plotting facilities), these measurements can be made in situ in process plant. Practices for carrying out these tests are given in ASTM G61. There is some controversy regarding this technique.
Galvanic current Measurement of the galvanic current between two different metals can be easily measured using a zero resistant ammeter ‘94-8. This method can have specific application, e.g. to provide a signal indicating failure of a protective coating in a process vessel. Commercial probes are available for industrial monitoring. Measurements of current using ‘same-metal’ electrodes are utilised for electrochemical noise’ measurements (see section below). Alternating Current
EIectrochemical noise This is a non-perturbation method and is defined as random low frequency low amplitude fluctuations either of the potential or current in a corroding system. Analysis of the corrosion potential noise can provide information relating to both the mechanism and kinetics of the corrosion ~ c c u r r i n g ~ The ’ ~ *method ~ ~ ~ ~has ~ . been applied to industrial monitoring in power generation plant, cooling water systems and reinforcement in concrete, and the method can provide information concerning localised corrosion and loss of passivity. The corrosion process is observed as a series of events which all contribute to the overall corrosion rate. Measurement of rest potential fluctuations between two identical electrodes of potential fluctuations with respect to a fixed reference can be carried out. The electrochemical noise output spectrum is analysed using digitised data. The interpretation requires electrochemical expertise, and the method is therefore usually provided as a specialised service. Development of this technique by CAPCIS (UMIST, Manchester, UK), has led to an instrument system utilising several electrochemical techniques (dx. and ax.) from a multi-element probe. Electrochemical noise was able to operate in an acid-condensing environment with small amounts of liquidz2.The combination of data using several electrochemical techniques enabled identification of the corrosion mechanism in this application. Impedance Some of the errors arising from the use of linear polarisation resistance led to interest and development in a.c. systems. 49 9, 24-26. An early development used a fixed a.c. frequency and a commercial instrument was produced in the UK. Inaccuracies still occurred, however, and were due to the electrode impedance which is fequency dependent. Electrode reactions have a capacitance component, in addition to resistance, resulting in a requirement to measure the impedance. However, the total impedance comprises values for the reaction, solution, diffusion and capacitance. Measurements at different frequency are more reliable, particularly where high solution resistances occur. Simplifications for industrial monitoring have been developed consisting of two measurements, i.e. at a high (10 kHz) and low frequency (0- 1 Hz). The high-frequency measurement can identify the
CORROSION MONITORING AND INSPECTION
19: 167
electrolyte resistance and the other measurement relates to the corrosion process. The advantage of the method is that measurements can be made in more resistive corrodents (crude oil and concrete) compared with linear polarisation measurements. However, corrosion expertise is required for both operation and interpretation, and most industrial applications, therefore, are provided as a specialised service. The method is referred to as ‘electrochemical impedance spectroscopy’ (EIS), by Man~field’~. Hydrogen
The hydrogen pressure-probe for detection of corrosion-produced hydrogen in industrial plant has been de~cribed’.~*~.*’. Since 1976, other manufacturers have produced equivalent versions of this probe type. Developments since 1976 have led to vacuum probesa, and the ‘patch pressure probe’ where a saddle of steel is welded to the outside of the pipe or vessel and a suitable pressure gauge mounted in the saddlez9,thus avoiding penetration of the pipe wall. Electronic instrument systems are now available using a portable electrochemical cell that can be strapped to the pipe wall. 3&3’. A solid electrolyte probe which overcomes problems arising from using corrosive acids (required as the electrolyte medium) is available3’. These electronic probes work by oxidising hydrogen atoms on a fixed potential surface. All the above systems for monitoring the presence of hydrogen have given good performance in industrial applications. However, correlation with actual corrosion is not always easy and, at best, can only provide qualitative information. Guidance related to hydrogen probes is provided in NACE document IC1 84. Thin Layer Activathn
Thin layer activation (TLA) has a long experience in monitoring or measuring wear and erosion. A small quantity of radioisotope tracer is introduced into the metal surface which can be either a coupon or component. Metal loss due to corrosion (provided the corrosion product is non-adherent) can be detected remotely with high A number of industrial applications have been described including successful subsea in~tallation~~. Double layer activation has been used in the laboratory for estimation of both shape and growth of pits in stainless steel. Advantages of TLA are that (a) the concept is easily understood, (b)interpretation is relatively easy and (c) the system installation can be noninvasive. Test Heat Exchangers
Test heat exchangers can be fitted to side-stream circuits in process plant which simulate actual temperatures existing in the plant 35. Condenser tubes can be removed and corrosion assessed.
19: 168
CORROSION MONITORING AND INSPECTION
Spool Pieces
Small lengths of piping (approx. 1 m in length) can be fitted to plant using appropriate flanges and can provide valuable data as to plant corrosion characteristics. The spool can only be removed at shut-down. An alternative procedure is to install the spool on a by-pass loop. Chemical and Bacteriological Analysis
The contribution to corrosion monitoring by a well-planned chemical analysis programme is often underestimated. The most important analyses related to corrosion behaviour are: oxygen, hydrogen, trace elements, iron, manganese, corrosion inhibitor, other treating chemicals (biocides/oxygen scavenger), and pH. Guidelines for iron analysis (counts) in oil and gas production are given in NACE Standard T-1C-7 (latest revision). Sampling points for analysis should be planned in order to give the whole record of analysis, e.g. in oil production from the well via processing facilities to the export facilities. Correlation with plant parameters such as temperatures, pressures, flow rates and, stream compositions can provide valuable records. Emphasis on chemical analysis should be incorporated in the design of process plant. Automatic analysis systems especially those using ion-sensitive electrodes with recording of data are recommended. The initiation of corrosion due to bacteria is well recognised in a number of areas, in particular in oil production, the pulp and paper industry and municipal waste systems36. Testing for sulphate reducing bacteria (SRB)is expensive and time consuming, and considerable effort has been devoted to improve testing methods. Guidance for tests relating to oilfield practice are given in APl* RP-38 and by the Institute of Corrosion. Methods of monitoring include bacterial analysis (via filters), side-streamsand vessels and specially designed corrosion coupons 37*38. Rapid methods for assessing bacteria include ATP? photometry, fluorescence microscopy and radiorespirometry. However, these methods are not regarded as suitable for routine use”. Commercial test kits are now available for sensing bacteria-a recent test developed by DuPont can be used for field testing and it is claimed that results can be obtained in 15 minutes39. The Robbins device3’ comprises a 25mm diameter pipe (stainless steel) with a series of carbon-steel studs - any sessile bacteria will deposit on the studs and can be quantitatively recovered and analysed. Variations of the Robbins device are commercially available for insertion into plant using standard access fittings. Inspection
Visual Inspection and Mensuration This is regarded by Abramchuk’’ as so fundamental that it should be a logical prelude to most other inspection methods. A properly executed visual inspection can: *American Petroleum Institute. ?ATP = Adenosine Triphosphate.
CORROSION MONITORING AND INSPECTION
19: 169
assist in failure analysis; indicate the need for further exploration; help to define the search area if further exploration is warranted; aid in suggesting techniques for further exploration. aid in determining the measures needed to prevent the recurrence of damage to equipment, or else minimise it; 6. reduce the possibility of installing faulty fabrications by ensuring that the right materials and procedures are used and that workmanship is of the proper quality.
1. 2. 3. 4. 5.
Obvious signs of possible damage can include rust staining, bulging, cracked or distorted insulation, and hot-spots that are indicative of possible corrosion damage. The value of visual observation has been emphasised by Hobinm with a caveat concerning direct eye observation regarding access, fatigue and variability. Internal inspection can only be carried out when plant is shut down, but can determine the condition of many components. The equipment utilised can range from simple types, such as callipers, pit gauges, scrapers, mirrors and magnifying glasses and endoscope systems, to complex equipment utilising miniaturised TV cameras and fibre optics. Description of devices used for inspection in the nuclear industry are described by Kovan ‘I. Mechanical gauges (using spring-loaded feelers) for recording corrosion inside oil or gas well tubing have been marketed for many years. Although there are some disadvantages (restriction to off-line and damage to corrosion inhibitor films), much useful data can be obtained from these instruments. Ultrasonics Use of ultrasonics involves the transmission of very high frequency sound waves through the metal whose thickness needs to be known 1. 2, 4.40*42,43 . An advantage of this technique is that access is only required to one side of the vessel or pipe wall (Fig. 19.58). A short burst of energy is transmitted via a transducer probe into the metal using the pulse-
A Fig. 19.58
Ultrasonic pulse-echo for estimation of wall thickness
19: 170
CORROSION MONITORING AND INSPECTION
echo system. The time taken for the sound to traverse the thickness of metal and return to the probe is usually displayed on an oscilloscope although digital read-out of metal-wall thickness is generally provided. Modern thickness-gauging instruments are portable, measuring thicknesses from 1 * O to 300.0 mm to an accuracy of *O* 1 mm". A problem associated with plant inspection is the coupling of the probe to the metal under examination. The exterior surface of plant can be covered in rust which causes problems in coupling the probe to the metal and results in spurious and inaccurate results. Often operators move the probe to an area where adequate coupling is obtained, circumventing the real objective of the measurement. It is often impossibleto obtain any coupling on severely corroded steel surfaces, e.g. the interior of flash distillation plant chambers. Temperature can destroy the piezoelectric properties of the probe, although techniques for cooling probes (delay blocks), and development of temperature-resistant piezoelectric materials, are extending the temperature range. Differences of up to 5 % in thickness can occur between hot and cold readings. Surface preparation before measurement is important, as sound travels about twice as fast in steel as in paint. Many years accumulation of paint can indicate a thicker steel wall than actually exists. Pitting and other forms of localised corrosion are not detected easily, requiring complex equipment with separate probe assemblies for resolution of pits. Although there are shortcomings with corrosion measurement using ultrasonics, the contribution of the technique in an overall corrosion monitoring programme is considerable. This was confirmed in the surveys of users of equipment*. The use of ultrasonics, combined with computer data handling and display has assisted not only in the inspection process but in overall record keeping. The availability of the microprocessor has led to the development of multiprobe assemblies designed for specific applications, e.g. trolleys containing many probes used for inspection of oil-tank floors and sides of storage tanks and marine vessels. The incorporation of multiprobes in crawler-vehicles that can traverse vertical walls (oil storage tanks and ships hulls) can result in considerable savings, e.g. erection of scaffolding, compared with conventional hand monitoring. Developments using ultrasonics related to corrosion detection have been 'zinscan' (digitally-based system with a scanning rig), and 'P-scan' (system incorporating a programmed scanning pattern using magnetic tape for storage and projection of images). The latter has been adapted for corrosion inspection of subsea pipelines. An ultrasonic probe has been developed for condenser tube inspection using a mirror to direct ultrasound to the tube wall from a probe that is passed through the tube. Inspection can only be carried out off-line. The availability of probes that can be welded to the outer walls of piping or process vessels providing on-line thickness measurement is an interesting development which would seem to bridge the technologies of corrosion monitoring and inspectiona. Ultrasonics has been utilised in several versions of pipeline inspection vehicle (see section below). The use of ultrasonics for detection of hydrogen damage in steels has been reported4s4.
CORROSION MONITORING AND INSPECTION
19: 171
The use of electromagnetic acoustic transducers (EMAT) obviates the coupling problems already referred to, and has been applied successfully to the inspection of boiler tubes. Ultrasonic time of flight diffraction (TOFD), developed by the Harwell Laboratory4’, is utilised to ‘fingerprint’ flaws (cracks) in process plant. Subsequent examination at, say, six month intervals can indicate any growth or extension of the crack. It is claimed that changes in crack height of 0-5mm or less be estimated. An overall assessment as to the reliability of inspection techniques (NDT), including ultrasonics, is given by Silk4*. Radiography With its background of success in detecting weld defects and cracks, radiography (using X-rays or gamma-rays) can successfully reveal generalised and localised corrosion in plant 2. 4*40*42.43.Advantages are that pipe lagging need not be removed and permanent records are obtained which permit comparison with subsequent exposures. Several boiler tubes, for example, can be examined using one radiation source and spot checks can be made to units on-stream. Disadvantages are the radiation hazards, the time required to complete exposure, and the fact that access is required to both sides of the item to be examined. Developments have inciuded television fluoroscopy (allowing imaging in real time), X-ray tomography (using a computer programme to calculate and display an X-ray image absorbed in different directions) and flash radiography (short pulses of X-rays). Flash radiography has proved useful in detecting corrosion under pipe lagging49. ‘9
Eddy currents The examination of non-ferrous tubing using external coils is a well-tried and successful inspection technique, owing mainly to the pioneering work of Forster in Germany. The adoption of this method for in-situ inspection of condenser tubes, by mounting eddy-current coils in probes (or bobbins) that can be inserted in condenser tubes, was a logical development of the technique. Suitable apparatus was developed in the immediate post-war period more or less independently by several oil and chemical companies. The principle of operation has been described in the literature 4, 42. In-situ inspection is concerned with corrosion in its many forms, such as pitting or more general attack and thinning. The double-coil probe, however, does not provide any output if general uniform thinning should occur in a length of tube exceeding the probe length. This can be overcome by winding a different coil or coil factor and the centre pole-piece may be moved to obtain a magnetic impedance between coils of equal numbers of turns. Probes are available for a wide range of alloys and tube sizes, ranging from 6.3 mm diameter to around 50mm diameter, which are suitable for tubes from 11 1 mm 0.d. x 1 - 8mm wall thickness to 57 mm 0.d. x 1-6mm waI1. The probe is propelled through the tube by either winching or compressed air. Large and small holes, pits, larger areas of localised attack and cracks are easily detected. In addition, areas of selective attack such as intergranular corrosion or dezincification can be identified. An example of a tape recording artificial defects in a standard 25.4 mm 0.d. x 2.6 mm wall thickness (Type 316L) stainless steel tube is shown in Fig. 19.59. ‘3
-
19: 172
CORROSION MONlTORlNG AND INSPECTION
Fig. 19.59 Artificial defects in stainless steel tube
KEY ( I ) 50% wall reduction. (2) 10070 wall reduction. (3) Circumferential slot 0-254mm wide x 12.7 mm 1 x 50qo gauge depth. (4) Longitudinal slot 0.254 m m wide x 12.7 m m 1 x 50% gauge depth. ( 5 ) I a 5 9 m m diameter through hole. (6) 1.59 m m diameter hole to 50% gauge depth. (7) 0.78 m m diameter through hole. (8) 0.78 m m diameter hole to 50% gauge depth
Most eddy current equipment is custom designed for a particular application. Signals can be digitised allowing computer signal processing. Many materials can be tested, and they include copper, cupro-nickel alloys, brasses, stainless steels, zirconium, zircaloy, tungsten, molybdenum, lead, beryllium and titanium. It is usual to check the calibration of the instrument with a tube having calibrated defects. The baffles supporting the condenser tubes may mask any corrosion occurring close to the baffles.
Infrared Methods Commercial instrumentation for recording infrared radiation has been available for some years and has been explored by the electrical power industry in the UK for assessing corrosion in boiler tubes at power-station shut-down. An external heat source is played onto the outside of boiler tubes at the same time as cold water is circulated inside the tubes. Hot spots due to poor heat conductivity caused by excessive corrosion product indicated areas of high corrosion. Acoustic Emission Acoustic emission (AE) describes the technique in which sound generated by a metal under stress can be d e t e ~ t e d ~ , ~The -~*. method can provide indication of cracking or other defects in process equipment. Advantages of the technique are applicability to on-line monitoring (claims that cracking can be detected and arrested on-stream have been made) and that tests can be made relatively quickly. The equipment required has been described by Halmshaw" who indicates the controversial claims for this technique. There is no doubt that successful applications of AE are now being reported for corrosion detection, in particular an application regarding stress corrosion cracking4'. AE is provided as a service by specialist firms.
CORROSION MONITORING AND INSPECTION
19: 173
Magnetic saturation This method has been adopted for several versions of intelligent vehicle (instrumented devices that are propelled through pipelines to assess both internal and external corrosion, see section below). Magnetic flux exclusion (MFE) has been used for detecting pitting in oil storage tank floors by the Harwell Laboratory”. The instrument specification required that reiiable indication be given of any underfloor corrosion pit in carbon steel plate a diameter greater than that produced by a 120” cone penetrating 3 mm into 6 mm plate. Intelligent Vehicles (for pipelines) The oil and gas industry has for a long time used devices which are passed through pipelines (propelled by the product being transported) for (a)cleaning the interior of pipelines and removal of paraffin waxes and (b) pipeline inspection using photographic, TV and mechanical gauging equipment 5 ’ ~ s 2 . The consequences of high pressure pipeline failure can be catastrophic (occurrences have been recorded in both the USA and eastern Europe), and much effort has been devoted to the development of intelligent vehicles or PIG (PIG = pipeline inspection gauge) using various NDT techniques for inspection. A recording of data is made during a pipeline run, data within specification is discarded, and the tapes analysed at a base c ~ m p u t e r ~ ~ ’ ~ ~ . Reliability and confidence in those vehicles is continuously improving and they will be utilised to a greater extent in the future because of more stringent legislation being applied to pipeline operators in several countries. Factors to be considered are the products in the pipeline (gadliquid), facilities for loading and retrieval of the vehicles (called pig-traps), condition of line prior to inspection (dents or damage that would prevent passage of the vehicle), economics and size of pipeline. The British Gas version uses magnetic saturation, and can identify corrosion on both internal and external pipe surfaces. The specification for seamless pipe is: General corrosion Pitting corrosion
Detection sensitivity 0-2t 0-4t
Sizing accuracy 10.It *O.lt
where t = nominal pipe wall thickness (seam welded pipe), and pitting corrosion is defined as corrosion affecting a surface area of pipe contained within a square of dimensions 3t x 3t. Intelligent vehicles have been developed with arrays of ultrasonic probes for pipeline inspection -one commercial version contains 5 12 ultrasonic sensors. Under development are intelligent vehicles for crack detection. An elasticwave version (developed by British Gas and the Harwell Laboratory) is currently being evaluated in a test-loop. This vehicle has successfully detected stress-corrosion cracks in the test-loop. The Gas Research Institute (USA) is sponsoring development work with intelligent vehicles at the Battelle Columbus Division (Ohio). Facilities for testing vehicles were commissioned in 199lS2.
19:174
CORROSION MONITORING AND INSPECTION
Industry Requirements, Philosophy and Case Histories Oil and Gas Production This sector is a major user of corrosion monitoring equipment, in particular for offshore fields where ramifications of corrosion and consequent maintenance are far more serious and costly compared with onshore production. Carbon steel is used for approximately 70-80% of production facilities. The development of a field is assessed on a defined corrosion risk which may not be correct, leading to serious corrosion. In addition, a reservoir may become more corrosive as the field is extracted owing to (a) increased water content, and (6) eventual ‘souring’ of the field (hydrogen sulphide production). As well as corrosion in oil and gas streams, there are other applications such as various water circuits and injection of treated sea-water into the reservoir. The structures used (platforms) require monitoring in addition to sub-sea pipelines, satellite wells and other equipment (e.g. manifolds) on the sea floor. Corrosion inhibitors are widely used in internal-streams (from the reservoir and many of the downstream operations). Corrosion monitoring can provide valuable data for assessing the effectiveness of the inhibitors used and for optimising dosage rates. User experience of monitoring techniques in oil and gas production has been reviewed55and indicates success and failure for the same methods by different operators. A survey of current monitoring practice in UK offshore fields has been published 56, and other experience related to oil/gas producdocument has been prepared by CEA tion has be r e p ~ r t e d ~ ”A ~ *draft . Task Group E2-5 providing guidelines for monitoring sea-water injection systems. Oil Refining Substantial corrosion monitoring is used in oil distillation and other refinery units. Refineries are major users of corrosion inhibitors depending on the processes and materials adopted. The utilities necessary for the refinery operation, e.g. steam and cooling water, also require monitoring. MillerS9has provided a review of one particular company philosophy regarding corrosion monitoring in oil refining. Loushinm has outlined the benefits from computer application in the overall monitoring and inspection programme. Chemical and Petrochemical Production One of the earliest applications for corrosion monitoring is in chemical and petrochemical production. In contrast to the oil industry, more exotic metallurgy is used requiring different monitoring data, e.g. monitoring of potential for stainless steel to indicate passivity or activity. Localised corrosion is a common mode of failure and still presents a challenge to monitoring technology. Chemical plants also have large requirements for utilities, e.g. steam and cooling water. Case histories in chemical plant monitoring have been p ~ b l i s h e d ~ , ~ ’ - ~ . Boilers Plants always experience corrosion problems with boilers, and monitoring of corrosion has always presented a challenge relating to both access and simulation of exact conditions, e.g. condensation conditions. Applications in the monitoring of condenser tubes have been reported 65-66.
CORROSION MONITORING AND INSPECTION
19: 175
Industrial Cooling Water Systems Waters used for recirculating cooling systems can either be scaling or corrosive. Corrosive waters are treated with corrosion inhibitors which require monitoring for overall assessment of the treatment programme. Pipelines Pipelines carrying wet gas and crude oil present a corrosion hazard and are protected accordingly by coatings and/or inhibitors. Limitations of corrosion monitoring arise from sampling, in relation to the sampling and interval, and access problems for subsea pipelines (major trunk lines). The use of inspection by intelligent vehicles which are sent through pipelines is increasing but is still limited to inspection only with long intervals between each inspection (1-3 years). The performance of these intelligent vehicles is dependent on the type of vehicle utilised. Civil Engineering The corrosion of concrete reinforcement (rebar) is a serious problem which is attracting much research effort. Solutions being examined include epoxy coatings, exotic metallurgy and/or adoption of cathodic protection. Monitoring and inspection is a major challenge in this area, as is the monitoring of the effectiveness of the chosen corrosion prevention method. Most major civil engineering structures may be affected. A review of techniques for monitoring corrosion of rebar in concrete has been published6’. A state of the art survey of possible, methods for monitoring rebar in concrete concluded that electropotential measurements were the only practicable NDT aid to diagnosis of corrosion in the field@. Miscellaneous There are many interesting applications that arise from time to time that are outside the main stream of industry described above. Examples include: desalination plant; reactor cooling water circuits; automobile body corrosion (in situ); marine (vessels, piling, harbour installations); aircraft (in situ); packaging; and cavitation monitoring.
Future Developments A thrust can be expected in development of the electrochemical methods (reliability, operation and interpretation), TLA and custom-built NDT systems for specific requirements. The disadvantage of using separate sensors rather than the actual plant has been stated. Methods that can use plant for data are required, and developments in magnetic finger-printing may contribute here. The use of expert systems combined or in conjunction with corrosion monitoring and inspection techniques will provide an attractive synergistic approach to the control of corrosion in process plant in the futurea, providing the corrosion monitoring techniques used are reliable and sufficiently rugged for industrial use. The inability to monitor corrosion in inaccessible areas (oil/gas-well tubing) and for specific applications (corrosion under lagging) are examples of deficiencies in the overall technology. C. F. BRITTON
19: 176
CORROSION MONITORING AND INSPECTION
REFERENCES I. Britton, C. F., ‘Corrosion Monitoring and Chemical Plant’, in Corrosion 2nd. Edn., edited by L. Shreir. Butterworths-Newnes, London (1976) 2. Britton, C. F. and Tofield, B. C., ‘ER‘ective Corrosion Monitoring’, Muterials Performance, 27 No. 4, 41-44, April (1988) 3. Edeleanu, C., ‘Corrosion Monitoring for Chemical Plant’, Corrosion Technology, 2 , 7, July (1955) 4. Industrial Corrosion Monitoring, Department of Industry, Committee on Corrosion, HMSO, London (1978) 5. Britton, C.F., ‘Corrosion Monitoring’, Chemistry in Britain, 17 No. 3, 108-11 1 , March (1981) 6. Farina, C., Faita, G., Casolo, U. and Olivani, F., ’Industrial Corrosion Monitoring’, Werkstoffe und Korrosion, 30. 858-861 (1979) 7. Anon, ‘On-Line Monitors Attack Pr6cess Corrosion’, Processing, 34, 34-35, February (1986) 8. Moreland, P.J., and Hines, J.G., ‘The Concept and Development of Corrosion Monitoring’, Conf. Cormion ‘78,NACE. Paper 164 (1978) 9. Richardson, J. A., ‘Innovations in Techniques for Corrosion Monitoring’, Cony. Advances in Materials Technologyfor Process Industries Needs,Atlanta (1984) 10. Houghton, C. J., Nice, P. 1. and Rugtveit, A. G., ‘Automated Corrosion Monitoring for Downhole Corrosion Control’, Materials Performance, 24 No. 4. 9-17, April (1985) 1 I . Dillon, C. P., Krisher, A. S. and Wissenberg, H.‘Plant Corrosion Tests’, Handbook on Corrosion Testing and Evaluation, Wiley (1971) 12. Abramchuk, J., ‘Basic Inspection Methods’, Materiak Protection, 1 No. 3, 60, March ( 1962) 13. Mansfield, F.,‘Don’t Be Afraid of Electrochemical Techniques-But Use Them With Care’, Corrosion, 27, 12, 856-868, December (1988) 14. Baboian, R.. (ed.), ‘Electrochemical Techniques for Corrosion’, Corrosion, Sponsored by Unit Committee T-3L. NACE (1976) 15. Stern, M., ‘A Method for Determinining Corrosion Rates from Linear Polarisation Data’, Corrosion, 14 No. 9,44Ot-444t, September (1958) 16. Callow, L. M.. Richardson, J. A. and Dawson. J. L., ‘Corrosion Monitoring Using Polarisation Resistance Measurements, (i) Techniques and Correlation, (ii) Sources of Error’, Br. Corros. J., 11 No. 3. 123-139 (1976) 17. Moreland, P. J. and Rowlands, J. C., ‘Techniques and Instrumentation for Polarisation Resistance Measurements’, Br. CorrosJ., 12 No. 2. 72-79 (1977) 18. Cao, Ch., Song,Ch. and Wang, Y.,‘Differential Polarisation Resistance’, Corrosion, 45 No. 2, 99-102,February (1989) 19. Danielson, M. J., ‘Analysis of Errors in Using The Two Electrode and Three Electrode Polarisation Resistance Methods In Measuring Corrosion Rates’, Corrosion, 36, No. 4, 174-178 , April (1980) 20. Mansfield. F., ‘Some Errors in Linear Polarisation Measurements and Their Correction’, Corrosion, 30 No. 3, 92-96,March (1974) 21. Jasinski, R. J. and Efird. K. D., ‘Electrochemical Corrosion Measurements in Crude Oil’, Materials Performance, 43 No. 8, 476-478,August (1987) 22. Dawson, J. L., Hladky. K.and Eden, D. A.. ‘Electrochemical Noise-Some New Developments in Corrosion Monitoring’, Conf. UK Corrosion (1984) 23. Hladky, K. and John, D.G., ‘Corrosion Monitoring Using Electrochemical Noise’, 2nd. Int. Conf. on Corrosion Monitoring and Inspection in the Oil, Petrochem. and Process Industries, London, Oyez Scientific and Technical Services Ltd.. London (1984)
24. Neufeld, P. and Queenan, E. D., ‘Frequency Dependence of Polarisation Resistance Measured with Square Wave Alternating Potential’. Br. Corros. J.. 5.72-75, March (1970) 25. Fontana, M. G., Corrosion Engineering, 3rd edn., McGraw-Hill, pp 194-8 (1986) 26. Dawson. J. L., Callow, L. M., Hlady. K. and Richardson, J. A.. ‘Corrosion Rate Determination By Electrochemical Impedance Measurement’, Conf. On-Line Surwillance and Monitoring of Process Plant. London, Society of Chemical Industry (1977) 27. Fincher, D. R., Nestle, A. C. and Marr, J. J., ‘Coupon Corrosion Rates Versus Hydrogen Probe Activity’, Materials Performance, 15. 1. 34-40 (1976) 28. Vennett, R. M..‘Corrosion Monitoring in Oil Field Operations Using a Vacuum Hydrogen Probe’, Muterials Performance, 16. 8 , 31-41 (1977)
CORROSION MONITORING AND 1NSPECTlON
19: 177
29. ‘Hydrogen Monitor Assemblies’, Rohrback-Cosasco Systems, Leaflet-Hydrogen Patch
Assembly, p 1.601A 30. Thomason, W. H.. ‘Corrosion Monitoring with Hydrogen Probes In The Oilfield’. Moterials Performonce, 23. 5. 24-29, May (1984)
31. Martin, R. L. and French, E. C., ‘Corrosion Monitoring in Sour Systems Using Electrochemical Hydrogen Patch Probes’ Conf. Soc. of Pet. Engrs. of AIME, Sour Gas Symposium, Texas, (1977) 32. Lyon, S. B. and Fray, D. J., ‘Detection of Hydrogen Generated by Corrosion Reactions using a Solid Electrolyte Probe’. Moteriols Perfortnonce, 24.4, 23-25, April (1984) 33. Asher, J., Conlon, T. W., Tofield, B. C. and Wilkins. N. J. M., ‘Thin Layer Activation-A New Plant Corrosion Monitoring Technique’, Proc. Conf . ‘On-Line Surveillance and Monitoring’, entitled On-Line Monitoring of ContinuousPlants, Ellis Horwood, London ( 1983) 34. Asher, J., Harwell Laboratory (private communication) 35. Handbook of Industrial Water Conditioning. Betz Laboratories lnc., Trevose, Pa., USA, p 181 (1980) 36. Tatnall, R. E., ‘Fundamentals of Bacteria Induced Corrosion’, Moteriols Performance, 20, 9, 32-38, September (1981) 37. Ruseska, I.. ‘Biocide Testing Against Corrosion-Causing Oil Field Bacteria Helps Control Plugging’, Oil and Gas J . , 253-264, 8 March (1982) 38. Corrosion Control Engineering Joint Venture, Task Group EM, Review of Current Practices for Monitoring Bacteria Growth in Oilfield Systems, ICom Document 001/87 (1987). 39. Severin, J., Conoco (private communication). 40. Hobin, T. P.. ‘Survey of Corrosion Monitoring and the Requirements’, Brit. J. of NDT, 284-290, November (1978) 41. Kovan, R., ‘Reaching Those Remote Corners of Reactors’, Atom, 388, 8-11, February (1989) 42. Halmshaw, R., Non-destructive Testing, Arnold, London (1987) 43. Callister, W. G., ‘Corrosion Monitoring by Non-Destructive Testing as an Inspection Tool and Means of Loss Prevention’, Bull. Inst. Corrosion Science and Tech. Issue 52, pp 2-12, August (1975) 44. Fothergill, J. R., Willis, P. and Waywells, S. ‘Development of High Temperature Ultrasonic Transducers for under Sodium Viewing Applications’, Brit. J . of NDT, 31, 5, 259-264, May (1989) 45. Birring, A. S., ‘Ultrasonic Detection of Hydrogen Attack in Steels’, Corrosion, 45 No. 3, 259-263 (1989) 46. Maggard. M. G., ‘Detecting Internal Hydrogen Attack’. Oilond Cos J., 90-94,IO March ( 1980) 47. Anon., Monitoring Flow Growth, NDT Centre, Harwell Laboratory (1989) 48. Silk, M. G.. ‘Can Non-Destructive Inspection Be Reliable?’ Atom. 3’17.47, March (1988) 49. Allinson, M., ‘Flash Radiography’, (E2/2 Unit Committee), UK Corrosion ‘87, (1987) 50. Saunderson, D. H., ‘The MFE Floor Scanner-A Case History’, Colloq. ‘The Economics of Non-Destructive Evaluation’, Inst. of Elect. Engrs., January (1988)
51. Kiefner, J. F., Hyatt, R. W. and Eiber, R. J., ‘Tools Locate, Measure Dents and Metal Loss’, Oil and Gas J., 30-38, 7 April (1989) 52. Kiefner, J. F.. Hyatt, R. W. and Eiber. R. J.. ‘Metal Loss, Crack Detection Tools Targeted’, ibid., 69-71, 24 April (1989) 53. Harle, J. C., ‘Magnetic-Flux Pig Run Successful on TAPS. Oil and Cos J . , 72-75, 10
October (1988) 54. Shannon, R. W. E. and Argent, C. J., ‘Maintenance Strategy Set by Cost Effectiveness’, Oil and Cos J . , 41-44, 6 February (1989). 55. Britton. C. F.. T h e Selection, Evaluation and Testing of Oil Field Corrosion Inhibitors’, 56. 57.
58. 59.
Seminar ‘The Development and Use of Corrosion Inhibitors’, Oyez Scientific and Technical Services Ltd., London (1983) Task Group E2-I. ‘Survey of Corrosion Monitoring Practices in the North sea’. I Corr ST/NACE (1983) Comeau, B. D., and Marsden, C. J., ‘Unexpected Field Corrosion Leads To New Monitoring with Revised Predictive Model’, Oil ond Cos J., 45-48. 1 June (1987) Walker, C. K.and Maddox, G. C., ‘Corrosion Monitoring Techniques and Applications’. Moteriols Performance, 28, 5 , 64-70 (1989) Miller, D. R., Begernan, S. R. and Lendvai-Lintner, E., ‘Current and Future Applications
19: 178
CORROSION MONITORING AND INSPECTION
of On-Line Surveillance and Monitoring Systems in the Petroleum Industry’, Chem. and Ind.. 782-785, 17 October (1983) 60. Loushin, L. L., ‘First Application of Artificial Intelligence for Corrosion Control in the Petroleum Industry’, Materials Perfomance, 27, 6, 77-83. June (1988) 61. Bergstrom, D. R., ‘Case Histories-Electrical Resistance Probes Control Corrosion in Chemical Industry’. Materials Performance, 20. 9. 17-20. September ( I 98 1) 62. Harrell, J. B., ‘Corrosion Monitoring in the CPI’, Chemical Eng. Prog., 57-61, March (1978) 63. Arnold, C. C., ‘Using Real Time Corrosion Monitors in Chemical Plants’, Chemical Eng. Prog., 74, 3, 43-46, March (1978) 64. Liening, E. L.. ‘Industrial Applications of Corrosion Probes (Case Histories)’, Muterials Performance, 16, 39-41, 9 September (1977) 65. Ronchetti. C., Buzzanca. G. and Diacci. E., ‘A New System for a Continuous Monitoring of Steam Condenser Corrosion’, ClSE Report, CISE-NT 81.094 (1981) 66. Parker, J. G. and Roscow J. A., ‘Method for the Assessment of the Quality of Surface Films Formed on the Cooling Water Side of Copper-Based Alloy Condenser Tubes’, Br. Corros. J., 16, 2, 107-110 (1981) 67. McKenzie. S. G.. ‘Techniques for Monitoring Corrosion of Steel In Concrete’, Seminar ‘Corrosion In Concrete-Monitoring, Surveying and Control by Cathodic. Protection’, Global Corrosion Consultants, Telford (1986) 68. Arup and Partners, Electroporential Mapping of Corrosion in Concrete, HMSO (OTH 88286)
BIBLIOGRAPHY
Industrial Corrosion Monitoring. Department of Industry. Committee on Corrosion, HMSO, London (1978) Controlling Corrosion (6) Monitoring. Department of Industry, Committee on Corrosion, HMSO, London (1977) ‘Corrosion Monitoring in Industrial Plants Using Nondestructive Testing and Electrochemical Methods’, Proc. Symposium Sponsored by ASTM Committee E-7 and G-1, Montreal, Canada, May 1984, ASTM Special Technical Publication 908 ‘Internal Corrosion Control and Monitoring in the Oil, Gas and Chemical Industries’ Proc. Seminar, London, March 1987, Global Corrosion Consultants, Telford, UK ‘On-Line Surveillance and Monitoring of Process Plant’, Proc. Symposium, London, September 1977, Society of Chemical Industry, London Inspection of Chemical Plant, (Pilborough), Leonard Hill Books, London (1971) ‘Corrosion Monitoring and Inspection in the Oil, Petrochemical and Process Industries’, Proc. Conference, February 1984, Oyez Scientific and Technical Services, London ‘Corrosion Monitoring in the Oil, Petrochemical and Process Industries’, J. Wanklyn, (ed.), Proc. Conference, Oyez Scientific and Technical Services, London (1982) In addition to the above the annual corrosion conferences (NACE and UK Corrosion), usually sponsor sessions on corrosion monitoring which are reported in the respective proceedings. The British Institute of Nondestructive Testing (Northampton) and the American Society of Nondestructive Testing (Ohio) publish transactions and sponsor conferences/symposia on a regular basis. Standards and Recommended Practices relating to corrosion monitoring are published by both ASTM and NACE. The Institute of Corrosion (UK) publishes European practices and experience via the NACE Technical Committe system.
19.4 Inspection of Paints and Painting Operations
Improvements in process and quality control made significant contributions to the transition from iron to steel as the major ferrous construction material over a century and a half ago. For most of that time red lead was relied upon, and not without a remarkable degree of success, as the rustinhibitive pigment in anti-corrosive paints. In the last twenty years, however, there has been a similar dramatic change from such simple paints as red lead to synthetic polymer coatings which have as complex a technology as steel manufacture itself. Improved processes and quality control have helped to establish these new coating materials but the care necessary for successful use has to be appreciated. Sections 11.1 and 11.2 have shown how necessary it is to remove millscale before coating and how scale-free surfaces may still retain seeds of further corrosion even when apparently cleaned well. The percentage of premature failures with sophisticated systems is still high, even on apparently well-prepared surfaces and there is a strong case for effective inspection at each stage of coating operations.
Painting of Structural Steel It is not always easy to apply the concepts of quality control which have become routine on coating production lines to single structures. For instance, in ordinary steelwork fabrication shops the service conditions for which a significant proportion of the throughput is required, simply do not call for high performance systems to be applied. This generally means at least two levels of quality in one works. The problem is further complicated by the multiplicity of techniques involved and the near-uniqueness of every steel design. Nevertheless, many conditions of environment, use, maintenance and safety, exist where it is essential to produce long-life protective coating for structural steel. Only by continuous inspection of surface preparation and coating application, however, can high performance from modern systems be achieved with certainty.
19: 179
19: 180
INSPECTION OF PAINTS AND PAINTING OPERATIONS
Case Histories and Cost To illustrate the above, consider two case histories ' where no specialist inspection was provided. Figure 19.60 illustrates a steel surface which should have been blast cleaned to a high standard before being coated at works with a zinc-rich epoxy primer and on site with two-pack intermediate and finishing coats. After exposure for 18 months in a marine environment, flaking millscale from beneath the paint was observed, and a survey showed that the paint
Fig. 19.60 Rusting surface after marine exposure for 18 months; no inspection during the coating process
Fig. 19.61 Same section shown in Fig. 19.60 photographed 4 years after remedial work; full inspection during recoating on site
INSPECTION OF PAINTS AND PAINTING OPERATIONS
19:181
film thickness varied between 50 and 140pm. Some parts had not been primed. Although the remedial re-blast cleaning of all surfaces had to be carried out in-situ,and a suitable system applied, the resulting protection (Fig. 19.61) after exposure for 4 years was still of the high order expected. The only difference between the two contracts was that full-time specialist inspection was given at every stage of the site remedial work. Figures 19.62 and 19.63 were taken at the time of investigation into a failure of lOOt of steel which should have been coated with a nearly maintenance-free system. Zinc metal spray and four coats of paint were specified but, for reasons of economy, special measures for inspection were not taken. Within months of erection areas were flaking off where millscale had not been removed (Fig. 19.62) and other areas were blistering. Figure
Fig. 19.62 Result of millscale not being removed from steelwork before zinc metal spraying
Fig. 19.63 Random check of the surface of the steelwork shown in Fig. 19.62 showing the surface completely oxidised under the coating
19 :182
INSPECTION OF PAINTS AND PAINTING OPERATIONS
19.63 shows one of the many areas investigated where there were no visible signs of failure until the whole system was removed; the dark surface in the cut-out area represents the rusted metal found underneath the metal spray. Subsequent investigation revealed that the steel members had been too long to be housed completely in the workshops concerned. The blastcleaned surface had been allowed to stand outside overnight and had been metal sprayed the following morning. No one had visited the works before the order had been placed to see if the contractor would be able to cope with this very long steelwork. The benefit of the protective system specified had been lost for good. The compensation finally paid could never recompense the owners fully as in situ repair was made very difficult by restricted access.
Prior Inspection of Works and Site Facilities Before a works painting or metal spraying contract for steelwork is awarded, the workshops concerned should be inspected by qualified personnel. It is essential that some check is made to ensure that the necessary facilities and equipment for carrying out all the terms of the order or specification are available. Cleanliness of the painting areas and order in the paint store usually indicate the seriousness with which works personnel take protectivecoating operations. Prior inspection should also be made on site to ensure scaffolding and platforms have been erected to give the operators easy access to the work, Le. at the correct level to, and distance from, the surface to be treated; temporary covers and equipment, including compressors, should be included in the survey. The safety of operators, the rigging of life-lines for scaffolding, and the proper removal of dust and solvent vapours should be discussed with the management.
Type of Inspection The British Standard Code of Practice for protecting iron and steelwork from corrosion, BS 5493:19772 contains a section on inspection which may be helpful. This type of inspection is entirely unnecessary for very simple painting schemes. In such situations an occasional visit by an inspector may be sufficient to ensure acceptable results once initial standards of surface preparation and paint application have been established. For the many systems which may be specified today, some will require the maximum level of surface preparation and demand continuous inspection at every stage. Others will require less onerous policing. It is up to the client to determine the appropriate level of inspection for the job and to make financial provision for this.
INSPECTION OF PAINTS AND PAINTING OPERATIONS
19 :183
Quality of Inspection This will depend partly on the experience and personal integrity of the inspector and partly on the organisation to which he belongs. Whatever the type of inspection ordered, the inspector himself should always be experienced in the processes concerned. The organisation behind the inspector must be sufficiently competent in coating technology to give him the backing he requires, to answer his queries by telephone if necessary to prevent delays (although anticipation is the hallmark of the competent organisation) and to provide consultants to investigate and report where necessary. Some of the larger owner organisations, e.g. government departments, public utilities and major contractors do have such staff but, where a long-life protective-coating contract is awarded, it is becoming more and more frequent for them to employ independent inspection firms who specialise in this field and can even provide teams of inspectors where necessary.
Independent Inspection Organisations On its appointment to provide inspection for a coating contract, the organisation concerned might be expected to include in their service the following: 1. Constructive criticism of the coating specification,bringing to the attention of the client aspects that may be difficult to implement fully. 2. Initiation of preliminary discussions on the contractor’s proposed methods of working. 3. Close liaison with the coating manufacturer on materials and their use. 4. An introductory schedule showing precisely how and at what stages inspection will be carried out and reported. Inspec