Advances in Polymer Processing: From Macro- to Nano-Scales

i Advances in polymer processing ii Related titles: Mechanical evaluation strategies for plastics (ISBN 978-1-85573-...

Author: Sabu Thomas | Weimin Yang

77 downloads 2273 Views 90MB 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

DOWNLOAD PDF

i

Advances in polymer processing

ii

Related titles: Mechanical evaluation strategies for plastics (ISBN 978-1-85573-379-4) Thermoplastics, being non-linear viscoelastic, impose constraints on testing which are absent in elastic and plastic materials. End products manufactured from them are often anisotropic, complicating the relationships between laboratory test data and service performance. This new book explains recently developed testing strategies for providing service-pertinent data within a limited budget. It relates the structure of the tests and the functions that they serve to the intrinsic nature of the mechanical properties of thermoplastic materials. Polymer nanocomposites (ISBN 978-1-85573-969-7) Polymer nanocomposites are a class of reinforced polymers with low quantities of nanometric-sized clay particles which give them improved barrier properties, fire resistance and strength. Such properties have made them valuable in components such as panels and as barrier and coating materials in automobile, civil and electrical engineering as well as packaging. Polymer nanocomposites provides a comprehensive review of the main types of polymer nanocomposite and their properties. Biodegradable polymers for industrial applications (ISBN 978-1-85573-934-5) The vast majority of plastic products are made from petroleum-based synthetic polymers that do not degrade in a landfill or in a compost-like environment. Therefore, the disposal of these products poses a serious environmental problem. An environmentally conscious alternative is to design/synthesise polymers that are biodegradable. In this authoritative book, fundamental concepts concerning the development of biodegradable polymers, degradable polymers from sustainable sources, degradation and properties and industrial applications and their importance are reviewed. Details of these and other Woodhead Publishing materials books can be obtained by: • visiting our web site at www.woodheadpublishing.com • contacting Customer Services (e-mail: [email protected]; fax: +44 (0) 1223 893694; tel.: +44 (0) 1223 891358 ext. 130; address: Woodhead Publishing Limited, Abington Hall, Granta Park, Great Abington, Cambridge CB21 6AH, UK) If you would like to receive information on forthcoming titles, please send your address details to: Francis Dodds (address, tel. and fax as above; e-mail: [email protected]). Please confirm which subject areas you are interested in.

iii

Advances in polymer processing From macro to nano scales Edited by Sabu Thomas and Yang Weimin

CRC Press Boca Raton Boston New York Washington, DC

WOODHEAD

PUBLISHING LIMITED

Oxford

Cambridge

New Delhi

iv Published by Woodhead Publishing Limited, Abington Hall, Granta Park, Great Abington, Cambridge CB21 6AH, UK www.woodheadpublishing.com Woodhead Publishing India Private Limited, G-2, Vardaan House, 7/28 Ansari Road, Daryaganj, New Delhi – 110002, India Published in North America by CRC Press LLC, 6000 Broken Sound Parkway, NW, Suite 300, Boca Raton, FL 33487, USA First published 2009, Woodhead Publishing Limited and CRC Press LLC © 2009, Woodhead Publishing Limited The authors have asserted their moral rights. This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. Reasonable efforts have been made to publish reliable data and information, but the author and the publishers cannot assume responsibility for the validity of all materials. Neither the author nor the publishers, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording, or by any information storage or retrieval system, without permission in writing from Woodhead Publishing Limited. The consent of Woodhead Publishing Limited does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. Library of Congress Cataloging in Publication Data A catalog record for this book is available from the Library of Congress. Woodhead Publishing ISBN 978-1-84569-396-1 (book) Woodhead Publishing ISBN 978-1-84569-642-9 (e-book) CRC Press ISBN 978-1-4398-0149-9 CRC Press order number: N10035 The publishers’ policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured from pulp which is processed using acid-free and elemental chlorine-free practices. Furthermore, the publishers ensure that the text paper and cover board used have met acceptable environmental accreditation standards. Typeset by Replika Press Pvt Ltd, India Printed by TJ International Limited, Padstow, Cornwall, UK

v

Contents

Contributor contact details Preface

xiii xix

Part I Introduction 1

Recent advances in polymer processing: the state of the art, new challenges and opportunities

3

A T SUNNY and S THOMAS, Mahatma Gandhi University, India and W YANG, Beijing University of Chemical Technology, China

1.1 1.2 1.3 1.4 1.5

Introduction Polymer processing: an overview Recent advances in multicomponent polymeric materials Concluding remarks and new challenges References

3 5 7 10 11

2

Non-Newtonian fluid mechanics and polymer rheology

13

K E GEORGE, Cochin University of Science and Technology, India

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13

Introduction Non-Newtonian behaviour Newtonian shear flow Shear flow of a power law fluid Parameters influencing non-Newtonian behaviour Elongational flow Elastic effects in polymer melt flow Polymer blends Polymer nanocomposites Rheometry Solution viscosity Molecular rheology Polymer processing

13 14 16 18 18 25 25 27 29 30 37 37 38

vi

Contents

2.14 2.15

Applications References

43 44

3

Polymeric materials: elastomers, plastics, fibers, composites, nanocomposites and blends

47

A T SUNNY and S THOMAS, Mahatma Gandhi University, India

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 4

Overview of polymeric materials Elastomers Plastics Fibers Composites (macro, micro and nanocomposites) Blends (micro and nano blends) Conclusions References

47 55 57 63 64 67 69 69

Compounding and mixing of polymers

71

E K SILVIYA, S VARMA, and G UNNIKRISHNAN, National Institute of Technology Calicut, India and S THOMAS, Mahatma Gandhi University, India

4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 5

Introduction Basic principles of compounding Plastic compounding Rubber compounding Criteria for selection of ingredients Ingredient functions Fillers: fibres to nanoparticles Design of compounds Machinery for compounding Compounding for specific properties Compound management Quality control Conclusions Future reading References

71 72 73 82 85 85 90 94 94 99 102 103 104 104 104

Screw extrusion

106

J W SIKORA, Lublin University of Technology, Poland

5.1 5.2 5.3 5.4 5.5

Introduction Screw plasticizing system Construction of the screw Construction of the barrel References

106 109 124 129 140

Contents

6

Reactive extrusion of polymers

vii

143

R M Jeziórska, Industrial Chemistry Research Institute, Poland

6.1 6.2 6.3 6.4 6.5

Introduction Functionalisation of polyolefins Polyblends containing functionalised polyolefins Summary References

143 144 155 169 170

Part II Moulding technology 7

Injection moulding of polymers

175

W YANG and J ZHIWEI, Beijing University of Chemical Technology, China

7.1 7.2 7.3 7.4 7.5 7.6 7.7 8

Introduction Conventional injection moulding Hot runner injection moulding Gas (water)-assisted injection moulding Microinjection moulding Future trends References

175 176 190 192 195 199 200

Rotational moulding of polymers

204

F G TORRES, Catholic University of Peru, Peru

8.1 8.2 8.3 8.4 8.5 8.6 8.7 9

Introduction Polymer sintering models Rotational moulding of composite and nanocomposite materials Rotational molding of biobased materials Mechanical properties of rotomoulded materials Rotational foam moulding References

204 206 215 222 222 222 223

Blow moulding of polymers

226

Y MARCO, ENSIETA – LBMS, France and L CHEVALIER, Université Paris-Est, France

9.1 9.2 9.3 9.4

Introduction Process description and industrial part design requirements Mechanical strength of poly(ethylene terephthalate) (PET) plastic bottles Numerical simulation of the process and microstructural changes during stretch blow moulding

226 230 246 258

viii

Contents

9.5 9.6

Conclusion and future trends References

280 281

Part III Alternative processing technologies 10

Liquid resin and polymer solution processing

289

C DONG, Curtin University of Technology, Australia

10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9 11

Introduction Viscosity of resin Curing Liquid resin and composite processing Resin flow and defect generation Dimensional variations Future trends Sources of further information and advice References

289 289 290 293 297 304 309 309 309

Calendering of polymers

312

E MITSOULIS, National Technical University of Athens, Greece

11.1 11.2 11.3 11.4 11.5 11.6 11.7 11.8 11.9 12

Introduction Lubrication approximation Two-dimensional analysis Three-dimensional analysis Viscoelastic effects Calendered sheet and film defects Conclusions Acknowledgement References

312 318 336 343 345 346 348 348 348

Thermoforming of polymers

352

P J MARTIN, Queen’s University Belfast, UK

12.1 12.2 12.3 12.4 12.5 12.6 12.7 12.8 12.9 12.10 12.11

Introduction Developments in thermoforming processes Products and markets Polymers for thermoforming Advances in tool materials Developments in process simulation Advances in instrumentation and control Measurement of thermoformability Future trends Sources of further information and advice References

352 353 357 359 366 369 372 375 377 380 381

Contents

13

Polymer processing using supercritical fluids

ix

384

C L HIGGINBOTHAM, J G LYONS and J E KENNEDY, Athlone Institute of Technology, Ireland

13.1 13.2 13.3 13.4 13.5 13.6 13.7 13.8 14

Introduction What is a supercritical fluid? Brief history of supercritical fluids Polymer applications of supercritical fluids Polymer processing Recycling Conclusions References

384 384 386 386 390 397 399 400

Radiation processing of polymers

402

V RAO, Mangalore University, India

14.1 14.2 14.3 14.4 14.5 14.6 14.7 14.8

Introduction Types of radiation sources used Radiation chemistry of polymers Radiation-induced chemical reactions Industrial applications Conclusions Sources of further information and advice References

402 403 408 414 423 431 432 432

15

Novel processing additives for extrusion and injection of polymers

438

O L KULIKOV and K HORNUNG, Universität der Bundeswehr München, Germany and M H WAGNER, TU Berlin, Germany

15.1 15.2 15.3 15.4 15.5 15.6 15.7 15.8 15.9 15.10

Introduction Flow instabilities, sharkskin, slip Conventional processing additives Novel processing additives Interaction of polymer processing additives with polymer modifiers Future trends in polymer processing additives Conclusions Sources of further information and advice Acknowledgements References

438 439 448 453 469 470 470 471 471 471

x

Contents

Part IV Micro- and nanotechnologies 16

Processing of macro, micro and nanocomposites

479

Y L LU, Y-P WU and L-Q ZHANG, Beijing University of Chemical Technology, China

16.1 16.2 16.3 16.4

16.5 16.6 17

Background of rubber reinforcement Melt blending method Latex compounding method In-situ polymerization of metallic salts of unsaturated carboxylic acid (MSUCA) during peroxide curing of elastomers Microstructural evolution of rubber-based nanocomposites during storage and the curing process References

479 480 493

Micromolding of polymers

552

501 526 545

D YAO, Georgia Institute of Technology, USA

17.1 17.2 17.3 17.4 17.5 17.6 17.7

Introduction Classification of micromolding processes Microinjection molding Hot embossing Micromold making Emerging topics in micromolding References

552 556 558 564 570 572 574

18

Reactive polymer processing and design of stable micro- and nanostructures

579

A V MACHADO and J A COVAS, University of Minho, Portugal and V BOUNOR-LEGARE and P CASSAGNAU, Université de Lyon, France

18.1 18.2 18.3 18.4 18.5 19

Micro- and nanostructures Reactive extrusion Reactive processes for stable micro- and nanostructured morphologies Equipment References

579 583 587 604 615

Processing of carbon nanotubes and carbon nanotube based nanocomposites

622

V M KARBHARI, University of Alabama in Huntsville, USA and C LOVE, Naval Research Laboratories, USA

19.1 19.2 19.3

Introduction Structure of carbon nanotubes Processing methods for nanotube based polymer nanocomposites

622 624 629

Contents

19.4 19.5 19.6 19.7

Nanotube alignment Properties and characteristics Future trends References

xi

639 643 646 648

Part V Analysis of the moulding process and post-processing technologies 20

Online monitoring of mold flow in polymer processing

655

D KAZMER and S JOHNSTON, University of Massachusetts Lowell, USA

20.1 20.2 20.3 20.4 20.5 20.6 20.7 20.8

Introduction Fundamentals On-line simulation Multivariate quality control Future trends Sources of further information and advice Acknowledgments References

655 656 660 668 677 678 679 679

21

Computer modeling and simulation of polymer processing

681

T C LIM, SIM University (UniSIM), Singapore

21.1 21.2 21.3 21.4 21.5 21.6 21.7

Introduction Interatomic potentials Monte Carlo approach Molecular dynamics approach Continuum and semi-continuum approaches Conclusions References

681 682 688 689 693 695 695

22

Joining, machining, finishing and decorating of polymers

698

P KDV YARLAGADDA, Queensland University of Technology, Australia S BUTDEE and A SUEBSOMRAN, King Mongkut’s University of Technology, Thailand

22.1 22.2 22.3 22.4

Joining process and applications Machining process Finishing process and applications References

698 706 709 713

Index

715

xii

xiii

Contributor contact details

(* = main contact)

Chapter 1

Chapter 2

Ms Anu Tresa Sunny and Professor Dr Sabu Thomas * School of Chemical Sciences Mahatma Gandhi University Kottayam Kerala India 686 560

Dr K. E. George Department of Polymer Science and Rubber Technology Cochin University of Science and Technology (CUSAT) Thikakkara, Cochin Kerala India 682022

E-mail: [email protected] [email protected]

Professor Dr Yang Weimin College of Mechanical and Electrical Engineering Beijing University of Chemical Technology Beisan Huan East Road 15# Chaoyang District Beijing 100029 China E-mail: [email protected]

E-mail: [email protected]

Chapter 3 Ms Anu Tresa Sunny and Professor Dr Sabu Thomas* School of Chemical Sciences Mahatma Gandhi University Kottayam Kerala India 686 560 E-mail: [email protected] [email protected]

xiv


Chapter 4

Chapter 7

E. K. Silviya, S. Varma and Dr G. Unnikrishnan* Polymer Science and Technology Laboratory National Institute of Technology Calicut Kerala India

Professor Dr Yang Weimin * and Jiao Zhiwei College of Mechanical and Electrical Engineering Beijing University of Chemical Technology Beisan Huan East Road 15# Chaoyang District Beijing 100029 China

E-mail: [email protected] [email protected]

E-mail: [email protected]

Professor Dr Sabu Thomas School of Chemical Sciences Mahatma Gandhi University Kottayam Kerala India 686 560

Chapter 5 Dr Janusz W. Sikora Department of Polymer Processing Lublin University of Technology 36 Nadbystrzycka St. 20-618 Lublin Poland E-mail: [email protected] [email protected]

Chapter 6 Dr R. M. Jeziórska Industrial Chemistry Research Institute ul. Rydygiera 8 01-793 Warszawa Poland E-mail: [email protected]

Chapter 8 Professor F. G. Torres Department of Mechanical Engineering Catholic University of Peru Av. Universitaria Cdra. 18 Lima 32 Peru E-mail: [email protected]

Chapter 9 Y. Marco* ENSIETA – LBMS 2 rue François Verny 29806 Brest Cedex 9 France E-mail: [email protected]


L. Chevalier Université Paris-Est Laboratoire MSME 5 boulevard Descartes Champs sur Marne 77454 Marne la Vallée cedex France E-mail: [email protected]

xv

Chapter 12 Dr P. J. Martin School of Mechanical and Aerospace Engineering Queen’s University Belfast Ashby Building Stranmillis Road Belfast BT9 5AH UK E-mail: [email protected]

Chapter 10 Dr C. Dong* Department of Mechanical Engineering Curtin University of Technology GPO Box U1987 Perth, WA 6845 Australia E-mail: [email protected]

Chapter 11 Professor E. Mitsoulis School of Mining Engineering and Metallurgy National Technical University of Athens Zografou 157 80 Athens Greece E-mail: [email protected]

Chapter 13 Dr C. L. Higginbotham,* Dr John G. Lyons and Dr James E. Kennedy Materials Research Institute Department of Polymer Engineering Athlone Institute of Technology Dublin Rd Athlone Co. Westmeath Ireland E-mail: [email protected]

Chapter 14 Professor Vijayalakshmi Rao Department of Materials Science Mangalore University Mangalagangothri – 574199 Karnataka India E-mail: [email protected]

xvi


Chapter 15 Dr O. Kulikov* and Professor K. Hornung Institut für Strömungsmechanik und Aerodynamik LRT-7-1 Universität der Bundeswehr München (University of the Federal Armed Forces of Germany in Munich) Werner – Heisenberg – Weg 39 85577 Neubiberg Germany

Professor Y.-P. Wu and Professor L.-Q. Zhang* Key Laboratory of Beijing City on Preparation and Processing of Novel Polymer Materials Beijing University of Chemical Technology Beijing 100029 China E-mail: [email protected]

Chapter 17 E-mail: [email protected]

Professor M. H. Wagner Fachgebiet Polymertechnik und Polymerphysik TU Berlin Fasanenstr. 90 10623 Berlin Germany

Donggang Yao School of Polymer, Textile and Fiber Engineering Georgia Institute of Technology Atlanta Georgia USA E-mail: [email protected]

Chapter 16

Chapter 18

Professor Y.-L. Lu Key Laboratory for Nanomaterials Ministry of Education Beijing University of Chemical Technology Beijing 100029 China

Professor A. V. Machado* and J. A. Covas IPC-Institute of Polymers and Composites Department of Polymer Engineering University of Minho Campus de Azurém 4800-058 Guimaraes Portugal E-mail: [email protected]


V. Bounor-Legare and P. Cassagnau Université de Lyon Lyon, F-69003 France

Chapter 19 Dr V. M. Karbhari* University of Alabama in Huntsville 366 Shelbie King Hall Huntsville, AL 35899 USA E-mail: [email protected]

C. Love Naval Research Laboratories Alt. Energy Section Code 6113 Washington DC 20375-5342 USA

xvii

Chapter 22 Professor P.K.D.V. Yarlagadda* Director of Smart Systems Research Theme Faculty of Built Environment and Engineering Queensland University of Technology Queensland Australia E-mail: [email protected]

Professor Dr S. Butdee Director of IMSRC Department of Production Engineering Faculty of Engineering King Mongkut’s University of Technology North Bangkok Thailand

Chapter 20 Dr D. Kazmer* and S. Johnston Department of Plastics Engineering University of Massachusetts Lowell 1 University Avenue Lowell, MA 01845 USA E-mail: [email protected]

Chapter 21 Dr T.-C. Lim School of Science and Technology SIM University (UniSIM) 461 Clementi Road S599491 Singapore E-mail: [email protected]

Professor Dr A. Suebsomran Department of Teacher Training in Mechanical Engineering Faculty of Technical Education King Mongkut’s University of Technology North Bangkok Thailand

xviii

xix

Preface

Multi component polymeric materials such as polymer blends and polymer composites from macro to nano scales have emerged as a new class of materials showing promising industrial application potential and have been the focus of an ever-growing interest in the past few years due to their quite exceptional performance. Many new multi component polymeric materials have been developed during the past two decades. The large variety of polymers available on the market today is the result of blending, that is, combining various polymers or adding micrometer-scale or larger fillers, such as minerals, ceramics, and metals (or even air). Processing techniques are critical to the performance of polymer products, which are used in a wide range of industries. Over the past decade, the utility of nanoparticles as additives to enhance polymer performance has been established and now provides additional opportunities for many diverse commercial applications. Of late the interrelationship between processing, morphology and properties has become one of the important research topics in both industry and academia. This book provides a comprehensive review of polymer processing, focusing on recent developments in techniques and materials. Thermosets, thermoplastics, elastomers, foams and nanocomposites have been discussed; multiphase systems are considered from macro to nano scales. Recent developments in the established techniques are reviewed, which include extrusion technologies, injection moulding and blow moulding, in addition to recently developed processing technologies, such as those using supercritical fluids, micromoulding and reactive processing. Finally post-processing technologies are discussed, as well as the analysis of the moulding process. In summary, this book intends to broaden and integrate the knowledge of the behaviour of polymeric materials through the better understanding of the inter-relationship between processing, morphology, structure and properties. This aspect is very important for the large tonnage production of polymeric products with consistent quality. Sabu Thomas Weimin Yang

xx

1 Recent advances in polymer processing: the state of the art, new challenges and opportunities A T S U N N Y and S T H O M A S, Mahatma Gandhi University, India and W Y A N G, Beijing University of Chemical Technology, China

Abstract: This chapter presents a brief introduction to polymer processing, focusing on recent developments in techniques and materials, followed by an overview of recent advances in the processing of multicomponent polymeric materials, future trends, new challenges and opportunities. Multicomponent polymeric materials have emerged as new materials showing promising industrial application potential and have been the focus of an ever-growing interest in the past few years due to their quite exceptional properties. Hence the development of the various processing–structure–properties relations, a complete evaluation of the effects of the different processing techniques on the material properties, the optimization of the quality control of the starting materials, and the development and improvement of the manufacturing technologies used with these materials in order to obtain constant quality products all require a lot of attention in the future. Key words: polymer processing, polymer blends, composites, nanocomposites.

1.1

Introduction

Many new multicomponent polymeric materials have been developed during the past two decades. The large variety of polymers available on the market today is the result of blending, that is combining various polymers or adding micrometer-scale or larger fillers, such as minerals, ceramics and metals (or even air). The number of scientific papers, industrial patents, scientific meetings and conferences devoted to this class of materials is sufficient witness to their strategic importance. Processing techniques are critical to the performance of polymer products, which are used in a wide range of industries. Multiphase polymer systems such as blends and composites provide unique advantages for enhancing material value. As with any multiphase material, performance is critically dependent on the structure or morphology in the final part. Significant progress has been made in modeling the influence of processing on the morphology of multiphase systems. However, substantial challenges remain. 3

4


Polymeric materials have experienced an important development over the past 10 years in laboratories across the world. A good example in this respect is poly(ethylene terephthalate) (PET) whose production has an annual growth rate of 10%, mostly due to its excellent properties as a packaging material for pressurized beverages, food, medicines and other products. This expectation is supported by the fact that in developing countries 30–50% of the food is spoiled because of bad packaging, while in Western Europe this amount is only 1–2% [1]. A very recent study (Financial Times, April 2008) [2] compared India and the UK in this respect and reported similar results (50% for India and 3% for the UK). A wide range of nanofillers, compatibilizing agents and polymers have been successfully used in the preparation of various composites and blends, and new methods of synthesis directly applicable in the industry, such as the melt intercalation process, have been developed. The substantial improvements in mechanical and physical properties brought by blends and composites have widened the use of polymers in industry. For example, their improved mechanical and thermal properties might extend the use of polymers for under-the-hood applications in the automotive industry, and their excellent barrier properties combined with good transparency make them ideal for packaging applications. The development of effective nanocomposites is directly linked to the availability and the properties of the nano reinforcements, which should be able to be dispersed in a polymeric matrix at a nanoscale level. Over the past decade, the use of nanoparticles as additives to enhance polymer performance has been established and now provides additional opportunities for many diverse commercial applications. Polymer nanocomposites (PNCs), i.e. nanoparticles (spheres, rods and plates) dispersed in a polymer matrix, have garnered substantial academic and industrial interest since their inception in about 1990. With respect to the neat matrix, nanoparticle dispersion has been shown to enhance physical (e.g., barrier, erosion resistance and reduced flammability), thermomechanical (e.g., heat distortion temperature, thermal expansion coefficient and stiffness) and processing (e.g., surface finish and melt strength) characteristics. Low-volume additions (1–10%) of isotropic nanoparticles, such as titania, alumina and silver, and anisotropic nanoparticles, such as layered silicates (nanoclays) or carbon nanotubes, provide property enhancements with respect to the neat resin that are comparable to those achieved by conventional loadings (15–40%) of traditional micrometer-scale inorganic fillers. The lower loadings facilitate processing and reduce component weight. Most important, however, is the unique value-added properties and property combinations that are not normally possible with traditional fillers, such as reduced permeability, optical clarity and selfpassivation, and flammability, oxidation and ablation resistance. Beyond maximizing nanoparticle dispersion, however, the morphology of these

Recent advances in polymer processing

5

materials is often uncontrolled, yielding isotropic nanofilled systems, not necessarily spatially engineered, designed and tailored composite materials [3]. In recent years the interrelationship between processing, morphology and properties has become one of the most important research topics in both industry and academia.

1.2

Polymer processing: an overview

There are many potential definitions of polymer processing. One was given by Tadmor and Gogos in their classic textbook: polymer processing concerns ‘operations carried out on polymeric materials or systems to increase their utility’ [4]. The goal of polymer processing is to increase the value of the polymer or formulation. Polymer processing can be viewed as a tool to achieve the desired shape, properties and performance for a polymer article. It has broad applicability from packaging to aerospace. It is the performance of a material which generates value. Thus, for each particular application, one should look for the critical performance characteristics for this application. How can this performance be achieved by a combination of the material and the process? Let’s take a particular example, the body and wings of the F-117A stealth fighter/bomber. What are the critical performance characteristics for this application? A few of them are listed below. • • • •

High strength/weight ratio Small radar signature Temperature resistance Chemical resistance

⇒ ⇒ ⇒ ⇒

Strong and light materials for aerospace Must be a ‘stealthy’ aircraft Hot engines and exhaust gases Intermittent contact with rain, jet fuel, etc.

How are these characteristics achieved? The body and wings of these stealth aircraft are constructed primarily of an epoxy/carbon fiber composite which is shaped and manufactured using a process called hand lay-up. For example, the starting material may be preforms in the shape of flat sheets which consist of uniaxial carbon fibers impregnated with a low or medium molecular weight epoxy. These sheets are stacked and formed in the desired shape of the wings and/or body and then cured to polymerize the epoxy. Both the materials used and the processing operation are critical for achievement of the desired performance. In this sense the stealth aircraft is a poor example. But the high performance epoxy/carbon composites are very expensive. The most important property of any material is its price. The most important property of any process is its cost. In addition, the hand lay-up process is very expensive because it requires a large amount of highly skilled labor. In this case, high costs are justified, due to the concurrent high costs of the engines and electronics on the aircraft. In addition, consider the military and political costs of having a pilot shot down behind enemy lines.

6


The shaping of the polymer through a variety of potential processing operations – fiber spinning, blow molding, thermoforming, injection molding, etc. – is where classical polymer processing starts. The vast majority of models for processing operations such as injection molding, single screw extrusion, twin screw extrusion and die flows consider the material to be a homogeneous continuum. Substantial advances have been made in modeling these processes and this approach is quite successful for the determination of velocity, stress and temperature fields along with useful macroscopic characteristics like pressure drop and flow rate. Commercial software packages are routinely used for equipment and process design. Unfortunately, with rare exceptions, these models do not even consider the morphology of the material and its interaction with the process. One of the major advances in polymer process modeling will be the inclusion of morphological characteristics, including both changes during the process itself and morphology of the final part. Large amounts of both experimental and numerical work are required before this goal can be achieved and substantial efforts are currently being made in a number of research laboratories. From a more encompassing perspective, this major advance is just one of the steps required to complete the process–structure–property connection for multiphase polymer systems. Computer simulations have recently been able to address many fundamental flow problems which include complicating features like inertial effects, multiple components that may be only partially miscible, and viscoelasticity. Molecular dynamics is probably the most fundamental numerical approach to fluid problems and has been applied to problems like the determination of appropriate boundary conditions for the interaction of a fluid with a wall (Koplik and Banavar, 1998) [5]. For simulations of flows on a scale much larger than the particle scale, however, molecular dynamics is both impractical and wasteful. There are numerous examples of multicomponent polymer systems where key performance characteristics depend critically on the material morphology. For instance, the phase domain morphology in a multicomponent polymer blend has a critical influence on the mechanical properties of the material. Wu showed the effect of dispersed phase size on the impact strength of nylon/rubber blends [6]. This morphology is controlled by the properties of the constituent polymers and the processing conditions utilized in the manufacturing operation. In general, interest is not limited to the size scale of the morphology, because control of domain shape can impart useful properties, such as reinforcing effects for cylindrical domains. More recently, advances in synthetic techniques and the ability to readily characterize materials on an atomic scale have led to interest in nanometer-size materials. Since nanometer-size grains, fibers and plates have a dramatically increased surface area compared to materials of conventional size, the chemistry of these nanosized materials is altered compared to conventional materials.


7

Most modeling capabilities are limited to predictions of droplet deformation and breakup in infinitely diluted, monodisperse Newtonian systems [7]. This is a severe deficiency, considering the fact that most commercially relevant systems involve significant coalescence, high dispersed phase concentrations and non-Newtonian fluids. Lack of fundamental understanding and applicable models impedes technological progress in this field. For example, new mixing devices in the polymer industry are currently designed on a trial-and-error basis. Successful development of a model capable of accurately simulating multiphase flows in polymer processing would dramatically advance our understanding of these processes as well as our ability to design and optimize them.

1.3

Recent advances in multicomponent polymeric materials

One fundamental question which has to be addressed first is: what inspires polymer scientists and industrialists to focus on polymer blends and composites rather than synthesizing new materials? Indeed, there are several reasons, among which the two most important are as follows. First development costs are lower; the design of new materials with special properties for demanding applications is always more expensive than the cost of the constituent existing polymers and the mixing process. The second reason is maximum diversification and increased use of existing polymers. Attention should be paid to the fact that combining of polymeric components in a certain ratio might result in a product with optimal properties for demanding applications. Additionally, a remarkably broad spectrum of properties can often be accomplished with polymer blends. Thus, blending seems rewarding with respect to the price/quality ratio as well as flexibility in realizing properties, compared to polymer synthesis. Besides these, other benefits that can be obtained by polymer blending include (i) improving specific properties, viz. impact strength or solvent resistance, and (ii) offering the means for industrial and/or municipal waste recycling. Furthermore, blending also benefits the manufacturer by offering (i) improved processability, product uniformity and scrap reduction, (ii) quick formulation changes, (iii) plant flexibility and high productivity, (iv) reduction of the number of grades that need to be manufactured and stored, (v) inherent recyclability, etc. [8]. Despite all these positive aspects, there are at least two big problems that should be alleviated to achieve the desired properties. These are the inherent immiscibility and incompatibility of polymeric materials, and a comprehensive and coherent approach towards these problems is essential. Also in the case of multicomponent thermosetting blend systems, volume shrinkage associated with curing is a serious concern as it may lead to the development of low performance materials having poor adhesion, shape

8


distortion, microcrack delamination, etc. A clear understanding of the interrelationship between miscibility, morphology and volume shrinkage is essential for the development of high performance materials having superior properties. Important renaissance in polymer processing in recent times has occurred mainly because of the developments of microfibrillar composites (MFC) [9– 12], electrically conducting polymer blends [13–15], nanostructured polymer blends [16–21], biodegradable polymer blends [22, 23], high temperature polymer blends [24] and polymer blends as biomaterials. Nanostructured blends very often exhibit unique properties, which are directly attributed to the presence of structural entities having dimensions in the nanometer range. The idealized morphology of these polymer blend systems is characterized by the molecular-level dispersion of the phases, which leads to a considerable enhancement in the mechanical, electrical and optical properties. Composite materials combine beneficial properties of their component materials that are not obtained in one component by itself. Generally composites are defined as materials consisting of two or more distinct phases with a recognizable interface or interface boundary, in other words, as a combination of two or more materials (reinforcing elements, fillers and composite matrix binders) differing in form or composition on a macroscale. Nanomaterials, characterized by having at least one dimension in the nanometer range, can be considered to constitute a bridge between single molecules and infinite bulk systems. Besides individual nanostructures involving clusters, nanoparticles in the form of arrays and superlattices are of vital interest to the science and technology of nanomaterials. Polymer nanocomposites are polymers that have been reinforced with small quantities (less than 10%) of nanosized filler particles of size below 100 nm. The dispersed phase can be inorganic particles, minerals, modified clays, etc. [1, 25]. Nanocomposites have been found to exemplify even more positive attributes than their predecessors do and thus an understanding of what occurs when nanocomposites of a polymer and inorganic components are produced is significant [1, 25]. Polymeric nanocomposites have been an area of intense industrial and academic research for the past 15 years. No matter the measures, articles, patents or research and development findings worldwide, progress in the use of PNCs has been growing exponentially. For example, the total number of hits for ‘polymer’ and ‘nanocomposite’ on SciFinder (Chemical Abstract Service (CAS) of the American Chemical Society) from 1988 to 2005 is >9400, the yearly number having approximately doubled every two years since 1992. Recent market surveys have estimated global consumption of PNCs at tens of millions of pounds (>$250m), with a potential annual average growth rate of 24% to almost 100 million pounds in 2011 at a value exceeding $500–800m [26–29]. A significant number of excellent review


9

papers, e.g. on clays [30–36] and carbon nanotubes [36–38], and books [39– 43] are available that chronicle and summarize the status of various nanoparticle–polymer combinations and the broad scientific and technological challenges still to be overcome. The structure and properties of nanomaterials differ significantly from those of atoms and molecules as well as those of bulk materials, and the synthesis, structure, energetics, response, dynamics and a variety of other properties and related applications form the theme of the emerging area of nanoscience, in each of which there is a large chemical component. Raw material producers, converters and end-users have therefore to tackle both compounding and processing issues. Surface modification of nanofillers with organic surfactant and adaptation of compounding conditions (high shear, high residence time, special screw profile design in case of melt compounding, for example) may help to get rid of most compounding issues. Research groups have made significant progress in that field. More research efforts are still required to identify the processing conditions that allow maintaining the dispersion and avoiding nanoparticles from aggregating again in the manufactured products. In conclusion the high potential of nanocomposites has already been demonstrated at the lab scale. It is now time to bridge the gap between scientific challenges and industrial stakes. The key issue is currently to maintain nanoparticle dispersion during the industrial-scale processing of nanocomposites on machines and equipments that are used to manufacture polymer composite parts on a regular basis. Efforts have to be oriented in this direction. This book provides a comprehensive review of polymer processing, focusing on recent developments in techniques and materials. Thermosets, thermoplastics, elastomers, foams and nanocomposites are all discussed; multiphase systems are considered from macro to nano scales. Developments in established techniques are reviewed, such as extrusion technologies, injection molding and blow molding, in addition to recently developed processing technologies, such as those using supercritical fluids, micromolding and reactive processing. Finally, post-processing techniques are discussed, as well as analysis of the molding process. The book intends to broaden and integrate knowledge of the behavior of materials through the development of the various processing–structure–properties relations, a complete evaluation of the effects of the different processing techniques on the material properties, optimization of the quality control of the starting materials, and the development and improvement of the manufacturing technologies used with these materials in order to obtain large series parts in a sector, which demands constant quality products.

10

1.4


Concluding remarks and new challenges

There is a great future and scope in the field of multiphase polymer blends and composites, especially in compatibilized systems, as multicomponent systems offer an extraordinarily rich range of new materials with enhanced characteristics in terms of mechanical, chemical and optical performance. Given the extensive variety of nanoparticles now commercially accessible (clays, carbon nanotubes, quantum dots, metals, silica, titania, zirconia, various other oxides, etc.), the potential combinations of polymers and nanoparticles, and thus the tailorability of the property suite, are essentially endless. Major revenues are forecast from large commercial opportunities, such as automobile, coatings and packaging, where lower cost, higher performance resins would improve durability and design flexibility while lowering unit price. The diversity in scientific investigation, technology advancement, processing innovations and product development is staggering. Processing techniques are critical to the performance of polymer products, which are used in a wide range of industries. In light of global polymer production, which from oil alone exceeds 200– 450 billion pounds annually, nanoparticle additions to plastics afford one of the commercially largest and most diverse near-term applications of nanotechnology. Since the first reports in the early 1990s the term ‘polymer nanocomposite’ has evolved to refer to a multicomponent system, where the major constituent is a polymer or blend thereof and the minor constituent exhibits a length scale below 100 nm. As such, the term is sometimes used as a synonym for inorganic–organic hybrids or molecular composites, or to encompass mature commercial products, such as filled polymers with carbon black or fumed silica. The numerous reports of large property changes with very small ( M c

2.17

log η0

where K1 and K2 are constants that depend on the molecular structure. Entanglements are an essential feature of polymer chains. They contribute greatly to the properties of polymer melts and solids. For most polymers Mc lies between about 10 000 and 40 000 and increases with increasing chain stiffness. Critical molecular weights of some commercial polymers are listed in Table 2.2.9 For polymer solutions the change in slope occurs at a particular value of c.Mw, where c is the polymer concentration, rather than Mw alone. As the shear rate is increased, the average entanglement density decreases. Since the viscosity is strongly dependent on the number of entanglements, viscosity falls.

3.4

Mc

log Mw

2.6 Dependence of zero shear viscosity on molecular weight.

22

Advances in polymer processing Table 2.2 Critical molecular weights of some commercial polymers

2.5.3

Polymer

Mc

Linear polyethylene Polystyrene Polymethyl methacrylate Cis-polyisoprene 1,4-polybutadiene

4000 30000 28000 10000 5000

Molecular weight distribution (MWD)

The transition from the constant zero shear region to the shear thinning region is more abrupt and occurs at a higher shear rate for the narrow molecular weight distribution polymer than for the broadly distributed material. Increasing Mw/Mn also has the effect of decreasing the slope of the shear thinning region. Narrow MWD polymers are known to show melt flow instabilities more easily than their broad MWD counterparts. The viscosity curves of two polymers having the same weight average molecular weight but different MWDs are shown in Fig. 2.7.8

2.5.4

Chain branching

Many industrially important polymers are branched. The branches can be long or short and these in turn can have secondary branching. The branches can be randomly spaced along the backbone chain or several branches can originate from a single point to give a star-shaped molecule. Short branches generally do not affect the viscosity of a molten polymer very much. On the other hand long branches can have a significant effect. Branches that are long but are still shorter than those required for entanglements decrease the zero shear viscosity when compared to a linear polymer of the same molecular weight.11,12 However, if the branches are long enough to participate in entanglements, the branched polymer may have a viscosity much higher than that of a linear polymer of the same molecular weight at low rates of shear.13,14 In general, branched polymers also show more shear thinning than comparable linear polymers.

2.5.5

Effect of additives

Properties of polymers can be modified to a great extent by manipulating their structures. Still, very few polymers are used in their chemically pure form. Additives that are widely used affect rheological properties of the base polymer to which they are added even when they are not added for that purpose.

Number of molecules


23

Resin-B

Resin-A

100

101

102 103 Molecular weight (a)

104

104

Viscosity (Pa.s)

Resin-B Resin-A 103

102 100

101

102 Shear rate (s–1) (b)

103

104

2.7 Dependence of molecular weight distributions on the flow curve: (a) molecular weight distribution; (b) flow curves (sketch courtesy of D.E. Delaney, Dynisco Polymer Test).

Fillers While fillers generally enhance the viscosity of polymers, the extent of increase depends upon the type of filler and shear. The effect of different types of fillers on the viscosity–shear stress curve of a polymer is shown in Fig. 2.8.15 High aspect ratio fillers such as long glass fibres produce significant improvement in viscosity at low shear, but the effect is not so pronounced at high shear. In the case of non-interacting low aspect ratio fillers, the enhancement is more or less the same at low and high shear. In the case of interacting low aspect ratio fillers, the effect is similar to that of high aspect ratio fillers, i.e. significant improvement in viscosity at low shear, with a

24

Advances in polymer processing High aspect ratio filler Agglomerated low aspect ratio filler

Viscosity

Low aspect ratio filler

Base polymer

Shear stress

2.8 Effect of different types of fillers on the viscosity–shear stress behaviour of polymer.

much reduced effect at high shear. The elastic response of filled melts is usually reduced relative to that of the base polymer. However, fillers of high aspect ratio may form a network themselves within the melt and produce anomalous elastic response.16 There may be a dramatic increase in elongational viscosities when fibrous fillers are added to polymer melts.17 The viscosity of filled systems can be easily determined for a Newtonian fluid by using the Einstein equation18,19 for low concentration and low shear:

η = ηs (1 + 2.5φ)

2.18

where ηs is the viscosity of the suspending liquid and φ the volume content of the solids. This equation does not take into account the shape or size of the dispersed particles. At higher filler content, they have to be taken into account. An example is the Krieger–Dougherty equation:20

φ  –[ η ]φm η = ηs  1 + φm  

2.19

where φm is the maximum packing fraction and [η] the intrinsic viscosity. Plasticizers Plasticizers are low molecular weight liquids used for improving the processability of polymers and flexibility of products. Plasticizers act by spacing out the polymer molecules. Their most obvious effect is to reduce


25

viscosity, but they also tend to reduce the elastic modulus of the melt, thereby increasing the elastic response at a given stress. The effectiveness of a plasticizer depends on its concentration, compatibility and viscosity.13 Low molecular weight polymers are frequently effective plasticizers.

2.5.6

Effect of pressure

Even though the effect of pressure on viscosity is not as large as the effect of temperature, pressure effects can become significant in processes such as injection moulding where high pressures are employed. Several theories support the idea that viscosity is strongly dependent on the free volume of the system.21–25 Since free volume is directly influenced by pressure, viscosity also depends on pressure.26–28 Viscosity generally increases with increasing pressure and an exponential relation such as

η0 = A.eBP has been found to be useful. polymer.

2.6

2.20 29,30

A and B are constants characteristic of the

Elongational flow

For polymer processes such as fibre spinning, blow moulding, thermoforming, certain extrusion die flows, etc., the major mode of deformation is elongational. The elongational viscosity (µ) of a liquid is much higher than the shear viscosity and for a simple Newtonian fluid Trouton31 has suggested

µ 0 = 3 η0

2.21

This relation holds good for polymer melts in the initial Newtonian region. Thereafter, regions of strain hardening or strain thinning are observed for different materials. The shear and elongational viscosities for two types of polystyrenes are shown in Fig. 2.9. In the region of the Newtonian plateau, the limit of 3 is observed.32 The elongational viscosities as a function of tensile stress for various thermoplastics in common processing conditions are shown in Fig. 2.10.32 The strain-thinning part is due to the reduction in entanglement density by links being slipped off the ends of chains. These are short-lived entanglements near the chain ends. The increase in µ with strain rate is an effect of the long-lived entanglements giving rise to strain hardening.

2.7

Elastic effects in polymer melt flow

Polymeric fluids are also viscoelastic in nature, i.e. they exhibit both viscous and elastic properties. For purely Hookean elastic solids the stress corresponding

26

Advances in polymer processing 5.108

Viscosity, η˙ ,µ (Pa.s)

µ 0 = 1.7·108 Pa-s

Elongational test

µ 0 = 1.6 ·108 Pa-s

108

η0 = 5.5·107 Pa-s

5.107

η0 = 5 ·107 Pa-s

Shear test

107

T = 140°C

5.106

Polystyrene I Polystyrene II

106 102

5.102 103

104 105 Shear/tensile stress τ, σ (Pa)

5.105

2.9 Shear and elongational viscosities of two types of polystyrenes. 6 LDPE

Viscosity, log µ (Pa.s)

5 Ethylene–propylene copolymer

4

PMMA POM

3 PA66

2 3

4 5 Tensile stress, log σ (Pa)

6

2.10 Elongational viscosities of polymers as a function of tensile stress.

to a given strain is time independent, but with viscoelastic materials the stress dissipates over time. Viscoelastic fluids undergo deformation when subjected to stress; however, part of such deformation is gradually recovered when stress is removed.


27

The response of a viscoelastic material to an applied stress or strain depends on the rate of application of this stimulus. When the structure of a body is perturbed by external stimuli, the time it takes to reach another equilibrium state or return to its original rest state is characteristic of that structure. A useful parameter to determine the extent of elasticity is the Deborah number De defined by Reiner:33

De = λ tp

2.22

where λ is the relaxation time of the polymer and tp a characteristic process time. In the case of flow through a die, the characteristic process time can be defined by the ratio of die dimension to average speed through the die. A Deborah number of zero represents a viscous fluid and a high Deborah number an elastic solid. As the Deborah number becomes greater than 1, the polymer does not have enough time to relax during the process, resulting in possible extrudate dimension deviation or irregularities such as extrudate swell, shark skin or even melt fracture. Even though the Deborah number provides a reasonable qualitative description of material behaviour, polymers generally cannot be characterized by a single response time. A more realistic description involves the use of a distribution or a continuous spectrum of response times. If a Newtonian fluid is forced through a capillary, the extrudate would swell to a diameter larger than that of the capillary on exit from the die. A viscoelastic material shows a much more pronounced swell on extrusion than a Newtonian material. The reason for this is the generation of forces normal to the shearing direction. Normal stresses are an essential component in polymer processing.

2.8

Polymer blends

Polymer blending is a common technique for manipulating the rheological behaviour. In the case of miscible blends where the constituent polymers mix intimately on a molecular level to form a homogeneous material, the rheological properties, like other properties, are usually a weighted average of the properties of the component polymers. An example of this is provided by blends of poly(phenylene oxide) (PPO) and polystyrene.34 In the case of high viscosity incompatible polymers, the blend may have significantly lower viscosity than either of the constituents as in the case of the blends of polycarbonate and poly(4-methylpentene-1) shown in Fig. 2.11.15 This behaviour is attributed to the weakness of shear planes at the interface between the phases.35 A completely different response may be observed in the blends of low viscosity melts where the blend may have a higher viscosity than either of the constituents as in the case of the blends of polypropylene

28


Viscosity(Ns/m2)

Poly(4-methyl pentene-1)

Polycarbonate

103

102

50/50 blend

104

105 Shear stress (N/m2)

106

2.11 Variation of viscosity with shear stress for polycarbonate, poly(4methyl pentene-1) and their 50/50 blend.

and nylon-66 shown in Fig. 2.12.15 This effect is probably due to the large interfacial area produced in a dispersion at a 1 µm level. The flow of such a dispersion requires that the droplet shape be deformed to an ellipsoid, the increase in surface area accounting for the increase in viscosity. Generally, it is not easy to predict the rheology of polymer blends since it depends on so many factors such as compatibility, concentration of the components, viscosity of the components, morphology of the blends, history of blending operations, etc.36 Melt blending is the most widely using technique of blend preparation. Melt blending of two immiscible polymers naturally results in a dispersion of the minor phase component within a matrix of the major component. The dispersed phase droplets tend to minimize their surface area to volume ratio by adopting a spherical or near-spherical shape, and a high shear mixer like a twin screw extruder can generate a fine dispersion with droplet diameters of

Molecular relaxation

ASP1

>

ASP1

(b)

8.2 (a) Shape recovery of non-spherical particles due to stress relaxation; (b) particle shape recovery and relaxation mechanism.

This shape recovery effect, shown in Figs 8.2(a) and (b), produces important differences in the sintering behaviour of ground polymer powders compared with some as-polymerized powders (Greco and Maffezzoli, 2004). This effect does not apply to some high molecular weight polymers, such as UHMWPE

210


(Wu et al., 2002). However, in the analysis of the sintering of HDPE, shape recovery can be considered as an initial stage in the overall sintering process. Equivalent sintering factor One of the main parameters used to characterize the initial stage of the process is the sintering factor. It expresses a relationship between the sintering neck and the initial radius of a spherical particle. The analysis presented here takes into account the non-spherical character of the particles considering an equivalent radius (see Fig. 8.3). Equivalent sintering factor = X a0

8.3

where: X = length of the sintering neck a0 = initial equivalent radius (see below). Initial equivalent radius The initial equivalent radius a0 of a particle is determined, assuming that it is spherical. These radii are calculated from real projections of the area which is considered equivalent to a circular area. The radius of each particle is then calculated. For a two-particle system, an equivalent ‘system-radius’ can be defined: a0 =

r1 + r2 2

8.4

where r1 and r2 are the radii for particles 1 and 2 in the two-particle system, respectively.

X

req

8.3 Scheme of the parameters used in sintering studies; grey areas show the shapes of non-spherical particles, approximated by spheres (req = equivalent radius; X = sintering neck).


211

Area An approximate projected area is calculated from the images obtained in the tests. With this value, the apparent radii of the particles are calculated as follows:

R=

a proj π

8.5

where R is the radius of each particle and aproj is the corresponding projected area. The analysis presented above can be used to assess the sintering process of non-spherical polymer particles. Optical measurements can be performed using standard image analysis software such as Image-Pro Plus. Evolution of the sintering neck and sintering ratio The sintering process of two representative particles in a controlled temperature oven may be used to study the evolution of the sintering neck and the sintering ratio. At the beginning, no visible physical change is observed. After a critical time is reached, the particles change their shape, undergoing a decrease in surface area. The non-spherical particles start to display a smoother surface, becoming more spherical in shape, and accelerating the speed of neck formation. Figure 8.4 shows the evolution of the sintering ratio of HDPE at 110° and 130°C. For comparison purposes, data from the literature for linear low

1.00

0.80

X/a0

0.60

HDPE-T = 110°C experiment

0.40

HDPE-T = 130°C experiment LLDPE

0.20

LLDPE PC 0

0

2

4

6

8 10 Time (min)

12

14

16

18

8.4 Evolution of the neck formation process as a function of time for different materials.

212


density polyethylene (LLDPE) and polycarbonate (PC) have been included (Kontopoulou et al., 1998). As can be observed from Fig. 8.4, most polymers show similar tendencies in the evolution of the sintering neck. Sintering time and temperature have a significant influence on the shape of the curves. In some cases it is possible to achieve rates of x/a0 = 1.16, which is close to the values found for ideal spherical particles (x/a0 = 1.26). The shape and disposition of the particles influence the rate. Shrinkage Dilatometry and microscopy have been used in the past (Kontopoulou and Vlachopoulos, 2001) to study shrinkage and its relationship with sintering. In the present work, microscopy and image analysis were used to determine shrinkage. In the past, some simple expressions have been used to relate shrinkage to sintering parameters. German and Munir (German and Munir, 1975; German, 1998) have used the following equation to relate shrinkage to the neck development rate: 2 ∆L =  X  L0  2D

8.6

where: ∆L/L0 = contraction (change in particle length divided by the original particle length) X = length of sintering neck D = particle diameter X/2D = neck development rate. Figure 8.5 shows the variation of shrinkage with the square of the development rate. With increasing development rate, the shrinkage increased steadily up to a sintering factor of 0.5, where it reached a plateau. The maximum shrinkage measured for this experiment was 13.6%. The type of behaviour observed in Fig. 8.5 is not in agreement with previous results for metals as described by equation 8.6 (German and Munir, 1975; German, 1998). This is due to the fact that shrinkage occurs by different mechanisms with metals and polymers (German and Munir, 1975; German, 1998, Exner and Arzt, 1983). Evolution of the surface area One of the fundamental characteristics of the sintering process is the decrease in surface area brought about by capillary forces and molecular relaxation (German and Munir, 1975; Eshelby, 1949; Rozenzweig and Narkis, 1981). To describe this process, it is useful to define a specific area, namely its


213

14.00 12.00

[(L – L0)/L0]%

10.00 8.00 6.00 4.00 Experimental data

2.00

Equation 8.6 0 0

0.20

0.40

0.60 2 (X/2D)

0.80

1.00

1.20

8.5 Variation of shrinkage with the square of the development rate for polyethylene particles.

surface area divided by its weight. Three of the techniques used to measure specific area are gas absorption, gas permeability and quantitative microscopy (Batel, 1971; Kontopoulou and Vlachopoulos, 1999; Allen, 1998). Quantitative microscopy was used in the experiments discussed here. During the onset of sintering of HDPE particles, they do not suffer any considerable change or deformation for the first 2 or 3 minutes. This might be due to the time required for the particles to reach thermal equilibrium. After a sintering time of 4 or 5 minutes a marked decrease in surface area occurs, as the particles begin to adopt a more spherical shape. For a 300 µm particle the projection of the initial area was 0.433 mm2. For most twoparticle systems studied, the reduction in area is most important in the time interval 4–6 min, where 70–80% of the change occurs. During this stage the material undergoes deformation, resulting in a decrease in the surface area of the two particles and the growth of the coalescing neck connecting them (Bellehumeur et al., 1996; Torres et al., 2004). The amount of change in surface area due to shape recovery is considerably smaller than that caused by the formation of the sintering neck in the first stages of sintering. German and Munir (1975) used the following equation for metallic powders in order to relate the decrease in surface area to the sintering neck rate: M ∆S = – K  X  s S0  D

8.7

where ∆S/S0 is the change in surface area divided by the original particle area, and X/D is the sintering neck rate.

214


The tendency observed by German and Munir is shown in Fig. 8.6 and compared with experimental data from this research. The type of behaviour observed in Fig. 8.6 is not in agreement with the previous results for metals described by equation 8.7. The initial decrease in surface area in the case of polyethylene is caused mainly by molecular relaxation. This initial stage is not present in the case of metals in the same way as with polymers (German and Munir, 1975; German, 1998; Exner and Arzt, 1983). 0.45

Relative area loss (S – S0)/S0

0.40 0.35 0.30 0.25 0.20 0.15 0.10 HDPE: 130°C

0.05

Equation 8.7

0 0

0.10

0.20

0.30

0.40

0.50

X/D

8.6 Comparison of the relative area loss for polyethylene particles. Diamonds indicate experimental points and the broken line indicates the values predicted with equation 8.7. Stereoscope

Camera

T°C Chronometer Thermocouple

Sample

0:00 Glass

Hot air

X

Thermal isolation Heater

Q

Heat chamber

t

Sintering chamber Data acquisition

8.7 Sintering rig used for the study of a small group of particles.


215

Densification The onset of sintering is first observed among those particles in contact with the mould wall. A thin layer of material starts to melt. No major changes are observed in those layers apart from the mould wall. Heat is transferred by conduction between the different layers of particles. This mechanism induces the formation of a new layer of coalescing particles closer to the mould wall. The air between the particles is forced away as each layer sinters and shrinks, before the next layer starts to melt. However, some of the air remains trapped on its way out, forming bubbles. These bubbles are undesirable because they adversely affect the mechanical properties of the final moulding. It is possible to observe three simultaneously occurring phases during the process: • • •

A lower layer (at the mould wall) of fused HDPE (already coalesced) A middle layer of coalescing particles An upper layer that shows no signs of coalescence.

At the end of the process, when the material had been fully melted, the overall thickness showed an important reduction compared to its initial value. This is usually accompanied by bubbles entrapped in the material. In order to eliminate or reduce them, it would be necessary to extend the processing cycle or increase the processing temperature, although the scope to do this is limited by degradation of the polymer. The introduction of a compacting mechanism might become necessary in order to eliminate bubbles from the final product.

8.3

Rotational moulding of composite and nanocomposite materials

8.3.1

Fibre reinforced polymer systems

No theoretical study has been offered yet to describe the sintering process of polymer particles in the presence of reinforcing fibres or fillers. On the other hand, in the metals and ceramics field, the classification of the different types of sintering processes seems to be well established. Thümmler and Oberacker (1995) differentiate single component sintering from multicomponent sintering in metals and ceramics. More specifically, for the latter case, they divide the processes into solid phase sintering and liquid phase sintering. They also describe the specific stages characteristic to each of these processes, and the prevailing transport mechanisms associated with them. For metals and ceramics, these mechanisms can be surface diffusion, evaporation and condensation, volume diffusion and grain boundary diffusion. For the discussion of this type of sintering processes, they consider coalescence

216


as a secondary mechanism which enforces the ‘coarsening’ of the structure by building up larger particles during liquid phase sintering. For the description of the fibre reinforced systems we proceed as we did with unreinforced systems (see Section 8.2). According to Torres et al. (2006), for non-spherical particles, the projected area of the particles involved in the analysis is used. Thus, the equivalent radius is equal to: ap 8.8 π where req is the particle radius and ap is the projected area of a particle. When two particles with different geometry are considered, one equivalent radius can be approximated for the two-particle system. This is mainly used for purposes of comparison with spherical particle systems. Then the equivalent radius for the system can be defined as: r + r2 req–s = 1 8.9 2 where r1 and r2 are the approximated radii for each of the particles. The projected area of a particle can be determined by means of digital area estimation procedures (see Fig. 8.3). Since the initial particle areas vary from one particle to another, it is necessary to express the results in a normalized form. The relative surface area can be defined as: a(t ) 8.10 a rel ( t ) = a0 where arel (t) is the relative area at time t, a(t) is the actual surface area at time t, and a0 is the initial particle area at time t = 0. req =

Studies of a small group of particles and fibres Figure 8.7 shows the typical configuration of a rig for the study of sintering in a temperature control chamber. In this case a chamber is used instead of the more typical ‘hot stage’ disposition. The temperature-controlled chamber produces similar conditions to those encountered in real processing situations. Visualization of the sintering sequences is carried out using standard digital photographic and video cameras. Studies of a polymer/fibre bed In this case, Torres et al. (2006) used a sintering chamber that allows for viewing the cross-section of a sintering bed. A stereomicroscope was fitted in a perpendicular position with regard to the sample. The rig can be seen in Fig. 8.8. The oven temperature could be regulated up to 200ºC.


217

8.8 Sintering chamber used for the polymer/fibre bed studies.

8.9 Sintering sequence of two polymer particles and one sisal fibre (oven temperature 110°C) at 25×.

Study of the sintering of two non-spherical polymer particles in contact with one fibre This set of studies has shown some of the possible mechanisms for the formation of bubbles and voids during the melt densification of a polymer powder bed when filled with natural fibres. As can be seen from Fig. 8.9, the two particles studied in the sintering chamber start in a position where the gap between them can be considered to be zero. As the process continues (t = 7 min 25 s), an important separation between the particles is observed. When the particles cool down (Fig. 8.9, right) they still display a considerable separation between them. For a more detailed assessment of the evolution of this process, Fig. 8.10 shows the variation of the gap between particles expressed in millimetres. Figure 8.10 considers only the heating process and does not include the values obtained during cooling of the particles. The variation of relative surface area is also displayed in the same figure. As can be seen from Fig. 8.10, particle 2 (right in Fig. 8.9) shows a higher reduction in relative area than particle 1 (left in Fig. 8.9). The gap between the particles starts increasing at around t = 230 s and then continues to grow up to the end of the heating process up to almost 0.4 mm.


0.40 0.35

Relative area, A/A0

1.0

0.30 Particle 1

0.8

0.25 0.20

0.6

0.15

0.4

Particle 2

0.10

Gap between particles

0.2

Gap (mm)

218

0.05 0

0 0

100

200

300 Time(s)

400

500

8.10 Relative area of the particles including cooling time (6 min). 1.1

Relative area, A/A0

1.0 0.9 Particle 1

0.8 0.7 0.6

Cooling

0.5 Particle 2 0.4 0.3 0

100

200

300

400 500 Time(s)

600

700

800

900

8.11 Effect on cooling on the relative area of the particles.

It is clear from Figs 8.9 and 8.10 that a single fibre can prevent two particles from coalescing when placed one in contact with the other. The size of the gaps encountered during the process is comparable in size to the actual particle size. For larger numbers of particles, larger voids can be expected and have also been verified in other experiments, as well as in the rotational moulding process (Torres and Aguirre, 2003). The behaviour observed in Fig. 8.10 is in contradiction with the behaviour found for two polymer particles without the presence of a fibre (Torres and Carrillo, 2003). In that case, the two particles would coalesce and form a larger particle. One explanation for this behaviour would be the differences in wetting and surface forces between polymers and natural fibres. Further studies are being carried out at POLYCOM (Catholic University of Peru) in order to determine accurately the interfacial shear strengths and surface forces between these components. Figure 8.11 shows the variation in relative area of the particles shown in Fig. 8.10; however, in this case the effect of cooling is included. The change


219

in the particles area due to shrinkage can be estimated in the range 5–10%, with regard to the end of the heating process. This might also be a source for minor voids and cavities; however, it is of a smaller magnitude compared to the gap occurring between the particles. Study of the sintering of a natural fibre filled polymer powder bed The present group of studies claims to produce the most similar conditions to those encountered in the rotational moulding process of natural fibre reinforced thermoplastics. However, the same treatment can also apply to other processing situations in which fibre filled polymer powders are consolidated with a free surface. Figure 8.12 shows a sequence of melt densification studies with many polymer particles and fibres. At time t = 0 s, a dry powder bed can be observed where the fibres are distributed more or less evenly. At time t = 18 min 17 s, partial areas of transparent melted plastic can be observed together with some white non-transparent regions and some fibre as well. In the time range t = 26–30 min, most of the polymer is already melted and transparent, so the fibres can be more easily observed. After cooling, the final thickness of the polymer–fibre system can be observed and some small dark regions indicate the presence of voids. Figure 8.13(a) shows a sequence of the sintering of a multiparticle polymer bed in the presence of sisal fibres. At time t = 0 s, the individual polymer particles can be observed in the presence of two sisal fibres. In the middle and right micrographs it is possible to observe how some of the particles have started to coalesce. In the areas adjacent to the fibres, however, one can

0 min 0 s

13 min 30 s

18 min 17 s

20 min 0 s

26 min 0 s

32 min 2 s

37 min 20 s

Air cooled 10 min

8.12 Sintering sequence for HDPE powder with 5% w/w sisal fibres (ca. 5 ×).

220


0 min 0 s

2 min 05 s

3 min 10 s

(a)

(b)

8.13 (a) Sequence of the sintering of a multiparticle polymer bed in presence of sisal fibres (10); (b) detail of the sequence of the sintering of a multiparticle polymer bed in presence of sisal fibres (40×).

observe that the presence of cavities is more obvious than in regions far from the fibres. Figure 8.13(b) shows a detail of the sequence of the sintering of a multiparticle polymer bed in presence of sisal fibres. The reduction of pores in the region between the two fibres can be noted. Natural fibre reinforced thermoplastics Natural plant fibres have been used in the past as a reinforcing material for different types of matrices (Mohanty et al., 2000; Nabi Saheb and Jog, 1999; Rowell et al., 1997; Bledzki et al., 1998). In recent years, attention has been paid to their use as a reinforcing material for thermoplastics. In particular, the automotive industries have shown interest in the advantages that this type of fibre-reinforced system can provide (Mohanty et al., 2000; Nabi-Saheb and Jog, 1999). The advantages of biofibres over traditional fibre reinforcements, such as glass fibres, are: low cost, low density (good specific properties), reduced wear in processing equipment, high toughness, biodegradability and ‘ecological friendliness’ (since they can be produced from renewable resources). Torres et al. have introduced the rotomoulding process for natural fibre reinforced polymer systems (Torres et al., 2002; Torres and Aguirre, 2003).


221

They used polyethylene reinforced with sisal, jute and cabuya fibres cut in lengths in the range 5–10 mm. The process starts with loading a mould with polymer powder and fibre together. Then the mould is rotated in two axes at relatively low speeds (usually 10–30 rpm in industrial operation) while being heated, so that particle consolidation can occur. Then the mould is cooled down and the product is extracted from it. Rotation in the first axis is achieved with an electric DC motor, so that speed could vary over the processing range. Rotation in the second axis is achieved by means of a lightweight gearbox that transmitted the movement from the first axis with a controlled transmission rate. One of the drawbacks of using natural fibres is their poor resistance to moisture absorption. Wang et al. (2007) have tried several chemical treatments, including mercerization, silane treatment, benzoylation and peroxide treatment, in order to remove the non-cellulosic components of flax fibres used to produce LLDOE and HDPE composites. This study showed that chemical treatments improved the surface properties of flax fibres and, thus, the tensile properties of the composites produced.

8.3.2

Particle reinforced composites

Particles can be used to reinforce rotational moulded composites. Yan et al. (2002) have used glass beads to reinforce polyethylene. As in other processing techniques, a uniform distribution of the filler is required in order to improve the mechanical properties of the products. The Halpin–Tsai–Niesen and Nicolais–Narkis models were used by the same authors in a later paper to predict the tensile modulus and tensile strength of the composite when the reinforcement is uniformly distributed (Yan et al., 2006). Rusu and Rusu (2007) have obtained composites reinforced with graphite microparticles by means of reactive rotational moulding. In this process the in situ anionic ring-opening polymerization of epsilon-caprolactam monomer is carried out in the presence of the reinforcement in order to obtain Nylon 6/graphite composites. Although the impact strength decreases with increasing graphite concentration, flexural strength and flexural modulus were improved.

8.3.3

Polymeric nanocomposites

Rotational moulded nanocomposites are not reported in the literature. However, Martin et al. (2003) have studied polyethylene-layered silicate nanocomposites as possible new candidates for rotational moulding. The nanoclay used, namely modified montmorillonite and modified synthetic silicate, dispersed in the polymer by twin-screw melt compounding. The tests carried out showed that the nanocomposite materials produced had acceptable processability for rotational moulding.

222

8.4


Rotational moulding of biobased materials

Current work at POLYCOM (Catholic University of Peru) includes the development of starch-based polymers for rotational moulding. In order to process starch as a thermoplastic, the gelatinization process of starch needs to be carefully controlled. When a pre-gelatinized starch paste is introduced into the mould, the process resembles the rotational moulding of plastisols. Homogenization and bubble removal remain the main problems found in this process. The introduction of nanofillers such as silicates and biobased whiskers is also currently under investigation.

8.5

Mechanical properties of rotomoulded materials

In order to characterize the mechanical properties of rotomoulded parts, Torres and Aragón (2006) have described a variety of tests that can be carried out on final rotomoulded products. Mechanical tests include tensile, compression, impact, deep drawing and recovery tests, as well as environmental stress cracking resistance (ESCR) and shrinkage tests. Mechanical characterization shows that the properties of natural fibre-reinforced composites were in some cases superior to those of the unreinforced polymer products. Tensile strength results can be used to determine the optimal fibre content for which fibre-reinforced composites show the best properties. Beyond this optimal fibre content, the mechanical properties tend to decrease.

8.6

Rotational foam moulding

The rotational moulding process can be modified to produce foams instead of hollow pieces in an uninterrupted rotational cycle by using ordinary rotational moulding equipment in conjunction with a blowing agent/polymer mixture (Beall, 1998). As in the regular rotational moulding process, the foam moulding begins with loading a shell-like mould with blowing agent/polymer mixture. Then, the mould is rotated and heated in order to obtain a homogeneous continuous polymer melt. After the polymer melt is formed, the blowing agent starts decomposing and the foaming expansion begins. The foaming process comprises three steps: cell nucleation, cell growth and stabilization (Pop-Iliev et al., 2003a). Most of the studies made on rotational foam moulding have tried to assess the feasibility of the process, and to investigate the parameters and conditions influencing the final product quality, i.e. the desired cell population, cell density and average cell size. The heating profile, heating rate, heating time, rotational speed, the careful selection of the resin and the blowing agent are the key factors in the rotational foaming process (Pop-Iliev et al., 2003a).


223

Polyethylene (Archer et al., 2004; Liu and Yang, 2001; Liu and Tsai, 2004; Liu et al., 2004) and polypropylene (Pop-Iliev et al., 2003a, 2003b) in both powder and pellets are by far the most used materials for this process.

8.7

References

Abdullah MZ, Bickerton S and Bhattacharyya D (2005), ‘Enhancement of convective heat transfer to polymer manufacturing molds’, Polym Eng Sci, 45 (1), 114–24. Abdullah MZ, Bickerton S and Bhattacharyya D (2007a), ‘Rotational molding cycle time reduction through surface-enhanced molds: Part A – Theoretical study’, Polym Eng Sci, 47 (9), 1406–19. Abdullah MZ, Bickerton S and Bhattacharyya D (2007b), ‘Rotational molding cycle time reduction through surface-enhanced molds: Part B – Experimental study’, Polym Eng Sci, 47 (9), 1420–29. Abu-al-Nadi DI, Abu-Fara DI, Rawabdeh I and Crawford RJ (2005), ‘Control of rotational molding using adaptive fuzzy systems’, Adv Polym Tech, 24 (4), 266–77. Allen T (1998), Particle Size Measurement, London, Chapman and Hall, 45–58. Archer E, Harkin-Jones E, Kearns MP and Fatnes AM (2004), ‘Processing characteristics and mechanical properties of metallocene catalyzed linear low-density polyethylene foams for rotational molding’, Polym Eng Sci, 24 (4), 638–47. Batel W (1971), Einfuehrung in die Korngroessentechnik, Berlin, Springer. Bawiskar S and White J (1994), ‘Comparative study of warpage, global shrinkage, residual stress, and mechanical behaviour of rotationally molded parts produced from different polymers’, Polym Eng Sci, 34 (10), 815–20. Beall GL (1998), Rotational Moulding: Design, Materials, Tooling and Processing, Cincinnati, OH, Hanser/Gardner Publications. Bellehumeur CT, Bisaria MK and Vlachopoulos J (1996), ‘An experimental study and model assessment of polymer sintering’, Polym Eng Sci, 36 (17), 2198–2207. Bellehumeur CT, Kontopoulou M and Vlachopoulos J (1998), ‘The role of viscoelasticity in polymer sintering’, Rheologica Acta, 37 (3), 270–78. Bledzki AK, Reihmane S and Gassan J (1998), ‘Thermoplastics reinforced with wood fillers: A literature review’, Polym Plast Tech Eng, 37 (4), 451–68. Chaudhary BI, Takacs E and Vlachopoulos J (2002), ‘Ethylene copolymers as sintering enhancers and impact modifiers for rotational molding of polyethylene’, Polym Eng Sci, 42 (6), 1359–69. Cramez MC, Oliveira MJ and Crawford RJ (2003), ‘Optimization of the rotational moulding process for polyolefins’, Proc Instn Mech Engrs Part B: J Engineering Manufacture, 217, 323–34. Crawford RJ (1997), Rotational Moulding of Plastics, Chichester, John Wiley & Sons. Crawford RJ and Nugent PJ (1992), ‘A new process-control system for rotational molding’, Plast Rub Comp Proc Appl, 17 (1), 23–31. Crawford RJ and Scott JA (1987), Plast Rub Comp Proc Appl, 7, 85–99. Eshelby JD (1949), ‘Discussion of seminar on the kinetics of sintering’, Met Trans, 185, 796–813. Exner HE and Arzt E (1983), ‘Sintering processes’, in Cahn RW and Haasen P, Physical Metallurgy, Amsterdam, Elsevier, 1885–1912. Frenkel J (1945), ‘Viscous flow of crystalline bodies under the action of surface tension’, J Phys, 9 (5), 385–91.

224


Gao P and Mackley MR (1994), ‘A general model for the diffusion and swelling of polymers and its application to ultra-high molecular mass polyethylene’, Proc Roy Soc, 444 (1921), 267–85. German RM (1998), Sintering Theory and Practice, New York, John Wiley and Sons. German RM and Munir ZA (1975), ‘Morphology relations during surface-transport controlled sintering’, Met Trans B, 6 (2), 289–94. Greco A and Maffezzoli AJ (2004), ‘Powder shape analysis and sintering behavior of high-density polyethylene powders for rotational molding’, J Appl Polym Sci, 92 (1), 449–60. Guillén-Castellanos W, Lin W, Bellehumeur CT and Weber M (2003), ‘Effect of processing history on the sintering of ethylene copolymers’, Int Polym Process, 18, 87–90. Kontopoulou M and Vlachopoulos J (1999), ‘Bubble dissolution in molten polymers and its role in rotational molding’, Polym Eng Sci, 39 (7), 1189–98. Kontopoulou M and Vlachopoulos J (2001), ‘Melting and densification of thermoplastic powders’, Polym Eng Sci, 41 (2), 155–69. Kontopoulou M, Takács E, Bellehumeur CT and Vlachopoulos J (1998), ‘Resins for rotomolding: considering the options’, Plast Eng, 54 (2), 29–31. Liu GB, Park CB and Lefas JA (2004), ‘Production of low-density LLDPE foams in rotational molding’, Polym Eng Sci, 38 (12), 1997–2009. Liu SJ and Tsai CH (2004), ‘An experimental study of foamed polyethylene in rotational molding’, Polym Eng Sci, 39 (9), 1776–86. Liu SJ and Yang CH (2001), ‘Rotational molding of two-layered polyethylene foams’, Adv Polym Tech, 20 (2), 108–15. Martin D, Halley P, Truss R, Murphy M, Jackson O and Kwon OY (2003), ‘Polyethylenelayered silicate nanocomposites for rotational moulding’, Polym Int, 52 (11), 1774– 79. Martins JA, Cramez MC, Oliveira MJ and Crawford RJ (2003), ‘Prediction of spherulite size in rotationally molded polypropylene’, J Macromol Sci Phys Part B, 42 (2), 367– 85. Mohanty AK, Misra M and Hinrichsen G (2000), ‘Biofibres, biodegradable polymers and biocomposites’, Macromol Mater Eng, 276 (3–4), 1–24. Nabi Saheb D and Jog JP (1999), ‘Natural fiber polymer composites: a review’, Adv Polym Tech, 18 (4), 351–63. Pop-Iliev R, Rizvi GM and Park CB (2003a), ‘The importance of timely polymer sintering while processing polypropylene foams in rotational molding’, Polym Eng Sci, 43 (1), 40–54. Pop-Iliev R, Liu FY, Liu GB and Park CB (2003b), ‘Rotational foam molding of polypropylene with control of melt strength’, Adv Polym Tech, 22 (4), 280–96. Rowell RM, Sanadi A, Caulfield DF and Jacobson RE (1997), ‘Utilization of natural fibers in plastic composites: problems and opportunities’, in Leão AL, Carvalho FX and Frollini E, Lignocellulosic Plastic Composites, São Paulo, USP & UNESP, 23–51. Rozenzweig N and Narkis M (1981), ‘Sintering rheology of amorphous polymers’, Polym Eng Sci, 21 (17), 1167–70. Rusu G and Rusu E (2007), ‘Biodegradable anionic poly(esteramide)s. Physico-mechanical properties’, J Optoelectron Adv Mat, 9 (4), 958–64. Shih-Jung L and Chung-Yuan H (1999), ‘Factors affecting the warpage of rotationally molded parts’, Adv Polym Techn, 18 (3), 201–07. Teoh SH and Lau CY (1999), ‘Computer controlled rotational molding of a hollow femur for 3-D photoelastic analysis’, Int Polym Process, 14 (4), 370–76.


225

Thümmler F and Oberacker R (1995), Introduction to Powder Metallurgy, London, The Institute of Materials. Torres FG and Aguirre M (2003), ‘Rotational moulding and powder processing of natural fibre reinforced thermoplastics’, Int Polym Process, 18 (2), 204–10. Torres FG and Aragón CL (2006), ‘Final product testing of rotational moulded natural fibre-reinforced polyethylene’, Polym Testing, 25 (4), 568–77. Torres FG and Carrillo M (2003), Polymer Processing Society 19th Annual Meeting PPS19, Melbourne, Australia. Torres FG, Aguirre M and Aguila G (2002), ‘Powder processing and Rotational Moulding of Natural Fibre Reinforced Thermoplastics’, Proceedings of 18th Annual Meeting of the Polymer Processing Society, Guimaraes, Portugal. Torres FG, Cubillas ML and Dienstmaier JF (2004), Polymer Processing Society 20th Annual Meeting S–20. Torres FG, Carrillo M and Cubillas ML (2006), ‘Melt densification of natural fiber reinforced polymer powder beds’, Polym Polym Comp, 14 (7), 651–59. Wang B, Panigrahi S, Tabil L and Crerar W (2007), ‘Pre-treatment of flax fibers for use in rotationally molded biocomposites’, J Reinforced Plast Comp, 26 (5), 447–63. Wang WQ and Kontopoulou M (2004), ‘Rotational molding of polypropylene/ultra-lowdensity ethylene-α-olefin copolymer blends’, Polym Eng Sci, 44 (9), 1662–69. Wu JJ, Buckley CP and O’Connor JJ (2002), ‘Processing of ultra-high molecular weight polyethylene: modelling the decay of fusion defects’, Trans I Chem E Part A, 80, 423– 31. Yan W, Lin RJT, Bickerton S and Bhattacharyya D (2002), ‘Rotational moulding of particulate reinforced polymeric shell structures’, Int Conf on Advanced Materials Processing (ICAMP 2002), Singapore. Yan W, Lin RJT and Bhattacharyya D (2006), ‘Particulate reinforced rotationally moulded polyethylene composites – Mixing methods and mechanical properties’, Comp Sci Tech, 66 (13), 2080–88.

9 Blow moulding of polymers Y M A R C O, ENSIETA-LBMS, France and L C H E VA L I E R Université Paris-Est, France

Abstract: This chapter aims at showing the micro/macro interactions for polymer processing based on the blow moulding principle and how mastering the material microstructure can help in designing the final product. We want to cover the major aspects of design and tuning of process parameters for a blow moulded part, taking into account the microstructure. Moreover, coupling between process, material and final part is crucial for any designer and even truer for plastic parts. We therefore wish to present a product–process–material association, in order not to dissociate these crucial points. After a short introduction on the several processes using this common principle, we will therefore focus on the stretch blow moulding process, described for PET bottle manufacturing which is the main product and most used material. The reader may find in this chapter information on process technology, standard industrial tests, coupling between process, induced microstructure and final properties, and examples of how numerical simulations can help the parts designer. Key words: stretch blow moulding, crystallization, biaxial testing, WAXD, PET bottles.

9.1

Introduction

‘Blow moulding’ covers several industrial processes used to manufacture thermoplastic hollow parts. Their common basic principle is very similar to that of the old art of glass blowing. It starts from a plastic part which is already hot or has to be heated above its glass transition temperature (Tg). This is the parison or preform, whose shape is usually a closed cylinder. In this heated state, the material is easy to strain and able to accept very large deformations without rupture. Then, the material is blown against the walls of a cooled mould in order to quickly get the material under its Tg, to maintain the moulded shape. Mechanical stretching of the part can be done by air only or can be helped by an internal stretching rod. The part can then be removed for mechanical finishing, if necessary, and following operations. This basic principle lends itself to many techniques ranging from extrusion blow moulding to twinsheet thermoforming, and new processes will surely be proposed in the future. The three main manufacturing processes are extrusion blow moulding, injection blow moulding and stretch blow moulding. We will describe here their main steps and give an overview of their main characteristics. 226


227

In extrusion blow moulding, a cylindrical parison is produced by an extruder, using a squared die head. The material is therefore already hot and flows under gravity and extruder pressure. When the shape extruded is long enough, a two-part mould is closed around it, welding the tube ends, and a knife cuts the parison from the continuous flow. The mould is then moved sideways and a blow needle blows the welded parison into the shape of the chilled mould. After cooling, the mould opens and the part is ejected and trimmed. When the parts produced are big, several difficulties are encountered: the temperature gradient between the bottom and the top will be high and can prevent proper welding and blowing; extremely good melt strength in the resin is required to avoid creep and overstretching; the longer time spent at high temperature can lead to oxidization and high core/skin differences for semi-crystalline materials, etc. Intermittent extrusion blow moulding is then chosen, using either a reciprocating screw extruder (following a principle close to that of an injection screw) or accumulator head machines. It is therefore possible to accumulate enough resin for one part, so that the part ‘shot’ can be pushed out rapidly right before the mould closes around it. Extrusion blow moulding is illustrated in Fig. 9.1. Injection blow moulding presents two main steps, as the thermoplastic material is first injected in a heated mould around a core rod (as for injection moulding using a metallic insert). The injected preform looks like a laboratory test tube and its upper part presents the final shape of the future bottle and only the thick body will be modified in the following blowing step. The preform is transferred, on the rod, into a mould. During the transfer, the preform can be either conditioned to keep a temperature above Tg or cooled down during the injection step and reheated. Once in the mould used for blowing, the preform is expanded against the cold walls of the mould by compressed air blown throughout the opened rod core. After cooling, the mould opens and the bottle is stripped from the rod. We described here this process for packaging parts, but its characteristics of high volume production of small containers at fast cycle times also make it appropriate for small technical mouldings in engineering resins, for example like CVJ boots Extruder Air hose Hot knife

1

Parison

2

3

9.1 Extrusion blow moulding process.

4

228


(automotive drive joint covers) in polyester elastomers. Injection blow moulding is illustrated in Fig. 9.2. Like injection blow moulding, ‘Stretch blow moulding’ (also called reheat stretch blow moulding or injection stretch blow moulding) starts from an injected preform. For nearly all industrial applications this preform is cooled down in the injection mould and ejected. It is now an intermediate part that can be stored and even transferred to other plants. Before the blowing step, the preform is reheated above the material’s Tg (using infrared heaters mainly) and transferred into a two-halves blowing mould. The main difference with injection blow moulding is that now the preform is not only blown but also axially stretched by a stretching rod. The final bottle consequently exhibits increased strength due to biaxial molecular orientation. After cooling, the mould opens and the bottle is ejected. Stretch blow moulding is illustrated in Fig. 9.3. In 2002, blow moulding processes consumed about 10 wt% of all plastics worldwide (compared with extrusion 36 wt% and injection 32 wt%) and all kinds of hollow shapes can be obtained, allowing the manufacture of very different parts for very different markets. Even the limitations of the conventional blow moulding process, which is not well suited for long hollow parts in three dimensions, can be overcome by the use of 3-D extrusion blow moulding techniques (see, for example, the blow moulding processing manual by Dupont®). Still, the main market for blow moulding, as for the global plastics market is packaging (40 wt% for the global market), with a major product, the blow moulded bottle, sharing for example 55% of the US market for plastic containers (data from Impact Marketing Consultants, 2006). Indeed, 1

2

3

Rod

4

Preform

9.2 Injection blow moulding process.

1

Preform

2

3

Infrared heater Stretching rod

9.3 Stretch blow moulding process.

4

5


229

the market for blow moulded parts can be basically divided as follows. Extrusion blow moulding is widely used to produce containers of various sizes and shapes and is also adapted to make irregular complex hollow parts, such as those needed for automobiles, office automation equipment, pharmaceutical sectors, etc. The injection blow moulding technique is extensively used in the production of bottles for the food, beverage and pharmaceutical industries. Injection stretch blow moulding is more specifically used as soon as good mechanical or barrier properties require the benefits of bi-orientation. It is not possible in a single chapter to deal comprehensively with the numerous processes covered by ‘blow moulding’. We will therefore restrict ourselves to the major aspects of design and tuning of process parameters of a blow moulded part, taking into account the microstructure. Moreover, coupling between process, material and final part is crucial for any designer and even truer for plastic parts. We therefore wish to present a product– process–material association, in order not to dissociate these three crucial aspects. That is why we will focus in the following on the stretch blow moulding process, described for its main production and classical material, the PET bottle. The reasons are economic, technical and scientific: •

•

•

PET represents about 60% of US plastic bottle production (38% for HDPE, 2% for PP; data from American Chemistry Council, 2005) and used for injection blow moulding and stretch blow moulding. This material also presents the highest growth rate of any major plastic. Stretch blow moulding shows well the common principle and difficulties encountered in all blow moulding process and adds some specific problems due to the reheating step and biaxial stretching which make it more difficult to tune. The rate of stretch blow moulding machines is also among the highest in blow moulding manufacturing (up to 60 000 bottles per hour). Lastly, compared to other extrusion blow moulding processes, stretch blow moulding will provide products showing better dimensional quality and clarity. This book aims at showing the micro/macro interactions and how mastering the material microstructure can help in designing the final product. Stretch blow moulding is obviously the more interesting process from this point of view as the parts produced exhibit biaxial molecular alignment and crystallinity ratio evolution. The final material can then be optimized in order to enhance the mechanical and chemical properties and to make the final bottle meet the design requirements. That is the reason why stretch blow moulding leads lighter bottles than simpler blow moulding techniques, or can answer specific requirements like gas barrier properties for carbonated drinks and beer, or can resist the high temperature encountered during hot filling of pasteurized products.

230


In Section 9.2, we will describe the material, the process and the specific case of the PET bottle (markets served, specific requirements). Then, in Section 9.3 we will focus on the testing and design of the final product (standard tests, physical properties, numerical simulations for design assessment). In Section 9.4, we will present the numerical simulations of the process itself and investigate the evolution of the material microstructure during stretch blow moulding. We will then conclude on the design of blow moulded parts and on the future technical and scientific developments.

9.2

Process description and industrial part design requirements

We will describe here the details of the three items cited above: the material, the products and the process. First, we will discuss the material with its microstructure and typical behaviour in its different states, then the markets served and the associated requirements, and lastly the technology of the stretch blow moulding process, the process parameters and the tuning needed. We will conclude with the contribution of the numerical simulations to industrial part design, highlighted for normalized test simulation and for process simulation. For both cases, it is required to model the material behaviour and to characterize the model parameters, with a common industrial goal of making the cheapest bottle (i.e. the lightest preform).

9.2.1

Material

Overview Poly(ethylene terephthalate) (PET for short) is a semi-crystalline thermoplastic (saturated polyester family). It exhibits excellent tensile and impact strength, chemical resistance, clarity, processability, colour ability, quite good thermal stability (see Caldicott, 1999) and good recyclability (see Awaja and Pavel, 2005). Discovered in 1941 by Dickinson and Whinfield, it was first used (and is still used today!) for textile fibres (Dacron®, Tergal®) and bi-oriented films (Mylar®). In the 1970s the first stretch blow moulded bottles using PET were produced in the United States (Pepsi, 1975). This market then experienced an incredible growth, as detailed in Section 9.2.2, that explains the ever growing consumption of PET, from 2900 million tons worldwide in 1992 to more than 9 billion tons in 2002. Synthesis and chemical considerations PET is obtained by a polycondensation process from the association of dimethyl terephthalate with ethylene glycol. Figure 9.4 shows the PET repeating unit


O

H H

O

C

O C

C

231

O C

H H

9.4 Poly(ethylene terephthalate) repeating unit.

using both schematic and molecular representation (obtained using Rasmol®). The chain conformation stays in the benzene ring plane and can be either cisor trans-. It is usually considered that only the trans- conformation allows crystallization. For applications such as fibres and sheets, PET is used with quite low molecular weight (MW) and an intrinsic viscosity (IV) of about 0.6 dl/g. Stretch blow moulding requires higher MW and viscosity (IV between 0.75 and 0.85 dl/g). To give an example, a classical value of MW for carbonated soda is about 26 000 g/mol (9921W Eastman). For extrusion grade, the required MW and viscosity is even higher (IV up to 1.05 dl/g). One may find both PET homopolymer and PET co-polymers (the most common monomers are given in McDowell et al., 1999). The former leads to better gas barrier properties and presents a higher crystallinity ratio, allowing a better heat resistance, but the temperature range for the process is narrow, increasing the difficulty of the process. The latter is also used for PET bottles because of its lower crystallinity, improved ductility, better process ability and better clarity (Scheirs, 1998). The chemical nature of PET induces water sensitivity, which is to be taken into account. Water reduces MW through a hydrolysis reaction and acts as a plasticizer, lowering the glass transition temperature. Moisture contamination should be below 0.02% to avoid the MW reduction (Scheirs, 1998). The effect of water can be reduced substantially by proper drying but it is still important to evaluate water absorption for hot filling applications. As PET is used for food and beverage applications, its potential contaminants are of importance. The main one is acetaldehyde, exhibiting an apple taste and produced by the degradation of the macromolecules, under overshearing during injection or by hydrolysis. The classical limitations for bottles are of 4–6 ppm for soda bottles and 2–3 ppm for water bottles. As for water, the high volatility of acetaldehyde means that it can be minimized by processing under vacuum or by drying (Scheirs, 1998). The reader can refer with profit

232


to a recent review by Awaja and Pavel (2005) for more precise data on chemical aspects. Microstructure and thermo-mechanical aspects PET is a semi-crystalline thermoplastic. Its microstructure therefore exhibits a mix of an amorphous phase and a crystalline one, with a given ratio ranging from nearly zero to about 45% (fully crystallized PET). Figure 9.5 is an illustration of amorphous and semi-crystalline phases for a spherulitic crystallization. The more usual unit cell encountered for PET crystals is the one determined by Daubeny and Bunn (1954). As this book is also aimed at students, we will recall here some basics of polymer physics in order to make the following perfectly clear. As for every thermoplastic polymer, its thermo-mechanical behaviour can be understood by the study of molecular interactions. The first ones are the covalent links which are the strongest (about 450 kJ/mol for a single carbon link) and which hold the atoms of each macromolecule together. Then come the weak links (Van der Waals, hydrogen links, for example) that link the macromolecules to each other. These links are weaker (from 10 to 50 kJ/mol) but play a very important role as the chains are very long (multiple sites for weak links) and very numerous. These weak links act as gravitational forces: if the chains are close, the link is stronger than if they are distant. For PET, these links are mainly created between the benzene rings (see Fig. 9.4). In the following, we will use Figs 9.6 and 9.7 to illustrate the microstructure evolution, the temperature transitions and the states of the material with the associated mechanical behaviour. Figure 9.6 presents a typical DSC analysis for an initially amorphous specimen (see, e.g., Alves et al., 2002, Kong and Hay, 2002) and Fig. 9.7 presents the evolution of the Young modulus (or an indicator of the stiffness of the material) and the viscosity with temperature, obtained by DMTA (see, e.g., Rwei, 1999; Da Silva and Bretas 2000). For both charts, the material is

Amorphous phase

Crystalline phase 10 µm

9.5 Illustration of amorphous and crystalline phases.


233

Energy (mW/mg) 0.35 0.30

Exothermic reaction

Fusion peak

0.25 0.20

Glass transition

0.15 0.10

Crystallization peak

0.05 0 Tg ~79°C

–0.05

50

Temperature (°C) 100

150

200

250

9.6 Typical DSC analysis. Rubbery plateau

Liquid state

Degradation

Dynamic elastic modulus (MPa)

Glassy plateau

Tg

Temperature (°C)

Tf

Td

9.7 Evolution of the elastic modulus vs temperature.

initially nearly amorphous (very low crystallinity ratio) and under Tg. In this initial state, the chains can be oriented or not and the weak links hold the chains together. In the amorphous phase, all chains are distant from each other, and in crystallized zones they are close to each other, ordered in a crystal lattice. This is the glassy state, and no mobility of the chains is allowed. The material is rather elastic (isotropic or orthotropic depending on the chain orientation) and brittle, with a Young modulus ranging from 1300 MPa to 4000 MPa depending on the crystallinity ratio (as crystals acts as nano-fillers) and on the chain orientation. When we heat the material from the solid state, the first links that will break under heating are the weak links between distant chains. This transition is called the glass transition (Tg is about 79°C for amorphous PET, up to

234


100°C for fully crystallized PET) and affects only the amorphous phase (as the chains are closer in the crystalline phase their weak links are stronger). The chains of the amorphous phase are able to move, orient and to get closer to each other. We are now on the rubbery plateau, the material is viscoelastic with a low stiffness and a high ductility, and, depending on the temperature and strain speed, can exhibit a remarkable strain hardening effect. We will investigate deeper this material behaviour and its microstructural explanations in Section 9.4.1. This state of the material is of huge interest for industrial processing because it can be elongated to very large strains without rupture, under limited forces, and its microstructure is able to evolve. Moreover, the strain hardening effect is a real gift for the stabilization of processes like stretch blow moulding and for producing parts with nearly constant thickness. We will come back to the microstructure modification which is likely to occur in the rubbery state in the following paragraph. If we go on with the rise of temperature, at a given temperature the thermal energy is high enough to also break the weak links between the macromolecules in the crystalline phase. It is the melt temperature (Tm is about 250°C) which does not depend on crystallinity ratio or chain orientation. This melting of the crystals is also associated with the liquid–liquid transition. Above this temperature, the material is therefore fluid and exhibits a classical non-Newtonian (shear-thinning) rheological behaviour. If the temperature imposed increases more, it will now start to break the covalent links. This is the degradation temperature (Td is about 330°C) and the macromolecules will start to be damaged, with a drop in MW, mechanical properties, optical properties, etc., and with the creation of degradation products like acetaldehyde. If we now come back to the range of temperature between Tg and Tm, the chains of the amorphous phase are able to change their conformation and the thermal energy is low enough not to destroy the crystals that could be created. If enough time is given, the macromolecules will be able to get closer and to organize in crystalline zones, helped by the Brownian motion. This is the first way for the PET to crystallize, and is called thermal crystallization or quiescent crystallization. It can be observed during either slow cooling from the melt state or slow heating above Tg from the solid state. The obtained microstructure is spherulitic, which is classical for semi-crystalline polymers (Branov et al., 1970; Bassett, 1981; Haudin and Monasse, 1996). Crystalline lamellae are aligned along the radial direction, separated by amorphous sectors (see Fig. 9.5). A molecular chain can have several crystalline and amorphous zones, the shape of the spherulites depending on temperature (Benatmane, 1992) or on thermal gradient (Haudin and Monasse, 1995). The size of these spherulites ranges from 10–6 to 10–4 mm which is enough to opacify the material. As can be seen in Fig. 9.6, there is an optimal temperature for crystallization kinetics which is called the crystallization temperature. One can also imagine from Fig. 9.6 that the kinetics of thermal crystallization


235

are low for the blow moulding temperature range. This is true for isotropic amorphous material (see Marco et al., 2002a; Hieber, 1995) but several studies have shown that these kinetics are very sensitive to chain orientation (Asano and Seto, 1973; Smith and Steward, 1974; Fisher and Fakirov, 1976; Gupte et al., 1983) and that it cannot be totally neglected during stretch blow moulding. This slow and isotropic crystallization, leading to a drop of clarity, is to be avoided for classical applications but is still crucial for hot filling applications, as we will see in Section 9.2.3. The other means of PET crystallization by strain induction. The material stretching aligns macromolecules, gets the chains closer and induces a change from cis- to trans-conformation (Shen et al., 1991; Spiby et al., 1992; Lapersonne et al., 1992) which induces a partial crystallization. Benzene rings tend to align in a plane parallel to the principal directions of the mechanical solicitation. The chains are organized first in micellar structures (Haudin and Monasse, 1996) and then in crystalline lamellae. The limited dimensions of these structures (1 to 9 nm) keep the good transparency of the stretched and crystallized PET. Figure 9.8 summarizes the different steps and proposes a mechanism of crystal formation through a nematic and smectic mesophase (Asano et al., 1999). This strain-induced crystallization has been observed

Elongation

Amorphous state

Crystalline phase

Crystallographic pattern

Crystal

Molecule

Micelles

Lamellae

9.8 Different steps in the microstructure evolution of PET during elongation.

236


for various strain states: simple shear (Titomanlio et al., 1997; Pople et al., 2000), uniaxial tension (Salem, 1992; Lapersonne et al., 1991; Marco, 2003), and plane, biaxial or sequenced tension tests (Cakmak et al., 1986; Chang et al., 1993; Marco et al., 2002b), and for different kinds of solicitations: constant force, constant speed or constant strain rate tension tests (Le Bourvellec et al., 1986; Le Bourvellec and Beautemps, 1990; Vigny et al., 1997; Salem, 1998). Molecular mechanisms seem to be the same in all these studies but the induced microstructures shows differences in crystallinity ratio, amorphous and crystalline chain orientation and crystal size. This description of the material and its thermo-mechanical properties shows how much thermal and mechanical history are crucial to understanding what will be the induced microstructure and therefore enhances the strong coupling between process parameters, material changes and final properties. As illustrated in Section 9.2.3, we can also measure the difficulty of the task when it comes to predicting the induced microstructure for stretch blow moulding, where the history of elongations and thermal solicitation are complex.

9.2.2

Plastic bottle markets

In the 1970s the first blow moulded bottles using PET were produced in the United States. This kind of two-litre bottle was the first major application for PET, dedicated to carbonated soft drinks. Since then, PET has been incredibly successful and has imposed its qualities in numerous markets. Here are a few milestones: • •

• • • • • • •

1975 Pepsi launched the first two-litre bottle (with a PP cup glued to the bottom to ensure stability) and PET imposed its clarity and CO2 barrier properties. 1980 Mineral water, oils and flat beverages were also bottled in PET. New shapes had to be found to counter the lack of internal pressure. Stretch blow moulding is crucial for better mechanical properties and higher dimensional precision. 1981 The first machine producing 15 000 bottles per hour. 1986 PET was used for hot filling, using specific heat-resistant bottles. 1989 The first machine producing 40 000 bottles per hour. 1990 The ‘petaloid’ shaped base (see Fig. 9.21 in Section 9.3.3) replaced the glued base cup of carbonated sodas and was very successful for 500 ml containers. 1994 Preferential heating allowed blowing more complex shapes, including non-symmetrical bottles. 1995 The first two-cavity machine for small containers at a rate of 50 000 bottles per hour. 1997 The first machine allowing blowing, filling and closing.


237

The main goal today is still to make ever lighter bottles (the lightest 0.5 litre water bottle weighted 17 grams in 1998, 9.9 grams in 2007). Since the late 1990s and 2000, coating technologies and multilayer preforms have been developed to allow the packaging of new products (beer, pharmaceuticals, distilled spirits, automotive fuel tanks, etc.) in PET containers. Another recent advance was to produce panel-less hot fill containers (panels have been necessary until now, to counter the post-filling vacuum). Figure 9.9 gives in four charts some data on the economics of the plastic bottle market and its place compared to other packaging materials for beverages. Figure 9.9(d) presents the main markets served by PET packaging. Non-food applications cover custom containers for household, pharmaceuticals, industrial and personal care products. For all these markets, the PET bottle shows numerous qualities for a container: • • • • • • • •

Its mechanical strength allows a great variety of shapes, which is crucial for brand identification (for example, distilled spirits are increasingly bottled in PET bottles for this reason). Its low weight eases its transportation, from a customer and an industrial point of view. Its good optical properties (transparency and gloss) enhance the product. Its impact resistance is important for both customer security and transportation. It is easy to reseal. Its recyclability is a major argument in terms of both cost and the environment. Its barrier properties, dedicated to CO2 in its early days, are now very effective for nearly every kind of product. It is suitable for hot filling and aseptic applications.

The reader will find numerous details on the qualities and applications of PET bottles on the main manufacturers’ websites (Krones, Sidel, SIG, SIPA, Smiform, Amcor, etc.).

9.2.3

Technology of stretch blow moulding process

Figure 9.3 showed the main steps in stretch blow moulding and we present in this section the parameters of the different steps. The values of the parameters given in the following are to be taken as orders of magnitude only and cannot be used directly for processing. Injection of the preform Pellets of PET are first dried in a heating hopper (at 175°C for 4 hours) and then injection moulded using a classical injection machine. Humidity rate

Injection blow single machine process 0

20

40

60

80

100 % of respondent HDPE copolymer (HIC) Polypropylene

100 80 60 40 20 0

HDPE homopolymer (dairy) PET

Metal

Plastic

2007 Glass

Paper

(b) US beverage container demand (source freedonia Inc.)

(a) Top 5 blow moulding processes and materials in North America (source survey Plastic News 2003)

Other drinks 7% Beer 3% Juice 3% Non-food 5% Food 6% Thermoformed packaging 6%

120 Billions of units

2002

1997

LDPE

100 80 60 40

Carbonated soft drinks 38%

20 0

2002

1997 Metal

Plastic

2007 Glass

Paper

(b) US beverage container demand (source freedonia Inc.)

9.9 Four charts on economic aspects.

Water 32% (d) World market for PET packaging 2006 end use projections (source Pira International Ltd, Surrey, UK)


Stretch blow separate preform production process

Billions of units

Intermittent accumulator head monolayer Intermittent reciprocating screw monolayer

238

120

Continuous shuttle monolayer


239

has to be lower than 0.02%. An injection temperature of 270°C is usual; the number of cavities depends on the closure unit but is classically about 50 preforms per injection cycle. The mould is regulated at 8–15°C in order to cool the material very quickly from the melt. The induced thermal crystallization (see the point on microstructure in Section 9.2.1) is therefore very limited and the crystallinity ratio is lower than 3%. Another point to be checked is to limit the shear rate to prevent material degradation and acetaldehyde creation. Classical preform shapes can be found in Fig. 9.14 in Section 9.3.2. Detailed descriptions of the PET injection moulding process and parameters can be found elsewhere (Brydson, 1995; Rosato, 1997; Eastman, 1995). At the end of this step, preforms are used to feed the blow moulding unit, the principle of which is shown in Fig. 9.10. Preform heating Like most polymers, PET has low thermal conductivity. Heating techniques using convection or conduction not only require a long heating time but also cause heterogeneity in the microstructure between the skin and the core of the material. An alternative is radiation heating with infrared waves, which is the method commonly used for the industrial process. For a good wavelength (the optimal one for PET is about 1.2 µm), infrared waves will provide homogeneous heating even of thick preforms (about 3–4 mm). As the wavelength depends on the power given to the infrared heater (often 2–3 kW quartz lamps) and on the heating temperature, perfect power regulation is

Infrared heaters Moulds rotating unit

Infrared heaters

Bottles exit

Preforms infeed Moulds close around preforms and open to release the blown bottle

9.10 Stretch blow moulding process, from heating to packaging.

240


needed. A so-called reheating module therefore looks like a tunnel (about 2 metres long), with six to nine infrared lamps ensuring the precision of the longitudinal heating profile on one side and a reflecting wall on the other side. When the preforms are transferred in the first heating zone (see Fig. 9.10) they are turned upside-down in order to protect the neck from radiation, as this part of the preform is definitive and should not be modified by heating and blowing steps. As a preform passes in front of the lamps, it rotates around its axis in order to ensure homogeneous radial heating. At the end of each reheating module, the temperature is controlled by an infrared camera. Classical values for this temperature range from 95 to 105°C, which takes into account the cooling occurring during the short time between the exit of the reheating module and the closure of the blowing mould on the bottle. This is an average value but, as the different zones of the preform will not be stretched in the same manner during the stretching and blowing phase, it is necessary to regulate each infrared lamp separately to optimize the quality of the final bottle. When production starts, several zones are therefore defined with a pencil on each preform in order to check on the blown bottle what should be modified. In conclusion, it is crucial (and difficult!) to optimize the heating of the preform, because the range of acceptable temperature is narrow (if too low, the preform will break during stretching; if too high, it induces crystallization or the preform flows), because heating depends on many parameters (ambient temperature, preform storage conditions, reflections, ventilations, heater regulation, etc.) and because it should be precise for longitudinal temperature as well as for temperature gradient along the thickness (for example, the inner temperature should be higher than the outer one as the inner zone will be more stretched during blowing). Stretching and blowing steps At the exit of the oven, the preforms are seized by the neck, turned upsidedown and placed in a two-halves mould. The mould opening and closing, as well as the mould locking, are controlled by cams mounted on a common shaft. The blowing nozzle is introduced in the neck and ensures the guidance of the stretching rod. The bottle is then manufactured in three steps: 1. An axial stretching is imposed by the rod’s vertical displacement (Fig. 9.11). The zone under the neck is classically at a higher temperature and stretches first. The material is therefore moved downward. 2. At the same time as the preform is stretched axially, a light blowing (8 bars) is imposed to start radial elongation (Fig. 9.11). The delay between the first and second steps is a crucial parameter to balance the material share between the upper and lower parts of the bottle and to ensure thickness homogeneity. Waiting too long will lead to too thick a bottom


241

LP LP

Step 1

Step 2

HP

Step 3

9.11 Three steps in stretch blow moulding of PET.

and too thin upper part, while on the contrary, a too-short delay will lead to thicker upper zones and a thinner bottom. If blowing comes before stretching, the bottle will present a thin bottom with an injection point not well centred, as the stretching rod will not have ensured its radial position. As the PET exhibits a strain hardening effect during elongation, for a well-chosen temperature and strain speed the preform is stretched progressively. This remarkable effect explains the success of PET for stretch-blow moulding as it stabilizes a potentially unstable process and leads to quite constant thickness. As the material touches the mould, it cools quickly and sticks or slips along the walls, depending on its temperature. This step lasts till the material is pinched between the rod and the mould (Fig. 9.11). 3. As the axis of the blown preform is warranted, the bottle is blown quickly at higher pressure (30 to 40 bars) to obtain the bottle details and to ensure efficient cooling on the mould walls (Fig. 9.11). Finished bottles are gripped in the moulds before they open and are released at bottle exit. Industrial machines can be designed with a linear or rotary set of blowing units and their main assets are their productivity (output rates can reach more than 60 000 bottles per hour) and versatility (a wide variety of shapes for diverse applications: heat resistant bottles, wide mouth packages, flat or complex shaped packages; a wide range of sizes: from 0.25 litre up to 3 litres can be produced on standard equipment, and specially engineered machines can accommodate containers up to 10 litres in size; and easy change of material: production of PET or PP packages at similar rates with no modification

242


of machine configuration required). The reader may find more pictures and information in technical (and commercial) papers on industrial websites (Sidel, for example, presents detailed documents on SBO machines). We have described the main steps and the classical parameters for the major and most classical kind of stretch blow moulding processes, dedicated to water and carbonated soft drinks. In the following paragraphs, we will describe briefly three more specific points for bottle manufacturing: how to get complex shapes; how to increase heat resistance; and how to improve the barrier properties. How to get complex shapes: preferential heating process Specific features helping brand identification are one of the key factors explaining the use of PET for containers. But some applications (household products, sauces, distilled products) lead to complex shapes (wide panels and non-axisymmetric shapes). The blowing of a preform that has experienced a classical heating generates an axisymmetric bubble. If the final product is a flat bottle, a zone of the bubble is quickly blown and cooled on the mould, creating a ‘cold zone’, as the other zones are blown forward. This creates a longitudinal thick zone with a material showing a low draw ratio, and on the contrary, weaker zones in the end zones of the bottles. If a preform with a lower diameter is used, the cold zone is limited but generates high internal stresses, inducing the buckling of the panels. A technical solution, proposed by Sidel in 1994, is to heat longitudinal sectors of the preform in different ways. This is realized by the synchronization of the preform rotation in the oven with alternation of black and white zones on the reflector panels. At the exit of the heating modules, the preform then shows longitudinal sectors that have been preferentially heated. The preform is then transferred in the mould so as to present the hotter zones in front of the smaller dimension of the mould. Blowing then generates an ellipsoidal bubble (the hotter zone stretches more easily) and the material sticking and cooling first on the mould is also thinner. On the contrary, the polar zones in front of the higher radial dimension will be stretched from thicker zones of the bubbles and present the same thickness as the hotter zone of the bottle. There again, the strain hardening effect of the PET helps the process stability. Using this process, the defaults previously mentioned vanished and about 20% of material cost can be saved. Moreover, the versatility of the production can also be increased and a single preform can be used to manufacture up to six different complex bottles. How to increase heat resistance: heat crystallizing processes Several applications require heat resistance of the plastic bottle: hot filling is realized at a temperature between 75°C and 95°C, ensuring aseptic conditions


243

for both food and non-food (medical) products; pasteurization is realized on a cold-filled bottle that is heated several times in a heating tunnel; and some bottles are not recycled and may be reused (for example, office water fountains), which means washing at temperatures that can reach 75°C. The problem is that the Tg for amorphous PET is about 79°C. The quick cooling of the bottle on the mould has locked the chain orientation and induced crystallization, but it is an unstable microstructure: allowed to move again, the chains of the amorphous phase will tend to relax until they reach a statistical conformation (no preferential orientation) – see, for example, the yoghurt pot which is often made of amorphous plastic to limit energy consumption during reheating before thermoforming. An interesting experiment that can be carried out at home is to submit a yoghurt pot to a temperature higher than its Tg (110°C, typically) but lower than its liquid–liquid transition. The structure is then unable to melt, but after a while the chains of the amorphous phase (here 100%) will relax the orientation given by the process and maintained by the cooling under the Tg, and the structure will get back to its shape before the thermoforming, that is to say a plastic plate cut from an extruded part. The structure is then determined by the material thermo-mechanical ‘shape memory’. Consider now a PET bottle that has not been heat-stabilized: the first strains will occur at about 60°C, and after 90°C the chains of the amorphous phase will relax and get back to conformations close to their initial one. The product obtained then exhibits a shape between the shape of the bottle before heating and the shape of the preform used for the bottle manufacture. Nevertheless, you will never get back to the shape of the preform, as the bottle is partly crystallized (about 30% for classical applications) and as the Tg affects only the amorphous phase. The crystallization is not reversible until the fusion temperature is reached, which means the melt of the material. To increase heat resistance of PET bottles, the principle is to take advantage of the higher Tg induced by PET crystallization. Classical PET bottles are already crystallized by strain-induced crystallization during the stretching and blowing steps. This ratio can be controlled by draw ratio, draw rates and temperatures but cannot lead to a high enough crystallinity ratio for given dimensions of preform and final bottle. PET bottles for heat-resistant applications will therefore be highly crystallized by thermally induced crystallization, with the constraint of keeping optical clarity (see Section 9.2.1). Two main processes have been proposed by industry to answer this old problem (e.g., the 1984 patent by Sidel). The first is to induce the crystallinity directly in the blowing mould. The mould is regulated at temperatures between 80°C and 160°C while a high blowing pressure is maintained, which allows also the internal stresses to relax. The induced crystallinity ratio then depends on the time left (which limits production output rates). The bottles exit the

244


mould after brief blowing of cold air to get under the new Tg. The second solution is a two-step process, with a classical blowing unit, a specific reheating oven and another blowing unit. This process leads to higher heat resistance (up to 95°C). A first bottle is blown, whose volume is about 1.5 times the final volume and which is crystallized at about 25%. The bottle is then reheated, which induces a crystallinity ratio up to 45% and a recovery of bottle dimensions (length and diameter) to half of the initial ones. The neck is often also crystallized till it opacifies (see Fig. 9.12). The last step is another blowing in moulds regulated at 100°C (to relax internal stresses) to obtain the final shape. How to improve the barrier properties Gas-barrier properties are very important for beverage containers made of PET, as for those made of glass and metal. It is crucial to prevent oxidation of the ingredients by oxygen gas seeping in through the bottle, which can change the flavour of the product (beer, fruit juice, etc.), as well as to prevent carbon dioxide gas from escaping (the main limitation for carbonated beverages). Bottle permeability is more sensitive as the volume decreases, because the ratio between the volume and the surface area of the bottle is

9.12 The opaque aspect of the neck is due to thermal crystallization.


245

lower, which increases gas exchange and reduces shelf-life. Gas-barrier enhancement technologies are consequently in high demand (see, for example, Yamamoto et al., 2005). Several means are possible to improve the gasbarrier properties of PET bottles. The first is to improve the process parameters in order to optimize the thickness distribution and to reduce the amorphous zones which present lower barrier properties. Still, the improvement is limited to about 10%. Another way is to manufacture multi-layer bottles. These bottles are blown from co-injected preforms. The additional layers (EVOH or PA based polymers) are very thin but effective (about 8% of EVOH is enough to multiply barrier properties by five). However, co-injection remains difficult to master even today and limits this solution. The last major solution is external or internal coating. Started in the late 1990s, external coating is realized by spraying and curing various materials (for example, amine epoxy). Internal coating is more recent and one of the most promising processes is a DLC (diamond-like carbon) coating: see, for example, Ikeyama et al. (2007) or Boutroy et al. (2006).

9.2.4 Industrial and scientific concerns The description of the material’s features and of the injection stretch blow moulding process show several industrial and scientific concerns. Figure 9.13 is extracted from a Sidel document (SBO Universal Technical Document) and illustrates well the steps of a bottle design. The customer explains the concept he wishes to enhance, sometimes with no clear ideas of what can be done or what shape would be the best. Designers propose several numerical shapes and the specifications dedicated to the final bottle are defined. Structural analyses are then achieved, and when the final shape is validated, the preform shape is proposed, prototype moulds are machined, and the blowing feasibility is tested. The blown bottles are then tested and validated. This figure underlines the two main difficulties. First of all, the material properties are strongly affected by the parameters of the process (stretch rod speed, blowing pressure, preform field of temperature, delay between stretching and blowing, draw ratios, mould thermal regulation, etc.). Because predictive models are missing, industrial products are submitted to several tests (described in Section 9.3), which leads to high design costs as the bottles have to be blown. The second point is to optimize the design of the shape of the preform, knowing the shape of the final bottle. Industrial know-how is crucial today in this step but is unlikely to lead to a ‘good first time’ bottle for complex shapes or reach a very low weight. Scientific concerns arise naturally from these two points: •

How can a constitutive model be proposed to describe accurately the mechanical behaviour of the material during stretching and blowing (shape and thickness evaluation)?

246

Advances in polymer processing Start of production

Production moulds and custom parts

Laboratory control

Prototype mould

Blowing feasibility testing

Computer image Marketing samples

Specificaitons

Bottle shape design (2D and 3D drawings

Structural analysis (FEA)

9.13 Steps in the bottle design process.

• •

How can one understand and predict the final microstructure induced by the process? How can the microstructure be linked to physical properties?

In Section 9.3 we will describe the structural analysis and the laboratory controls proposed to answer the industrial product validation. In Section 9.4 we will focus on the scientific concerns.

9.3

Mechanical strength of poly(ethylene terephthalate) (PET) plastic bottles

We focus here on the final product: the bottle. It is the main point for the customer (final geometry) and is the starting point for the designer. Numerical simulation is of great help for industrialists in order to test the design of a new bottle before mould realization. Industrial tests such as the ‘burst test’, ‘top load’ etc., can be simulated, and the bottle validation thereby requires fewer experimental results. It appears that microstructural evolution in


247

poly(ethylene terephthalate) during a stretch-blowing process has an important influence on final mechanical properties. Good agreement between simulation and experiments is found only if appropriate induced properties are used for simulation. Anisotropic moduli, for example, must be used to simulate the ‘top load’ test. Here we present: (i) the classical tests preformed on bottles to qualify the product; (ii) some tensile measurements made on stretch-blow moulded samples to highlight the important influence of the process on physical properties; and (iii) a comparison between simulations and experimental results for several tests.

9.3.1

Standard industrial tests

Industrial tests (burst test, top load test, Carter test, permeability test, creep test, etc.) carried out on bottles and associated with the minimal performance to be reached allow the manufacturer to validate a bottle design and the processing parameters. Each test involves specific physical properties: strength, rigidity, permeability, transparency, etc., and satisfaction by all tests is a guarantee of product quality. Creep test 24 h at 38°C In order to make sure that the geometry of bottles for carbonated soft drinks is stable during storage while waiting for customers, a classic creep test is managed on full bottles at 38°C during 24 h. No excessive deformation must appear, otherwise the content level decreases and has a negative effect on the customer. The bottom of the bottle is especially studied because the deformation of the petaloid feet can induce instability of the bottle. Top load test Another validation test managed on PET bottles consists in applying a compressive loading on an empty bottle to test its axial resistance. The observed behaviour is approximately linear during a first step of the test, then a critical load leads to instability and the bottle collapses. Numerical simulation can easily replace this test and can be managed directly on the numerical model of the bottle, since the mechanical properties (especially the elastic modulus) of the material are well known. Burst test To test the strength of the bottle, the burst test consists in applying an internal pressure and increasing this pressure until cracks appear. Rupture can appear

248


in the cylindrical region of the bottle where the stress (proportional to RP/e where R is the bottle radius, e the bottle thickness and P the internal pressure) is highest, but cracks also frequently appear in the neck or in the bottom of the bottle. Carter test A normalized test is used to make sure that no rupture will appear during the lifecycle of the bottle. The petaloid base of the bottle, submitted to an internal pressure of 5.3 bar, is put in a 0.2% soda solution and no craze must occur before 10 minutes to validate the bottle. This test accelerates the damage process that occurs over a much longer duration.

9.3.2

Physical and mechanical properties of bottles

We focus here on a mechanical study based on four different preform geometries, injected with the same processing conditions. From the preform to the bottle, different elongation ratios can be defined: Longitudinal ratio: λ L = L  1 = e l  bi-orientational ratio: λ R λ L = D E λ E Circumferential ratio: λ R =  d

Here L and l are the curvilinear lengths of the bottle and the preform, D and E are the diameter and mean thickness of the bottle, and d and e are the diameter and mean thickness of the preform. Four different shapes of preform have been blown in the same bottle mould (Fig. 9.14) to get different biorientation ratios (Fig. 9.15). In this section we will study the influence of the draw ratios on PET moduli in the cylindrical central part of the bottle. The elongation ratios in the longitudinal and circumferential directions are given in Table 9.1. It is worth noting that lR values are quite similar because the different radii of the preforms are identical. The temperature and blowing parameters are identical from one preform to another. The only differences between those preforms come from their thickness and length. Rigidity of PET Specimens were cut from stretch-blow moulded bottles in both longitudinal and circumferential directions. We carried out tension tests at slow speed (V = 1 mm/s) using those specimens. Several tests have been achieved for each condition to check the good reproducibility of the results. Less than 5% deviation is observed from one test to another in the first part of the behaviour

Type 21

9.14 Geometry of the four studied preforms.

Type 31

Type 41


Type 11

249


φ27

R15

40

5

21

250

70

Niveau 605 ml

φ69.5±0.5

8 4.44

R10

φ68.5±0.5

4.44

91.91

228.8±1

R85

R10

24

R49.21

φ49

φ69.5±0.5 R6.5

9.15 Final bottle geometry (five-feet petaloid). Table 9.1 Elongation ratio

λL λR 1/λE

Type 11

Type 21

Type 31

Type 41

2.38 3.66 8.7

3.06 3.79 11.6

3.13 4.04 12.7

3.28 4.14 13.6

(elastic and starting of strain hardening). The measured Young’s moduli are reported in Table 9.2. We can see that these moduli are all greater than the one measured on injected specimens (Ea = 1200 MPa). The low experimental scattering (less than 5%) supports the conclusion that stretch blow moulding generates a large increase of the PET modulus. This testing session was completed by measuring moduli on ‘preblows’. These preblows were obtained by blowing the preform before longitudinal elongation had finished. Therefore, we could get more values of λL, but only


251

Table 9.2 Induced elastic modulus. EU is the elastic modulus along a 45° direction from the L or T direction

EL(MPa) ET (MPa) EU(MPa) νLT GLT (MPa)

Type 11

Type 21

Type 31

Type 41

1440 3020 2170 0.38 743

1790 3050 2280 0.38 774

1905 3100 2295 0.38 773

2015 3480 2320 0.38 759

(a)

3500

MPa 2500

3000

2000

2000

1500

1500

1000

1000

500

500

0

MPa

(b)

2500

1 2 3 4 5 6 Young’s modulus in L direction vs. L/L0

0 1 2 3 4 5 Young’s modulus in T direction vs. R/R0

9.16 Young’s modulus in longitudinal and circumferential directions.

one λR value was obtained because the same shape of preform and mould was used for all these preblows. All these results are plotted in Fig. 9.16. In Fig. 9.16(a) we can see that there is a maximum value of longitudinal modulus around λL equal to 4.5; no more benefit can be gained by increasing this parameter. This optimum value is not reached for the circumferential modulus (Fig. 9.16(b)). Greater stiffness in that direction can be obtained by increasing the λR value. A second conclusion a rises from comparing the two best-fitted curves: for a similar elongation ratio (λ = 4, for example) we notice a big difference between longitudinal (EL = 2100 MPa) and circumferential modulus (ET = 3200 MPa), which leads to an orthotropic behaviour. It can be analysed as an effect of the strain biaxiality or more likely as a strain rate effect which is different in the longitudinal and circumferential directions (about 10 times greater in the circumferential direction). Direct measurements on blown bottles Measurements have been made on two-litre Coca-Cola bottles using a digital correlation technique. The cylindrical part of the bottle was painted black

252


and sprayed with white spots (see Fig. 9.17). Internal pressure was increased from 0 to 7 bars and pictures were taken every bar. The bottle diameter was measured on each image and presented a linear evolution from 1 to 4 bars but non-linear as 5 bars were reached and afterwards. The last observation is linked to the yield stress of the PET, and the first nonlinearity can be explained by the initial shape which is not perfectly circular as illustrated in Fig. 9.18. The value of the diameter measured from the first picture must be corrected as soon as the pressure increases. The correct value is φ0 = 92.4 mm. The digital correlation technique limited to the area of the picture where the bottle surface is normal to the camera direction gives both circumferential and longitudinal strains εrr and εzz (see Fig. 9.19). The measurement dispersion on the strain measure is evaluated as 3 × 10–4, which is clearly below the measured strains. Longitudinal and circumferential stresses can be estimated respectively as σz = PR/(2e) and σr = PR/e from the internal pressure P. For R = 92.4 mm and e = 0.35 mm, identification of the Young’s modulus could then be achieved from Hooke’s relations:

ε tt =

σ tt – νσ zz σ – νσ tt ; ε zz = zz E E Pressure

CCD camera

d0

φ0

Bottle diameter (mm)

9.17 Cylindrical zone evolution under pressure. 97 96 95 94 93 92 91 90 0

1

2 3 4 5 Internal pressure (bar)

9.18 Bottle diameter evolution vs internal pressure.

6

7


253

0.12 0.10

ett ezz

Strains

0.08 0.06 0.04 0.02 0 0

2

4 Internal pressure (bar)

6

8

9.19 Circumferential and longitudinal strains measured by the digital correlation technique.

These two relations lead to two different values of the Young’s modulus (3970 MPa or 1530 MPa). This confirms the anisotropy of the material observed from tension test results. From an orthotropic elastic behaviour, one can identify the elastic characteristics from the blowing of the bottle:  1  ε z   Ez   =  νz  εt   –  Ez

νt Et 1 Et

–

  σ z    σ t   

where Et = 3650, MPa, νt = 0,47, Ez = 2430 MPa and νz = 0.31. These values are identical to the maximum values identified from tension tests managed on specimens cut from blown bottles. Elastic strength and rupture stress Yield stress σe and rupture stress Rm were measured during tension tests. Specimens cut from the bottles blown using the four different types of prefoms were tested and the yield stress and the stress at rupture were determined from the above relations:

σe =

Fe F ∆LR ; R = (l + ε R ) m ; ε R = L0 S0 m S0

Anisotropy is lower for these characteristics and for each property, σe or Rm, the results measured for longitudinal or circumferential specimens plotted respectively versus λL and λR are superposed (Fig. 9.20). A large dispersion (more than 100% in some cases) is observed on ultimate stress and breakout strain. Yield stress appears to be a monotonous function of the elongation ratio. The rupture stress highlights an optimum value for an elongation up to 3.5. One can explain the decreasing tendency of the ultimate stress by the orientation

254

Advances in polymer processing 180.00 160.00

Yield stress (MPa)

140.00 120.00 100.00 80.00 60.00

y = 13.532x2 – 43.924x + 92.502 R2 = 0.9607

40.00 20.00 0.00 1.00

1.50

2.00

2.50 3.00 Elongation ratio

3.50

4.00

4.50

400.00

Rupture stress (MPa)

350.00

y = – 33.417x3 + 238.05x2 – 450.91x + 342.13 R2 = 0.5745

300.00 250.00 200.00 150.00 100.00 50.00 0.00 1.00

1.50

2.00

2.50 3.00 Elongation ratio

3.50

4.00

4.50

9.20 Influence of elongation ratio on yield stress and rupture stress.

of the macromolecules that reduces the ductility of PET at ambient temperature (εr ≤ 0.3). The dispersion observed comes from the brittle mode of rupture.

9.3.3

Numerical simulations and examples of industrial part design

Numerical simulations help the designer to define the optimal shape (minimal thickness for permeability, preform geometry, petaloid shape of the bottom, etc.) but necessitate that mechanical properties of the bottle are measured. We will illustrate this contribution on some industrial and academic examples. In this section we will simulate the ‘top load’ test, at ambient temperature, with an industrial finite element software (Ideas). Firstly, we will compare


255

the simulated and experimental values of the critical buckling load. Accurate values of shear modulus GLT for the simulation have been evaluated by doing complementary tests on specimen cut from the bottle along the 45° direction. Assuming orthotropic behaviour and a constant value of the Poisson ratio, we can calculate the shear modulus by the relation: GLT =

1 4 – 1 – 1 – 2ν LT EU ET EL

All values given in Table 9.2 are used to describe the elastic behaviour of the cylindrical and conical parts of the bottle. On the top and bottom parts, we assumed that the material was isotropic and both values of Ea (1200 MPa) and νa (0.38) are used in the finite element simulation. The mesh was constituted of 4180 shell elements with about 15 000 d.o.f. and was generated from CAD files of the Sidel Process Engineering ‘Bottle specification’ department. The thickness was measured on blown bottles but could be easily obtained through a blow-moulding simulation. Boundary conditions were imposed on the feet of the petaloid bottom (all d.o.f. set to 0). On the top part of the mesh we imposed a displacement along Z to simulate the ‘top load’ test. As a classical treatment of buckling problems, we first solved the elastic problem of compression of the bottle as a linear study, and we then preformed a second, non-linear simulation taking into account large displacements in order to determine the critical load and the deformed shape of the bottle. Table 9.3 compares experimental values of the maximum load during the ‘top load’ test with the values obtained by finite element simulation for the different preforms. A good correlation is observed between experiment and FE simulation, both on the global value of the critical buckling load (less than 10% difference) and on the deformed shape of the bottle. We have a very good agreement between prediction and simulation. To conclude this section, we can say that industrial software can provide good prediction of bottle performance as far as accurate values for mechanical properties can be specified. The problem is that these properties cannot yet be predicted because, as far as we know, no behaviour law takes into account the evolution of microstructure during the blow-moulding simulation.

Table 9.3 Experimental and simulated top load comparison

Measured load (kg) Calculated load (kg) Error (%)

Type 11

Type 21

Type 31

Type 41

28.5 29.5 5%

35.7 32.3 10%

33.4 30.2 10%

38.5 35.6 8%

256


Yield stress or rupture stress gives important information to predict the strength of a bottle under internal pressure, as in the burst test, for example. Bottles are full of water and pressure increases until the bottle blows up. Rupture pressure is given in Table 9.4 and compared to a theoretical pressure estimated from the relation

σ θθ = PR e Other terms of the stress tensor (σzz and σrr) are smaller than σθθ and can be neglected in the estimation of the ultimate pressure. If rupture appears in the cylindrical central zone of the bottle, the ultimate pressure is obtained by the relation PR = R ⇒ P = e R m e R m Estimation is quite good for bottle type 11, whereas for other bottles the rupture appears near the injection point and the estimation of the stress field necessitates a finite element simulation on the 3-D shape of the bottle. Such calculations have been managed using a finite element code (MSC® Software). Difficulties arise because one must model the thickness distribution which is difficult to measure in the petaloid bottom. Another difficulty is due to the estimation of the local rigidity: feet have not been submitted to a fast biaxial loading like the central zone and the increase of rigidity is not as important as in the central zone. Figure 9.21 shows the thickness distribution (left) which decreases regularly from the injection point to a radius from where the thickness is supposed constant. In the zone where the thickness varies, the PET is modelled as an isotropic material with a Young’s modulus of 1200 MPa. Elsewhere, the material is modelled as orthotropic and the characteristics used for the simulation are identified on the blown bottle. The equivalent strain value near the injection point is about 1.3% for a pressure of 5.3 bar. One can note that the injection zone is submitted to a nearly equibiaxial strain:

Table 9.4 Ultimate pressure compared to analytical estimation. Rupture appears in the bottom near the injection point for bottles marked with a * and not in the cylindrical zone

Measured pressure (bar) Calculated pressure (bar) Error (%)

Type 11

Type 21

15.5 17.2 10%

13.3* 8.6 35%

Type 31 12.6* 10.2 20%

Type 41 13.1* 9.5 27%


257

9.21 Thickness modelling (left) and equivalent strain (right).

9.22 Initial and final pictures of a petaloid bottom.

ε ε =  0  0

0 ε 0

  ⇒ ε = 2ε eq  ~ – 2 ε  0 0

This means that according to the finite element simulation (FES for short), the strain value ε is only about 0.65%. With a 1200 MPa Young’s modulus, that pressure leads to an 8 MPa stress, corresponding to an internal pressure of 5.3 bar. The ultimate stress of amorphous PET is about 100 MPa, which would therefore lead to the imposition of an internal pressure equal to 100 × 5.3/8 = 66 bar to reach material rupture, which is clearly out of proportion. The strain value predicted by finite element simulation has been compared with an experimental strain measurement made from correlating digital images near the injection point. The displacement field components u and v are obtained by comparing the two photographs in Fig. 9.22, and strains are calculated using the relations:

ε xx =

∂u  ∂v ∂u  , ε = ∂v , ε = 1  +  ∂ x yy ∂ y xy 2  ∂ x ∂ y 

258


Principal strains are plotted in Fig. 9.23 and we must focus only on the central region near the injection point because other values are not correct, as the bottom is not a plane surface, so that the w component has an influence on strain which is not corrected here. One can see that experimental values are evaluated to εΙ = εΙI = 4% which is clearly higher than 0.65%. With a 1200 MPa modulus, stress is equal to 48 MPa and the applied pressure (here 5.3 bar, as previously for FE simulation) may be multiplied by 100/48 to generate rupture. This estimation leads to about 11 bar which is a compatible value with burst test results. Improvements have to be made to achieve more accurate numerical simulations, especially for the estimation of the thickness distribution and of the local elastic modulus. These two parameters have an important influence on the stress estimation in the injection point zone where rupture appears. This is still an open research subject and we will present some developments to improve these predictions in the following section.

9.4

Numerical simulation of the process and microstructural changes during stretch blow moulding

As the final shape of the bottle is defined according to final mechanical requirements, it is time now to define the geometry of the preform, which is a key factor for mechanical and economic objectives. The behaviour of the material under representative conditions of temperature and mechanical solicitations is needed and can thereafter be used for process numerical simulations. The industrial and scientific concerns here are to be representative of the material flow during the process and to be able to predict accurately the induced microstructure.

9.4.1

Mechanical behaviour of PET under uniaxial and biaxial tension

In Section 9.3 we have presented the mechanical behaviour of PET at service temperature, that is to say under the Tg and in its glassy state. We are dealing here with the rubbery state of PET, that is to say between Tg and Tf, which is so useful for forming processes as the forces needed to stretch the material are low, the material is able to accept large deformations without damage and, more importantly, its microstructure is likely to change. These microstructural changes are important for two reasons: they induce the crucial strain hardening effect stabilizing the stretching and blowing process (as described in Section 9.2), and they help tune the material reached in the final bottle (mechanical, optical and barrier properties as described in Section

Principal deformation II

y coordinate x coordinate

x coordinate

259

9.23 Principal strains measured from petaloid bottom of PET bottle.


y coordinate

Principal deformation I

260


9.3). We will focus, of course, on the range of temperature used for blow moulding. The mechanical behaviour of PET above its Tg has been widely studied over the past 20 years for various kinds of loading. Figure 9.24 presents experimental curves obtained for uniaxial tension tests achieved for different temperatures and tension speeds (Marco et al., 2002a). Tension tests have been carried out at different temperatures (from 80 up to 105°C) for a wide range of initial elongation speeds (from 0.02 to 2 s–1). Each specimen is stretched from 50 mm up to 200 mm (λ = 4). We assume in this part that local strain and stress can be deduced from global displacement and force. We compared local strain evaluated from markers on the specimen with global values and validated this classical assumption except during the quick transient phase of necking propagation. The uniaxial logarithm deformation ε and Cauchy stress σUT come out therefore as:

λF λ = L and ε = ln λ ; σ UT = F = L0 S S0 with L and S the actualized length and cross-section, L0 (50 mm) and S0 (40 mm2) the initial values, and F the experimental force. Figure 9.24 highlights the classical opposing effects of temperature and tension speed variation on the PET behaviour and on strain hardening effect (starting with an elongation of 2.5 and 3). Each graph presents the mean behaviour curve and the lower and upper bound for one temperature and for one tension speed condition. One can see that dispersion occurs essentially during the strain hardening phase of the test and this dispersion is important (about 20% of the mean stress value). For the considered temperature range, the mechanical response exhibits: • • •

a strong viscous dependency, illustrated by the tension speed sensitivity an elasticity, illustrated by the partial stress relaxation visible as soon as elongation stops a very visible strain hardening effect for restricted ranges of temperature and tension speed, starting when a critical elongation is reached (depending on temperature and tension speed).

Uniaxial solicitation affords precious information for model identification and validation but, due to the high anisotropy of molecular chains and because of the complex strain paths experienced by the polymer during the industrial processes, it is obviously necessary to investigate the mechanical response for different testing conditions. Plane strain tension or biaxial tension is then achieved in order to validate accurate models for process numerical simulations and to understand better the influence of thermo-mechanical solicitations on microstructure evolution. Still, this is not an easy task, taking into account that the material has to be heated, that induced deformations are often non-

Cauchy stress (MPa)

T = 80°C, V = 15 mm/s

T = 100°C, V = 15 mm/s

T = 105°C, V = 15 mm/s

30

30

30

30

20

20

20

20

20

20

10

10

10

10

10

10

0

0.5

0

1

0

T = 80°C, V = 50 mm/s

0.5

0

1

0

0.5

0

1

0

T = 90°C, V = 50 mm/s

T = 85°C, V = 50 mm/s

30

30

30

20

20

20

20

10

10

10

10

0

0.5

0

1

T = 80°C, V = 100 mm/s

0

0.5

0

1

0

0.5

1

0

1

0

0.5

1

0

0

0

0.5

30

30

30

30

20

20

20

20

20

20

10

10

10

10

10

10

0

0

0

0

0

0

0.5 1 Log strain

0

0.5 1 Log strain

0

0.5 1 Log strain

0

0.5 1 Log strain

0

0.5 1 Log strain

9.24 Uniaxial tension tests for various temperatures and tension speeds.

261

0

1

1

30

0.5 1 Log strain

0.5

T = 90°C, V = 100 mm/s T = 95°C, V = 100 mm/s T = 100°C, V = 100 mm/s T = 105°C, V = 100 mm/s

T = 85°C, V = 100 mm/s

30

0

0

T = 95°C, V = 50 mm/s

30

0

0.5


Cauchy stress (MPa)

T = 95°C, V = 15 mm/s

30

0

Cauchy stress (MPa)

T = 90°C, V = 15 mm/s

T = 85°C, V = 15 mm/s

30

262


homogeneous and that sequenced strain paths are the most relevant for injection blow moulding. Numerous technical solutions have been proposed to achieve biaxial tension tests on polymers. The first kind is based on industrial machines or uses an experimental set-up very close to the industrial process: • •

•

Film stretching in two directions, using industrial machines (Faisant de Champchesnel et al., 1997; Vigny et al., 1997) and combining constant speed or constant force tension tests. Blowing of initially flat (and usually circular) polymer sheet using hydraulic or pneumatic pressure. In this kind of testing, the polar zone is usually observed because it exhibits nearly equi-biaxial elongations. These setups are quite common and easy to use. See Treloar (1944), Hart-Smith (1966), Ogden (1972), and more recently Feng (1976) and Verron (1997) for elastomeric and thermoplastic materials. Stretching and blowing of a cylinder. The original tests were achieved by Alexander (1971) on latex, and this principle was used more recently by Benjeddou et al. (1993) to characterize numerous rubbers. More specifically, some stretching and blowing tests have been achieved, on industrial machines (Schmidt, 1995; Rodriguez-Villa, 1997) or in a laboratory set-up (Cakmak et al., 1985; Haessly and Ryan, 1993; Gorlier, 2001). This kind of testing can be realized in a mould or by letting the bubble inflate freely.

Other tests use mechanical set-ups less close to industrial solicitations: •

•

•

Biaxial plane tension testing. The plane specimen can be stretched in one or another of the two principal directions, with independent forces and speeds. The specimen can be thick (Obata et al., 1970; Meissner, 1987; Sweeney et al., 1997; Marco, 2003) or thin (film-like) (Chandran and Jabarin, 1993; Chang et al., 1993; Buckley et al. 1996; Mathews et al., 1997). These tests are rare because it is difficult to ensure homogeneous strains and because the grips are complex to design and to use. Plane compression testing. The principle is to compress a prismatic specimen along one direction, while keeping constant its original length along one of the orthogonal directions. These tests have been used for elastomers (Arruda and Boyce, 1993) but also for PET (Bellare et al., 1993; Boyce et al., 2000). Plane tension (or pure shear) testing. This test is widely used and is very common for rubbers because it is easy to achieve. The principle is to use a specimen with a small height compared to its width and to impose a stretching in the height direction. It is commonly supposed that the solicitation is close to plane strain (or pure shear), even if this assumption becomes quickly false for too high elongations (Chevalier and Marco, 2002).


263

In the following, we will present curves obtained for biaxial solicitations. The biaxial tests were carried out on a triaxial testing machine named Astree (LMT-Cachan) using cross-shaped specimens (see Fig. 9.25). The evolution of the strain field could then be determined thanks to a home-developed cross-correlation technique (Chevalier et al., 2001), implemented in MatlabTM. For more detailed description, the reader is invited to refer to Marco (2003) and Marco and Chevalier (2008). In the equi-biaxial tension test, the square specimen is simultaneously stretched in both directions 1 and 2 with the same displacement, actuator control allowing the centre of the specimen to be kept motionless. The principal elongations λ1 and λ2 in the central region of the specimen are measured equal and validate the equi-biaxial state in this zone. Assuming that the stress components are quasi-uniform in the diagonal section, we can estimate the biaxial Cauchy stress σBT from the biaxial loading force F: Grips position

Edge radius

50 mm

Heated and stretched zone

Clamping holes (b)

(a) F1

U 1 = U2 F2

F2 F1

λ=1

λ = 1.5

λ=2 λ=3 (c)

9.25 (a) Biaxial tension testing machine; (b) specimen design; (c) different step of the deformed shape.

264


σ BT =

2F eL

where e is the current specimen thickness and L the current length of the specimen diagonal. Figure 9.26 illustrates, for biaxial tension tests, that the material exhibits the same characteristic behaviour with opposing influences of temperature and speed and the same strain hardening effect. One can see that the strain hardening effect occurs sooner and that the level of stress observed for biaxial tension tests is classically higher than for uniaxial tension tests. Similar results are obtained for the plane strain tension test, which leads to stress levels between simple tension and biaxial tension levels. In Section 9.4.2 we will briefly review several models used to simulate the stretch blow moulding process. We will describe in Section 9.4.3 the microstructure evolution using in-situ and ex-situ measurements and the physically based explanations that can be proposed for these mechanical responses.

9.4.2

Numerical simulations of the process

At the process temperature the PET highlights a very strong elastic behaviour. Interrupted free-blowing tests show that the blown bottle collapses when internal pressure is stopped. Since the behaviour exhibits a strong strain hardening effect, several authors have proposed the use of hyperelastic modelling, as in Verron (1997), for example. This is clearly not satisfactory because the influence of speed is not taken into account. The effect of speed can be taken into account as in Buckley and Jones (1995), Gorlier (2001) and × 107

× 107

Tension speed 100 mm/s

3

1.8

80°C 2.5

Cauchy stress (MPa)

Test temperature 90°C

2

100 mm/s

1.6 1.4

2 90°C

30 mm/s

1.2

1.5

1 ✩

0.8 1

100°C

✩

0.6 0.4

0.5

✩ ✩ ★ ★

0.2 0 1

1.5

2

2.5 3 3.5 Elongation

4

4.5

0

1

1.5

✩ ★

✩ ★

✩ ★

✩ ★

15 mm/s

★

★

2 2.5 Elongation

3

3.5

9.26 Biaxial behaviour: (a) influence of temperature; (b) influence of speed.


265

Marco (2003), for example, by using elastic characteristics depending on the strain rate, which makes no sense in terms of a microscopic mechanism. Viscoplastic modelling Considering the monotonous evolution of the strain during the blowing process and the quick cooling of the material when coming in contact with the mould, several authors use simple viscoplastic modelling, for example Schmidt (1995), Champin (2007) and Chevalier and Marco (2006). Identification of the material characteristics can be made from the experimental data presented previously from uniaxial and biaxial tensile tests managed on PET specimens at a temperature slightly higher than the glass transition temperature Tg. The strain hardening effect observed during tension can be related to the straininduced modifications of the microstructure of PET. According to the viscoplastic model: K(ε) = K0 exp (aε 3 + bε 2 + cε + d) where K is the factor related to hardening; it varies exponentially with the equivalent strain. The characteristic values of the material parameters for this model are K0 = 0.333 MPa.s m, m = 0.4, a = 3.65, b = –7.6, c = 6.64 and d = –0.099. Considering low speed in uniaxial tension tests (V = 15 mm/s), one can study the influence of the temperature on the strain hardening effect. From mean behaviour curves plotted in the same log chart, it appears that a shift aT depending on temperature allows an approximate superposition. In that case, the strain hardening function K(ε,T) for a temperature T can be obtained from the K90(ε) function which is the strain hardening function coming from 90°C uniaxial tension tests: K ln  K  = ln  90°  + a T ⇒ K ( ε , T )  K0   K0  = exp ( a T ) K 90° ( ε ); log( a T ) = –

C1 ( T – 90° ) C2 + T – 90°

with C1 = –1.62 and C2 = –26.8°C. Shift factor evolution aT is identified on the test temperature range by a classical WLF relation. It is important to note that the influence of temperature is negligible near the glass transition temperature (near 80°C) but much more significant when the temperature rises. As a consequence, if thermal regulation presents a dispersion of a few degrees, this will not have much effect on the PET behaviour if the temperature is near the glass transition but it will have a much greater effect at higher values. The best solution should be to use viscoelastic modelling coupled with

266


microstructure changes but this modelling is still an open research subject. Even though several authors have made propositions, the use of such modelling in numerical simulations of the industrial process remains difficult and unstable. Numerical simulations and validation Numerous authors have proposed numerical simulations of the stretch blow moulding process (e.g. Champin, 2007; Yang et al., 2004) using finite elements. Figure 9.27 shows an FE mesh of a part of the preform. Taking into account the importance of geometrical transformations undergone by the specimen (elongation goes from 1 to 4), much distortion of elements appears and frequent remeshing is needed. The choice of the use of the natural elements method rather than the finite elements method for numerical simulations may be interesting, so that the complete evolution may be done on the same node set. The viscoplastic model is extended to a 3-D formulation with both definition of the classical equivalent strain rate and a less classical definition of the equivalent strain:

= 2ηD – pI with η = Kγ˙ m –1 and K = K 0 e ( a ε

3 +b ε 2 +c ε +d )

1

γ˙ = (2D:D ) 2

ε = sup ( ε i ), i = 1, 2, 3 i

Taking into account the very high viscosity of PET, body and gravitational forces may be neglected. In the general 3-D incompressible case, we have to solve the following problem: r r r div( ) = 0 in Ω ; ⋅ n = F on ∂Ω ; div( V ) = 0 in Ω r The variational formulation related to this problem is as follows: find V r cinematically admissible and V ∈ H1(Ω), p ∈ L2 (Ω) such that for all r* V ∈ H 01 (Ω) and all P* ∈ L2(Ω):

9.27 Finite element mesh of a portion of preform (17 000 elements, 4000 nodes) (Champin, 2007).


267

r r  – :D * dV + F ⋅ V * dS = 0 r r  Ω ∂Ω with D * = 1 (grad ( V * ) + grad ( V * ) T )  r 2  p * div ( V ) d V = 0  Ω

∫

∫

∫

We introduce the material behaviour: – ˙ m –1 D : D * dV +  Ω 2 Kγ  r  p * div ( V ) d V = 0  Ω

∫

∫

r p div ( V * ) d V +

∫

Ω

∫

∂Ω

r r F ⋅ V * dS = 0

For the simulation of free-blowing we use the axisymmetry assumption (ASA). We solve the problem using a pressure and velocity formulation. For incompressible material, numerical problems arise: the numerical gap between values of the pressure terms and the velocity terms is too important and leads to an ill-conditioned problem. An alternative is to use the technique of penalization: the hydrostatic pressure is replaced by a term proportional to the divergence of the velocity, as in the following expression: r 2ν p = –αη div ( V ) with α = (1 – 2ν )

ν is the equivalent of the Poisson ratio in elasticity. To reproduce a behaviour close to incompressibility, authors usually take ν = 0.49. The constitutive law and the weak formulation of the problem become: r σ = 2ηD + αηdiv ( V ) I –

∫

Ω

∫ + ∫

2 Kγ˙ m –1 D : D * dV +

Ω

∂Ω

r r α Kγ˙ m –1 div( V) div(V * ) dV r r F ⋅ V * dS = 0

The ASA and cylindrical coordinates (r, θ, z) are used to simplify and solve the problem above: –

∫

S

∫ + ∫

r 2 Kγ˙ m –1 D : D * dS +

r r rα Kγ˙ m –1 div( V) div(V * ) dS

S

∂S

r r rF ⋅ V * dl = 0

where r is the radius (i.e., the distance between a point x and the symmetry axis). The problem is discretized using the C-NEM shape functions (Cosson et al., 2007) and solved by the Newton Raphson method. This leads to the following non-linear system:

268

Advances in polymer processing m –1    T T BV ) 2 (2B T B + α A T A ) dS  V =   rMk (2V B ~ ~ ~ ~  S  

∫

r  rN T ⋅F dl  ∂S 

∫

with:    B( x ) =  ~     

φ1,r φ1 / r 0 1 φ  1,z 2

0 0 φ1,z

… … …

 1   φ  2  1,r

…

φn,r φn / r 0  1   φ  2  n,z

      1    φn,r   2  0 0 – φ n ,z

A(x) = [φ1,r + φ1/r φ1,z … φn,r + φn/r φn,z] With the C-NEM, simulations (600 nodes) of the free blowing of a PET preform were carried out with a unique set of nodes. Those simulations highlighted the importance of the initial temperature distribution along the preform to estimate the final thickness of the bottle. In Fig. 9.28 we can see two simulations, one with a uniform temperature repartition and another with a higher temperature near the neck than along the rest of the preform. Taking into account the elongation rod, one can use this simulation tool to manage an optimization of the stretch-blow moulding process. For example, the study of the influence of the initial temperature on the thickness homogeneity (Fig. 9.29 implies taking into account heat flow between the bottle and the mould to improve bottle design. Our simulations showed that an isotropic representation of the PET behaviour is not sufficient to model the effect of the material orientation. Work is currently in progress to include orthotropic behaviour, where hardening is a function of the principal strain directions and of the principal strains. This may be done by inverse identification from numerical simulation compared with the ‘movie’ of the process (see Fig. 9.30). Besides these purely numerical problems, another objective is to give a microstructural explanation of the behaviour. The material is strongly modified during the process, as illustrated both on industrial parts and in academic specimens using several means for microstructural investigations. A coupled approach is therefore needed, which is the future for stretch blow moulding.

9.4.3

Microstructural changes induced by stretch blow moulding

Microstructural parameters and experimental techniques In the first part of this section, we wish to recall the parameters that are classically used to describe the microstructure and the major techniques


269

90 80 70

–0.2

–0.3

80

–0.35

70

–0.4

60

–0.6

–0.4

60

–0.45 50

–0.5

40

–0.55

50

–0.8

40

–0.6 30 20

30

–0.7

20

–0.75

10

–1

–0.65

–0.8 0 –10

0 10 (a)

20

–1.4

10 0

–20

–1.2

–20

80

0 10 (b)

20

1 0.8 0.6 0.4 0.2 0 –0.2 –0.4 –0.6 –0.8 –1

70 60 50 40 30 20 10 0 –20 –10 0 (c)

–10

10 20

9.28 Simulation of free blowing: (a) uniform temperature; (b) heterogeneous temperature; (c) initial preform.

used. We will not describe these classical techniques here in detail and the reader is referred to associated literature. The crystallinity ratio (Xc) describes the average weight of crystalline and amorphous phase of the material. This parameter can be evaluated by a rule of mixtures considering specific values of each phase. It might be density (density measurements), specific volume, specific heat or specific enthalpy (differential scanning calorimetry measurements), etc. This ratio can also be evaluated more roughly still by using X-ray diffraction and comparing the diffracted signal for amorphous and fully crystallized samples. Let us recall that the ultimate crystallinity ratio reaches only 45% and that plastic bottles usually exhibit a ratio of about 35%.

270

4.5 160

160

4 3.5 2.5

140


Thickness distribution for small ∆L

3

140

2 1.5 1

120

120

0.5 0 100

0

20

40

T

60 ∆L

80 100 120 140 160 100

∆T

80

80

60

60

4 40 Thickness mm

3.5

20

0 0

10

20

30

40

50

Thickness distribution for large ∆L

3

40

2.5 2 1.5 1 0.5 0

20

0

20

40

60

80 100 120 140 160

0 0

20

40

9.29 Influence of the initial temperature with orthotropic modelling.


271

9.30 Evolution of the bottle shape during stretch blow moulding (Champin, 2007).

The orientation of the macromolecular chains is a crucial parameter, due to the high anisotropy of the chains, whether the chains are in the amorphous or the crystalline phase. This orientation plays a major role for crystallization (thermal or mechanical) and for macroscopic properties (mechanical or barrier properties). Optical (birefringence) and spectroscopic techniques (IR dichroism, whether transmitted or reflected) provide average orientation values. IR dichroism is also useful because it can discriminate between the orientation in amorphous and crystalline phases, in trans or cis conformations. X-ray diffraction is also widely used to evaluate chain orientation in amorphous or crystalline phases. Crystal size and morphology measurements are of course important in evaluating the mechanical properties from a given microstructure, using for example classical modelling for short-fibre composites. X-ray diffraction is here the more powerful means and allows evaluating the crystal dimensions along the crystal lattice directions. Important information that can also be deduced from crystal size is their volume, and therefore, knowing the crystallinity ratio, the crystals volumic population can be evaluated. In the following, we will use in-situ measurements obtained from wideangle X-ray diffraction (WAXD). To help the reader to analyse the pictures, we present in Fig. 9.31 two typical patterns. On the left we see a wide circular ring, which is associated with a non-oriented (all directions are diffracting the same way, no preferential orientation is to be seen and therefore the signal is a circular signature) and amorphous structure (no defined spaces are imposed between the chains and the incident beam is deviated on different locations). On the right we see the signal diffracted by a well-oriented (a

272

Advances in polymer processing X-ray beam X-ray beam Crystallized and oriented (along the arrow) microstructure

Nearly amorphous and isotropic microstructure

2-D diffracted pattern

2-D diffracted pattern

3-D view 3-D view

9.31 WAXD typical patterns: (a) amorphous and isotropic material; (b) crystallized and oriented material. Table 9.5 Crystallinity ratio measured on 0.5 litre bottles Preform geometry Zone studied

A

B

C

D

Upper zone Body Bottom

19% 17% 19%

23% 20% 19%

25% 18% 19%

22% 23% 19%

preferential orientation exists as no circular rings are to be seen) and crystallized structure (the spots are thin and their location is a signature of the crystal lattice). Measurements on bottles A first indicator of microstructural modifications is the crystallinity ratio, which is used for tuning industrial processes (see, for example, heat-stabilized bottles in Section 9.2.3). Table 9.5 gives an example of the crystallinity ratios measured for a 0.5 litre bottle for different zones and different preform geometries. However, the mechanical anisotropy induced by the process is clearly shown in numerous studies (see, for example, Chevalier, 1999) and was illustrated in Section 9.3. It would seem at first glance that because of its very low thickness, the bottle would exhibit a homogeneous microstructure and that the mechanical anisotropy could be easily related to an anisotropic microstructure, induced by the sequenced path. But a closer investigation of the microstructure reveals that the WAXD pattern obtained throughout the


273

bottle wall shows induced crystallization but reveals no evidence of anisotropy. Measurements by IR dichroism (Cole et al., 1999) and by optical and X-ray diffraction (Cakmak et al., 1984) as well as more recent FTIR investigations (Everall et al., 2002; Smith et al., 2006) have revealed substantial orientation gradients both along the bottle length and through the bottle thickness. Figure 9.32 is adapted from the study of Smith et al. (2006) and illustrates the classical trends measured. Measurements on specimens for uniaxial tension tests The measurements reported on bottles show that the induced microstructure is complex and strongly dependent on the thermo-mechanical history experienced by the material during the process. In order to understand better the coupling between the process parameters, the behaviour of the material during the process and the induced microstructure, it is necessary to impose simpler solicitations on a simpler geometry, through still representative of the industrial case. As for mechanical investigation, the first step is to achieve uniaxial tension tests, with either in-situ or ex-situ measurements. Until recently, microstructural studies were mainly based on ex-situ measurements (Salem, 1992, 1998; Le Bourvellec et al., 1986; Ajji et al., 1996). These studies exhibited several important points for strain-induced crystallization: a critical elongation (about 2 or 2.5) is needed (Ajji et al., 1996, Pearce et al., 1997) and the crystallinity ratio can reach 30%. Nevertheless, the microstructure is

Highly preferred orientation along longitudinal direction

No preferred orientation

Slightly preferred orientation along orthoradial direction increasing towards the bottle base

Slightly preferred orientation along longitudinal direction

Orthoradial orientation is higher for inner wall than for outer wall

Noticeable orientation switch into longitudinal direction (for outside wall only)

9.32 Trends of microstructure for bottle regions.

274


still evolving after the solicitation completion and the crystallization kinetics are strongly dependent on chain orientation (Smith and Steward, 1974; Gupte et al., 1983; Mahendrasingam et al., 2000b). A very effective quenching protocol is therefore needed to prevent the chain relaxation and to afford representative results (Gorlier et al., 2001; Marco and Chevalier, 2008). That explains why the first in-situ measurements (Blundell et al., 1996; Mahendrasingam et al., 1999), which remain quite rare in the literature, gave opposite results to former studies: crystals became visible only after the end of the solicitation. It was shown more recently that for lower strain rates ( 0.3 3 dt

10.10b

As an example, the curing kinetics for an epoxy resin system EPON 862/W is

(

dα = 398829.16 (1 – α ) exp – 8479.67 T dt

)

10.11

The curing degree vs. time at different temperatures is shown in Fig. 10.3.

10.4

Liquid resin and composite processing

10.4.1 Hand lay-up Hand lay-up, the simplest fabrication process, is an open mould process. In hand lay-up, the mould, which is usually made of reinforced plastic, defines the shape of the outer surface. As shown in Fig. 10.4, the mould is first coated with a wax to ensure removal after curing. A layer of gel coat is then sprayed on to the mould to form the outermost surface of the product. The gel coat is allowed to cure for several hours but remains tacky so subsequent resin layers adhere better. Alternate layers of catalysed resin and reinforcement material are applied. The ratio of resin to glass is usually 60 to 40 by weight, but varies by product. Each reinforcement layer is ‘wetted out’ with resin,

150°C

1.0

130°C 0.8

0.6

α

110°C 0.4

0.2

90°C

0.0 0

20

40

60 t (min)

80

100

10.3 Curing degree vs. time for EPON 862/W.

120

294


Roller Fibre mats impregnated with resin Mould

10.4 Hand lay-up process.

Chopper Roving

Gun

Resin

Catalyst

Mould

10.5 Spray-up process.

and then rolled out to remove air pockets. The process continues until the desired thickness is achieved. Hand lay-up is a room temperature curing process.

10.4.2 Spray-up Spray-up is another open-moulding composites fabrication process that uses mechanical spraying and chopping equipment for depositing the resin and glass reinforcement, as shown in Fig. 10.5. Resin and chopped glass can be deposited simultaneously or separately to the desired layer thickness on the mould surface (or on the gel coat that was applied to the mould). This process allows a greater production rate and more uniform parts than does hand lay-up, and often uses more complex moulds. This process involves the same initial steps (up through application of the gel coat) as used in hand lay-up. Following gel coat application, the polyester resin is applied with a spray gun that has a glass chopper attachment. Layers are built up and rolled out on the mould as necessary to form the part. The spray gun has separate resin and catalyst streams which mix as they exit the gun. However, compared to hand lay-up, more resin is typically used to produce similar parts by spray lay-up because of the inevitable over-spray of resin during application.


295

Resin spray-up is commonly used in open moulding processes in the fibre-reinforced plastics/composites (FRP/C) and boat building industries. Spray-up is also a room temperature curing process.

10.4.3 Resin casting In the resin casting process, the resin and catalyst are mixed and poured into various types of moulds at or near room temperatures. The part is cured through chemical reaction and usually with the addition of heat, as shown in Fig. 10.6. Commonly used materials include epoxies, polyurethanes and silicones. The moulds can be made of silicone, metal or epoxy/polyurethane.

10.4.4 Resin transfer moulding Because of its relatively low equipment and tooling costs, short cycle times and excellent design flexibility, closed-mould liquid composite moulding (LCM) processes, which include resin transfer moulding (RTM) and vacuum assisted resin transfer moulding (VARTM) processes, are replacing openmould processes such as hand lay-up and spray-up. The RTM process can be generally divided into four steps as shown in Fig. 10.7 (Gutowski, 1997). In the first step, dry reinforcements are cut and/or shaped into preformed pieces and then placed in a prepared mould cavity. This is usually called preform loading. After the mould is closed and clamped tightly, resin is injected into the mould cavity, where it flows through the reinforcement preform, expels the air in the cavity, and ‘wets out’ or impregnates the reinforcement. This step, which is considered the most critical in the RTM process, is called mould filling. When excessive resin begins to flow out of the vent area of the Pour resin

Male mould

Female mould

10.6 Resin casting.

296


Step 1: Load preform

Step 2: Inject resin

Step 3: Cure

Step 4: Demould

10.7 Resin transfer moulding process.

mould, resin injection is stopped and the curing step begins. Curing can take from several minutes to several hours. When curing is complete, the component is removed from the mould. This final step is called demoulding.

10.4.5 VARTM Vacuum-assisted resin transfer moulding (VARTM) refers to a variety of related processes, which represent the fastest growing new moulding technology. VARTM-type processes and standard RTM differ in that VARTM draws resin into a preform through use of a vacuum, rather than positive pressure. The primary advantage of VARTM is that it does not require high heat or pressure. For this reason, VARTM operates with low-cost tooling, making it possible to inexpensively produce large, complex parts in one shot. This has a great potential for the aerospace industry. Conventional composite parts for commercial aircraft are fabricated by an autoclave process with unidirectional carbon fibre prepreg. Because an autoclave is required to cure the materials, the parts are very expensive (Takeda et al., 2005). In the VARTM process, as shown in Fig. 10.8, only one rigid mould piece is used and the other mould piece is a vacuum bag. A highly permeable distribution medium is incorporated into the preform as a surface layer. During infusion, the resin flows preferentially across the surface and simultaneously through the preform thickness, enabling large parts to be fabricated. Current applications include marine, ground transportation and infrastructure parts.

10.4.6 Other liquid resin processing methods Other liquid resin processing methods include resin injection moulding (RIM), resin film infusion, etc. (Gutowski, 1997). In the resin injection moulding


297

Vacuum bag To resin reservoir

Peel ply and distribution layer

Sealant tape

To vacuum pump

Fibre fabric

Mould tool

10.8 VARTM process.

process, a rapid-cure resin and a catalyst are injected into the mould in two separate streams; mixing and the resultant chemical reaction both occur in the mould instead of in the dispensing head. The automotive industry is increasingly combining structural RIM (SRIM) with rapid preforming methods to fabricate structural parts that do not require a Class A finish. In the resin film infusion process, dry fabrics are laid up interleaved with layers of semisolid resin film supplied on a release paper. The lay-up is vacuum bagged to remove air through the dry fabrics, and then heated to allow the resin to first melt and flow into the air-free fabrics, and then after a certain time, to cure.

10.5

Resin flow and defect generation

10.5.1 Flow of resin in porous materials The flow of resin in porous materials is described by Darcy’s law as

 k xx u     1  v  = – µ  k yx    w   k zx

k xy k yy k zy

 ∂p  k xz   ∂x    ∂p  k yz      ∂y  k zz   ∂p   ∂z 

10.12

where u, v and w are the velocity components in the x, y and z directions, respectively; p is the fluid pressure, µ is the viscosity, and the kij’s are the components of the permeability tensor. Equation 10.12 assumes Newtonian flow. Since the Reynolds number is usually much smaller than 1 (Tucker, 1993), resin flow can be regarded as Newtonian flow. The velocity is the apparent fluid velocity, which is the product of the actual fluid velocity and φ, the porosity of the fibre reinforcement. The porosity is related to the fibre volume fraction vf as

φ = 1 – vf

10.13

298


The permeability tensor in Equation 10.12 is symmetrical can be diagonalized by coordinate transformation so that  k xx   k yx   k zx

k xz   k11   k yz  = T  0   0 k zz  

k xy k yy k zy

0 k 22 0

0   0  TT k 33 

10.14

where the kii’s are the permeability in the principal directions, and T is the transformation matrix. T is given by  l11  T =  l 21 l  31

l12 l 22 l32

l13   l 23  l33 

10.15

where the lij’s are the direction cosines of unit vector i, with reference to Fig. 10.9. Equation 10.14 can be expanded as  k xx   k yx   k zx

k xy k yy k zy

k xz   k yz   k zz 

2 2 2 l11 k11 + l12 k 22 + l13 k 33   =  l11 l 21 k11 + l12 l 22 k 22 + l13 l 23 k 33   l11 l31 k11 + l12 l32 k 22 + l13 l33 k 33

l11 l 21 k11 + l12 l 22 k 22 + l13 l 23 k 33 2 2 2 l 21 k11 + l 22 k 22 + l 23 k 33

l 21 l31 k11 + l 22 l32 k 22 + l 23 l33 k 33

l11 l31 k11 + l12 l32 k 22 + l13 l33 k 33   + l 21 l31 k11 + l 22 l32 k 22 + l 23 l33 k 33   2 2 2 l31 k11 + l32 k 22 + l33 k 33 

10.16

Composite parts are often thin relative to their size and the velocities can be assumed to be constant in the thickness direction, in which case Equation 10.16 is reduced to

 k xx   k yx

2 2 k xy   l11 k11 + l12 k 22  =  k yy   l11 l 21 k11 + l12 l 22 k 22

l11 l 21 k11 + l12 l 22 k 22   2 2 l 21 k11 + l 22 k 22 

10.17

As shown in Fig. 10.10, l11 = cos θ, l12 = sin θ and l22 = cos θ. Thus,


299

3 1 2

z y

x

10.9 Principal directions vs. global coordinate system.

2

θ 1

y

x

10.10 Principal directions vs. coordinate system.

k11 + k 22 k – k 22 + 11 cos2θ 2 2 k – k 22 = k yx = 11 sin2θ 2 k + k 22 k – k 22 = 11 – 11 cos2θ 2 2

k xx = k xy k yy

10.18

where θ is the angle between the principal direction 1 and the x-axis. When the thickness of the composite part is neglected and the resin flow is considered in 2-D, resin flow can be classified as channel flow and radial flow. A typical channel flow is shown in Fig. 10.11. In 1-D, Darcy’s law is written as

300


dp u= –k µ dx

10.19

The closed-form solution for the mould filling time is given by Cai (1992a, 1992b) as t fill =

φµl 2 2 k∆p

10.20

where l is the distance from the injection gate to the flow front and ∆p is the pressure difference between the injection gate and the flow front. In a radial flow, resin is injected from the centre of the preform. For an isotropic preform, k11 = k22, the flow front will be circular and the fluid moves in the radial direction. The closed-form solution is given by Adams et al. (1988) as t fill =

 φµ  2  Rf2  R ln – Rf2 + R02  4 k∆p  f  R02   

10.21

For an orthotropic preform, k11 ≠ k22, and the flow front will be an ellipse. Figure 10.12 shows the resin flow fronts of different fibre preforms. The permeability of fibre preforms can be measured using either channel flow or radial flow. In each case, measurement can be based on either saturated flow measurements or monitoring the progression of the flow front.

10.5.2 Effect of preform deformation The first type of deformation takes place during the preforming stage where fibre fabric is made to conform to the shape of a complex mould. Fibre orientations after deformation are significantly different from those of the original fabric and the fibre volume fraction also changes. The dominant mode of in-plane deformation for fibre fabric is shear deformation.

Flow front

10.11 Channel flow apparatus for measuring preform permeability.


301

10.12 Radial flow in fibre preforms: (a) isotropic preform; (b) orthotropic preform; (c) orthotropic preform aligned at 45°.

Resin-rich zone

10.13 Formation of a resin-rich zone.

The second type of deformation is the compaction of fibre preforms when the mould is closed. Gutowski et al. (1986) developed a model to predict the relationship between the applied pressure and the fibre volume fraction. Other investigators (Robitaille and Gauvin, 1999; Rudd et al., 1993; Trevino et al., 1991) proposed empirical models. Resin-rich areas are usually caused by fibre preform distortion due to compaction of fibre preform and resin infusion. Holmberg and Berglund (1997) studied the manufacture and performance of RTM U-beams. As the mould closes, the reinforcement tends to pull tight around corners and leaves a resin-rich area, as shown in Fig. 10.13. The third type of deformation occurs near the flow front where the reinforcement is being compressed by the pressure of the advancing resin.

302


Some studies are the work of Gong (1993), Farina and Preziosi (2000) and Ambrosi and Preziosi (1998). The fourth type of deformation is induced by the resin flow near the inlet, as shown in Fig. 10.14, where three regions can be distinguished: a fibre-free region, a saturated fibre bed and an unsaturated fibre bed. The deformation of the preform away from the wall opens a new channel flow.

10.5.3 Voids Voids can be caused by a number of factors. Dry spots or macrovoids are caused by edge flow (race-tracking) and improper vent locations. When the fibre preform is loaded into the mould, it is impossible to obtain a perfect fit, and there will be gaps between the fibre preform and the edge of the mould. The resin will flow faster along the gap because of less resistance. This phenomenon is often called ‘race tracking’ and likely causes dry spots, as shown in Fig. 10.15. Voids can also be caused by the compression of reinforcement, as illustrated in Fig. 10.16 (Holmberg and Berglund, 1997). At the inner radius the reinforcement is compressed and results in a reduced permeability. The low

Saturated fibre bed

Fibre-free region

Dry fibre bed

10.14 Deformation of fibre preform due to inlet pressure.

10.15 Race tracking.


303

Resin flow direction

Dry spots or voids

10.16 Dry spots or voids caused by the compression of reinforcement.

permeability makes fibre wet-out difficult and an air pocket or void region may form. These can often be eliminated by mould optimization or control (Jiang et al., 2002; Liu et al., 1996; Trochu et al., 2006). On a much smaller scale, microvoids are also formed because the heterogeneous nature of fabric reinforcement leads to non-uniform flows and the formation of microvoids. These voids reduce the strength of the material and should be eliminated. Several studies have been carried out to understand the basic mechanisms of microvoid formation and its governing parameters (Chang and Hourng, 1998; Hamidi et al., 2004; Kang et al., 2000; Parnas et al., 1994; Patel et al., 1995; Patel and Lee, 1995; Rohatgi et al., 1996). The mechanism can be summarized as that the resin near flow front flows at different speeds depending on the microstructure of the reinforcement and air is trapped by the transverse flow. Patel and Lee (1995) showed that fibre preforms contain two distinguishable pore structures: micropores consisting of gaps between fibre filaments inside fibre tows, and macropores consisting of much larger gaps between tows. There are several ways in which the competing macro and micro flows can cause microvoids. One example is during transverse flow through an array of fibre bundles: air is trapped inside due to the higher flow resistance, as shown in Fig. 10.17 (Kang et al., 2000; Parnas et al., 1994). Microvoids can also occur when resin flows in the fibre direction. The micro-flow front is ahead of the macroflow front because of the capillary action, which is often called fingering. If the fibre bundles have transverse stitches, when the flow front reaches the stitches a flow perpendicular to the fibre direction develops and air is trapped

304

Advances in polymer processing Fibre bundles

Voids

10.17 Microvoid formation during flow in the transverse direction. Macro-pores

Fibre tows

Voids

10.18 Microvoid formation during flow in the fibre direction.

between the fibre bundles, as shown in Fig. 10.18 (Chang and Hourng, 1998; Patel et al., 1995; Rohatgi et al., 1996).

10.6

Dimensional variations

Unlike metals, which can be carved, bent or stamped into the desired shape, for composites the material forms as the part forms. The constituent materials of a composite react differently to the changes in environmental conditions encountered during processing. Chemically, the reinforcing fibres do not experience significant change during the process cycle. The thermoset polymer matrix, on the other hand, contracts during crosslinking by as much as 6% (Yates et al., 1979). The shrinkage vs. time curves for two epoxy resins cured at 120°C are shown in Fig. 10.19 (Luck and Sadhir, 1992). One very important observation made from these curves is that approximately 60% of the total shrinkage occurs prior to the gel point. Even though the epoxy resins shrink about 4%


305

5 DGEBA/DETA 4

Volume shrinkage (%)

DGEBA/PA

3

2 Gel point 1

0

0

2

4

6 8 Curing time (h)

10

12

14

10.19 Shrinkage vs. time curves for two epoxy resins at 120°C.

when fully cured, only about 40% of this shrinkage (1.6%) will occur after gelation to produce internal stresses. As well as chemically induced deformations, there are thermally induced deformations during processing. The reinforcing fibres show very little thermal deformation during cool-down along the axis of the fibre. On the other hand, the polymer matrix has a higher thermal expansion coefficient, typically an order of magnitude or more. Because the constituent materials of composites must be well bonded and uniformly deform to maintain the continuum after processing, these deformations are balanced internally within the composite and residual stresses are induced. Based on these analyses, it is concluded that the factors causing dimensional variations include the volumetric shrinkage of the resin during curing and the mismatch in the coefficients of thermal expansion of the matrix and the fibre. Other factors include tool/part interaction and processing defects. Due to the anisotropic nature of fibre reinforced materials, the mechanical properties and coefficients of thermal expansion are fibre orientation sensitive. Thus, stacking sequence is a significant factor affecting dimensional variations. For asymmetric stacking sequences, the residual stresses are not balanced along the thickness direction and moments of force are induced. These moments of force can cause large warpage such as a ‘saddle’ shape, as shown in Fig. 10.20. For symmetric laminates, the residual stresses are balanced along the thickness direction and deformations are mostly due to the difference of

306


10.20 Warpage of a [0/90/0/90±45±45/0/90/0/90] E-glass/epoxy panel.

αT > α I αT

r

180 – φ 180 – φ′

r′

αI

φ′ φ

10.21 Spring-in of an angled composite part.

CTE in the in-plane direction and the through-thickness direction. For example, for an angled composite part as shown in Fig. 10.21(a), the CTE in the through-thickness direction is much larger than that in the in-plane direction. Thus, a decreased angle (φ′ < φ) is observed as shown in Fig. 10.21(b), which is commonly called ‘spring-in’. The closed-form solution for ‘spring-in’ can be derived as follows. As shown in Fig. 10.21, the included angle is 180 – θ and the arc length is s =


307

r (180 – φ). When the temperature change is ∆T (negative), the arc length changes to s′ = r(180 – φ)(1 + αI∆T) and the radius becomes r′ = r(1 + αI∆T) The included angle after ‘spring-in’ is

1 + α I ∆T 180 – φ ′ = s ′ = (180 – φ ) 1 + α T ∆T r′ Hence, the ‘spring-in’ is 1 + α I ∆T ∆φ = φ ′ – φ = s – s ′ = 180 – φ – (180 – φ ) r r′ 1 + α T ∆T

i.e.

∆φ = (180 – φ )

(α T – α I ) ∆T 1 + α T ∆T

10.22

Since 1 + αT ∆T ≈ 1, Equation 10.22 can be reduced to ∆φ = (180 – φ)(αT – αT)∆T

10.23

The dimensional variation induced in the curing process can be reduced by stress relaxation during the post-curing process. This is illustrated using the spring-in (Svanberg and Holmberg, 2001). As shown in Fig. 10.22, further curing of the resin takes place during the post-curing cycle. During heating to the post-cure temperature, the part will expand and result in a decrease in

After cure 1 3

Spring-in

6

2

5 After post-cure 4 RT

Tg

Post-cure

10.22 Dimensional variation during post-curing process.

308


spring-in angle. At the glass transition temperature (Tg), the resin will change from a glassy to a rubbery state. Any frozen residual stress will be released and results in an increase in spring-in. At temperatures higher than Tg, the CTE of the resin in the rubbery state is much larger than in the glassy state. Therefore, the spring-in angle will decrease at a higher rate. At the post-cure temperature, further cure of the resin will induce chemical curing shrinkage that increases the spring-in. Finally, during cool-down, the spring-in will increase to its final value. The deformations of composites such as warpage and spring-in increase the secondary machining requirements and cost. They also cause matrix microcracks, and thus weaken the mechanical properties. The deformations often increase the difficulty for the assembly process of composite components. Thus, effective dimensional control is highly desirable to improve the part quality, reduce the cost and ensure the tolerance requirements for assembly. Two approaches are mainly used to minimize the dimensional variations in composites processing. The first approach is deformation compensation, where the moulds are modified to account for the dimensional variations generated due to processing. The second method is design optimization. The design variables such as thickness, curvature radius, stacking sequence, etc., can be varied within some range to minimize the induced dimensional variation. A special type of dimensional variation occurs in the VARTM process. Typically, thickness gradients and variations result from the infusion pressure gradient during process and material variations. Pressure gradient is the driving force for resin flow and the main source of thickness variation. After infusion, an amount of pressure gradient is frozen into the preform, which primarily contributes to the thickness variation (Li et al., 2006). Studies have shown that debulking can be used to eliminate the part-to-part dimensional variation and quality difference (Robitaille and Gauvin, 1998, 1999). Gama et al. (2001) investigated two processing options to improve the dimensional tolerances during VARTM processing. The influence of vacuum debulking on final part has been studied. However, vacuum debulking cannot eliminate the thickness gradient along the infusion direction of the part. Gama et al. (2001) compared several processing scenarios including ‘Open-Open’, ‘Close-Open’ and ‘Closed-Micro-Flow’. The ‘Close-Open’ processing scenario provides superior compaction and dimensional tolerance over the ‘Open-Open’ scenario. One major disadvantage of the ‘Close-Open’ approach is that significant resin starvation may occur that reduces mechanical properties. The ‘Closed-Micro-Flow’ processing scenario is a new approach that has reduced the thickness gradient in a VARTM part while retaining better mechanical properties as compared to the ‘Open-Open’ processing scenario.


10.7

309

Future trends

With the stringent requirement of environmental protection, closed-mould processes are replacing open-mould processes. Closed-mould processes including RTM and VARTM offer many advantages, including flexibility and low cost. Future work will be focused on the improvement of materials and processes, especially for the aerospace industry. Computer tools including CAD/CAE/CAM will be employed to reduce the cost. Better process control will be applied to reduce the part-to-part variation and improve the quality.

10.8

Sources of further information and advice

Gutowski, T. G. (1997) Advanced Composite Manufacturing, John Wiley & Sons. Åström, B. T. (1997) Manufacturing of Polymer Composites, Chapman & Hall. Composites Part A: Applied Science and Manufacturing, Elsevier, Oxford, UK. Polymer Composites, Society of Plastics Engineers (SPE), Brookfield, CT, USA. Journal of Composite Materials, Technomic, Lancaster, PA, USA.

10.9

References

Adams, K. L., Russel, W. B. & Rebenfeld, L. (1988) Radial penetration of a viscous liquid into a planar anisotropic porous medium. International Journal of Multiphase Flow, 14, 203–215. Ambrosi, D. & Preziosi, L. (1998) Modeling matrix injection through porous preforms. Composites, Part A, 29, 5–18. Bogetti, T. A. & Gillespie, J. W., Jr (1991) Two-dimensional cure simulation of thick thermosetting composites. Journal of Composite Materials, 25, 239–273. Cai, Z. (1992a) Analysis of mold filling in RTM process. Journal of Composite Materials, 26, 1310–1377. Cai, Z. (1992b) Simplified mold filling simulation in resin transfer molding. Journal of Composite Materials, 26, 2606–2630. Chang, C. Y. & Hourng, L. W. (1998) Study on void formation in resin transfer molding. Polymer Engineering and Science, 38, 809–818. Farina, A. & Preziosi, L. (2000) Non-isothermal injection molding with resin cure and perform deformability. Composites, Part A, 31, 1355–1372. Gama, B. A., Li, H., Li, W., Paesano, A., Heider, D. & Gillespie, J. W., Jr (2001) Improvement of dimensional tolerances during VARTM processing. 33rd International SAMPE Technical Conference, Seattle, WA, Society for the Advancement of Material and Process Engineering. Gong, H. (1993) Pressure distribution in resin transfer molding with a non-rigid fiber preform. Journal of Materials Processing Technology, 37, 363–371.

310


Gonzalez-Romero, V. M. & Casillas, N. (1989) Isothermal and temperature programmed kinetic studies of thermosets. Polymer Engineering and Science, 29, 295–301. Gutowski, T. G. (1997) Advanced Composite Manufacturing, John Wiley & Sons. Gutowski, T. G., Wineman, S. J. & Cai, Z. (1986) Applications of the resin flow/fiber deformation model. Proceedings of the 33rd National SAMPE Symposium, 245–254. Hamidi, Y. K., Aktas, L. & Altan, M. C. (2004) Formation of microscopic voids in resin transfer molded composites. ASME Transactions: Journal of Engineering Materials and Technology, 126, 420–426. Holmberg, J. A. & Berglund, L. A. (1997) Manufacturing and performance of RTM Ubeams. Composites Part A, 28, 513–521. Jiang, S., Zhang, C. & Wang, B. (2002) Optimum arrangement of gate and vent locations for RTM process design using a mesh distance-based approach. Composites Part A, 33, 471–481. Kamal, M. R. (1974) Thermoset characterization for moldability analysis. Polymer Engineering and Science, 14, 231–239. Kamal, M. R. & Sourour, S. (1973) Kinetics and thermal characterization of thermoset cure. Polymer Engineering and Science, 13, 59–64. Kang, M. K., Lee, W. I. & Hahn, H. T. (2000) Formation of microvoids during resintransfer molding process. Composite Science and Technology, 60, 2427–2434. Lee, W. I., Loos, A. C. & Springer, G. S. (1982) Heat of reaction, degree of cure, and viscosity of Hercules 3510-6 resin. Journal of Composite Materials, 16, 510–520. Li, J., Zhang, C., Liang, R. & Wang, B. (2006) Modeling and analysis of thickness gradient and variation in VARTM processes. SAMPE 2006, Long Beach, CA, Society for the Advancement of Material and Process Engineering. Liu, B., Bickerton, S. & Advani, S. G. (1996) Modeling and simulation of resin transfer molding (RTM) – gate control, venting and dry spot prediction. Composites Part A, 27, 135–141. Loos, A. C. & Springer, G. S. (1983) Curing of epoxy matrix composites. Journal of Composite Materials, 17, 135–169. Luck, R. M. & Sadhir, R. K. (1992) Shrinkage in conventional monomers during polymerization, in Sadhir, R. K. & Luck, R. M. (eds), Expanding Monomers. Boca Raton, FL, CRC Press. Parnas, R. S., Salem, A. J., Sadik, T. A. K., Wang, H. P. & Advani, S. G. (1994) The interaction between micro- and macro-scopic flow in RTM preforms. Composite Structures, 27, 93–107. Patel, N. & Lee, L. J. (1995) Effects of fiber mat architecture on void formation and removal in liquid composite molding. Polymer Composites, 16, 386–399. Patel, N., Rohatgi, V. & Lee, L. J. (1995) Micro scale flow behavior and void formation mechanism during impregnation through a unidirectional stitched fiberglass mat. Polymer Engineering and Science, 35, 837–851. Robitaille, F. & Gauvin, R. (1998) Compaction of textile reinforcements for composite manufacturing. II: Compaction and relaxation of dry and H2O saturated woven reinforcements. Polymer Composites, 19, 543–557. Robitaille, F. & Gauvin, R. (1999) Compaction of textile reinforcement for composites manufacturing. III: Reorganization of the fiber network. Polymer Composites, 20, 48– 61. Rohatgi, V., Patal, N. & Lee, J. L. (1996) Experimental investigations of flow-induced microvoids during impregnation of unidirectional stitched fiberglass mat. Polymer Composites, 17, 161–170.


311

Rudd, C. D., Rice, E. V., Bulmer, L. J. & Long, A. C. (1993) Process modelling and design for resin transfer moulding. Plastics, Rubber and Composites Processing and Applications, 20, 67–76. Svanberg, J. M. & Holmberg, J. A. (2001) An experimental investigation on mechanisms for manufacturing induced shape distortions in homogeneous and balanced laminates. Composites Part A, 32, 827–838. Takeda, F., Nishiyama, S., Hayashi, K., Komori, Y., Suga, Y. & Asahara, N. (2005) Research in the application of the VARTM technique to the fabrication of primary aircraft composite structures. Mitsubishi Heavy Industries, Ltd. Technical Review, 42. Trevino, L., Rupel, K., Young, W. B., Liou, M. J. & Lee, L. J. (1991) Analysis of resin injection molding in molds with replaced fiber mats. I: Permeability and compressibility measurements. Polymer Composites, 12, 20–29. Trochu, F., Ruiz, E., Achim, V. & Soukane, S. (2006) Advanced numerical simulation of liquid composite molding for process analysis and optimization. Composites, Part A, 37, 890–902. Tucker, C. L., III (1993) Fundamentals of computer modeling for polymer processing, in Advani, S. G. (ed.), Flow Phenomena in Polymer Composite. Cambridge, Woodhead Publishing. Yates, B., McCalla, B. A., Phillips, L. N., Kingston-Lee, D. M. & Rogers, K. F. (1979) The thermal expansion of carbon fiber-reinforced plastics. Part 5: The influence of matrix curing characteristics. Journal of Material Science, 14, 1207–1217.

11 Calendering of polymers E M I T S O U L I S, National Technical University of Athens, Greece

Abstract: Calendering is a process used in many industries for the production of rolled sheets of specific thickness and final appearance. Calendering of molten polymers is a process for the production of continuous sheet or film by squeezing the melt between a pair of heated counter-rotating rolls. This chapter reviews research on modelling and experimental investigation of calendering. The lubrication approximation theory (LAT) is outlined together with its predictions for the general case of pseudoplastic and viscoplastic fluids. Two- and three-dimensional analyses are also considered based on the finite element method (FEM). These analyses are able to predict vortex patterns and helical flows which agree with experimental results. Key words: calendering, lubrication approximation, FEM analysis, pseudoplastic fluids, viscoplastic fluids, sheet thickness, surface defects.

11.1

Introduction

11.1.1 The calendering process The term ‘calender’ is derived from the Greek kylindros (cylinder), and according to Webster’s International Dictionary, it means ‘to press (as cloth, rubber, paper) between rollers or plates in order to make smooth and glossy or glazed or to thin into sheets’. Thus, calendering (sometimes also written as ‘calendaring’) is a process used in many industries, such as the paper, plastics, rubber and steel industries, for the production of rolled sheets of specific thickness and final appearance. In particular, the calendering of molten polymers is a process for the production of continuous sheet or film by squeezing the melt between a pair of heated counter-rotating rolls. The modern technological developments in the calendering of thermoplastic materials are offshoots of the art of fabric and rubber calendering, which dates back to the early 1800s. Elden and Swan1 present a short history and a detailed account of the major technological developments up to about 1968. Industrial calenders consist usually of 3–6 rotating rolls in a Z or L arrangement. The basic forming operation is completed by the calender itself and is normally followed by additional treatment of the plastic sheet or film produced. A typical calendering layout is shown in Fig. 11.1 (from Marshall2). 312


313

Extruder

Thickness gauge

Banbury mixer

Metal detector

Edge trimmer

Mill

Wind-up

Cooling drums

11.1 A typical calendering layout.

The molten material is fed to the calender rolls from a Banbury mixer and a two-roll mill system or from a large extruder. The rolls are internally heated. With a four-roll configuration there are three nip regions or areas of contact between two rolls, which reduce the polymer melt thickness gradually while making a wider film. The difference in speed between two adjacent rolls is termed the friction. The application of friction in combination with temperature gives increased shear mixing and allows the transfer of the melt from one roll to the next. In front of each nip a polymer bank of material is formed which feeds the nip. The residence time in this bank can be quite long, because for each rotation of the roll, only a small portion of the bank actually enters the nip. Thus, the material in the bank is exposed to high temperatures for long times, whereas the material that passes through the nip does so for short times but at high shear rates. Additionally, properties such as adhesion to the rolls and some form of melt strength are important as the material exits the last nip, and is drawn down to the final sheet thickness on take-off rolls cooling the film to room temperature. The final nip region is the most important for hydrodynamic analysis for determination of flow variables, such as pressures and forces which affect the final film properties. Calender sizes range up to 90 cm in diameter and 250 cm in width, with polymer throughputs up to 4000 kg/h, according to Tadmor and Gogos,3 who give further details on the machinery involved. Roll speeds may be as high as 2 m/s for certain thin flexible films (thickness less than 0.1 mm). Calendering lines are very expensive in terms of capital investment in machinery. Film and sheet extrusion are competitive processes because the capital investment for the extruder itself is only a fraction of the cost of a calender. However, the high quality and volume capabilities of calendering lines make them advantageous for many types of products, especially for temperature sensitive materials. Polyvinyl chloride (PVC) is the major polymer that is calendered, while elastomers maintain a heavy volume of calendered sheets. New trends see the move from PVC towards ethylene–styrene interpolymers (ESI) and thermoplastic olefins (TPO) for improved recyclability. It should be further

314


emphasized that polymer blends and composites are also calendered and offer new challenges due to the complicated nature of the materials involved.

11.1.2 Literature survey Mathematical modelling The quality of the sheet or film produced is determined by the flow and heat transfer phenomena in the gap between two rotating rolls. It is thus of great importance to determine the velocity and temperature profiles as the melt is squeezed between the rolls (see Fig. 11.2), the pressure distribution on the roll surfaces, the roll-separating force and the torque and energy input. One of the earliest attempts to model this process was published by Ardichvili4 as early as 1938. Finston5 included thermal effects. It is, however, a rather realistic Newtonian and Bingham flow analysis by Gaskell6 that spearheaded further developments in modelling. An attempt to include viscoelastic effects was presented by Paslay.7 The major conclusions of these early investigations were summarized by Marshall.2 McKelvey8 reworked Gaskell’s model for Newtonian flow and extended it to include non-Newtonian (i.e. shear-thinning) effects. Brazinsky et al.9 presented an analysis for power-law fluids and Alston and Astill10 for a hyperbolic tangent (tanh) viscosity model. Some inaccuracies in McKelvey’s8 power consumption calculations were corrected by Ehrmann and Vlachopoulos.11 The asymmetric problem was treated analytically using bipolar cylindrical coordinates by Ehrlich and Slattery12 and by Takserman-Krozer et al.13 Renert14 provided an approximate solution for the pressure profile in the nip for the power-law model. Reher and Grader15 presented a numerical solution for the non-Newtonian problem. Some experimental results and numerical calculations on temperature development due to viscous dissipation were reported by Torner.16 Viscoelasticity was taken into account in the investigations of Tokita and White,17 Chong18 and White.19 y x z

11.2 Schematic representation of producing a sheet by squeezing material between two rotating calender rolls.


315

Kiparissides and Vlachopoulos20 developed a finite element analysis for symmetric and asymmetric calendering (different roll speeds, different roll diameters). Viscous dissipation effects were included in finite difference solutions of the energy equation in bipolar coordinates by Bekin et al.21 and by Kiparissides and Vlachopoulos22 in rectangular coordinates. The nonisothermal problem was also treated by the method of orthogonal collocation by Dobbels and Mewis.23 Agassant and co-workers24–26 presented a method for the calculation of the average temperature rise and several calculations of separating forces, torques and the critical conditions for the appearance of defects in calendered sheets. Non-isothermal calculations were also carried out by Dimitrijew and Sporjagin27 by the finite difference method and by Woskressenski et al.28 by the finite element method. Seeger et al.29 presented a finite element solution and calculated velocity and stress profiles for a power-law fluid. An isothermal model with slip was developed by Vlachopoulos and Hrymak30 using the lubrication approximation and a Runge–Kutta solution. Other investigations were carried out by Chung for Bingham plastic fluids31 and compressible fluids,32 and by Suto and co-workers33–35 for Newtonian and viscoelastic fluids. The vast majority of the above work is based on the Lubrication Approximation Theory (LAT) of Reynolds. The textbook by Middleman36 nicely summarizes this and presents the most important findings up to 1977. Lifting this approximate assumption leads to a truly two-dimensional (2-D) analysis, as was done by Mitsoulis et al.37 and Agassant and Espy.38 These two works have shown very interesting results with intricate patterns dominated by large vortices in the melt bank before the rolls. Zheng and Tanner39 also solved the 2-D problem for Newtonian, power-law and viscoelastic fluids, with particular emphasis on finding the point where the fluid detaches from the rolls. They used a boundary element method and the criterion that the fluid detaches from the rolls at the point where the tangential traction becomes zero. Again, all works prior to 1990 have been summarized in the textbook by Agassant et al.40 In 1990, Tseng and Sun41 carried out a full 2-D nonisothermal finite-difference investigation for PVC coupled with experiments, and showed how to use the simulations to deduce a new roll design. More recently, Levine et al.42 used a 2 12 -D approach to calculate the side spreading of the calendered sheet while employing the lubrication approximation for the flow through the nip. Mewes et al.43 have done work on the simultaneous calculation of roll deformation and polymer flow, while the full threedimensional (3-D) problem has been solved by Luther and Mewes44 for power-law fluids. A renewed effort has appeared recently to study viscoplastic materials in calendering and to find out the effect of yield stress. A series of papers by Sofou and Mitsoulis45–47 have used LAT and carried out full parametric

316


studies for pseudoplastic and viscoplastic fluids obeying the Herschel–Bulkley model of viscoplasticity with or without slip at the wall. This model, upon appropriate simplifications, can be reduced to the Newtonian, power-law and Bingham plastic models. The results have shown interesting trends and yielded/ unyielded regions. However, the latter are misleading as pointed out by Sofou and Mitsoulis45 and by Mitsoulis,48 and they are a direct consequence of using LAT and its inherent approximations. Experimental investigations Very few experimental investigations have been published in the open literature. For a long time, Bergen and Scott’s49 measurements on a plasticized resin were the only basis for comparison of the proposed mathematical models. Direct comparisons could not be made, however, because of limited information on the rheological properties of the resin. Unkrüer50 carried out a detailed experimental study of calendering of polyvinyl chloride (PVC) and polystyrene (PS). Using colour tracers he observed a complex flow pattern with three recirculation regions in the melt bank and flow in the axial (cross-machine) direction as shown in Fig. 11.3. In Unkrüer’s thesis50 there is a limited amount of rheological data of the melts, and thus a direct comparison of model predictions and experimental measurements on pressure distribution is impossible. Agassant26 carried out an extensive experimental study on calendering (using PVC and a silicone oil for certain measurements) and compared his results to model predictions for separating forces and torques. Visualization studies of the complex flow pattern (similar to Unkrüer’s50) in the melt bank were also included in Agassant’s work as well as studies on the origin and appearance of calendered sheet defects.25 An investigation on the surface irregularity of a polypropylene (PP) sheet was carried out by Prentice.51 A limited amount of experimental data was also given by Hatzmann et al.,52 Kopsch53 and Suto and co-workers.33–35 Vlachopoulos and Hrymak30 presented pressure distribution and torque measurements for rigid PVC carried out by Chauffoureaux,54 as well as measurements of apparent viscosity and slip velocity for a wide range of temperatures. Experimental results for the melt bank and the pressure distribution for a rigid PVC are found in Agassant and Espy.38 Tseng and Sun41 presented experimental data on an industrial PVC calendering line. The work of Mewes and co-workers43,44 also includes experimental data on polymer melts and deformation of calender rolls, while Kalyon et al.55 presented experimental data on detachment of polymers from the calender rolls and defects associated with a range of operating conditions.


V

–λ

317

λ

0

V

(a)

Z

W1 2

W2 2

Y V2

Z

λ

O 2 H0

2H

X

V1

Pressure (kN/mm2)

(b)

1.0 0.8 0.6 0.4 0.2 Centreline

6c m 5.5

cm

(c)

11.3 (a) Flow pattern in the melt bank; (b) flow in the axial (crossmachine) direction; (c) pressure profiles at different axial positions (according to Unkrüer50).

318

11.2


Lubrication approximation

11.2.1 General considerations Very viscous fluids such as polymer melts, foodstuffs, paper pulp, and other materials used in calendering, flow at low Reynolds numbers, e.g. Re = 10–4 to 10–2, and the creeping flow approximation is applicable.36 Thus, the conservation equations for steady flow are ∇ · (ρ v ) = 0

11.1

0 = – ∇p + ∇ · τ

11.2

ρ C p v ⋅∇ T = ∇⋅ ( k ∇T ) + p∇⋅ v + τ : ∇v

11.3

where ρ represents the density, v the velocity vector, p the pressure, τ the extra stress tensor, Cp the specific hear, T the temperature and k the thermal conductivity. Fluid compressibility may be important in certain processes (see Tadmor and Gogos 3 ), but not in calendering. Chung’s results 32 are for uncharacteristically high calendering pressures. Cp and k can be assumed constant. Thus the above equations reduce to ∇· v =0

11.4

0 = – ∇p + ∇ · τ

11.5

ρ C p v ⋅∇T = k∇ 2 T + τ : ∇v

11.6

Further assuming that the fluid does not spread as it enters the gap between the rolls, we may write equations 11.4–11.6 for two dimensions, where x is the direction of flow and y is perpendicular to the roll axis, as

∂v y ∂v x + =0 ∂x ∂y

11.7

0= –

∂ p ∂τ xy ∂τ xx + + ∂x ∂y ∂x

11.8

0= –

∂τ yy ∂ p ∂τ yx + + ∂y ∂x ∂y

11.9

∂T  ∂2 T   ∂ 2T  ∂T ρC p  v x + vy  =k 2 +  ∂y   ∂x ∂y2   ∂x + τ xx

∂v y ∂v y ∂v x ∂v + τ xy x + τ yx + τ yy ∂x ∂y ∂x ∂y

11.10


319

It is also reasonable to assume that the flow in the gap will be nearly parallel so that ∂/∂x y0). We also note that τy = | dP/dx | y0. Thus, we obtain:

 U + n n+1  u=   n U – n + 1 

1

 1 dP  n [( y – y ) nn+1 – ( h – y ) nn+1 ], y > y 0 0 0  K dx  1

 1 d P  n ( h – y ) nn+1 , y ≤ y 0 0  K dx  for dP/dx > 0 11.20a

1  n+1 n+1 n   n 1 d P U – – [( y – y 0 ) n – ( h – y 0 ) n ], y > y 0 n + 1 K dx   u=  1 n+1  n  – 1 dP  n (h – y ) n , y ≤ y U + 0 0  n + 1  K dx  

for dP/dx < 0 11.20b Integration of the velocity profile gives the volumetric flow rate Q according to: 1

n+1 2 n[( n + 1) h + ny 0 ]  1 d P  n Q = 2Uh – ( h – y 0 ) n for dP/dx > 0 ( n + 1)(2 n + 1)  K d x 

11.21a


323

1

Q = 2Uh +

n+1 2 n[( n + 1) h + ny 0 ]  1 d P  n – ( h – y0 ) n ( n + 1)(2 n + 1)  K d x 

for dP/dx < 0

11.21b

In the case of the Bingham plastic model (n = 1), the above equation becomes: Q = 2Uh –

2 h + y0  d P  ( h – y 0 ) 2 for all dP/dx 3µ  d x 

11.22

From equation 11.21 the pressure gradient dP/dx can be found. In calendering the following dimensionless parameters are introduced:36

x′ =

2 x , h = H  1 + x  , y ′ = y , y ′ = y0 , 0 H0 0 H0 2 RH 0   2 RH 0

n

H Q P′ = P  0  , λ 2 = –1 K U  2UH 0

11.23

where λ is a dimensionless flow rate (or leave-off distance) and the rest of the symbols are defined in Fig. 11.4. After the appropriate manipulations, the following dimensionless pressure gradient is obtained: n –1

2 2 2 2 1 2 R (λ – x ′ ) (λ – x ′ ) n 2 H0  (1 + – y 0′ ) n +1 x ′ n  2 (1 + ) + x y ′ ′  n + 1 0  11.24

n dP ′ = –  2 n + 1   n  dx ′

where

y 0′ = τ *y dP ′ dx ′

dP ′ dx ′

11.25 0

Equation 11.24 incorporates both cases: for −∞ < x′ < −λ the pressure gradient is positive, and for −λ ≤ x′ ≤ λ the pressure gradient is negative. Integration of equation 11.24 requires boundary conditions for the pressure gradient dP′/dx′ and the pressure P′ as well. In the case of an infinite reservoir, the standard conditions are the so-called Swift conditions:36,39 P′ = dP′/dx′= 0 at x′ = λ∞

11.26a

at x′ = −∞

11.26b

P′ = 0

Then the pressure is given by the integral: 2n + 1  P′ =   n 

n

2R H0

∫

λ∞

x′

( λ ∞2 – x ′ 2 ) ( λ ∞2 – x ′ 2 )

n –1

 (1 + x ′ 2 ) + n y ′   n + 1 0 

n

1 dx ′ (1+ x ′ – y 0′ ) n +1 2

11.27

324


The extrema of the pressure distribution occur at x′ = −∞, ±λ∞. The dimensionless leave-off distance λ∞ corresponding to an infinite reservoir can be found from the above equation knowing that P′(x′ = – ∞). Therefore λ∞ can be found from the relation: 0=

∫

λ∞

( λ ∞2 – x ′ 2 ) ( λ ∞2 – x ′ 2 )

 (1 + x ′ 2 ) + n y ′   n + 1 0 

–∞

=

∫

λ∞

–∞

n –1 n

1 dx ′ (1+ x ′ – y 0′ ) n +1 2

I ( n , τ *y )dx ′

11.28

The above integral has no analytical solution for the general case of Herschel– Bulkley fluids. Therefore, a numerical solution must be found, based on some numerical algorithm for solving non-linear equations. Sheet thickness Once λ∞ is found as a function of n and τ *y , then all other quantities of interest are readily available. The exiting sheet thickness H from an infinite reservoir is given by: H = 1 + λ2 ∞ H0

11.29

For the case of calendering sheets with a finite thickness, the leave-off distance λ∞ is substituted by λ and −∞ by – x ′f . The exiting sheet thickness H is given by:

H = 1 + λ2 H0

11.30

while the thickness of the entering sheet Hf enters the analysis according to the definition: 1/2  Hf  – 1 x ′f =   H0 

11.31

Yield-line location The location of the yield line y0 is derived for Bingham plastics as: 1  4 y 3 – 2 y 2 h + 2 h 3   dP  0 µ 3 0 3   dx  2 τy = 2UH 0  x – λ2 ± ( h – y0 ) 2 µ  2 H0 R 

11.32


325

where the pressure gradient dP/dx and the yield-line location y0 are connected via the yield stress, i.e., y 0 dP = τ y dx

11.33

This line separates the yielded from the unyielded zone and serves as a contour of the stress being equal to the yield stress between the viscous liquid part and the solid plastic part of the material. In dimensionless form the above equations become:

1 H 0 [(1 + x ′ 2 ) 3 – y ′ 3 ] d P ′  0 3 2R  dx ′  H0 [(1 + x ′ 2 ) 2 – y 0′ 2 ] 2R

= ( x ′ 2 – λ 2 ) ± Bn 2

11.34

with

y 0′ = Bn dP ′ dx ′

2R H0

11.35

In the above equations, the + sign corresponds to x′ < −λ and the − sign to −λ ≤ x′ ≤ λ. Equation 11.35 also serves as a link between the dimensionless yield line (or yield stress τ *y ) and the Bingham number, Bn. In the general case of the Herschel–Bulkley fluids, the yield-line location results in the solution of the general equation 11.24, which gives simultaneously both the pressure gradient dP′/dx′ and y 0′ . Operating variables The operating variables used in engineering calculations are also of interest:36 1. The maximum pressure, P

( n , τ *y ) =

P

( n , τ *y ) , is defined by:

2n + 1  Pmax ′ =   n  2 R/ H0

n

∫

λ

–λ

I ( n , τ *y ) dx ′

11.36

2. The roll-separating force per unit width, F/W(n, τ *y ), is defined by:

F ( n, τ * ) = y W

∫

λ

– x ′f

P( x ) d x = K  U   H0 

n

RF ( n , τ *y )

11.37a

 I ( n , τ *y ) d x ′  dx ′ 

11.37b

with F

(

( n , τ *y ) = 2 2 n + 1 n

)∫ ∫ n

λ

– x f′

  

λ

x′

326


3. The power input for both rolls, W

˙ ( n , τ *y ) = 2 WU

∫

λ

– x f′

W

˙ ( n , τ *y ) , is defined by:

τ x y | y = h( x ) dx

= WU K  U   H0 

n –1

R H0

2

11.38a E

( n,

τ *y )

with E

(

( n , τ *y ) = – 2 2 2 n + 1 n

)∫ n

λ

– x f′

I ( n , τ *y )(1 + x ′ 2 ) dx ′

11.38b

11.2.4 Method of solution In the general case of the Herschel–Bulkley fluids and an infinite reservoir, the solution for a given value of n and τ *y and an assumed value of λ must provide at each axial distance x′ the pressure gradient dP′/dx′ and the yieldline location y 0′ by solving equation 11.24. The local solution is found numerically by the modified regula falsi method.59 Once this is done, the local pressure P′(x′) is obtained by integrating dP′/dx′, starting from x′= λ and applying the boundary condition of equation 11.26a. The integration is done by using Simpson’s rule. Then follows integration of all pressures in the assumed domain (i.e., assumed λ) according to equation 11.26 to see whether equation 11.28 is satisfied within a specified tolerance (10−6). If this is not the case, a new value of λ is assumed, based again on the modified regula falsi method. Usually 10–20 iterations were needed for finding the correct λ-value. More details can be found in Sofou and Mitsoulis.45,46 In the case of a finite sheet of a given thickness ratio Hf /H0, the value of –∞ was substituted by – x f′ given by equation [11.31]. From the operating variables, the maximum pressure P ( n , τ *y ) occurs at –λ and is readily obtained by inspection. The roll-separating force F ( n , τ *y ) and the power input to both rolls E ( n , τ *y ) are obtained by the appropriate integration according to equations 11.37 and 11.38 using Simpson’s rule.

11.2.5 Results and discussion Pseudoplastic fluids We begin with typical calculation results for pseudoplastic fluids, both shearthinning (0 ≤ n ≤ 1) and shear-thickening (1 ≤ n ≤ 2). The well-known Newtonian value36 (for n = 1) for an infinite reservoir is λ∞ = 0.475 corresponding to H/ H0 = 1.226. The characteristic pressure distributions found in calendering of


327

Newtonian fluids are shown in Fig. 11.6 for different λ-values. These results can also be obtained analytically,36 so there is a direct comparison with any numerical method. As the leave-off distance λ increases so does the pressure distribution and the domain length – x f′ . The pressure-gradient distributions for different shear-thinning powerlaw fluids are shown in Fig. 11.7 for a fixed value of λ = 0.3, where the different regions of negative and positive pressure gradients become apparent, passing through zero at x′ = −λ. More shear thinning gives flatter profiles. It should be noted that the shear stress distribution follows the pressure-gradient distribution according to equation 11.18. The corresponding pressure distributions are shown in Fig. 11.8. More shear thinning gives more triangularlike profiles. The leave-off distance λ as a function of the entering sheet thickness Hf/ H0 is given in Fig. 11.9. Results are given for a wide range of entering sheet thickness ratios Hf/H0, from the limiting case of Hf/H0 = 1 (no thickness reduction) to the case of Hf/H0 = 1000 (approaching a sheet coming from an infinite reservoir). The well-known Newtonian values36 (for n = 1) of λ∞ = 0.475 and H/H0 = 1.226 are a starting point after which it is noted that shear thinning increases the values, reaching at the limit for n = 0, λ∞ = 0.578 and H/H0 = 1.334 (found analytically). On the other hand, shear thickening decreases the values, reaching for n = 2, λ∞ = 0.465 and H/H0 = 1.217. Thus, the maximum sheet thickness over the minimum gap can be at most 33.4% for 1.0

5

Analytical

0.7

Newtonian fluid

λ= 0.4

Pressure, P/Pmax(λ ∞ )

λ∞

0.8

44

=0 .47

0.9

0.6

λ = 0.358

0.5 0.4

λ = 0.3

0.3 0.2 0.1 0.0 –3.50

–3.00

–2.50

–2.00 –1.50 –1.00 Dimensionless distance, x ′

–0.50

0.00

11.6 Pressure distribution for different leave-off distances λ (Newtonian model).

0.50

328


Pressure gradient, (dP ’/dx’)/(dp/dx ’)|x=0

Power-law fluids

n n n n n n n

1.0

0.5

= = = = = = =

1.0 0.8 0.6 0.5 0.4 0.2 0.1

0.0

–0.5

–1.0 –1.2

–0.9

–0.6 –0.3 Distance, x’ = x/(2RH0)1/2

0.0

0.3

11.7 Axial distribution of dimensionless pressure gradient for different values of the power-law index n for pseudoplastic shearthinning fluids (λ = 0.3).45 The same shape curves are valid for the shear stress distribution.

extremely shear-thinning fluids (n → 0), 22.6% for Newtonian fluids (n = 1), and 21.7% for shear-thickening fluids with n = 2. Figure 11.9 shows that for Newtonian fluids the values corresponding to an infinite reservoir are attained for Hf/H0 ≥ 20, while shear-thinning fluids require even higher ratios as the degree of shear-thinning increases (e.g., for n = 0.1, Hf/H0 = 1000 is not enough to attain the asymptotic values). It is interesting to note that for a certain value of Hf/H0 ≈ 15.85, the results do not depend on the power-law index n (crossover points of Fig. 11.9). The corresponding values are λ ≈ 0.473 and H/H0 ≈ 1.224. After this point, shear thinning increases these values, reaching the highest values as n → 0 and Hf/ H0 → ∞. The reverse is true for Hf/H0 < 15.85. On the other hand, shear thickening decreases the values, reaching for n = 2, λ∞ = 0.465 and H/H0 = 1.217, and this value appears to be reached for Hf/H0 ≥ 10. Thus, as pointed out by Middleman,36 in calendering with finite sheets it is no longer correct to say, without qualification, that the effect of non-Newtonian behaviour is to increase the sheet thickness, as this depends on the thickness of the entering sheet as well. The three engineering quantities P, F and E, given in dimensionless form, are shown in Fig. 11.10 as a function of the power-law index n. As n increases


329

2.2

n n n n n n n

2.0 1.8 1.6

Pressure, P ’/ P ′x=0

1.4

= = = = = = =

Power-law fluids

1.0 0.8 0.6 0.5 0.4 0.2 0.1

1.2 1.0 0.8 0.6 0.4 0.2 0.0 –0.2 –1.2

–0.9

–0.6 –0.3 Distance, x ’ = x/(2RH0)1/2

0.0

0.3

11.8 Axial distribution of dimensionless pressure for different values of the power-law index n for pseudoplastic shear-thinning fluids (λ = 0.3)45.

0.6 Power-law fluids

Leave-off distance, λ

0.5 0.4

n n n n n n n

0.3 0.2 0.1

= = = = = = =

0.1 0.3 0.5 0.8 1.0 1.5 2.0

0.0 1

10

100

1000

H f /H 0

11.9 Dimensionless leave-off distance λ as a function of the entering sheet thickness Hf /H0 for different values of the power-law index n for pseudoplastic fluids.45

330

Advances in polymer processing Power-law fluids E

/10

10 J

P

1

0.1

0.0

0.2

0.4

0.6

0.8 1.0 1.2 1.4 Power-law index, n

1.6

1.8

2.0

11.10 Operating variables as a function of the power-law index n for pseudoplastic fluids calendered from an infinite reservoir.45

all quantities decrease monotonically in dimensionless form. The well-known values for Newtonian fluids (n = 1) from an infinite reservoir (λ∞ = 0.475) are P = 0.3802, F = 1.2259 and E = 4.516; they are also given in Middleman.36 Viscoplastic fluids The corresponding calculations for viscoplastic fluids cover a wide range of Bn values, i.e., 0 ≤ Bn ≤ 1000. Figure 11.11 shows typical calculation results for the pressure-gradient distribution obtained for λ = 0.3 with the Bingham plastic model. As was the case with pseudoplastic fluids, viscoplasticity tends to flatten the distributions. Similar profiles are obtained for the shear stress distributions because of equation 11.18. The corresponding results for the pressure distribution are shown in Fig. 11.12, where viscoplasticity tends to give triangular pressure distributions in the calender gap and also increase the flow domain. Yielded/unyielded regions Because of the yield stress, viscoplastic fluids have the characteristics of both viscous fluids and plastic solids. The yield line y0 separates the two


331

1.5 Bingham plastics

Pressure gradient, (dP ’/dx ’)/(dP ′/dx ’)|x=0

1.0

0.5

0.0

Bn = 0 Bn = 0.01 –0.5

–1.0 –1.2

Bn Bn Bn Bn Bn –0.9

= = = = =

0.1 1 10 100 1000 –0.6 –0.3 Distance, x’ = x/(2RH0)1/2

0.0

0.3

11.11 Axial distribution of dimensionless pressure gradient for different values of the Bingham number Bn for viscoplastic fluids obeying the Bingham model (λ = 0.3).45 The same shape curves are valid for the shear stress distribution.

regions, called yielded and unyielded, respectively. In calendering, such regions have been shown only schematically by Gaskell6 (see Fig. 11.13) and by Chung.31 A thorough parametric study under the key assumption of LAT has been given by Sofou and Mitsoulis.45,46 With regard to Fig. 11.13, in calendering finite sheets of viscoplastic materials, we observe that the entering and exiting sheets have a constant thickness and move with a plug velocity as a solid plastic sheet. Therefore, they constitute unyielded regions (shaded) just before coming into contact with the rolls and after leaving them. The lubrication analysis also shows that plug velocity profiles exist at distance λ (exit) and –λ. Therefore, it follows that the yielded/unyielded regions have qualitatively the shapes shown in Fig. 11.13. As the sheet first bites the rolls at − x f′ , the material next to the rolls softens and becomes yielded (unshaded) due to the shearing motion at the rolls. The core speed increases, while the shear rate decreases, thus allowing some of the softened (yielded) material to solidify again (unyielded). When x′ = –λ, the speed of the core is that of the rolls, and the core extends from roll to roll, being totally unyielded.

332

Advances in polymer processing 2.2 2.0

Bn = 0 Bn = 0.01

1.8

Bn Bn Bn Bn Bn

Pressure, P ’/ P ′|x=0

1.6 1.4

= = = = =

Bingham plastics

0.1 1 10 100 1000

1.2 1.0 0.8 0.6 0.4 0.2 0.0 –1.2

–0.9

–0.6 –0.3 0.0 Distance, x’ = x/(2RH0)1/2

0.3

11.12 Axial distribution of dimensionless pressure for different values of the Bingham number Bn for viscoplastic fluids obeying the Bingham model (λ = 0.3).45 U 2H f

λ

H0 y0,0

y0 –λ

y0

y 0

Solid

2H

x

x

Liquid

11.13 Schematic representation of the yielded/unyielded zones in calendering of viscoplastic materials exhibiting a yield stress according to Gaskell.6 The shaded regions are unyielded, representing a plastic solid.

After the material passes through the unyielded state at x′ = −λ, the core speed continues to increase, giving rise again to shear forces, which soften the material near the rolls (yielded regions, unshaded). This effect continues until at the nip (x′ = 0) the core velocity is a maximum, but the core thickness


333

is a minimum there (yield-line location, y0,0). Beyond the nip, the reverse process takes place, and by the time the section x′ = λ is reached, the solid (unyielded) core extends from roll to roll again, and the calendered material leaves the rolls at this section as a plastic solid. This is an interesting result, as pointed out by Gaskell.6 The quantitative analysis by Sofou and Mitsoulis45,46 shows these findings as evidenced in Fig. 11.14 for different Bn numbers and λ = 0.3. For Bn = 0, we have purely viscous flow and the area is all yielded. At the other extreme, for Bn → ∞, we have a purely plastic solid, and the area is all unyielded. Any other value of Bn between 0 and ∞ gives both yielded and unyielded regions. For the finite-sheet case shown in Fig. 11.14, the abrupt change of the unyielded region at the point of contact is due to the limitations of the lubrication approximation, which is not valid there. It remains to be seen by performing a full two-dimensional analysis whether the results are like those shown schematically in Fig. 11.14 (see critique on LAT further down). Similar results are obtained for the Herschel–Bulkley fluids, but the yielded/ unyielded regions are less rounded due to the presence of shear thinning.45 A more pronounced shear-thinning effect (smaller n) produces even flatter interfaces between yielded/unyielded regions. Sheet thickness Results have been obtained for a wide range of entering sheet thickness ratios Hf/H0, from the limiting case of Hf/H0 = 1 (no thickness reduction) to the case of Hf/H0 = 1000 (approaching a sheet coming from an infinite reservoir). For Bingham plastics, the results for the sheet thickness H/H0 are shown in Fig. 11.15 for various Bn values. The Newtonian values (for Bn = 0) of λ∞ = 0.475 and H/H0 = 1.226, corresponding to calendering from an infinite reservoir, are also obtained when Hf/H0 = 1000, as expected. Figure 11.15 shows that for Newtonian fluids these values are attained for Hf/H0 ≥ 20, while Bingham plastics, like shear-thinning fluids, require even higher ratios as the degree of viscoplasticity increases (e.g., for Bn = 1000, Hf/H0 = 1000 is not enough to attain the asymptotic values). It is interesting to note that for a certain value of Hf/H0 ≈ 15.85, the results do not depend on the Bn value (crossover points of Fig. 11.15). The corresponding values are λ = 0.470 and H/H0 = 1.221. After this point, viscoplasticity increases the values. Thus, as was also the case with pseudoplastic fluids, it is no longer correct to say, without qualification, that the effect of non-Newtonian behaviour is to increase the sheet thickness. For Herschel–Bulkley fluids, sample results are given in Fig. 11.16 for the entry sheet thickness Hf/H0 at a given leave-off distance λ = 0.3 and for different values of the power-law index n. The entry sheet thickness is increased as the degree of shear thinning is increased and as viscoplasticity increases.

334

Advances in polymer processing 3

3

Bn = 0.01

Bn = 10 2

Yield line, y 0′ = y0 / H0


2 1 0 –1 –2 –3

1 0 –1 –2 –3

–1.0 –0.5 0.0 0.5 Dimensionless distance, x’ = x/(2RH0)1/2 3


Bn = 0.1

Bn = 100 2



2 1 0 –1 –2 –3

1 0 –1 –2 –3

–1.0 –0.5 0.0 0.5 Dimensionless distance, x’ = x/(2RH0)1/2


3

Bn = 1

Bn = 1000 2

1 0 –1 –2



2

1 0 –1 –2

–3

–3



11.14 Progressive growth of the unyielded zones as the dimensionless Bingham number Bn increases in calendering of viscoplastic materials obeying the Bingham plastic model. The shaded regions are unyielded. The exiting sheet thickness is fixed so that λ = 0.3.45


335

0.6

Leave-off distance, λ

0.5

0.4

0.3

Bn = 0 Bn = 0.01

0.2

Bn Bn Bn Bn Bn

0.1

= = = = =

0.1 1 10 100 1000

0.0 1

10

100

1000

H f /H 0

11.15 Dimensionless leave-off distance λ as a function of the Bingham number Bn for different values of the entering sheet thickness Hf /H0 for viscoplastic fluids obeying the Bingham model.45

3.0 Herschel–Bulkley fluids λ = 0.3

H f / H0

2.5

2.0

n = 0.1 n = 0.3 n = 0.5 n = 0.8 n = 1.0 n = 1.5 n = 2.0

1.5

1.0 0.001

0.01

0.1 1 10 Bingham Number, Bn

100

1000

11.16 Dimensionless entry sheet thickness Hf/H0 as a function of the Bingham number Bn for different values of the power-law index n and for a fixed leave-off distance λ = 0.3 for viscoplastic fluids obeying the Herschel–Bulkley model.45

336


The results show a convergence at high Bn numbers. In precision calendering such small differences are taken into account in order to produce sheets with very small thickness tolerances. More results can be found in the papers by Sofou and Mitsoulis.45–47

11.2.6 Critique on the lubrication approximation The key assumption for obtaining the above results has been the lubrication approximation theory (LAT), which assumes locally fully developed flow and reduces the conservation equations by using only the axial velocity component. This assumption is known to give good results for the pressure distribution in calendering of power-law fluids37 and hence for all integrated quantities resulting from that. It is therefore reasonable to assume that it also does well for viscoplastic fluids, such as those used in the present work. However, a closer look at the physics of the viscoplastic problem, and in particular the interesting yielded/unyielded regions proposed by Gaskell6 and found also here, reveals that these cannot be so and that Figs 11.13 and 11.14 are contentious. The shaded areas cannot be rigid plugs, since the speeds at entry and exit are different, and there is no chance that a constant speed occurs at the centreline. It is therefore obvious that the introduction of LAT in lubrication flows with viscoplastic fluids is not valid for such quantities as the yielded/unyielded regions, since it leads to a ‘paradox’, as pointed out by Lipscomb and Denn.60 A 2-D analysis is therefore essential in obtaining the correct regions. However, it is not expected to change drastically the other results shown here, especially the pressure distribution and integrated quantities, since calendering is primarily a lubrication flow.

11.3

Two-dimensional analysis

11.3.1 Governing equations The need for a fully two-dimensional (2-D) model of calendering became apparent because of the inability of the lubrication approximation to describe the flow for large entrance angles and to account for recirculation phenomena in the melt bank. The flow field is shown in Fig. 11.17, and a Cartesian coordinate system (x, y, z) is used with its origin at the centre of the minimum gap. Mitsoulis et al.37 used their MACVIP finite element program61 to solve the 2-D creeping flow equations 11.7–11.10, while the rheological model is the generalized Newtonian fluid with a viscosity function given by the powerlaw model. In such a case, the components of the stress tensor are given by:

τ xx = 2η

∂v x = ηγ˙ xx ∂x

11.39

Calendering of polymers y A

337

R2 2H 1

V2 2 H0

2 H2

x R

V1 2H E

E S

R1

11.17 Flow domain and notation for the two-dimensional flow analysis.37

τ yy = 2η

∂v y ∂y

= ηγ˙ yy

11.40

∂v y   ∂v x τ xy = η  + = ηγ˙ xy ∂ x   ∂y

11.41

and the apparent viscosity η is a function of the magnitude |γ˙ | of the rate-ofstrain tensor given for a power-law model by: ( n –1)/2 1 η(γ˙ ) = K |γ˙ | n –1 = K  I 2  2 

1 2 2 2  = K  (γ˙ xx + γ˙ yy + 2γ˙ xy ) 2 

( n –1)/2

11.42

where I2 is the second invariant of the rate-of-strain tensor γ˙ with components

γ˙ xx = 2

∂v y ∂v y ∂v x ∂v x , γ˙ yy = 2 , γ˙ xy = + ∂x ∂y ∂y ∂x

11.43

The apparent viscosity will normally be a function of temperature and may also depend on pressure.3 Under usual calendering conditions, however, pressure dependence of viscosity will be insignificant.

11.3.2 Boundary conditions The above equations need a set of boundary conditions for their solution. These are: 1. On the moving roll surfaces, the linear roll speed is imposed as the tangential component of the velocity, i.e., vt = U, vn = 0.

338


2. At the entry of the incoming sheet, the tangential velocity profile is constant, i.e., vt = U, vn = 0. 3. On the free surfaces at entry and at exit, no surface tractions are imposed and the normal velocity is zero, i.e., v ⋅ n = 0. 4. For the temperature, the roll temperatures are assumed known and constant, on the lower roll, Tl, and on the upper roll, Tu, while at entry the temperature of the entering sheet is known and constant, Tm.

11.3.3 Results and discussion Pseudoplastic fluids The fully 2-D analysis has the major advantage of finding the free surfaces present in calendering, and especially the free surface of the melt bank, which from experiments38,50 is known to be unusual and complicated. Mitsoulis et al.37 developed such a method by starting with an initial guess for the free surface resembling typical photographs of the melt bank. Then the free surface is found numerically iteratively by constructing streamlines, which have the property of zero normal velocity across them ( v ⋅ n = 0 ). Figure 11.18 shows the results for the Newtonian fluid, where two huge, almost symmetric vortices are present before the fluid passes through the nip. The overall symmetry of the process has been slightly disturbed by the feeding of the entering sheet on the lower roll in the system. Otherwise the problem would have been perfectly symmetric. A comparison of the fully 2-D non-isothermal simulations using the powerlaw model with slip at the roll surface against the experiments by Chauffoureaux54 given by Vlachopoulos and Hrymak30 is shown in Fig. 11.19. It should be noted that for roll speeds of about 7.4 cm/s, the characteristic shear rates of 170 l/s, and the ratio R/H0 = 417. It is interesting to note the good agreement between LAT and 2-D simulations for the pressure distribution. 0.4 Newtonian fluid 0.3 1.0

y -coordinate, cm

0.2 –3

0.1

–2.5

–2.0

–1.5

–1.0 0

0.0 0

15

–0.1 –0.2 –0.3

3 1.0

1.5

25

1.0

20 –1.0

1.0

–1.0 –0.4 –2.7 –2.6 –2.5 –2.4 –2.3 –2.2 –2.1 –2.0 –1.9 –1.8 –1.7 –1.6 –1.5 –1.4 –1.3 –1.2 –1.1 –1.0 –0.9 –0.8 –0.7 –0.6 –0.5 x-coordinate, cm

11.18 Streamlines for a Newtonian fluid.37


339

7.5 7.0 6.5

FEM

no slip slip

Theory without slip

6.0

Pressure, P (MPa)

5.5 5.0 4.5 4.0

Theory with slip

3.5 3.0 2.5 2.0 1.5 1.0

Experiment (rigid PVC)

0.5 0.0 –6.5 –6.0 –5.5 –5.0 –4.5 –4.0 –3.5 –3.0 –2.5 –2.0 –1.5 –1.0 –0.5 0.0 0.5 1.0 Dimensionless distance, x*

11.19 Comparison between predicted and measured pressure distribution for the data given by Vlachopoulos and Hrymak.30

The discrepancies with experiments reaching 40% are due mainly to 3-D effects as shown in Fig. 11.3c. The non-isothermal analysis gives better results when taking into account the roll temperatures and the incoming sheet temperature. The full analysis has the ability to give detailed results (as expected) for all field variables. For example, Fig. 11.20a shows the streamlines for rigid PVC, where there are two vortices in the melt bank, one of which is dominant, as found out experimentally. These streamline patterns compare very well with the experimentally observed ones given by Agassant and Espy38 (see Fig. 11.20b). Figure 11.21 shows the temperature field, seen through isotherms. The highest temperature of 184ºC is in the eye of the big vortex and is within the limits for critical temperature of PVC, which must not exceed 190ºC to avoid thermal degradation. Viscoplastic fluids The full 2-D analysis for viscoplastic fluids has been recently undertaken by Mitsoulis.48 The case of symmetric calendering is analysed with an entering sheet having a known Hf/H0 value. Free surfaces are present both at entry and at exit and must be found iteratively. The entering sheet is assumed to exit from an extruder and is taken up symmetrically by the rotating rolls. The

340


1.0 0.3

1.0

0.6 –1

y -coordinate, cm

0.4 0.2

–1

0.0

.5

I0

.0

0 1.

3°C T 3 = 16

0

–0.6

I0

0.57

7 1.0 1.5 1.25

–0.2 –0.4

Roll4

T4 = 183°C

–0.5

Roll3

–1.0

–0.8 1.0

–1.0 –1.2

Rigid PVC

–5.7 –5.5 –5.3 –5.1 –4.9 –4.7 –4.5 –4.3 –4.1 –3.5 –3.7 –3.5 –3.3–3.1 –2.5 –2.7 –2.5 –2.3 –2.1 –1.5 –1.7 –1.5 –1.3 –1.1 –0.5 –0.7 –0.5 –0.3 –0.1 0.1 x-coordinate, cm

0.3

0.5

(a)

(b)

11.20 (a) Streamline pattern for calendering of rigid PVC with slip37 (for the calendering conditions and property data of Vlachopoulos and Hymak30); (b) experimental streamline pattern for rigid PVC according to Agassant and Espy.38 1.0

183 °C 18 182 3

0.3

y -coordinate, cm

0.6 0.4 18

0.2 18

0.0 18

–0.2 –0.4

18

18 3

2

0

182

175

172

170 163°C

Roll3

–0.8

–1.2

178 180

183°C

2

–0.6 –1.0

Roll4 184

4

2 17

°C 16

3°C

Rigid PVC

–5.7 –5.5 –5.3 –5.1 –4.9 –4.7–4.5 –4.3 –4.1 –3.5 –3.7 –3.5 –3.3 –3.1 –2.5 –2.7 –2.5 –2.3 –2.1 –1.5 –1.7 –1.5 –1.3 –1.1 –0.5 –0.7 –0.5 –0.3 –0.1 0.1 0.3 0.5 x-coordinate, cm

11.21 Isotherms (same conditions as in Fig. 11.20a).37

point of attachment is thus fixed (Hf/H0 given), and the point of detachment is obtained from LAT, which has been shown by Zheng and Tanner39 to be a good enough approximation for the full 2-D flow domain length. Results are shown for the streamlines in Fig. 11.22 for Hf/H0 = 23.6 and for different Bn numbers. While there are big vortices for the Newtonian


y / H0

20

341

(a) Bn = 0

0 –20 (b) Bn = 0.1

y / H0

20 0 –20

y / H0

20

(c) Bn = 1

0 –20

(d) Bn = 10

y / H0

20 0 –20

(e) Bn = 100

y / H0

20 0 –20

y / H0

20

(f) Bn = 1000

0 –20 –280

–240

–200

–160

–120 x / H0

–80

–40

0

40

11.22 Fully 2-D results for the streamline patterns as the dimensionless Bingham number Bn increases in calendering of viscoplastic materials obeying the Bingham plastic model. The entering sheet thickness is fixed at the attachment point so that Hf/H0 = 23.6. The vortex in the melt bank, evident for Newtonian and low Bn flows, disappears for Bn > 0.5.

case (Bn = 0), these become smaller as Bn increases and disappear for Bn > 0.5. Also the entering sheet becomes first thicker and then thinner than the Newtonian counterpart, in order to accommodate the different amounts of viscoplasticity for the same attachment point as set by assuming Hf/H0 = 23.6. The yielded/unyielded regions are very interesting as shown in Fig. 11.23 for the whole flow field, and in blow-up in Fig. 11.24 near the nip

342


y / H0

20

(a) Bn = 0

0 –20

(b) Bn = 0.1

y / H0

20 0 –20

(c) Bn = 1

y / H0

20 0 –20

(d) Bn = 10

y / H0

20 0 –20

(e) Bn = 100

y / H0

20 0 –20

(f) Bn = 1000

y / H0

20 0 –20 –280

–240

–200

–160

–120 x / H0

–80

–40

0

40

11.23 Fully 2-D results for the progressive growth of the unyielded zones as the dimensionless Bingham number Bn increases in calendering of viscoplastic materials obeying the Bingham plastic model. The shaded regions are unyielded. The entering sheet thickness is fixed at the attachment point so that Hf/H0 = 23.6.

region and exit. Most of the unyielded zones are evidenced in the entering and exiting sheets, which move as rigid plugs with a constant velocity (incidentally not the roll speed U). Unlike LAT predictions, the vast areas between the rolls are yielded, and only small unyielded islands appear in the nip around the centreline. However, the pressure distributions are not very different from LAT and are not shown here. Hence, all other engineering quantities are well predicted by LAT.

y / H0

y / H0

y / H0

y / H0

y / H0

y / H0

Calendering of polymers 2 1 0 –1 –2 –8

(a) Bn = 0

–4

0

4

8

12

16

20

24

28

(b) Bn = 0.1

2 1 0 –1 –2 –8

–4

0

4

8

12

16

20

2 1 0 –1 –2 –8

–4

0

4

8

12

16

20

24 28 (c) Bn = 1

24

28

(d) Bn = 10

2 1 0 –1 –2 –8

–4

0

4

8

12

16

20

2 1 0 –1 –2 –8

–4

0

4

8

12

16

20

2 1 0 –1 –2 –8

343

24 28 (e) Bn = 100

24

28

(f) Bn = 1000

–4

0

4

8

12

16

20

24

28

x / H0

11.24 Blow-up of the results of Fig. 11.23 in the nip region and near the exit. The shaded regions are unyielded. The entering sheet thickness is fixed at the attachment point so that Hf/H0 = 23.6.

11.4

Three-dimensional analysis

It is well known that the thermoplastic material spreads laterally as it enters the calender gap. This is due to the drag-induced pressure build-up, which produces flow both in the machine and cross-machine (axial) directions. This was clearly shown by Unkrüer,50 and apparently the combination of these forward and lateral motions (see Fig. 11.3) is responsible for the complex flow patterns observed. The complete three-dimensional analysis with the finite element method was carried out by Luther and Mewes.44 It is indeed a very difficult problem, because the three-dimensional boundary (Fig. 11.25) is not known a priori and must be determined iteratively. Luther and Mewes44 showed all the relevant information for a 3-D flow of an SBR fluid modelled with the power law and calculated the 3-D spiral flow of the melt in the axial

344

Advances in polymer processing P0

End of roll Grid refinement M˙ 2

P0

h1 M˙ 1 V1

Y X Z V2

Free surface Plane of axial symmetry

11.25 Finite-element grid for the three-dimensional analysis employed by Luther and Mewes.44 The free surface in the melt bank is unknown a priori. End of roll

Velocity, Vz –1.8 m/min –1.2 m/min –0.6 m/min 0.0 m/min

Y X Z Plane of axial symmetry d = 0.5 m f = 1.0 n = 0.24 h2 = 0.5 mm h0* = 0.8 0.24 K = 142 000 Pa s v2 = 5 m/min W = 0.5 m

11.26 Streamline patterns found in the three-dimensional analysis by Luther and Mewes44 for an SBR fluid obeying the power law. Note the spiral flow in the z-direction.

(z–) direction of the machine, as shown in Fig. 11.26. Thus, all features found experimentally by Unkrüer50 and shown in Fig. 11.3 were also found in the full 3-D numerical simulations. To date, the work by Luther and Mewes44 remains the most complete work on calendering for power-law fluids in 3-D.


11.5

345

Viscoelastic effects

As mentioned in the literature survey, few attempts were made to account for the effect of viscoelasticity. While the importance of a Deborah number has been cited, the results are rather inconclusive. Mitsoulis et al.37 carried out some finite element calculations using the Criminale–Ericksen–Filbey (CEF) constitutive equation,62 which has the general tensorial form D γ˙ 1 τ = ηγ˙ +  Ψ1 + Ψ 2  {γ˙ ⋅ γ˙} – 1 Ψ1 D t 2 2 

11.44

where η, Ψ1 and Ψ2 are the viscosity and the first and second normal stress coefficients, respectively. These are functions of the magnitude of the rateof-strain tensor | γ˙ |. The operator D /Dt gives the corotational or Jaumann derivative

γ˙ = D t

D

γ˙ ij ∂γ˙ ij ∂γ˙ ik = + vk + 1 {ω ik γ˙ kj – γ˙ ik ω kj} D t 2 ∂t ∂xk

D

11.45

where ωij is the vorticity tensor given by

ω ij =

∂v j ∂v i – ∂x j ∂xi

11.46

Results were obtained for elasticity levels up to Deborah number, De = 1, where the Deborah number is defined by

De =

Ψ1 V = θγ˙ 2η H 0

11.47

where θ is a relaxation time of the material. The results were only slightly different from inelastic calculations (especially the pressure distribution), and no definite trends could be detected. The only other viscoelastic calculations worth mentioning were carried out by Zheng and Tanner39 with the viscoelastic Phan-Thien/Tanner model, which is more non-linear than the CEF model and possesses memory. The calculations were pursued up to De = 6, but again the results were slightly different from their inelastic counterparts. In particular, the biggest changes were found for the swelling of the exiting calender sheet reaching up to 5%, while the pressure distributions, roll-separating force and torque were less than 1–2% compared with their inelastic counterparts. This is not surprising, because calendering is essentially a steady shear flow, and viscoelastic effects are not bound to contribute much to the flow in the nip region between the rolls. That viscoelasticity is not important in calendering, especially for the

346


pressure distribution, can be ascertained from an order-of-magnitude analysis of the effect of normal stresses.3,36 Starting from the momentum equation 0= –

∂τ xx ∂p ∂τ xy + + ∂x ∂y ∂x

11.48

and adding and subtracting ∂τyy/∂x and then rearranging we get

∂(τ xx – τ yy ) ∂τ xy ∂( P – τ yy ) + = ∂x ∂y ∂y

11.49

The first term is of the order of magnitude [τxx – τyy]/[R] and the second term is of the order [τxy]/[H0] and their ratio is of the order  τ xx – τ yy   H 0   τ xy   R   

11.50

Since R would typically be 102–103 times larger than H0 and the stress ratio (τxx – τyy)/τxy probably no larger than 10 before the onset of flow instability,36 we conclude that the normal stresses do not contribute much to the total pressure. It is possible, however, that the normal stresses are responsible for flow pattern rearrangements in the melt bank and for enhanced swelling at the exit from the rolls for highly elastic materials. This problem is still an open question and has not been adequately addressed or solved.

11.6

Calendered sheet and film defects

There has been little information on the origin and the characteristics of defects in calendered sheets. The discussion that follows is based entirely on Agassant’s26 comprehensive study on PVC. The calendered PVC defects can be classified into the following categories: 1. Dimensional non-uniformities: There are thickness variations in the machine or cross-machine direction due to the tendency of the rolls to bend under large separating forces. Compensations for roll deflections are provided by using crowned rolls having a larger diameter in the middle than at the ends, roll bending or roll skewing (see Meinecke63). 2. Structural anomalies: PVC exhibits certain particulate and crystalline structure changes under the influence of elevated temperature and stress, which may lead to formation of defects in calendered sheets or films (see Fig. 11.27). 3. Mattness: This is a micro-irregularity or loss of surface gloss that appears only on the surface that is not in contact with the roll after the sheet leaves the calender gap. This defect is apparently related to the phenomena of mattness, sharkskin and melt fracture in extrusion through capillary


347

(a)

(b)

11.27 Surface defects in calendered sheets: (a) irregular patterns, (b) V-shapes (according to Agassant26).

and slit dies.64,65 Agassant26 found that the onset of this defect occurs at a constant maximum wall shear stress value of about 0.5 MPa, which is about twice the critical stress for slit extrusion65 of polystyrene. 4. V-shapes: These are surface thickness irregularities (up to 3 µm) in the form of more or less regular partially open V’s with their vertices at the centre of the sheet. According to Agassant26 these V-shapes are due to undulating motions in the melt bank that propagate from the centre to the edges. 5. Air bubbles: Bubbles of air are captured in the recirculating melt bank, pass through the calender gap and become elongated air enclosures in the sheet produced. High pressures that develop in the gap sometimes prevent these bubbles from passing through.

348

11.7


Conclusions

Most mathematical models of calendering are based on the lubrication approximation theory (LAT) in one direction for Newtonian and shear-thinning materials. Slip at the wall is apparently necessary for the description of molten PVC flow through a calender gap. These models give reasonable predictions of pressure distribution, torque and power consumption. Two-dimensional finite element calculations permit the determination of the free surface and the recirculating flow pattern in the melt bank. Threedimensional flow occurs as the melt moves both in the machine and crossmachine directions. So far there has been only one such work44 showing very interesting results for Newtonian and shear-thinning fluids (albeit with missing information). Viscoelasticity apparently is not important in the determination of pressure and separating forces but may play an important role in vortex pattern formation in the melt bank and in the determination of the swelling of the exiting sheet after the nip. The problem of attachment and detachment of the sheet in the rolls has received much less attention with only one work39 apparently addressing the issue. It seems, however, that the Swift boundary conditions for the detachment of the sheet at the exit as assumed in LAT may be quite adequate. There have been very few experimental studies, and most of them refer to materials with a limited knowledge of melt properties. There is clearly a need for more experimental measurements of pressures, separating forces, torques and power consumption with polymers and other materials, which have been subjected to a thorough rheological characterization. Also, more research on the origin and appearance of sheet and film defects is needed. Finally, new challenges in the area of calendering elastomers and polymer blends and composites must be addressed. Virtually no calendering studies have been done for the latter, which due to their complicated nature, may present extra difficulties during the process.

11.8

Acknowledgement

Financial support from NTUA in the form of a grant for basic research, under the code name KARATHEODORI, is gratefully acknowledged.

11.9

References

1. Elden R A and Swan A D (1968), Calendering of Plastics, New York, American Elsevier. 2. Marshall D I (1959), ‘Calendering’, in Bernhardt E C, Processing of Thermoplastic Materials, New York, Van Nostrand Reinhold, 380–404. 3. Tadmor Z and Gogos C G (1979), Principles of Polymer Processing, New York, Wiley.


349

4. Ardichvili G (1938), ‘Versuch der Rationalen Bestimmung der Bombierung von Kalanderwalzen’, Kautschuk, 14, 23–25, 41–45. 5. Finston M (1951), ‘Thermal effects in calendering of viscous fluids’, J Appl Mech, 18, 12–17. 6. Gaskell R E (1950), ‘The calendering of plastic materials’, J Appl Mech, 17, 334– 337. 7. Paslay P R (1957), ‘Calendering of a viscoelastic material’, J Appl Mech, 24, 602– 605. 8. McKelvey J M (1962), Polymer Processing, New York, Wiley. 9. Brazinsky I, Cosway H F, Valle Jr C F, Clark R, Jones R and Story V (1970), ‘A theoretical study of liquid-film spread heights in the calendering of Newtonian and power law fluids’, J Appl Polym Sci, 14, 2771–2784. 10. Alston W W Jr and Astill K N (1973), ‘An analysis of the calendering of nonNewtonian fluids’, J Appl Polym Sci, 17, 3157–3174. 11. Ehrmann G and Vlachopoulos J (1975), ‘Determination of power consumption in calendering’, Rheol Acta, 14, 761–764. 12. Ehrlich R and Slattery J C (1968), ‘Evaluation of power model lubricants in an infinite journal bearing’, Ind Eng Chem Fund, 7, 239–244. 13. Takserman-Krozer R, Schenkel G and Ehrmann G (1975), ‘Fluid flow between rotating cylinders’, Rheol Acta, 14, 1066–1077. 14. Renert M (1966), ‘Asupra teoriei procesului de calandrare a materialelor plastice’, Materiale Plastice, 3, 132–136. 15. Reher E O and Grader L (1971), ‘Zur Berechnung einer Isothermen Doppel-WalzenKalanderStrömung Nicht-linear-plastischer Medien’, Plast Kautsch, 18, 597–603. 16. Torner R V (1974), Grundprozesse der Verarbeitung von Polymeren, Leipzig, VEB Deutscher Verlag für Grundstoffindustrie. 17. Tokita N and White J L (1966), ‘Milling behaviour of gum elastomers: experiments and theory’, J Appl Polym Sci, 10, 1011–1026. 18. Chong J S (1968), ‘Calendering thermoplastic materials’, J Appl Polym Sci, 12, 191– 212. 19. White J L (1969), ‘Extrusion of polymer melts and melt flow instabilities: III. Theoretical analysis of extrusion through a slit die’, Rubber Chem Technol, 42, 691– 704. 20. Kiparissides C and Vlachopoulos J (1976), ‘Finite element analysis of calendering’, Polym Eng Sci, 16, 712–719. 21. Bekin N G, Litvinov V V and Petrushanskiy V Yu (1976), ‘Method of calculation of the energy and hydrodynamic characteristics of the calendering of polymeric materials’, Intern Polym Sci Tech, 3, T55–T67. 22. Kiparissides C and Vlachopoulos J (1978), ‘A study of viscous dissipation in the calendering of power-law fluids’, Polym Eng Sci, 18, 210–214. 23. Dobbels F and Mewis J (1977), ‘Nonisothermal nip flow in calendering operations’, AIChE J, 23, 224–231. 24. Agassant J-F and Avenas P (1977), ‘Calendering of PVC: Prediction of stress and torque’, J Macromol Sci – Phys B, 14, 345–351. 25. Bourgeois J-L and Agassant J-F (1977), ‘Calendering of PVC-defects in calendered PVC films and sheets’, J Macromol Sci – Phys B, 14, 367–385. 26. Agassant J-F (1980), Le calandrage des matières thermoplastiques, Doctoral thesis, Paris 6, Université Pierre et Marie Curie. 27. Dimitrijew J G and Sporjagin E A (1977), ‘Nichtisothermer Prozess des Asymmetrischen Kalandrierens von Polymeren’, Plast Kautsch, 24, 484–491.

350


28. Woskressenski A M, Krassowski W N, Sewastjanow L K, Kohlert C and Reher E O (1979), ‘Nichtisotherme Analyse des Kalandrierens von Elastomeren mit Hilfe einer Prozessintensivierenden Zusatzeinrichtung (Keil)’, Plast Kautsch, 26, 92–108. 29. Seeger R, Schnabel R and Reher E O (1982), ‘Zur Thermomechanischen Analyse des Kalandrierprozesses Strukturviskoser Medien am Beispiel von PVC-Mischungen’, Plast Kautsch, 29, 406–418. 30. Vlachopoulos J and Hrymak A N (1980), ‘Calendering poly(vinyl chloride): Theory and experiments’, Polym Eng Sci, 20, 725–731. 31. Chung T-S (1980), ‘Analysis for the calendering of Bingham plastic fluids’, J Appl Polym Sci, 25, 967–970. 32. Chung T-S (1983), ‘Analysis of the calendering of compressible fluids’, J Appl Polym Sci, 28, 2119–2124. 33. Suto S, Yamaguchi K and Fujimura T (1980), ‘Prediction of sheet thickness of calendered viscoelastic material’, Nih Reo Gakk (J Soc Rheol Japan), 8, 103–112. 34. Suto S and Fujimura T (1980), ‘Prediction of pressures in calendering of viscoelastic materials’, Kobunsh Ronb, 37, 627–632. 35. Suto S, Suginuma S and Fujimura T (1983), ‘Flow freezing of polymer melts in nonisothermal calendering’, Kobunsh Ronb, 40, 17–23. 36. Middleman S (1977), Fundamentals of Polymer Processing, New York, McGrawHill. 37. Mitsoulis E, Vlachopoulos J and Mirza F A (1985), ‘Calendering analysis without the lubrication approximation’, Polym Eng Sci, 25, 6–18. 38. Agassant J-F and Espy M (1985), ‘Theoretical and experimental study of the molten polymer flow in the calender bank’, Polym Eng Sci, 25, 113–121. 39. Zheng R and Tanner R I (1988), ‘A numerical analysis of calendering’, J NonNewtonian Fluid Mech, 28, 149–170. 40. Agassant J-F, Avenas P, Sergent J-Ph and Carreau P J (1991), Polymer Processing: Principles and Modeling, Munich, Hanser Publishers. 41. Tseng A A and Sun P F (1990), ‘A finite difference study of roll design in calendering processing’, Intern Polym Proc, 5, 292–299. 42. Levine L, Corvalan C M, Campanella O H and Okos M R (2002), ‘A model describing the two-dimensional calendering of finite width sheets’, Chem Eng Sci, 57, 643– 650. 43. Mewes D, Luther S and Riest K (2002), ‘Simultaneous calculation of roll deformation and polymer flow in the calendering process’, Intern Polym Proc, 17, 339–346. 44. Luther S and Mewes D (2004), ‘Three-dimensional polymer flow in the calender bank’, Polym Eng Sci, 44, 1642–1647. 45. Sofou S and Mitsoulis E (2004a), ‘Calendering of pseudoplastic and viscoplastic sheets of finite thickness’, J Plast Film & Sheeting, 20, 185–222. 46. Sofou S and Mitsoulis E (2004b), ‘Calendering of pseudoplastic and viscoplastic sheets using the lubrication approximation’, J Polym Eng, 24, 505–522. 47. Mitsoulis E and Sofou S (2006), ‘Calendering pseudoplastic and viscoplastic fluids with slip at the roll surface’, J Appl Mech, 73, 291–299. 48. Mitsoulis E (2008), ‘Numerical simulation of calendering viscoplastic fluids’, J Non-Newtonian Fluid Mech, 154, 77–88. 49. Bergen J T and Scott G W Jr (1951), ‘Pressure distribution in the calendering of plastic materials’, J Appl Mech, 18, 101–109. 50. Unkrüer W (1970), Beitrag zur Ermittlung des Drucksverlaufes und der Fliessvorgänge im Walzspalt bei der Kalanderverarbeitung von PVC Hart zu Folien, PhD Thesis, TU Aachen, IKV.


351

51. Prentice P (1981), ‘Surface irregularities of calendered polypropylene’, Polymer, 22, 250–254. 52. Hatzmann G, Herner M and Müller G (1975), ‘Verarbeitung von PVC auf Kalandern. Formgebung im Walzenspalt’, Ang Makrom Chem, 47, 257–267. 53. Kopsch H (1975), ‘Viskositätsverhalten von PVC beim Kalandrieren’, Ang Makrom Chem, 47, 269–278. 54. Chauffoureaux J C (1980), Solvay & Cie S.A., Brussels, private communication to J. Vlachopoulos. 55. Kalyon D M, Gevgilili H and Shah A (2004), ‘Detachment of the polymer melt from the roll surface: calendering analysis and data from a shear roll extruder’, Intern Polym Proc, 19, 129–138. 56. Pearson J R A (1966), Mechanical Principles of Polymer Melt Processing, Oxford, Pergamon Press. 57. Bird R B, Dai G C and Yarusso B J (1982), ‘The rheology and flow of viscoplastic materials’, Rev Chem Eng, 1, 1–70. 58. Mitsoulis E, Abdali S S and Markatos N C (1993), ‘Flow simulation of Herschel– Bulkley fluids through extrusion dies’, Can J Chem Eng, 71, 147–160. 59. Gerald C F and Wheatley P O (1914), Applied Numerical Analysis, 5th edn, New York, Addison Wesley. 60. Lipscomb G G and Denn M M (1984), ‘Flow of Bingham fluids in complex geometries’, J Non-Newtonian Fluid Mech, 14, 337–346. 61. Mitsoulis E, Vlachopoulos J and Mirza F A (1984), MACVIP – A Finite Element Program for Creeping Viscoelastic Flows, PhD Thesis, Hamilton, Ontario, McMaster University. 62. Bird R B, Armstrong R C and Hassager O (1987), Dynamics of Polymeric Liquids: Vol. I, Fluid Mechanics, 2nd edn, New York, Wiley. 63. Meinecke E (1965), ‘Calendering’, in Encyclopedia of Polymer Science and Technology, Vol. 2, New York, Wiley, 802–819. 64. Petrie C J S and Denn MM (1976), ‘Instabilities in polymer processing’, AIChE J, 22, 209–236. 65. Vlachopoulos J and Chang T W (1977), ‘A comparison of melt fracture initiation conditions in capillaries and slits’, J Appl Polym Sci, 21, 1177–1187.

12 Thermoforming of polymers P J M A R T I N, Queen’s University Belfast, UK

Abstract: Thermoforming is one of the fastest growing of the major polymer processing techniques. From a relatively low technological base, advances in technology have helped to transform the process in recent years. Developments have included new highly automated processes that can create products ranging from micro-scale packaging to large hollow structural parts. These are now being made from complex blended or composite polymer formulations, or from new biodegradable polymers. The industry has also benefited from advances in tooling materials, instrumentation, test methods and process simulations. This chapter reviews these changes in technology and examines the future prospects for the process. Key words: thermoforming, materials, simulation, instrumentation, thermoformability.

12.1

Introduction

The term thermoforming has come to refer to any polymer process where a preformed polymer shape is reheated to a softened state, then further deformed before it is rapidly cooled to retain a new shape. In the vast majority of thermoforming applications the preform is an extruded sheet, and the deformation of a heated polymer sheet is the most familiar image of thermoforming. However, thermoforming processes are also employed very successfully in the forming of other preform shapes such as tubing for automotive fuel and brake lines. Thermoforming has a long history and dates back to the early days of large-scale industrial polymer processing in the 1930s (Throne, 1996). In these early days the process quickly established itself because of its ability to rapidly produce parts at very low capital costs. However, such advantages also gave the process an unfortunate reputation for producing low cost or inferior parts and for much of the twentieth century this was the prevailing view of thermoforming. Over the past 25 years this view has changed dramatically because of the rapid expansion in the use of polymer materials across an ever-widening range of applications and the major improvements in processing technology that have helped to transform the thermoforming industry. All polymer processes have benefited from advancing technology, but developments in thermoforming have been different for two fundamental reasons: 352


353

1. Most thermoforming processes are carried out with the material in the solid phase rather than the melt phase. This means that many of the advances in the major melt phase processes, such as injection moulding and extrusion, are not directly applicable to thermoforming. This is particularly true of most of the familiar laboratory test techniques and standard measurements like melt flow index (MFI). While some of these can be used to give an indication of thermoforming behaviour, more often they are not reliable because they are not measuring solid material responses. Some research work has specifically pursued technology purely for solid-phase thermoforming, but overall the published research on solid deformation behaviour is much less extensive and less advanced than that for the melt phase. 2. Thermoforming requires a preform (usually a sheet), which is normally produced by extrusion. Since thermoforming usually involves processing the polymer in the solid phase, much of its extruded structure, such as orientation and crystallinity, remains at forming temperatures and is a major factor in determining both the processing and the final properties of the part. This is particularly important when thermoforming semicrystalline polymers such as polypropylene and PET. Thermoforming is therefore very dependent on the preceding extrusion process and it is common for problems in thermoforming to ultimately lie with the earlier extrusion process. The purpose of this chapter is to give the reader an overview of the latest technological developments in thermoforming. Compared to the major polymer melt manufacturing processes, the body of research literature on thermoforming is small. However, technology has advanced rapidly through a combination of fundamental research, a small vibrant industry and a supportive network of technology-driven machine builders and material suppliers.

12.2

Developments in thermoforming processes

Early thermoforming processes were mainly variants of vacuum forming, where a heated sheet was drawn under vacuum pressure to take the shape of a cooled mould. As products became more sophisticated and required greater depth of draw, processes increasingly required the action of a mechanical plug to pre-stretch the sheet, creating the plug-assisted thermoforming process. This is now the most familiar and important industrial thermoforming process. However, there are many other thermoforming processes that are largely distinguishable by the different combinations of mechanical movement and applied air pressure that are used to create thermoformed products. The plug-assisted thermoforming process is illustrated in Fig. 12.1. Thermoforming takes place in two steps. Firstly, the heated polymer sheet is pre-stretched by

354


the movement of a mechanical plug that pushes material partly into the mould (plugging phase). Secondly, positive air pressure is applied from above, stripping the material from the surface of the plug and onto the mould wall where it cools (pressure phase). The principal objective of the plugging step is to try to distribute material more evenly in the part by using the plug to capture and push material towards the furthest extremities of the mould. The variation in the wall thickness distribution at three different plug temperatures for a simple thin-walled cup made by plug-assisted thermoforming is shown in Fig. 12.2. In this case the wall thickness was measured from the centre of the base of the cup through to its top lip. Although the initial aim is usually to try to achieve an even wall thickness distribution in the product, this is almost impossible to achieve in practice. For this example it is clear

Plug Air pressure Heated sheet

Final product

Mould

Start

Plugging phase

Pressure phase

12.1 Plug-assisted thermoforming.

1400 100°C

75°C

50°C

Wall thickness (mm)

1200 1000 800 600 400 200 0 0

20

40 60 80 100 Distance from base centre (mm)

120

140

12.2 Variation of wall thickness distribution with plug temperature in a thermoformed cup.


355

that by varying the temperature of the plug, it is possible to make large changes in the product wall thickness. Furthermore the distribution of material in the product is greatly influenced by the shape and properties of the plug and the various process settings used. These effects are well documented in the literature (Collins, et al., 2002; Martin and Duncan, 2007) and experimental research of this kind is helping thermoforming companies to make more effective use of their processes. The lack of controlled wall thickness distribution is a major limitation of all thermoforming processes.

12.2.1 The thermoforming industry The thermoforming industry worldwide is dominated by small to mediumsized companies (SMEs) and although there has been some consolidation in sectors such as packaging, the majority of companies are specialised producers of small ranges of products. The industry may be most conveniently divided into two main parts: 1. Thin-gauge thermoforming: This generally refers to the processing of extruded sheets of thickness less than 2.5 mm. Products are made from roll-fed sheet in very large volumes and at high cycle rates. The most important processing objectives for this sector are to maximise production speeds and to minimise material costs through downgauging. Packaging is the dominant application. 2. Heavy-gauge thermoforming: This generally refers to the processing of extruded sheets of thickness greater than 2.5 mm. Products are made in batches from cut sheet and cycle times are longer. The most important processing objectives for this sector are to maintain close control of temperature, on both heating and cooling, and to ensure high quality surface finish and replication of detail. Worldwide the majority of separate thermoforming companies are concentrated in the heavy-gauge category but typically these are very small organisations that employ fewer than 50 people. Larger companies tend to be thin-gauge thermoformers, particularly in the packaging sector, with some having thousands of employees across multiple sites. Production volumes for thin-gauge thermoforming can be very high with output measured in millions of products. In contrast heavy-gauge thermoforming usually involves much shorter production runs and larger parts may be produced in very small batches.

12.2.2 Advances in thermoforming machines Modern machinery for thermoforming has benefited greatly from general advances in machine technology such as automation systems, computer controls

356


and advanced motor drive systems. These are making it possible for thermoforming machines to become larger, faster, and at the same time more accurate and reliable. New thermoforming machines for packaging can now achieve output rates of greater than 100 000 products per hour and are approaching 50 production cycles per minute. Some specific advances in thermoforming machines are the use of highly controllable electric motor index drives, third motion plug activation (plug has separate drive), enhanced mould cooling systems and adjustable clamp frames. Heater panels that were traditionally powered by durable but slow-response ceramic electric or gas catalytic elements are increasingly being replaced by fast-response, programmable quartz halogen or tungsten lamps. In tandem with the advances in thermoforming processes, major steps forward have also been made in post-production operations such as part trimming, collection and recycling of skeletal waste, stacking, printing and packaging. Most of these operations are now wholly automated on new machines, eliminating the need for labour, and thereby significantly reducing costs. Advances include the use of laser and water jet systems for part trimming, robotics and automated movement systems for trimming and stacking, Thermoformed packaging has also benefited immensely from advances in printing technologies that mean that near-photo quality images can be reproduced on products. These advances have been taken further by eliminating the need for a separate downstream decoration stage. This is achieved by inserting pre-printed polymer labels into the mould cavity prior to thermoforming. This is known as in-mould labelling (IML) and it is now an increasingly common feature of thermoformed packaging (Grande, 2007). As the labels are initially flat before folding into the cavity, they can be produced with the highest print quality possible. The presence of the label also adds stiffness to the product, meaning that significant sheet downgauging is possible. For the matching lids for these IML containers, which have very shallow depths of draw, it is now possible to preprint the sheet prior to thermoforming and achieve equivalent print quality in the final combined product.

12.2.3 Twin-sheet thermoforming One of the newest variants of thermoforming is twin-sheet thermoforming, where two sheets are heated simultaneously, clamped and welded together around their periphery before being blown to either side of a mould to form a hollow part. The process is already well established for heavy-gauge applications such as pallets, truck bed liners and equipment panels. It has been shown to be particularly advantageous in manufacturing complex multilayer parts such as automotive fuel tanks (aus der Wiesche, 2004). There are few applications to date in thin-gauge thermoforming but the


357

process has much potential to develop further into markets presently occupied by blow moulding.

12.2.4 Micro-thermoforming Although micro-moulding processes, such as micro-injection moulding, have been around for many years and have well-established applications, it is only very recently that work on micro-thermoforming has been reported (Heckele and Schomburg, 2004). A number of methods are under development for applications in medicine and electronics using specially adapted features of conventional thermoforming processes (Giselbrecht et al., 2004; Truckenmueller et al., 2006). In another example the microforming process under development is a combination of thermoforming and compression moulding (Frick et al., 2008). At the micro scale one of the most difficult challenges is the method of heating used to soften the material and its means of control.

12.3

Products and markets

Over the past 20 years the volume of products made by thermoforming has increased steadily and the sophistication, complexity and quality of the parts produced has increased very markedly. This rise may be partially attributed to the general rise in the use of polymers across all industrial sectors and all polymer processes. However, in many areas thermoforming is now a real competitor to injection moulding, not just for lower costs but for equivalent quality, and through developments like twin-sheet thermoforming it is starting to make significant inroads into markets that were exclusively occupied by blow moulding.

12.3.1 Packaging products Packaging is the largest thermoforming market, with more than 75% of output devoted to this sector. Production of food packaging is a particularly strong and vibrant market, where the process competes very effectively against injection moulding. Changes in food packaging have been very rapid in recent years as any visit to a supermarket or food store will attest. As well as more sophisticated individual designs of packaging and quality of printing, thermoformed containers are now being used over a much wider range of products. For example, metal cans and glass jars are increasingly being replaced by lighter, more functional polymer-based packaging designs, and societal changes are leading to a much greater demand for convenience packaging. However, such applications are considerably more demanding in terms of the performance of the thermoformed product. Requirements include

358


greater shelf-life (barrier performance), tamperproof and resealable enclosures, low-temperature performance for frozen foods, and heat stability for processes like retort, pasteurisation, sterilisation, and microwave or conventional oven cooking. Performance requirements are even greater in the increasingly important medical packaging sector, which has very stringent barrier and sterilisation requirements. Environmental concerns have also started to change the packaging market quite significantly in recent years and this is being reflected in both the design of food packaging and the materials that are being thermoformed.

12.3.2 Technical parts Developments in industrial thermoforming applications have tended to be led by the blossoming automotive sector and to a lesser extent by aerospace applications. In both cases demands on part quality are extremely high and this has forced heavy-gauge thermoformers to greatly improve their processes. Automotive applications for thermoforming have grown particularly quickly and in this sector thermoforming is seeing some of its most advanced technical challenges. Structural components such as automotive bumpers have been manufactured by thermoforming for some time, but increasingly the materials being formed are reinforced through the addition of short glass or carbon fibres. These significantly enhance the mechanical properties of the polymer but this comes at the expense of its ease of thermoforming. Even greater forming difficulties are presented by the addition of layers of woven fibres or fabrics to the polymer, but such materials are now being successfully thermoformed (Lussier and Chen, 2002). Most of these new thermoformed composites are for automotive applications but examples are starting to appear in structural aircraft components (Conway and Clark, 2005) and in sporting goods such as kayaks and canoes. In many cases the forming process used is actually a combination of thermoforming with either stamping or compression moulding (Lebrun et al., 2004). Advances in laminating now mean that very high quality coloured surfaces can be thermoformed for interior and exterior automotive trim. These advances are particularly noticeable in the interior of modern vehicles, where thermoformed panels for dashboard components now convincingly give the surface appearance of either metal or wood (Sherman, 2005). Similar technology is also being exploited in the thermoforming of cases for consumer electronic products, such as flat-screen televisions, and for decorative bathing and shower enclosures. The performance of many of these thermoformed panels is also being enhanced further through the use of layered composite structures. As well as enhanced mechanical properties, multilayer structures are also being successfully exploited in the manufacture of components requiring a barrier to volatile chemicals such as in automotive fuel tanks and


359

fuel lines. A new process has also been recently reported for the manufacture of a lightweight, multilayer sandwich composite containing a thermoformed honeycomb core (Bratfisch et al., 2007).

12.4

Polymers for thermoforming

Twenty years ago styrenic polymers like acrylonitrile-butadiene-styrene (ABS) and polystyrene (PS) were the most important materials for thermoforming, with PS dominating the thin-gauge packaging sector and ABS the heavygauge industrial sector. Since then the range of materials used has expanded rapidly. Early thermoforming was carried out almost exclusively with amorphous polymers because of their ability to cope with the inadequacies of the forming processes at the time. Typically such materials have a broad softening temperature range over which they may be thermoformed. In comparison many semi-crystalline polymers exhibit rapid softening at their crystalline melting temperature, which leaves them with an extremely narrow temperature range over which they are soft enough to successfully thermoform. These differences are illustrated in Fig. 12.3, which shows the change in the stiffness (bending modulus) for a number of polymers with increasing temperature (Macauley et al., 1996). The data was obtained from DMTA tests. In the case of semi-crystalline polypropylene (PP) the temperature forming window is around 10°C, as the material must be formed in the region of sharp decline of modulus at its crystalline melting temperature (around 167°C). Amorphous polystyrene, on the other hand, exhibits a plateau

9.5

8.5

log E*

7.5

6.5

PP PMMA PC

5.5

PVC PS

4.5 –50

0

50 100 Temperature (°C)

150

12.3 Change in modulus with temperature for common polymers.

200

360


region beyond its glass transition temperature of around 100°C, at which its modulus is relatively stable and it is easy to thermoform. Similar patterns of forming windows may be observed with other amorphous and semi-crystalline polymers. These improvements in the understanding of forming windows for different materials coupled with greatly improved sheet heating systems on modern machines have helped to expand the range of materials that may be routinely thermoformed.

12.4.1 Advances in packaging materials For new food packaging applications polypropylene is now becoming the principal material of choice and the use of amorphous polymers such as PVC and polystyrene is declining for environmental and economic reasons respectively. PP is also significantly cheaper per tonne than its rivals, but until recently its narrow forming window and temperature sensitivity had been a serious deterrent for many processors. Most recently the use of PET for thermoforming has increased very sharply. This has been partly driven by its clarity (aPET), good mechanical properties and high temperature service (cPET), but the most powerful impetus has come from environmental pressure. Uniquely among the major commodity polymers, PET has an already established and economically viable route for recycling of household waste, and this has generated a surge in demand from food retailers eager to be seen to be using a packaging material that is perceived to be environmentally friendly. Pressures have been so high that over the last year the price of recycled PET (rPET) has reportedly exceeded that of virgin PET. With increasing environmental pressures and rising costs of raw materials, most packaging producers are now looking to use less material in their products. This drive to downgauge is being recognised by resin suppliers who are producing higher stiffness grades specifically for thermoforming. Property improvements are also being sought through blending of materials. For example, use of blended compositions and additives to enhance the natural clarity of PP in many packaging applications is an ongoing area of intense research interest (Sherman, 2003). Many processors also frequently blend miscible grades of polymers together to create hybrid compositions with enhanced properties. For example, homopolymer PP is often blended with copolymer PP to create a more rigid final material. Multilayer coextruded sheet is now commonly employed in thermoformed packaging and the technology of extrusion has advanced to a point where structures consisting of more than six layers may be routinely manufactured. Often these extrusion processes are carried out immediately prior to thermoforming or in-line in a combined extrusion and thermoforming operation. Typically the different layers in the sheet are designed to provide slightly different functions according to their location, e.g. printability, sealability


361

and permeability. Layers may also be included to hide slightly degraded inhouse recycled materials. Providing an effective barrier to oxygen by greatly reducing permeability is particularly important in many food packaging applications. One of the most effective barrier technologies is the introduction of a thin layer of ethylene vinyl alcohol (EVOH). As this is not chemically compatible or miscible with common materials such as PP, it is also necessary to add adhesive, or tie layers, that sandwich the EVOH on either side to provide a five-layer structure. An example of the structure of a typical PP/ EVOH coextruded sheet is shown in Fig. 12.4. This shows a photograph of the cross-section of a 1.27 mm thick sheet clearly indicating the relative thicknesses of the individual layers. In this case a thin white EVOH layer is sandwiched on either side by fine dark adhesive layers, a PP regrind layer on the left and then finally virgin PP capping layers on either side. Use of such technology has increased the typical product shelf-life for PP packaging from several days up to more than a year, and this change is now helping thermoformed packaging to replace more traditional canning and bottling technologies.

12.4.2 Advances in engineering materials For non-packaging applications ABS is the most widely used polymer in thermoforming because of its combination of excellent mechanical properties

EVOH and tie layers

PP

Regrind PP

PP

0.5 mm

12.4 Multilayer PP/EVOH barrier sheet for food packaging.

362


and ease of forming. It is particularly important in automotive applications where it is used for both structural and non-structural components. Other important applications are in panelling for aircraft, sporting goods and cases for domestic appliances. In more decorative applications, or where mechanical performance is less important, PS is a lower-cost alternative. Other major thermoforming polymers are polycarbonate, used as a replacement for ABS because of its toughness and durability, and PMMA, which is widely used in lighting and display panels due to its very high clarity. All of the main packaging materials like PP, PVC and PET are also finding wider engineering applications, and other polymers such as polyethylene, which was almost unheard of in thermoforming in the past, is now increasingly being used. Much of this growth in new markets has been driven by general improvements in thermoforming technologies and the much higher levels of confidence that companies have in their process. In addition, material suppliers have also introduced an ever wider range of grades of their polymers which are optimised specifically for thermoforming. Although the use of most of the traditional thermoforming polymers is continuing to rise at well above the average growth rate for all polymer processes, some of the most dramatic changes in thermoforming markets have come through the introduction of new blended polymers and new polymer composites. The number and complexity of blended polymers available for thermoforming has expanded very rapidly in recent years due to developments in compatibilisers. It is now possible to blend previously immiscible polymers such as polypropylene (PP) and polyphenylene oxide (PPO) to create thermoformable materials with much wider ranges of engineering properties. Other important examples are the increasing use of PC/ABS blends for automotive applications, and PMMA/ABS and PMMA/PVC for pools, spas and bathroom fittings. New research is also examining the potential for creating composite structures by carefully blending immiscible polymers. For example, small particles of one polymer in another can become small fibrils through orientation and stretching. This has been observed in a composite of immiscible poly(vinyl alcohol) (PVA) containing nanoparticles of polytetrafluoroethylene (PTFE) which gives enhanced mechanical properties and resistance to degradation (Aveila et al., 2004). One of the fastest growing groups of materials at the moment are thermoplastic polyolefins (TPOs), which are blended polymers based on PP and PE. These are now widely used in automotive applications and are notable because they are readily extruded with a high-gloss cap layer to create ‘class A’ surface finish automotive parts (Grande, 2006). TPOs were once difficult to process but new formulations are readily thermoformed (Hogan et al., 2007). Similar technologies for coextrusion or laminating of paint or decorative films are also used with ABS and PC.


363

12.4.3 Recycling Recycling of in-house scrap has been an important feature of most thermoforming processes for many years. Skeletal waste in packaging can account for 15–50% of the original sheet and it must be reused for economic reasons. Generally the left-over material or scrap products are simply shredded and reintroduced into the sheet extrusion stream. However, the quality of this ‘regrind’ material can be significantly lower than that of the original virgin polymer due to chain scission and degradation of properties (Incarnato et al., 1999). In the case of multilayer sheet or printed materials the quality of the regrind can be so poor as to render the material unfit for the original purpose. For example, barrier PP sheet containing EVOH can be very badly affected, particularly in terms of its clarity (Faisant et al., 1998). With food packaging, legislation also dictates that recycled material cannot come into contact with the packaged foodstuff and must therefore be hidden within virgin capping layers. External recycling of materials for thermoforming was almost unheard of until very recently, but it is now starting to become an important issue for all types of polymer processing. In Europe the European Commission has introduced a number of recycling directives and resulting landfill taxes that are beginning to drive the market. Other countries worldwide are starting to follow suit. These measures have had most effect on the packaging sector but other initiatives, such as the end-of-life vehicles (ELV) directive, are now starting to dictate how the automotive sector manufactures and recycles its products. For packaging current EU targets require a minimum of 60% of packaging waste to be recovered or incinerated and between 55 and 80% of packaging waste must be recycled by the end of 2008.

12.4.4 Biodegradable materials In terms of environmental impact the alternative to recycling of polymer materials is the use of new biogradable polymers made from renewable sources. These have only started to emerge as realistic alternatives to conventional polymers within the last five years but they have now grown into an area of intense commercial interest. This is being driven by the growing public awareness of environmental concerns and the desire to reduce the dependence on crude oil. In terms of chemistry there are many potential natural sources and types of biodegradable polymer, and in some cases simple forms of such materials have been used for many years. However, few of these materials have mechanical properties to rival those of conventional polymers and in many cases it would be extremely difficult to produce them economically and in equivalent volumes with existing technology. To date the most commercially successful of these new polymers is poly(lactic acid) (PLA), which is now being widely introduced as a replacement packaging

364


material. It is made from either corn starch or sugar cane and its clarity, thermoformability and mechanical properties are similar to those of polystyrene (Bosiers and Engelmann, 2003). Its current cost is higher than that of the mainstream polymers but this difference is expected to close, especially with increasing oil prices. Currently the greatest impediment to growth is its lack of availability. Despite very substantial recent investments worldwide in new synthesis plant, production volumes are still only a small fraction of those of conventional polymers and well below current market demand. In addition the same raw materials are also being increasingly used to produce biofuels, which has added to the shortages of supply. An alternative material made from similar sources is thermoplastic starch. This has also found some applications in packaging, but it is a much weaker and more delicate material than PLA. It is very moisture sensitive and degrades rapidly in the presence of water, which greatly restricts its applications. Control of its degradation rate is an important area of ongoing research (Avérous et al., 2001). Starchbased biopolymers have also been blended with conventional polymers like PP to create new thermoforming materials (Naitove, 2007). While these are not biodegradable, they have been of some market interest as they allow manufacturers to increase their use of materials from renewable sources. A different approach again is the use of oxo-biodegradable additives that catalyse the degradation of conventional polymers like PP. These can allow the polymer to slowly degrade in the environment but they suffer from the fundamental problem that the polymer is not from a renewable source and is therefore not carbon neutral. It is also worth highlighting that many materials that are described as biodegradable are only degradable under ideal composting conditions.

12.4.5 Foamed materials Thermoforming of polymer foams has been around for many years and has established applications in simple polystyrene food trays and automotive seats. However, the technology has yet to establish wider applications. This has been partly because of difficulties in forming cell structures with the correct size and distribution to survive the thermoforming process, but more importantly the relatively high cost of the extruded foam sheet has been a major disincentive (Dixon et al., 2001). Recently some new applications have started to emerge for the thermoforming of foam sandwich composites and in using foams to create a core in twin-sheet thermoformed parts.

12.4.6 Nanocomposites Over the past five years polymer nanocomposite materials have been one of the most intense areas of academic research. With this technology it is possible


365

to create new classes of materials with enhanced mechanical and electrical properties, for example to give clear packaging improved barrier performance without any reduction in clarity, or to create electrically conductive polymer components. A very large amount of research literature has explored the structure and properties of these materials, which are formed by embedding nanometre-sized particles of various materials in a polymer matrix. Much of the work reported has been concerned with the interaction of the nanoparticles and the polymer, and great difficulties have been encountered in creating optimum material structures by conventional polymer processing methods like injection moulding and extrusion. For this reason progress has been relatively slow. A small number of commercial applications for the new technology are starting to emerge but to date little of this work has had any impact on thermoforming. The most promising research related to thermoforming has concerned the effects of stretching on nanocomposite sheet. Recent work has clearly shown that biaxial stretching of polymer sheet loaded with a nanoclay filler does lead to significant changes in the distribution and structure of the nanoclay particles (Jéol et al., 2007; Rajeev et al., 2008). These effects are illustrated in Fig. 12.5, which shows TEM micrographs of a PP/montmorillonite nanoclay composite before and after biaxial deformation. The clay particles are clearly orientated by the stretching process and there is also evidence of improved intercalation and dispersal of the clay platelets. Related studies have shown that these changes are accompanied by increases in mechanical and barrier properties (HarkinJones et al., 2008). Recently reported work is also starting to examine the

12.5 TEM images of the structure of a PP/montmorillonite nanoclay composite before and after biaxial deformation (×28 500).

366


potential for creating new thermoforming materials by combining the potential benefits of biogradable polymers with nanofillers (Angellier-Coussy et al., 2008).

12.5

Advances in tool materials

Like all polymer processes, thermoforming has benefited significantly in recent years from advances in machining technologies and computer-aided design. These have enabled ever more complex mould features to be modelled and manufactured, and this has been a major factor in enabling thermoforming to move away from its traditional roots in low-cost packaging into higher precision and quality products for medical and automotive applications. A major advantage of thermoforming is that forming pressures are relatively low and thus moulds may be made from virtually any material. Non-metals such as wood and plaster are particularly common in low-volume production and in the manufacture of prototypes prior to full production. Some work has been reported on the use of the latest rapid prototyping techniques such as stereolithography and laser sintering (Levy et al., 2003) for thermoforming, but the relative expense of the models produced has so far limited these applications (Grande, 2008). Figure 12.6 illustrates an example of a prototype mould for a plastic cup (a) that was created by a rapid tooling technique. In this case the female cavity mould was produced from a shell of low melting point metal alloy (b) sprayed onto a machined male metal former (c), which was embedded in an epoxy resin/aluminium powder potting compound and machined to fit the tool cavity (d). The resulting tool was successfully used for many hours of continuous production without any noticeable reduction in product quality, and the same technique has been successfully used for rapid evaluation of other product concepts.

12.5.1 Mould materials As production speeds and volumes increase in thermoforming the need to quickly remove heat from the formed part becomes the most important consideration and moulds are generally made from machined aluminium alloys. In many cases such tooling is temperature controlled by the flow of a chilled fluid through specially designed cooling channels in the mould block. One unique feature that must be provided by tooling for thermoforming is the ability of trapped air to escape from between the deformed sheet and the mould wall. This is normally achieved by careful positioning of small vent holes across the features of the mould surface. However, new microporous mould materials have recently been introduced that enable air to pass freely through any part of the mould surface. This is claimed to produce parts with extreme levels of detail, accuracy and definition. Porous ceramic materials


(a)

(b)

367

(c)

(d)

12.6 Manufacture of a rapid prototype mould for thermoforming.

have been used in this way, but the most common microporous materials used in thermoforming consist of a formulation of aluminium powder in an epoxy binder that may be readily machined and retains much of the high thermal conductivity of the metal.

368


12.5.2 Plug materials Unlike moulds, most plugs for thermoforming are specifically designed to minimise heat losses from the sheet during contact. In addition many plugs are also designed to encourage sliding of the sheet during plug movement. These effects are illustrated in Fig. 12.7. As the plug moves downwards its base first contacts the heated sheet and as its movement increases more and more of the sheet comes into plug contact. Material not in sheet contact may deform freely, whereas on the plug surface the local deformation is greatly reduced by frictional resistance. At the same time all free surfaces of the sheet lose some heat due to convection and to a much lesser extent by radiation, and along the contacting surfaces heat may be transferred by conduction. The extent of deformation locally is therefore greatly dependent on the shape and properties of the plug material as these are critical in determining the local temperature and the magnitude of contact slip. The net outcome of these effects is that generally the material that contacts the plug retains much of its original thickness whereas the freely stretched regions undergo the greatest deformation and thinning. The encouragement of sliding of material between these regions is therefore very important in helping to give a more balanced distribution of material in the final product. Traditionally materials such as wood were commonly used but nowadays the majority of plugs are manufactured from polymers and polymer-based composites. Typically all of these modern materials have very low thermal conductivity and with careful preparation and polishing their surfaces can Plug displacement

Clamped sheet

Warm plug Conduction

Conduction

Key Sheet subject to:

Slip

Plug or mould contact Free extension

12.7 Effects of plug contact.

Convection

Slip Convection

Cool mould


369

provide low coefficients of friction. Generally plugs made from such materials are unheated prior to operation and then quickly reach an equilibrium temperature during production. In the thermoforming of some sensitive or delicate materials, electrically heated metal plugs are sometimes employed so as to ensure very precise temperature control. Some of the most commonly used polymer-based plug materials are nylon, polytetrafluoroethylene (PTFE), polyoxymethylene (POM) and syntactic foams. The latter are the most advanced materials as they are composites that are manufactured by adding glass or ceramic microspheres to an epoxy or polyester binder. Other components may be included in the formulation to enhance properties such as machinability, durability and stiffness, or to reduce surface friction or heat transfer. This flexibility has allowed the creation of a range of syntactic foams specifically for different thermoforming applications. However, while there is increasing use of such materials, there is as yet no widespread agreement across industry on the most suitable plug materials to use for any given application. Tradition and conservatism are still major influences on decision-making, but this view is increasingly being challenged through improvements in the technical understanding of the process. Recent research in thermoforming is now helping processors to better understand the role that the plug plays during the process (Collins et al., 2002). Experiments have shown great variations in the wall thicknesses for parts produced by different plug materials and plug geometries (Martin and Duncan, 2007). Work has also provided test methods for contact friction (Hegemann et al., 2003) and heat transfer effects relevant to thermoforming (Choo et al., 2008).

12.6

Developments in process simulation

Use of computer-based process simulation is now commonplace in all sectors of manufacturing and most of the major polymer processes have greatly benefited from the development of such software. However, this has not been the case with thermoforming, where developments have lagged significantly behind those in competing processes such as injection moulding. The relatively small size of the industry and its associated lack of technical resources have certainly been contributors to this, but one of the most important reasons has been the genuine lack of in-depth physical understanding of key elements of the thermoforming process. Despite these problems a significant number of researchers have attempted to construct process simulations and it has been one of the most active areas of academic research into thermoforming for many years. Such work has largely been driven by a strong desire from industry to minimise or eliminate the current need to resort to trial and error methods in production as the principal means of process development.

370


12.8 Prototype FE thermoforming simulation.

12.6.1 Practical process modelling Although it is possible to estimate the wall thickness of thermoformed products using simple mathematical formulae, truly advanced process simulations require the application of dedicated computer software. For thermoforming there is almost universal agreement in the literature that the best approach to simulation is through the use of finite element analysis (FEA).This computational technique divides the heated sheet and the tooling surfaces (plug and mould) into finite elements which are then subjected to an iterative process of displacement and loading that recreates the thermoforming process. Generally the polymer is treated as a viscoelastic or viscoplastic solid in this type of analysis, but there are also a smaller number of examples where a computational fluid dynamics (CFD) approach was taken and the polymer was treated as a liquid melt (Christopherson et al., 2001; Metwally et al., 2005). As the design of most thermoformed products is relatively simple, the geometrical aspects of modelling are usually quite straightforward and the construction of a basic process simulation can be completed very quickly. In many cases researchers have simply used commercial general purpose FEA packages to create the model architecture, but others have developed their own software codes (Karamanou et al., 2006). An example of the output from a prototype FE thermoforming simulation is shown in Fig. 12.8. Here the product is a simple plastic cup and the model was constructed using the commercial FEA package Abaqus, which is widely used for this type of analysis. The mould, sheet and plug are shown sectioned and at this point in


371

the analysis the plug is part-way through its downwards movement. The shaded bands on the sheet represent areas of different thickness, with thinner areas being darker. Analyses of this type have demonstrated that thermoforming can be readily simulated using FEA, but that the accuracy of the models is highly dependent on two key input parameters (McCool et al., 2006). These are: 1. The mathematical representation or model for the deformation properties of the polymer material. This is one of the most challenging research objectives as the model must be able to accurately replicate the response of the polymer under biaxial stretching at very large strains, high strain rates and high temperatures. Work in this area has spawned a substantial body of separate research dedicated solely to the development of mathematical models for the major polymer materials (Makradi et al., 2007). Often the models proposed are mathematically complex and different formulae have been suggested for the different polymer materials (Erchiqui et al., 2005). In the past this work suffered from a lack of adequate test methods for accurately measuring the response of a polymer under thermoforming conditions. However, with the development of new biaxial test systems (as described in Section 12.8.2), the understanding of the thermoforming behaviour of different polymers and the accuracy of material models is improving rapidly. 2. The treatment of contact between the polymer and other interacting surfaces during the process. These include its interaction with the surrounding air and the tooling surfaces (plug and mould). The principal interactions are heat transfer (convection and conduction) and the ability of the polymer to slide across the contacting surface (frictional effects), which is particularly important in the treatment of the plug. In much of the reported work these effects have received little attention and very basic relationships such as the simple Coulomb friction law and a single coefficient of friction have been used for simplicity. However, the improved understanding of the process, and particularly the role of the plug (highlighted in Section 12.5.2), are now enabling researchers to more accurately address these issues in process simulations.

12.6.2 Advances in simulation software Over the past 15 years a small number of thermoforming process simulations have been commercially released but most of this software has struggled to penetrate the market. At the moment there are only two commercially available software packages that are wholly designed for the development of multipurpose thermoforming process simulations. These are Blowview and TSim, which have each been under development for more than 10 years. In both cases the software is very user friendly and enables the operator to enter

372


their own product and process data, and then run their own process simulations. Blowview was developed by the National Research Council of Canada (NRC) and is available to members of its special interest group on thermoforming (SIGFORM). T-Sim was developed by Accuform from the Czech Republic and is available through a commercial licence. Use of Blowview is largely restricted to North America, whereas T-Sim has customers worldwide. For both packages one of the major obstacles to their wider acceptance is their relatively small databases of extruded sheet polymers that have been characterised and modelled for thermoforming. For new customers this limitation can mean that their specific thermoforming materials must be specially tested and characterised in order to make the most effective use of the software. There are many more published reports of the development of thermoforming process simulations for academic research purposes (Erchiqui, 2006), or for specific applications as widely spread as pharmaceutical blister packs (Christopherson et al., 2001) and acrylic baths (Dong et al., 2006). Most early work in simulating thermoforming was concentrated on vacuum forming because of its relative simplicity (Nam et al., 2000), but increasingly simulations are being reported for more advanced processes such as plug-assisted and twin-sheet thermoforming (aus der Wiesche, 2004), and in the modelling of advanced materials such as thermoplastic composites (Pham et al., 2005; Sandighi et al., 2008). At the same time simulation techniques have been applied successfully to wider aspects of thermoforming such as modelling of sheet heating prior to forming (Yousefi et al., 2002; Schmidt et al., 2003) and in the analysis of shrinkage (Hosseini et al., 2006) and warpage (Xu and Kazmer, 2001) in thermoformed parts. Work has also been reported on the use of artificial neural networks for the selection and optimisation of production settings (Yang and Hung, 2004).

12.7

Advances in instrumentation and control

Modern thermoforming machines now feature the latest computer displays and machine system controls. Software is smarter, more user friendly, and packed with features like memorised settings and built-in alarms. However, despite these advances thermoforming still lacks any reliable system for fully automated closed loop control. Instead fine process adjustments must be made by experienced operators, and often trial and error while in production remains the primary means of process control. The principal reason for this is the lack of instrumentation to effectively measure the quality of the process output. For most thermoforming operations the wall thickness distribution is the primary control variable along with part definition and surface quality. All of these are extremely difficult to measure on-line and instead companies tend to rely on post-production inspection procedures.


373

12.7.1 Temperature measurement and control Control systems are currently available for the heating systems used in thermoforming and these have been very important in enabling the successful thermoforming of semi-crystalline polymers. Typically a heating panel for thermoforming consists of an array of small individual radiant elements that are mounted above (and/or below) the sheet to be heated. By varying the power going to each element, the magnitude and distribution of the temperature in the sheet may be adjusted and thereby the thermoforming behaviour of the sheet may be changed. In many thermoforming machines thermocouples are fitted to each heating element so that the temperature distribution across the panel may be precisely controlled, and this is also useful in identifying heaters that are either faulty or failing. In recent years such systems have been improved further through the introduction of non-contact infrared sensors that are capable of detecting the resulting temperature distribution in the sheet. Initially these were introduced as a single sensor that could sample an area of the heated sheet prior to forming and were used simply to provide an indication of forming temperature. However, full scanning systems have now been introduced that can provide a detailed map of the temperature distribution across and along a moving sheet. Placement of infrared sensors in a thermoforming machine is typically not a straightforward task. Figure 12.9 shows an arrangement for a prototype sensing system in which a line of infrared sensors has been placed across the sheet between the heater panel and the forming tool. Typically space is very limited in most industrial machines, but a more important problem is the way that the sensors work. All forms of non-contact infrared sensors measure temperature by detecting

Heaters

Forming tool

12.9 An infrared sheet temperature sensing system.

374


the infrared radiation emitted by the hot body (sheet). This is dependent on the emissivity of the material which defines the proportion of the detected radiation that is reflected rather than generated by emission from the hot body. Emissivity can change with the precise material composition and surface finish, and the radiation detected can therefore vary slightly from one batch of material to another. In addition the sensors will pick up stray infrared radiation from other sources so their placement within or close to the heater panel is problematic.

12.7.2 Advanced instrumentation Better measurement of sheet temperature has certainly improved thermoforming processes but as temperature is only one of the factors that determines the processing behaviour of a material, it alone cannot be used to fully control the process. Other sensors that have been used in thermoforming include pressure transducers to measure the air pressure during the blowing phase of the process, and thermocouples embedded in the mould and plug materials to precisely monitor the cycle-to-cycle variation in the temperatures of the tooling. While use of such instrumentation has been reported in research studies, it is rarely used in industry. Another prototype technology that has been pioneered at Queen’s University Belfast is the use of a force transducer fixed between the plug and its drive shaft (as shown in Fig. 12.10) to measure the deformation response of the material during the process. Tests have shown that the force response recorded provides a good indication of the deformation properties of the material during thermoforming and can clearly identify cycle-to-cycle fluctuations and longer-term changes in the health of

Hollow plug stem Metal core Polymer skin

Thermocouple

Force transducer

Cartridge heater

12.10 Prototype plug force monitoring system.


375

the process (Harron et al., 2003). However, as the force response is also dependent on the shape and properties of the plug material, the system does not measure the deformation response directly. Nevertheless, data provided by the system has been found to be particularly valuable in the analysis and testing of thermoforming process simulations.

12.8

Measurement of thermoformability

There is now a wide range of laboratory test methods for polymer materials that are useful in determining their processing behaviour. Many of the most familiar test methods, such as viscometers and rheometers, measure the deformation response of polymer melts and are therefore very relevant to processes like extrusion and injection moulding. However, there are few equivalent tests that can measure the deformation properties of solid polymers and thereby can estimate the thermoformability of materials. Part of the problem in measuring thermoformability is in defining what exactly the term actually means. For example, is it the ease with which a material forms, its sag resistance, the size of its temperature forming window, its temperature sensitivity, its quality of mould replication, the magnitude of part shrinkage, or all of these?

12.8.1 Laboratory test methods for thermoforming One of the most useful standard laboratory test methods for thermoforming is dynamic mechanical thermal analysis (DMTA). This relatively straightforward technique may be used to subject a small heated polymer specimen to a tiny oscillating strain, and thereby map the change in modulus of the polymer with temperature (see Fig. 12.3). However, as the strain in the DMTA test is generally less than 1% compared with more than 100% strain in most thermoforming processes, the test has limited ability to detect differences in forming behaviour. Tensile tests may also be used to measure material response at higher temperatures and larger strains, but crucially these tests are very slow in comparison to processes and the mode of deformation is wrong. This is illustrated in Fig. 12.11, which shows the grid pattern in a formed sheet that results when a sheet marked with a 10 mm square grid is thermoformed into the shape of the pot shown. In the base of the pot (top surface in the photograph) the original square is clearly larger, indicating equal biaxial deformation, whereas in the sides the square has become a long rectangle, indicating that the deformation here is largely in one direction (Collins et al., 2002). As well as leading to variations in forming behaviour, these different modes of biaxial deformation create materials with different levels of biaxial orientation. For some polymers this can lead to large local variations in mechanical and other material properties that can significantly affect the performance of the final product.

376


12.11 Deformation modes in a simple thermoformed pot.

12.8.2 Development of biaxial testing machines The development of testing machines to measure the biaxial deformation response of polymers at elevated temperatures has been a major research theme for many years. Worldwide there are now a number of successful operational systems. Among the first to be developed were the T. M. Long stretcher (Chandran and Jabarin, 1993) and the flexible biaxial film tester at Oxford University (Buckley et al., 1996). Other, more recently developed machines are the flexible biaxial stretcher at Queen’s University Belfast (Martin et al., 2005) and the commercially available Karo IV film stretcher by Brückner Maschinenbau GmbH (Capt et al., 2003). In all cases the development of these machines has been driven by a need for researchers to understand the response of materials undergoing rapid biaxial deformation at elevated temperatures. Generally this information has been used to assist the development of mathematical models for polymers under biaxial stretching and these in turn have been utilised in the development of simulations for processes such as blow moulding and thermoforming. The machines have also been used to prepare specimens of materials with programmed levels of biaxial orientation and to compare the deformation properties of different polymers and polymer grades. The operation of the flexible biaxial stretcher at Queen’s University is shown in Fig. 12.12. A small square sheet specimen measuring 76 × 76 mm and up to 2.5 mm thick is clamped, heated to thermoforming temperature


377

12.12 The flexible biaxial stretcher at Queen’s University Belfast.

and then biaxially stretched at up to 1.5 m/s (32/s nominal strain rate). Deformation properties are recorded in the form of stress/strain curves as shown in Fig. 12.13, which highlights the major differences observed in the properties of four of the main polymer packaging materials when tested at typical thermoforming conditions (Martin et al., 2005). The biaxial stretching properties of materials have also been investigated using alternative methods such as bubble inflation tests (Laroche and Erchiqui, 1999), hot impact tests (Martin et al., 2000) and plug deformation tests (Hegemann et al., 2002; Dharia and Hylton, 2006).

12.9

Future trends

In the coming years the range and sophistication of thermoformed products will continue to expand and it is very likely that thermoforming will grow at the expense of competing technologies. This growth will be driven by the relative simplicity, efficiency and flexibility of the process, but this will be coupled with increasing market confidence as a result of the exploitation of new technologies. The opportunities for expansion are particularly strong in the further replacement of materials like glass, paper and metal in food packaging, and in the increasing displacement of rival processes like injection moulding for the manufacture of larger structural, decorative and high quality engineering components. Additional markets are also likely to emerge for the thermoforming of newer material formulations such as fibrous, woven and laminate thermoplastic composites, miscible or immiscible polymer blends, and high temperature, high performance polymers. Opportunities for these are strongest in the automotive and aerospace sectors, where there is already a powerful market desire to replace relatively heavy metal parts with lighter, cheaper and more functional components made from polymers. The

378

Advances in polymer processing Equi-biaxial stretching at nominal strain rate of 8/s

8 HIPS (135°C) PET (100°C)

7

LDPE (100°C) PP (155°C)

True stress (MPa)

6 5 4 3 2 1 0 0

0.2

0.4

0.6

0.8

1 Strain

1.2

1.4

1.6

1.8

2

12.13 The stress/strain response of four polymers under equi-biaxial deformation.

thermoforming industry will need to respond through the development of more powerful and controllable manufacturing processes. In some cases this will lead to the development of new hybrid processes that combine elements of thermoforming with other polymer processing techniques such as compression moulding or sheet stamping. However, developments in machine drives, heaters, tooling and control systems should also enable established thermoforming processes like plug-assist to be more widely exploited beyond their present applications. The relatively recently developed twin-sheet thermoforming process is also expected to increasingly compete with blow moulding, especially in future packaging applications. The major challenges for the thermoforming industry in the coming years will arise principally from a combination of global environmental concerns, legislation on recycling and the ongoing volatility in the price of crude oil. Over the next decade there is little doubt that environmental pressures will grow and increasingly dictate market requirements. Thermoformed packaging faces the most immediate challenge in creating products that reduce household waste and processes that are as energy efficient as possible. However, heavygauge processors, such as those in the automotive sector, will also face increasing pressure to create parts that are either recyclable or recoverable, and this may require the industry as a whole to develop new products that are made from either the same or readily separable polymer components. In addition, companies will need to seek economically viable routes for the


379

disposal of their own waste in an environmentally friendly way. The extent to which industry will need to change will be highly dependent on the attitude that both consumers and legislators will take in the future, and decisionmaking will be further coloured by rises in the costs of raw materials. New biodegradable polymers such as PLA have already entered the packaging market and it is anticipated that more will follow with greatly improved ranges of properties. Industrial production of such materials is expected to rise very dramatically in the next few years due to recent large investments by major producers in chemical plant worldwide. As the supply problems are reduced and costs fall, biodegradable materials are likely to find further applications, both in packaging and in wider industrial sectors. Thermoforming as a process is very well placed to exploit these new materials as it is regarded as less harsh and damaging than other polymer processes. However, the extent to which biodegradable materials penetrate the market will ultimately be greatly dependent on their relative costs and availability of supply, and there are lingering concerns about the ethics of using foodstuffs like corn starch as a source of polymer materials. If there are further significant and sustained rises in the costs of crude oil then demand for alternatives to conventional polymer materials will rise. Companies will also come under increasing pressure to minimise material usage through better part wall thickness distribution and downgauging, and material-saving technologies like thermoforming of foamed sheets will become of increasing interest again. It is also expected that new nanocomposite material formulations will emerge and may provide exciting new opportunities for thermoforming. Production speeds and efficiencies will continue to increase in thin-gauge thermoforming and this will be assisted by improvements in part design that enable sheet downgauging and quicker part cooling. Technology is also emerging to permit automated roll feeding of ever thicker sheets and it is likely that heavy-gauge processors will also see faster and more automated processes. New machines of all types will feature more sensors and better computer controls that will enable processors to more closely monitor and control the thermoforming operation on-line. Full process control will remain elusive but both the operation and control of heating and cooling systems will improve further. These developments will be accompanied by greater exploitation of virtual engineering techniques in the design and analysis of products and in the rapid manufacture of prototypes. It is also expected that process simulation software will finally gain a significant foothold in the thermoforming market and general acceptance by the industry. This will come as a direct result of the exploitation of recent advances in measurement technologies such as the development of the new biaxial testing machines. These and other measurement systems are now giving researchers a deeper understanding of the behaviour of polymers under thermoforming conditions, and this in turn will make process simulations more realistic, accurate and reliable.

380


12.10 Sources of further information and advice Thermoforming is featured in most general textbooks on polymer processing, but there is also a small number of books specifically dedicated to thermoforming. The most comprehensive of these is Technology of Thermoforming by James Throne. The same author has also recently updated his simpler guide to the process, entitled Understanding Thermoforming. Other important works are Thermoforming: a Practical Guide by Illig and Schwarzmann, Thermoforming: a Plastics Processing Guide by Gruenwald, Practical Thermoforming: Principles and Applications by Florian, and Thermoforming: Improving Processing Performance by Rosen. The largest professional body that embraces thermoforming is the Society of Plastics Engineers (SPE). This has a very strong and active Thermoforming Division which now boasts a membership of around 3000 worldwide and which runs a detailed programme of conferences, seminars and exhibitions. In the US it annually holds a major thermoforming exhibition and technical meeting each autumn and biannually its sister European Thermoforming Division holds a similar meeting in Europe. Technical papers on thermoforming are also regularly presented at the main annual conference of SPE (ANTEC), which is held in the USA, and at other smaller divisional conferences throughout the world. The Thermoforming Division also publishes a regular newsletter, Thermoforming Quarterly, that features technical articles and news on the latest technological developments. Other bodies that regularly publish thermoforming research work are the Polymer Processing Society (PPS) and the European Scientific Association for Material Forming (ESAFORM). Annually PPS holds one main and two regional conferences worldwide and generally each of these includes a technical session covering thermoforming. ESAFORM holds an annual meeting in Europe and work related to thermoforming is regularly presented. PPS and ESAFORM usually feature academic research, whereas the events organised by SPE tend to be more industrially focused, There are a small number of websites that provide useful general resources on thermoforming, although a greater number do provide a basic explanation of the process. Some of the most useful websites are the SPE Thermoforming Division (www.thermoformingdivision.com) and its sister European Thermoforming Division (www.e-t-d.org). The author James Throne also maintains a detailed on-line archive of his many articles, technical publications and presentations (www.foamandform.com). Several of the major trade magazines publish articles online and one of the most extensive resources is the thermoforming archive at the Plastics Distributor & Fabricator (www.plasticsmag.com). Resources relating to academic research may be found on the website of the thermoforming research group at Queen’s University Belfast (www.qub.ac.uk).


381

12.11 References Angellier-Coussy H, Torres-Giner S, Morel M, Gontard N & Gastaldi E (2008), ‘Functional properties of thermoformed wheat gluten/montmorillonite materials with respect to formulation and processing conditions’, J Appl Polym Sci, 107, 487–496. aus der Wiesche S (2004), ‘Industrial thermoforming simulation of automotive fuel tanks’, Appl Therm Eng, 24(16), 2391–2409. Aveila M, Errico M E & Rimedio R (2004), ‘PVA/PTFE nanocomposites: Thermal, mechanical and barrier properties’, J Mater Sci, 29(19), 6133–6136. Avérous L, Fringant C & Moro L (2001), ‘Starch-based biodegradable materials suitable for thermoforming packaging’, Starch, 53, 368–371. Bosiers L & Engelmann S (2003), ‘Thermoformed packaging made of PLA’, Kunst Plast Eur, 93(12), 21–23, 48–51. Bratfisch J P, Vandepitte D, Pflug J & Verpoest I (2007), ‘Development and validation of a continuous production concept for thermoplastic honeycomb’, J Sandwich Struct Mater, 9(3), 113–122. Buckley C P, Jones D C & Jones D P (1996), ‘Hot drawing of poly(ethylene terephthalate) under biaxial stress: application of three-dimensional glass-rubber constitutive model’, Polymer 37(12), 2403–2414. Capt L, Rettenberger S, Munstedt H & Kamal M R (2003), ‘Simultaneous biaxial deformation behaviour of isotactic polypropylene films’, Polym Eng Sci, 43(7), 1428–1441. Chandran P & Jabarin S (1993), ‘Biaxial orientation of poly(ethylene terephthalate). Part I: Nature of the stress–strain curves’, Adv Polym Tech, 12(2), 119–132. Choo H L, Martin P J & Harkin-Jones E (2008), ‘Measurement of heat transfer for thermoforming simulations’, Proc 11th ESAFORM Conf, Lyon, France, April 2008. Christopherson R, Debbaut B & Rubin Y (2001), ‘Simulation of pharmaceutical blister pack thermoforming using a non-isothermal integral model’, J Plast Film Sheeting, 17, 239–251. Collins P, Martin P J & Harkin-Jones E (2002), ‘The role of tool/sheet contact in plugassisted thermoforming’, Int Polym Proc, 17(4), 361–369. Conway D & Clark S D (2005), ‘Manufacture and evaluation of thermoplastic fuel tank access covers’, Int SAMPE Tech Conf, 2005. Dharia A & Hylton D (2006), ‘Novel method for rapid determination of thermoformability’, Therm Quart, 25(1), 10–15. Dixon D, Martin P J & Harkin-Jones E (2001), ‘The effect of material factors on the density and cell morphology of chemically foamed polypropylene’, Cell Polym, 20(6), 403–416. Dong Y, Lin R J T & Bhattacharyya D (2006), ‘Finite element simulation on thermoforming acrylic shets using dynamic explicit method’, Polym Polym Compos, 14(3), 307–328. Erchiqui F (2006), ‘A new thermodynamical approach for the simulation of thermoforming process using the quasi-static finite element method’, J Reinf Plast Compos, 25(3), 235–261. Erchiqui F, Gakwaya A & Rachik M (2005), ‘Dynamic finite element analysis of nonlinear isotropic hyperelastic and viscoelastic materials for thermoforming applications’, Polym Eng Sci, 45(10), 125–134. Faisant J B, Ait-Kadi A, Bousmina M & Deschenes L (1998), ‘Morphology, thermomechanical and barrier properties of polypropylene–ethylene vinyl alcohol blends’, Polymer, 39(3), 533–545. Frick A, Rochman A & Martin P J (2008), ‘A novel manufacturing technology for polymeric thin-walled micro components’, Proc PPS-24 Conf, Salerno, Italy, June 2008.

382


Giselbrecht S, Gietzelt T, Gottwald E, Guber A E, Trautmann C, Truckenmüller T & Weibezahn K F (2004), ‘Microthermoforming as a novel technique for manufacturing scaffolds in tissue engineering’, Nanobiotech, IEE Proc, 151(4), 151–157. Grande J A (2006), ‘NPE 2006 news wrap-up:thermoforming’, Plast Tech, 52(8), 96–99. Grande J A (2007), ‘Thermoforming: ready for in-mould labelling’, Plast Tech, 53(4), 56–65. Grande J A (2008), ‘Blow moulders and thermoformers try plastic rapid tooling’, Plast Tech, 54(7), 49–51. Harkin-Jones E et al. (2008), ‘Performance enhancement of polymer nanocomposites via multiscale modelling of processing and properties’, Plast Rubber Compos, 37(2–4), 113–123. Harron G W, Harkin-Jones E & Martin P J (2003), ‘Plug force monitoring for the control and optimisation of the thermoforming process’, Proc IMechE, Pt E: J Proc Mech Eng, 217(3), 181–188. Heckele M & Schomburg W K (2004), ‘Review on micro molding of thermoplastic polymers’, J Micromech Microengng, 14, R1–R14. Hegemann B, Kech A, Göschel U, Belina K & Eyerer P (2002), ‘Biaxial deformation behaviour of PET dependent on temperature and strain rate’, J Macro Sci, Pt B – Phys, 41(4–6), 647–656. Hegemann B, Eyerer P, Tessier N, Bush T & Kouba K (2003), ‘Polymer-polymeric friction at temperatures and rates simulating the thermoforming process’, Thermof Quart, 22/2, 10–13. Hogan T A, Hoenig S M, Walther B W & Walton K L (2007), ‘Changing the game in thermoplastic polyolefins (TPO) for cut sheet thermoforming’, Therm Quart, 26, 15– 18. Hosseini H, Beryshev B & Mehrabani-Zeinabad A (2006), ‘A solution for warpage in polymeric products by plug-assist thermoforming’, Eur Polym J, 42, 1836–1843. Incarnato L, Scarfato P & Acierno D (1999), ‘Rheological and mechanical properties of recycled polypropylene’, Polym Eng Sci, 39(9), 1661–1666. Jéol S, Fenouillot F, Rousseau A, Masenelli-Varlot K, Gauthier C & Briois J (2007), ‘Drastic modification of the dispersion state of submicron silica during biaxial deformation of poly(ethylene terephthalate)’, Macromolecules, 40, 3229–3237. Karamanou M, Warby M K & Whiteman J R (2006), ‘Computational modelling of thermoforming processes in the case of finite viscoelastic materials’, Comput Methods Appl Mech Engrg, 195, 5220–5238. Laroche D & Erchiqui F (1999), ‘Experimental and theoretical study of the thermoformability of industrial polymers’, J Plast Film Sheeting, 15, 287–296. Lebrun G, Bureau M N & Denault J (2004), ‘Thermoforming-stamping of continuous glass fiber/polypropylene composites: interlaminar and tool-laminate shear properties’, J Thermoplast Compos Mater, 17(2), 137–165. Levy G N, Schindel R & Kruth J P (2003), ‘Rapid manufacturing and rapid tooling with layer manufacturing (LM) technologies, state of the art and future perspectives’, CIRP Annals – Manuf Tech, 52(2), 589–609. Lussier D & Chen J (2002), ‘Material characterization of woven fabrics for thermoforming of composites’, J Thermoplast Compos Mater, 15(6), 497–509. Macauley N, Harkin-Jones E & Murphy W R (1996), ‘Extrusion and thermoforming of polypropylene: the effect of process and material variables on processability’, Proc ANTEC Conf, Indianapolis, USA, May 1996. Makradi A, Belouttar S, Ahzi S & Puissant S (2007), ‘Thermoforming process of amorphous


383

polymeric sheets: modeling and finite element simulations’, J Appl Polym Sci, 106, 1718–1724. Martin N, Lappin J F, Harkin-Jones E & Martin P J (2000), ‘The use of hot impact testing in the simulation of the plug-assisted thermoforming process’, Proc SPE Antec Conf, 2000, 783–787. Martin P J & Duncan P (2007), ‘The role of plug design in determining wall thickness distribution in thermoforming’, Polym Eng Sci, 47(6), 804–813. Martin P J, Tan C W, Tshai K Y, McCool R, Menary G, Armstrong C & Harkin-Jones E (2005), ‘Biaxial characterisation of materials for thermoforming and blow moulding’, Plasts, Rub Compos: Macromol Engng, 34(5), 276–282. McCool R, Martin P J & Harkin-Jones E (2006), ‘Process modeling for control of product wall thickness in thermoforming’, Plasts Rub Compos: Macromol Engng, 35(8), 340– 347. Metwally H, Portha K, Szulga S, Dozolme A & Marchal T (2005), ‘Numerical simulation of the thermoforming of an automotive dashboard’. SPE Conf Auto TPO, 2006, 87– 91. Naitove M H (2007), ‘New starch-based bioplastics arrive’, Plast Tech, 53(11), 47–49. Nam G J, Ahn K H & Lee J W (2000), ‘Three dimensional simulation of thermoforming process and its comparison with experiments’, Polym Eng Sci, 40(10), 2232–2240. Pham X, Bates P & Chesney A (2005), ‘ Modeling of thermoforming of low-density glass mat thermoplastic’, J Reinf Plast Compos, 24(3), 287–298. Rajeev R S, Harkin-Jones E, Soon K, McNally T, Menary G, Armstrong C & Martin P (2008), ‘A method to study the dispersion and orientation of nanoclay tactoids in PET matrix-focused ion beam milling combined with electron microscopy’, Mater Letters, 62, 4118–4120. Sandighi M, Rabizadeh E & Kermansaravi F (2008), ‘Effects of laminate sequencing on thermoforming of thermoplastic matrix composites’, J Mater Proc Tech, 201, 725– 730. Schmidt F M, Le Maoult Y & Monteix S (2003), ‘Modelling of infrared heating of thermoplastic sheet used in thermoforming process’, J Mater Proc Tech, 143–144, 225–231. Sherman L M (2003), ‘New polypropylene families for packaging, blends and composites’, Plast Tech, 49(11), 37–39. Sherman L M (2005), ‘Unusual TPOs displace other plastics in hard and soft auto parts’, Plast Tech, 51(3), 32–34. Throne J L (1996), ‘Thermoforming: Definitions, history, methods and equipment’, in Throne J L, Technology of Thermoforming. Munich, Hanser. Truckenmueller R, Giselbrecht S & Throne J L (2006), ‘Microthermoforming technology and applications’, Therm Quart, 25(2), 9–14. Xu H & Kazmer D (2001), ‘Thermoforming shrinkage prediction’, Polym Eng Sci, 41(9), 1553–1563. Yang C & Hung S (2004), ‘Modeling and optimization of a plastic thermoforming process’, J Reinf Plast Compos, 23(1), 109–121. Yousefi A, Bendada A & Diraddo R (2002), ‘Improved modelling for the reheat phase in thermoforming through an uncertainty treatment of the key parameters’, Polym Eng Sci, 42(5), 1115–1129.

13 Polymer processing using supercritical fluids C L H I G G I N B O T H A M, J G L Y O N S and J E K E N N E D Y, Athlone Institute of Technology, Ireland

Abstract: This chapter describes the use of supercritical fluids in modern polymer processing operations. The technologies required to use supercritical fluids in polymer processing operation in addition to the effects of supercritical fluids on both the polymer material and the final product are outlined. A brief history and the advantages of using supercritical fluids are presented first, followed by a discussion on the use of supercritical fluids in the synthesis, processing, blending, impregnation and reclamation of polymers. Key words: supercritical fluid, induced crystallisation, microcellular foam, polymer blending, chemical recycling.

13.1

Introduction

The use of supercritical fluids (SCFs) in the polymer industry presents an opportunity to create unique products both now and in the future as demand for and usage of the technology further develop. These fluids possess physicochemical properties, such as density, viscosity and diffusivity, which are intermediate between those of liquids and gases and which are continuously adjustable from gas to liquid with small pressure and temperature variations. Both the capability of supercritical fluids to replace toxic solvents and the ability to tune solvent characteristics for highly specific separations or reactions have led to the current scientific and industrial interest in supercritical fluids. They can be used for a variety of applications, such as analytical and preparative separations, organic, inorganic and polymer synthesis, waste management, as plasticising agents in material processing (nanomaterials, thin films and coatings), porous materials and chemical recycling (Arai et al., 2001; Sun, 2002; Bonnaudin et al., 2003).

13.2

What is a supercritical fluid?

The boiling point of a substance is raised by increasing the pressure. At a certain temperature the substance suddenly becomes a gas no matter how large the pressure is. The temperature and pressure at which this occurs are 384


385

called the critical temperature and critical pressure. A supercritical fluid is any substance at a temperature and pressure above its critical temperature and pressure (Fig. 13.1). A supercritical fluid has the unique ability to diffuse through solids like a gas, and dissolve materials like a liquid. Additionally, it can readily change in density upon minor changes in temperature or pressure. It also has highly tunable solvent behaviour facilitating easy separation. Other properties of a substance that change widely near the critical region are thermal conductivity, surface tension, constant-pressure heat capacity and viscosity. In comparing a liquid sample with a supercritical fluid sample of the same substance both possessing the same density, the thermal conductivity and diffusivity of a SCF are higher than those of the liquid, its viscosity is much lower, while its surface tension and heat of vaporisation have completely disappeared. The advantages of using a supercritical fluid can therefore be summarised as follows: 1. SCFs have similar solvating powers to liquid organic solvents but their higher diffusivities, lower viscosity and lower surface tension make them more effective in many cases. 2. Since the solvating power can be adjusted by changing the pressure or temperature, separation of analytes from solvent is fast and easy. 3. Since their density is pressure-tuneable, separation of substances from solvents is easy to achieve.

Pressure

Pc

Temperature

Tc

13.1 Phase diagram of a typical supercritical fluid.

386


4. By adding modifiers to a SCF (like methanol to CO2) its polarity can be changed for having more selective separation power. 5. Little harm is done to the environment in terms of residues from processes using SCF compared to volatile organic compounds and ozone-depleting substances. 6. SCFs are generally cheap, simple and safe to use. Disposal costs are much less, and in industrial processes the fluids can be simple to recycle. The main disadvantages of using a supercritical fluid in a processing operation are: 1. There is currently only limited phase equilibrium data available on the solubility of commercial polymers in supercritical media at various temperatures and pressures. 2. Owing to the relatively high pressures involved and the need for precise control of these pressures, the tooling for supercritical media assisted processing can be expensive. 3. Compression of solvent often requires elaborate recycling measures to reduce energy costs.

13.3

Brief history of supercritical fluids

The development and application of supercritical fluid technology has actually been quite slow; however, it is unlikely that the pioneering researchers into supercritical fluid properties in the early 1900s envisioned the wide-ranging impact supercritical materials would have later in the century. First discovered in 1879, supercritical fluids have been used for extraction applications since the 1950s. Starting in the 1960s, many research groups, primarily in Europe and then later in the US, examined SCFs for developing ‘advanced’ extraction processes. European researchers emphasised extraction from botanical substrates, for example spices, herbs, coffee, tea and so on, using predominantly supercritical carbon dioxide, and by the 1980s there were several large SCF extraction processes in operation in Germany, the UK and the US, for decaffeinating coffee and tea and extracting flavours and essential oils from hops, spices and herbs. The major driver for the development of the SCF process was in the elimination of residual solvents in the products, especially methylene chloride, which had previously been used to decaffeinate coffee. Extraction by SCF also gave improved flavour and aroma characteristics in comparison to standard extraction practices.

13.4

Polymer applications of supercritical fluids

Now that supercritical fluids themselves have been considered, discussion will begin on how these materials can be applied to various aspects of polymer processing.


387

13.4.1 Extraction and purification A number of polymers used in medical devices, for example silicone or polyester-based polymers, can contain residual raw materials or by-products of production. These impurities can be removed by dissolution and extraction using organic solvents such as hexane or methylene chloride. However, these solvents themselves often leave unwanted residues in the materials and can alter their characteristics. The use of a SCF can alleviate this problem. The fluid is compressed to elevated pressures above its critical pressure, to make it supercritical. The polymer is then exposed to the supercritical fluid and swells. As the free volume in the polymer is increased, the SCF can penetrate deeply into the matrix and the impurities are dissolved by the supercritical fluid. Volatile materials within the feed matrix will partition themselves within the supercritical phase and be removed with the SCF during the extraction from the feed system. The pressure is then quickly reduced and the supercritical fluid and the impurities diffuse out of the polymer. The advantages of using SCF over conventional liquid solvents for extractions include the fact that the solvating power of the SCF is easily controlled and manipulated by pressure and/or temperature, no harmful residues are left as solvents are non-toxic, the SCF is easily recovered and recycled from the extract due to its volatility, there is the ability to sometimes achieve separations that are not possible by traditional processes, and heat-sensitive materials can be extracted as low temperatures can be employed. The disadvantages centre on the cost of the initial experimental set-up and the hazards associated with using elevated pressures. Compared with traditional solvent extraction, a dry membrane treated by supercritical carbon dioxide (scCO2) extraction has much less shrinkage and greater water permeability, whereas the degree of crystallisation of a membrane extracted by scCO2 is slightly greater than that extracted by ethanol.

13.4.2 Use of supercritical fluids in polymerisation Supercritical carbon dioxide has been intensively researched in a wide range of polymerisation reactions and used as the continuous phase for numerous step-growth and chain-growth reactions. This includes free-radical, ionic and metal-catalysed process routes. However, its application is likely to be limited to specific special applications as its use is limited by the poor solubility of most long-chain polymers. The synthesis of fluoropolymers, polysiloxanes and soluble low molecular weight polymers can be carried out using SCFs. However, the ability to tailor the properties with scCO2 has enabled the synthesis of polymers with precise control of molecular weight and polydispersity, but with minimum contamination. The morphology can also be controlled in some cases.

388


Another issue associated with the use of scCO 2 as a solvent for polymerisation is the reaction pressure. This problem has been approached in a number of ways. For example, DiSimone and coworkers have developed methods for the continuous precipitation polymerisation of fluoropolymers (Saraf et al., 2002). In general, continuous SCF processes are more likely to be implemented than batch processes. Other recent breakthroughs in the applications of SCFs to polymer synthesis include dispersion and emulsion polymerisation, polymerisation of polymer matrices and the synthesis of dendrimers.

13.4.3 Impregnation and supercritical dyeing Polymer impregnation and dyeing involve the introduction of a guest doping solute into a host polymer matrix. It utilises the SCF properties of high diffusivity, low surface tension and ease of solvent recovery to prepare new polymeric materials (Tang et al., 2003; Xu et al., 2003). The SCF diffuses out of the polymer easily once the pressure is reduced to ambient values and the guest solute remains trapped in the matrix without the presence of CO2 in the finished product. Supercritical carbon dioxide has been successfully used to impregnate various dyes, drugs and metal complexes into polymer hosts. Dissolution of scCO2 in a polymer can also facilitate diffusion of monomers and catalysts/initiators within the polymer matrix. It is important to distinguish between two different mechanisms of supercritical fluid impregnation of additives into polymer matrices. The first involves the simple deposition of a compound soluble in an SCF into the polymer matrix. In this case a polymer matrix is subjected to an SCF containing a solute. When the pressure is lowered, CO2 molecules quickly leave the polymer matrix, leaving the solute molecules trapped inside the polymer matrix (Cooper et al., 1993; Kazarian et al., 1994). A different mechanism applies to impregnation of compounds having very low solubility in the SCF phase. In such cases the high affinity of these solutes for certain polymer matrices can result in the preferential partitioning of a solute in favour of polymer over fluid. The high partition coefficient of polar dye molecules has played a crucial role in the success of supercritical dyeing (Schnitzler and Eggers, 1999). Materials that are both miscible and immiscible can become dispersed. A material with little affinity for the matrix is dispersed and then trapped when it comes out of solution, but has no particular molecular attraction with the matrix. The scCO2 does not appear to aid compatibility in these cases, in contrast to the effect seen in blending by extrusion, for example. In all cases it is necessary that the matrix species are able to swell under scCO2.


389

13.4.4 Rapid expansion of supercritical fluid solutions (RESS) The usage of supercritical fluids in particle formation technology has given birth to a number of modified processes that use different nucleation and growth mechanisms of participating particles. The RESS process can be used to produce thin film coating, polymer fibres and fine graining when the polymer has some degree of solubility in the supercritical fluid. The process is based on the solubility difference of the polymer in supercritical fluids at high and low pressures, respectively. The polymer is dissolved in a supercritical fluid and this high-pressure solution is sprayed through a nozzle head and deposited continuously or a required over relatively large areas. The rapid expansion stage causes a sudden drop in the dissolving capacity of the solvent as the fluid comes out of its supercritical state, causing nucleation and growth of any low vapour pressure solute species that are present in the solution prior to expansion. As in many current SCF applications, the solvent of choice is carbon dioxide; however, as in other technologies, it may be necessary to modify the properties of the solvent to improve solubility. Therefore modifiers may be required especially to improve the solubility of polar molecules, since carbon dioxide is non-polar. Co-solvents may also be employed to enhance solubility.

13.4.5 Supercritical anti-solvent precipitation (SASP) It is also possible to use supercritical carbon dioxide where there is no solubility, in a complementary process to RESS called SASP. In this process, polymer is dissolved in a liquid solvent and the solution is sprayed into a high-pressure chamber where a supercritical fluid (anti-solvent) already exists, causing rapid contact between the two media. Droplets are formed and the original solvent dissolves in the carbon dioxide, leaving the insoluble material in powder form. This results in fast nucleation and growth, and consequently creates smaller particles. Like RESS this technique is currently employed in the pharmaceutical industry (Chattopadhyay and Gupta, 2002).

13.4.6 Solution enhanced dispersion by supercritical fluids (SEDS) This process is a modified version of the SASP process in which the liquid solution and supercritical fluid are sprayed together using a specially designed coaxial nozzle. Here, the supercritical fluid serves a multiple purpose in that it is used both as an anti-solvent and as a dispersion medium. The spontaneous contact of high-speed streams of a liquid solution and a supercritical fluid generates the finely dispersed mixture and prompt particle precipitation.

390


This process has been used for the production of biodegradable polymeric microparticles of poly( DL -lactide-co-glycolide), poly( L -lactide) and polycaprolactone (Ghaderi et al., 1999). Schiavonea et al. (2004) demonstrated that for budesonide and albuterol, SEDS provides an attractive particle engineering option for the development of blend formulations for inhalation drug delivery.

13.4.7 Particles from gas-saturated solutions (PGSS) PGSS is a process where a SCF is dissolved in a melted substrate (or substrates), or a solution of the substrate(s) in a solvent, or a suspension of the substrate(s) in a solvent, followed by a rapid expansion, at moderate pressures, of the saturated solution through a nozzle. Depending on the system, fine solid particles or liquid droplets are formed (Jung and Perrut, 2001). This process does not require that either the carrier or the core material be soluble in the SCF. However, the SCF must be highly soluble in the liquid phase. This makes the process suitable for polymers, which generally absorb a large amount of carbon dioxide. The PGSS process was successfully developed and optimised for the particle formation of edible fats. The most critical variable affecting the particle morphology and size was found to be CO2, followed by the melt temperature, which affected the particle morphology and powder bulk density strongly (Münüklü and Jansens, 2007). PGSS can also be used with suspensions of active substrate(s) in a carrier, leading to composite microparticles, making it promising for drug-delivery systems. By far and away the most widely used SCF of interest to polymer scientists is supercritical carbon dioxide which will now be considered in greater detail.

13.5

Polymer processing

13.5.1 Solubility of CO2 in polymers In polymer processing the scCO2 is present as the minor phase in the polymer system. Solubility data for supercritical CO2 in polymers above their glass transition temperature (Tg) or melting point (Tm) has become widely available in recent years, thanks to the many in-depth investigations in the literature. This data helps to provide useful guidelines for the determination of processing conditions for specific polymers, for example phase separation occurs in blending and microcellular foaming if the concentration of dissolved CO2 is not kept below the solubility limit. The quantity of CO2 dissolved in a polymer depends largely on the ability of the CO2 to interact with the basic sites in polymers. Solubility is also influenced by the positioning of these basic sites. The solubility of CO2 in the polymer is increased if the basic sites are easily accessible (i.e., if the basic sites are side groups, etc.).


391

13.5.2 CO2 induced plasticisation of polymers In general terms, plasticisation of polymers involves allowing the polymer chains an increase in mobility. The sorption of scCO2 into a polymer matrix results in the plasticisation of the amorphous component, giving a reduction in the Tg of the polymer which in turn results in a reduction in the viscosity. This effect is due to increased interchain distance and the interaction of CO2 with sites along the polymer chains. The magnitude by which the Tg is depressed is related to the amount of scCO2 in the matrix. This has a direct impact on processing of polymer melts. The use of scCO2 to plasticise polymer melts removes the need for the use of harmful processing aids and also allows materials that would previously have undergone thermal degradation to be processed at lower temperatures. The use of supercritical fluids in the processing of polymer melts can lead to changes in the mechanical properties of the materials. Most mechanical property changes during processing can be attributed to the plasticisation of the polymer by the supercritical fluid and the resultant drop in Tg. Some blended polymer materials have shown significant increases in modulus and strength when formed in a supercritical fluid assisted process; this is often due to the tuning of the morphology and degree of crystallisation of the material by the supercritical fluid. Changes in the elastic and creep moduli of materials when processed with supercritical fluids can occur in a range of materials. However, these changes and their magnitude are dependent on the solubility of the polymer(s) in the supercritical media and the supercritical materials’ ability to induce crystallisation in the system in question.

13.5.3 Induced crystallisation The phenomenon of scCO2 plasticisation of semi-crystalline polymers has important implications as it may induce the formation of crystallites in these materials. This induced crystallisation results from free energy considerations within the CO2–polymer system, where the scCO2 induced mobility of the polymer chains allows them to rearrange into kinetically more favourable configurations and may result in the formation of crystallites. This movement to the lowest energy configuration occurs in terms of both polymer conformation and three-dimensional structure. Once the system undergoes depressurisation, the CO2 rapidly escapes from the material, leaving the induced morphological changes frozen in place. The scCO2 induced crystallisation of poly(ethylene terephthalate) (PET) has received a lot of attention in recent years due to the widespread use of this polymer both in synthetic polyester fibres and in soft drinks bottles. The rate of crystallisation has been shown to increase with increasing pressure (Lambert and Paulaitis, 1991) and follows the Avrami equation. Many other

392


polymers have also been crystallised with scCO2, including poly(styrene), poly(ether ether ketone) and poly(carbonate) (Kazarian, 2000). scCO2 induced crystallisation offers a route for a degree of tunable morphological control of polymer systems.

13.5.4 Polymer blending After the discovery of the major commodity and engineering plastics materials in the early to middle part of the twentieth century, the cost of bringing a new polymer material to market began to rise dramatically. As a result, both the polymer industry and academia began to focus on developing polymer blends with novel and valuable properties, in order to enlarge the spectrum of available polymers. Various polymeric materials are known for specific or unique characteristics, and melt blending of polymers during extrusion is a useful way of combining the desired properties of different polymers. In the non-reactive route, a two-phase (or multi-phase) system is formed when two immiscible polymers (A and B) are mixed in the molten state. One phase is disperse (droplets), rich in component B, while the other phase is continuous, rich in component A. The ratio of the viscosities of each of the constituents determines the size of the phases. Specific interactions between the supercritical media and the functional groups of the polymer(s) play an important role in processing or blending operations. Should the supercritical media interact differently with each component of the blend (due to varying molecular structures) it will have a different effect on the Tg of each component. This, in effect, will result in each component being plasticised differently. In SCF-assisted extrusion, the reduction in viscosity is different for different polymers depending on the amount of scCO2 dissolved under given processing conditions (temperature, pressure and shear rate), allowing the blending process to be easily manipulated. The use of scCO2 leads to decreased shear thinning and a finer dispersion of the minor component. This offers the possibility of controlling the blend morphology by allowing the supercritical CO2 to shorten the length of extruder required for phase inversion (due to the lowering of the Tg) and fine tuning the droplet size of the minor component via the mixing of the extruder screws. A reactive route for producing polymer blends involves scCO2-assisted sorption of reagents (monomer and initiator) into and the subsequent polymerisation reaction within a host polymer. This route is efficient when monomers and initiator are soluble in supercritical CO2 and the solubility of CO2 in the host polymer is high enough to cause it to swell (Kazarian, 2000). If foaming of a polymer blend is not desired, venting of CO2 from the blend is necessary, which may cause demixing of the polymers. Polymer composites processing can also utilise supercritical fluid technology, and extensive research has taken place in this area recently due to the burgeoning


393

use of these materials in the electronic and medical industries. Supercritical fluid can be used to carry the monomer onto the fibres or particles to be used in the composite and to act as a plasticiser for the synthesised polymer matrix when the composite is formed by in-situ polymerisation of the monomer. Polymer composites can also be prepared by blending the polymer and the other component in the presence of supercritical media.

13.5.5 Microcellular foam A microcellular foamed polymer has a cell diameter of less than 10 µm and a cell density of >109 cells/cm3 (Fleming and Kazarian, 2005). The creation of microcellular foam, like many scCO2 processes, entails the formation of a single-phase solution. On venting the CO 2 by depressurisation, thermodynamic instability causes supersaturation of the CO2 dissolved in the polymer matrix and hence nucleation of cells occurs. The growth of the cells continues until a significant amount of CO2 escapes, the polymer passes through its Tg and the foamed structure is frozen in place. The large number of cells required to create a microcellular structure are attained by ensuring that the cell nucleation rate is higher than the diffusion rate of the blowing agent (scCO2) into the cells. It is the rate and extent of the pressure drop that controls the time period for cell nucleation and growth. Microcellular foams are commonly produced using scCO2 using either extrusion or injection moulding.

13.5.6 Extrusion In scCO2 extrusion, a modified extruder assembly (both barrel and screw) is used and the scCO2 is fed continuously into the molten polymer. In the barrel of the extruder the SCF is dispersed in the polymer as bubbles, which are subsequently elongated and broken down into smaller bubbles by the actions of the extruder. In this manner dissolution of the SCF into the molten polymer is achieved. Trexel’s MuCell process technology was the first to widely offer microcellular foaming using scCO2 extrusion equipment. The system is available only by licence from Trexel and can be fitted to new or existing extrusion equipment. To retrofit the MuCell process to a conventional extrusion system requires the purchase and fitting of the following equipment: 1. The Trexel SCF system and a custom-designed MuCell Injector Kit with sufficient capacity to deliver the supercritical fluid at the volume and pressure required for microcellular processing and foaming. 2. A MuCell screw specifically designed for one’s target application and extruder configuration. 3. A modified extrusion die with the appropriate flow configuration.

394


After the required equipment has been installed (as depicted in Fig. 13.2), the MuCell system operates based on three steps: 1. The Trexel SCF system provides the pumping and, in conjunction with the MuCell injector assembly, the accurate low-dose injection of the supercritical fluid directly into the extruder barrel. 2. Trexel’s proprietary injection system and screw designs effectively disperse and mix the SCF and the polymer. By completely dissolving the blowing agent within the polymer while still maintaining a pressure profile throughout the extrusion system to keep the blowing agent in solution, a homogeneous single-phase solution is created. This single-phase solution is a requirement to obtain sufficient nucleation sites to enable microcellular foaming at the exit of the die. 3. The molten solution is extruded through a die that has been modified using MuCell design parameters. This die will both maintain the pressure in the extruder to keep the blowing agent in solution, and provide the appropriate rapid pressure drop rate or high rate of change of solubility. Such modifications are geared toward optimising flow channels and pressure profiles in order to control the microcellular foam creation at the exit of the die.

13.5.7 Injection moulding As with their proprietary extrusion technology, a scCO2 injection moulding system is available under licence from Trexel Ltd (Fig. 13.3). Trexel has designed and developed a method for precise dosing of SCF into the barrel of the injection moulding machine as well as special screw designs to create the single-phase solution that is critical to achieve a uniform microcellular structure. Similar to the MuCell extrusion technology, to adapt the microcellular foam injection moulding process to a standard injection moulding platform the following components must be acquired: 1. A specially designed screw (typically 22:1 or 24:1 L/D) designed to create and maintain a single-phase solution. Screw design

Die design

13.2 MuCell configuration.

SCF system

Polymer processing using supercritical fluids 01

04

02

395

03 05

03 MuCell®-plasticising screw (L/D ratio: 24:1) 04 Homogenising section for gas/melt mixture 05 Plastics granules without blowing agent

01 Needle valve nozzle 02 Gas injector

MuCell® plasticising unit 04

03 02

01 Process sequence

01 Plasticising of plastics 02 Injecting of supercritical fluid into plastic melt

03 Mixing and dissolving of the fluid Homogenising of the miscellany 04 During injection process, start of nucleation and creation of microcellular foam structure

13.3 The MuCell injection moulding system.

2. Barrel drilled for either one or two SCF injector systems to match the screw length. 3. Shutoff nozzle to keep the melt under pressure throughout the entire cycle. 4. Screw position control modification to prevent screw decompression after charging. This is normally accomplished through minor modification to the hydraulic system. Many hydraulic machines and most electrical machines are already capable on standard systems. 5. SCF delivery system including SCF injectors and SCF dosing control logic, designed to deliver SCF to the screw at a precisely controlled rate and pressure. The two-phase solution of molten plastic and SCF will move into the mixing section of the screw where it is continuously divided and recombined by the

396


dynamic mixing action of this section, until the SCF completely diffuses into the melt, creating a single-phase solution. The single-phase solution preparation is the first of four main steps in the production of a microcellular injectionmoulded foam structure. The other three steps–cell nucleation, cell growth and shaping–take place during injection as the mould cavity is filled. Ergocell is the injection moulding process operated by Demag Ergotech for the production of microcellular foamed products. Owing to the possibility of patent infringement with the Trexel technology, any prospective purchasers of Ergocell equipment must also purchase a MuCell licence. In contrast to the MuCell process, the Ergocell technology injects the SCF into a fully homogenised melt in a specially designed module (see Fig. 13.4). The module for gas metering and mixing is installed upstream of a standard 20D plasticising unit. This module enables homogenising of the melt/gas mixture to be effected independently of plasticising. The cycle sequences in the Ergocell process essentially correspond to the sequences in the standard injection-moulding process. The decisive difference is in the gas delivery which takes place simultaneously with plasticising. As the screw draws in, melts and delivers material into the space in front of the screw and, in the process, is being pushed back against the back pressure, gas is fed into the melt from a gas metering station. Thus, the screw moves back at a speed that is a function of the plasticising capacity of the screw. Simultaneously, an amount of gas as preset by the operator is delivered into the melt. In contrast to the MuCell technology, which requires a modified screw assembly, the injection of the SCF into a module downstream of a conventional plasticisation unit in the Ergocell technology means that it can

Single phase (melt/gas)

Gas injection nozzle Melt

Injection piston

Screw Spline

Mixer ErgoCell-Module

13.4 Ergocell Coax module.

Plasticising cylinder


397

be easily removed, allowing the injection moulding equipment to be used in a conventional process when required. For injection moulding of microcellular components, higher injection speeds lead to more foaming, as there is a higher pressure drop in the tool. The microcellular foam process changes the fundamental cost structure of injectionmoulded components through four main sources of economic benefit: 1. Reduced operating costs through cycle time reductions of up to 50%, reduced scrap rates, and lower energy consumption (energy savings are based on reduced processing temperatures, etc., and are process dependent) 2. Lower capital costs through the purchase of smaller and fewer machines, and fewer and less expensive moulds 3. Reduced material costs through component density reduction, thinner design, and material substitution 4. The ability to mould thermoplastic parts that are dimensionally improved.

13.6

Recycling

The recycling of polymers has become an increasingly important issue from environmental and resource conservation aspects in the early years of the twenty-first century. This has led to the development of various recycling techniques, such as material recycling, thermal recycling and chemical recycling. In material recycling, the properties of the recycled polymers become inferior to that of virgin material, thus making further recycling nonbeneficial. Thermal recycling recovers energy by burning polymers, which can be applied to most kinds of polymers, but hazardous substances like dioxin may be produced. In relation to chemical recycling, the waste polymer is reduced to the feedstock of the polymer and can be recycled repeatedly. Every recycling technique has economic advantages, and they can be complementary to each other; however, it is chemical recycling which is more important in the view of the conservation of non-renewable resources (Sugeta et al., 2000). In the context of this chapter we will look at the effect of chemical recycling and in particular the emerging field of recycling of polymers using supercritical fluid technology.

13.6.1 Polymeric recycling using supercritical fluid technology Supercritical fluid technology using carbon dioxide (scCO2) in conjunction with the continuous twin-screw extrusion process opens many opportunities for new engineered processing techniques for the development of new products and new product concepts, such as recycling incompatible polymers such as PET or polycarbonate (PC).

398


The application of supercritical fluid for the recycling of polymers using conventional extrusion is a new and emerging technology. Many mixed polymeric recyclates cannot effectively be melt-processed without the use of expensive compatibilisers to improve their compatibility. This is due to their low entropy of mixing, which is a result of the high molecular weight associated with most polymeric chains. Another way of stating this problem is that, since a macromolecule interacts with its environment along its entire length, its enthalpy of mixing is very large compared to that of a small organic molecule with a similar structure; thermal excitations are often not enough to drive such a huge molecule into solution on their own. Owing to this uncommon influence of mixing enthalpy, polymers must often be of nearly identical composition in order to mix with one another, or they will need expensive compatibilisers. Supercritical carbon dioxide encourages the compatibility of recyclate polymers by lowering the melt viscosity. Supercritical water Supercritical water (SCW) is widely used in waste water treatment; however, recently its potential for polymer recycling has been investigated. Chemical recycling involves the fragmentation of polymers into smaller molecules by altering their chemical structures. The recyclate can then be used as a raw material for reproducing the original polymers or other chemical products. An example of this technique is the synthesis of ε-caprolactam which is used in the manufacture of nylon. This method eliminates the need for acid catalysts and the recovery of solvents. An illustration of the manufacturing technique and synthesis is given in Fig. 13.5 and 13.6 respectively. This technique is an environmentally friendly method for the synthesis of a raw compound for making nylon.

Microreaction system utilising supercritical water Supercritical water

Thermo-insulator 50µ L microreactor (400°C and 40 MPa)

Reaction time < 1 s Aqueous solution of cyclohexanone oxyme

Temperature rise < 0.05 s

Back-up pressure valve

Rapid cooling Thermocouple Yield ~ 100%

13.5 The synthesis of ε-caprolactam with microreaction system utilising supercritical water.

Polymer processing using supercritical fluids H N

O 230°C ~ 260°C (polymerisation)

(NH4)2SO4

ε-caprolactam

399

H N O

n Nylon-6

+ H2SO4 Materials recycling

OH N

Non-catalytic synthesis Supercritical water

Decomposition Supercritical water

Products

Usage · degradation

Cyclohexanone oxyme Supercritical route Conventional route

13.6 Environmentally friendly material recycling system utilising supercritical water – Nylon-6 synthesis.

High reaction rates are possible due to the high diffusivity and low viscosity of SCW. However, SCW processes are expensive to implement due to the highly corrosive nature of SCW and the high working pressures required by the process. Supercritical methanol Supercritical methanol has begun to be offered as an alternative to SCW for the chemical recycling of polymers. The attraction of using supercritical methanol is that when compared to supercritical water it has lower critical conditions and thus offers a more energy-efficient production process; and due to the boiling point of methanol being lower than that of water, this system allows for easier separation of the recycling products.

13.7

Conclusions

The use of supercritical fluids in the polymer sector affords the opportunity to add a new and exciting dimension to the processing of these materials. The use of supercritical CO2 as an inexpensive solvent in many polymer processing applications has already brought many benefits to the industrial sector, and as its use becomes more widespread, materials that had previously been designated as ‘un-processable’, due to their high viscosity or their thermal instability, can now be reinvestigated with the aid of supercritical fluids. SCF technology has not yet reached its potential within industry; however, considerable research into this field is ongoing and would indicate

400


that the number of applications and the usage of this technology are only likely to grow. The development of supercritical fluid-based processes as reactive media continue to be linked with the synthesis of materials with unusual physicochemical properties, which cannot be obtained with standard processes. SCFs have also recently been employed to address the drawbacks of existing scaffold fabrication methods. They can be used to create biocomposites and/ or preserve protein activity, and to manipulate scaffold porosity and architecture. Numerous processing strategies are emerging which rely on SCFs for either their plasticising or their antisolvent properties and in the near future SCFbased processes will begin to rival 3-D printing techniques. In polymerbased drug delivery, SCFs allow labile bioactive molecules, such as proteins, drugs and nucleic acids, to be introduced at the polymer processing stage – something that was not previously possible when using synthetic polymers. A current downside is the limited number of drugs that can be processed in scCO2, although additional solvents have been used in techniques where the SCF acts as the antisolvent to circumvent this, and as new drugs are developed the use of SCFs will become more widespread, suggesting an important role for this technology in future drug delivery applications. The wide range of applications clearly suggests an interesting and important future for SCFassisted polymer processing, but it also indicates the need for further investigation of specific interactions in polymer/SCF systems.

13.8

References

Arai Y, Sako T and Takebayashi Y (2001), Supercritical Fluids, New York, Springer. Bonnaudin N, Cansell F and Fouassier O (2003), Supercritical Fluids and Materials, Nancy, France, ISASF. Chattopadhyay P and Gupta R B (2002), ‘Supercritical CO2 based production of magnetically responsive micro- and nanoparticles for drug targeting’, Ind Eng Chem Res, 41(24), 6049–6058. Cooper A I, Kazarian S G and Poliakoff M (1993), ‘Supercritical fluid impregnation of polyethylene films, a new approach to studying equilibria in matrices; the hydrogen bonding of fluoroalcohols to (η5–C5Me5)Ir(CO)2 and the effect on C-H activation’, Chem Phys Lett, 206, 175–180. Fleming O and Kazarian S (2005), ‘Polymer processing with supercritical fluids’, in Kemmere M and Meyer T, Supercritical Carbon Dioxide, Weinheim, Germany, WileyVCH, 205–238. Ghaderi R, Artursson P and Carlfors J (1999), ‘Preparation of biodegradable microparticles using solution-enhanced dispersion by supercritical fluids (SEDS)’, J Pharm Res, 16(5), 676–681. Jung J and Perrut M (2001), ‘Particle design using supercritical fluids: literature and patent survey’, J Supercrit Fluids, 20, 179–219. Kazarian S G (2000) ‘Polymer processing with supercritical fluids’, Polymer Sci., 42, 78– 101.


401

Kazarian S G, Cooper A I and Poliakoff M (1994), ‘IR spectroscopy of intermolecular interactions in polyethylene films’, Opt Spectrosc, 76, 242–245. Lambert S M and Paulaitis M E (1991), ‘Crystallization of poly(ethylene terephthalate) induced by carbon dioxide sorption at elevated pressures’, J Superc Fluids, 4(1), 15– 23. Münüklü P and Jansens P J (2007), ‘Particle formation of edible fats using the supercritical melt micronization process (ScMM)’, J Superc Fluids, 40(3), 433–442 Saraf M K, Gerard S, Wojcinski II L M, Charpentier P A, DeSimone J M and Roberts G W (2002), ‘Continuous precipitation polymerization of vinylidene fluoride in supercritical carbon dioxide: formation of polymers with bimodal molecular weight distributions’, Macromolecules, 35(21), 7976–7985. Schiavonea H, Palakodatya S, Clarkb A, Yorkb P and Tzannis S (2004), ‘Evaluation of SCF-engineered particle-based lactose blends in passive dry powder inhalers’, Int J Pharm, 281 (1–2), 55–66. Schnitzler J and Eggers R (1999), ‘Mass transfer in polymers in a supercritical CO2atmosphere’, J Superc Fluids, 16, 81–92. Sugeta T, Okajima I, Sako T, Otake K and Kamizawa C (2000), ‘Decomposition and recycling of waste plastics using supercritical fluids’, Proc 5th Int Symp on Supercritical Fluids, Atlanta, GA. Sun Y P (2002), Supercritical Fluid Technology in Materials Science and Engineering, New York, Marcel Dekker. Tang M, Wen T Y, Du T B and Chen Y P (2003), ‘Synthesis of electrically conductive polypyrrole–polystyrene composites using supercritical carbon dioxide. II. Effects of the doping conditions’, Eur Poly J, 39(1), 151–156. Xu Q, Chang Y, He J, Han B and Liu Y (2003), ‘Supercritical carbon dioxide-assisted synthesis of poly(acrylic acid)/nylon 6 and polystyrene/nylon 6 blends’, Polymer, 44(18), 5449–5454.

14 Radiation processing of polymers V R A O, Mangalore University, India

Abstract: This chapter deals with the basic principles of radiation processing, types of radiation used, radiation chemistry, comparison of chemical processing and radiation processing, and the industrial applications of gamma rays and electron beams in the modification and enhancement of polymer properties. When a source of radiation from a gamma ray, electron beam or X-ray interacts with a polymeric material, its energy is absorbed by the material, and active species like radicals are produced, initiating various chemical reactions. These reactions-include crosslinking, degradation, grafting and curing. These radiation-induced reactions are responsible for many useful applications in plastic and rubber materials. Important properties of polymeric materials such as mechanical properties, thermal stability, chemical resistance, melt flow, processability and surface properties can be considerably improved by radiation processing. Key words: radiation processing, electron beam irradiation, radiation chemistry.

14.1

Introduction

Radiation processing is a well-established and commercial technology practised over the past 50 years for the modification and enhancement in the properties of polymeric materials. Radiation processing of polymers was introduced after World War II with the development of nuclear reactors. Modification in polymeric structure can be brought about either by conventional chemical means, usually involving silane or peroxide, or by exposure to radiation. Compared to conventional techniques, radiation processing is simple, faster, clean, effective, economical and environmentally friendly. Radiation processing directs energy to the polymer molecule with much greater precision and control. It does not require chemical catalysts or toxic chemicals or extreme physical conditions like high temperature and high pressure and can be carried out at ambient temperature. Further, irradiated materials do not themselves become radioactive. Throughput rates are very high and the treatment cost per unit of product is competitive with the conventional chemical processes. The application of radiation energy has proved its versatility and cost effectiveness in a number of chemical processes. Irradiated materials are usable immediately after processing. Thus radiation processing offers an alternative technique to conventional methods for developing new polymeric 402


403

products. Extensive research is being continued to explore the various applications of radiation processing technology. When radiation from a source of gamma rays, electron beam (EB) or Xrays interacts with a polymeric material, its energy is absorbed by the material and active species like radicals are produced, initiating various chemical reactions. These reactions include crosslinking, degradation, curing and grafting. These radiation-induced reactions are responsible for many useful applications in plastic and rubber materials. Important properties of polymeric materials such as mechanical properties, thermal stability, chemical resistance, melt flow, processability and surface properties can be considerably improved by radiation processing. Radiation processing has contributed significantly in industry for the development of special polymer products. Radiation processing technologies include radiation vulcanization of rubber latex, polymer recycling, production of heat shrinkable polymers and fibre reinforced composites, development of ion track membrane and micro device production (Clough and Shalaby,1991, 1996; Singh and Silverman, 1992; Woods and Pikaev, 1994; Ivanov, 1992; Charlesby, 1991; Makkuchi, 1999; Chmielewski et al., 2005; Chmielewski, 2006). Another achievement in radiation processing is the effective use of radiation-crosslinkable negative photoresist polymers and radiation-degradable positive photoresist polymers in microelectronics (Pethrick, 1991). Polytetrafluoroethylene (PTFE) micro powders are commonly processed by scissioning of the polymer by irradiation, thus it can be easily ground for use as additives in ink and lubricants (Lunkwitz et al., 1989; Neuberg et al., 1999). Radiation-induced grafting is beneficially utilized to modify the surface properties of polymers, such as permeation resistance, biocompatibility, adhesion, friction, hydrophilicity, hydrophobicity, etc. (Dworjanyn and Garnett, 1992). Ion exchange membranes and fuel cell/ battery separator films are made by grafting (Charlesby and Lawler, 1980; Holmberg et al., 1996). This chapter presents the basic principles of radiation processing, types of radiation used, radiation chemistry, comparison of chemical processing with radiation processing, and the industrial applications of gamma rays and electron beams in the modification and enhancement of polymer properties.

14.2

Types of radiation sources used

Various radiations such as X-rays, gamma and ultraviolet rays and electron beams are widely used for radiation processing. These radiations affect the molecular structure and macroscopic properties of polymeric materials. The energy ranges of various radiations are shown in Table 14.1. For radiation technologies, the main radiating sources employed are gamma rays and electron beam. Table 14.2 gives a comparison of electron beam and gamma rays (Drobny, 2005; Woo and Sandford, 2002).

404


Table 14.1 Frequency and wavelength of various radiation sources Radiation

Frequency (Hz)

Wavelength (µm)

Energy

Gamma rays X-rays UV IR Microwave Electron beam

1024–1019 1019–1017 1017–1015 1015–1012 1012–1010 1021–1018

10–7–10–6 10–6–10–3 10–2–1 1– 102 103–105 10–7–10–4

124 MeV to 124 keV 124 keV to 1.24 keV 1.24 keV to 12.4 eV 12.4 eV to 124 meV 124 meV to 1.24 meV 12.4 MeV to 12.4 keV

Table 14.2 Comparison of electron beam and gamma rays Characteristics

Electron beam

Gamma rays (60Co)

Energy (MeV) Maximum penetration (cm)* Dose rate Charge Rest mass Velocity

0.1–20 5.0 100 kGy/s –1 9.1 × 10–28 g 0.3–0.99c

1.3 50 10 kGy/h 0 0 c**

* At unit density. ** c = velocity of light.

In the process of interaction with the material, gamma rays transfer their energy to the material primarily by Compton scattering, i.e., scattering involving elastic collisions between incident photons and unbound or weakly bound electrons in which the incident energy is shared between the scattered electron and the deflected photon. These electrons recoil a short distance as unbound electrons, giving up energy to the molecular structure of the material as they collide with other electrons, causing ionization and free-radical formation. The scattered gamma ray carries the rest of the energy as it moves off through the material, possibly to interact again with another atomic electron. Since the probability of Compton scattering is low, gamma rays penetrate relatively deeply into the material. Accordingly, gamma rays deposit energy in material over relatively large volumes so that penetration is high, typically greater than 50 cm in unit-density material. The penetration power of cobalt-60 gamma rays is high. The intensity is reduced to one-tenth of its initial value after passing through 43.2 cm of a material of unit density (see Fig. 14.1). Gamma radiation penetrates approximately 10 times further into materials than 10 MeV electrons in the same material. The absence of both rest mass and charge gives gamma radiation a far greater penetration than accelerated electrons, whose penetration is primarily dependent on kinetic energy or the potential difference through which the electrons are accelerated.


405

100 90

Percentage depth-dose

80 70 60 50 Co γ radiation

60

40 30 20

Cs γ radiation

137

10 0 0

5

10

15

20

25

30

Depth (cm)

14.1 Percentage depth–dose curves for the irradiation of water by gamma rays. 180 160 140

Dose (%)

120 100 80 60 40 1

20 0

0

2

1

4

6

2

3 Depth (cm)

10 MeV

4

5

6

14.2 Percentage depth–dose curves for the irradiation of water by electron beam.

10 MeV electron beams typically penetrate approximately 5 cm in unitdensity material before losing their energy (see Fig. 14.2; Sarma, 2005). Another major difference between the two sources is the dose rate. A typical gamma irradiator delivers dose rates approximately between 5 and 20 kGy/h, while electron accelerators could deliver dose rates as much as

406


10 000 times higher. Under such dose rate conditions, significant thermal effects could arise in order to modify the material’s reaction pathways. Secondly, due to this huge dose rate difference, irradiation times are also vastly different. While a gamma facility delivers its dose over several hours, the electron accelerator would take mere seconds for delivery of the same dose. Electron beam radiation is generated in a high vacuum (typically 10–6 torr) by a heated cathode. The electrons emitted from the cathode are then accelerated in an electrostatic field applied between cathode and anode. When an electron beam enters a material, the energy of the accelerated electrons is greatly altered. They lose their energy and slow down almost continuously as a result of a large number of interactions, each with only a small energy loss, on average about 30 eV per collision. As high-energy electrons penetrate the surface they collide with atomic electrons of the material. These electrons, in turn, recoil and collide to set more electrons in motion so that from a relatively few electrons penetrating the surface, there results a multiplicity of electrons depositing energy in the material, primarily by the production of ions and free radicals. This results in higher doses being delivered to depths below the surface where the primary beam and its recoil electrons can no longer produce ionization. Thereafter, the electrons quickly lose their remaining energy, primarily by soft interactions with atomic electrons (excitation) and radiative losses. High-energy electrons are produced in electron beam accelerators. The electron beams attain high kinetic energies ranging from keV to several MeV in the accelerators. The electron beams generated by accelerators are monoenergetic in nature. Such beams are unidirectional and can be made to impinge straight on the materials and to sweep from one end to the other. The required dose is delivered to the product by transporting the product under the beam at definite speeds. These beams are generally preferred in applications to deliver high doses at faster rates in thin products. The absorbed dose is maximum just below the surface of the irradiated material and rapidly falls at greater depths in the material. Two types of radiation sources are more widely used: gamma rays emitted by cobalt-60 and electron beams generated by accelerators operating over a broad range of energies from 0.3–10 MeV up to 20 MeV. These energies are not high enough to initiate nuclear reactions; hence the irradiated material does not exhibit any radioactivity. In recent years, electron beam accelerators, as a source of high-energy ionizing radiation, have emerged as a preferred alternative for industrial radiation processing as they offer technological and commercial advantages over gamma rays. Industrial electron beams with energies in the 150–500 keV range are generally used in applications where low penetration is needed, such as curing of surface coatings, adhesives and paints. Accelerators operating in the 1–5 MeV range are used where more penetration is required as in crosslinking (Berejka, 1995). Electron beam


407

machines can also be used to generate X-rays, where more penetration is needed, although this conversion has relatively low efficiency. Ion beams are also being used in polymer processing. In the last 10 years, there has been increasing interest in the use of energetic ions to selectively modify the original structure of polymers by inducing crosslinking, grafting or multiple bond formation. Ion beam induced modifications in electrical, optical, mechanical and thermal properties have been noticed in polymeric materials. The interaction of the ion with the material is the deciding factor in the modification of materials. When an energetic ion penetrates the material, it loses energy mainly by (i) elastic collisions with the nuclei, known as nuclear energy loss, which dominates at an energy of about 1 keV/amu, and (ii) inelastic collisions with the atomic electrons of the matter, known as electronic energy loss, which dominates at an energy of about 1 MeV/amu or more. In the inelastic collision (cross-section ~10–14 cm2) the energy is transferred from the projectile to the atoms through excitation and ionization of the surrounding electrons. The amount of electronic loss in each collision varies from tens of eV to a few keV per angstrom (Å) (Kanjilal, 2001). Heavy ions with energies so high that the electronic energy loss process dominates are referred to as swift heavy ions. Swift heavy ions produced highly damaged zones called ion tracks in polymers, which have various applications. The ion tracks formed during heavy ion irradiation on polymers comprise highly damaged zones, nearly cylindrical with a length comparable to that of the ion range (Apel et al., 1998). These ion tracks are found to incorporate ions and medium-sized molecules such as impurities or dopants and dyes (Fink et al., 1995). The damaged zone can be etched by suitable chemicals to get pores (etched tracks) with diameters ranging from a few nm to a few µm. By careful selection of projectile, target, etchant and etching conditions, the etched track can be tailored towards any required shape, such as cylindrical, conical or hyperbolic, transmittent (in thin foils) or non-transmittent (Fink et al., 2005). The pore density can be changed by the ion fluence, which is controlled by either ion current or irradiation time or both. The pore density can be varied from 1 per cm2 to 106 per cm2 for micropores and up to 1011 per cm2 for nanopores. These nanopores and micropores find diverse applications as templates for the synthesis of micro-and nanostructures (Apel, 2001). Deposition of metallic nanowires and hollow nanotubes has been carried out into the pores (Toimil Molares et al., 2001; Fink et al., 2003). Many advanced applications of etched tracks have already been reported, such as the formation of nanosized or microsized diodes, tunnelling structures, field effect transistors, devices to control the perm selectivity of electrolytes, temperature sensitive valves, miniaturized magnetic field sensors, etc. (Fink et al., 2005). For example, by the grafting of thermoresponsive gels onto the walls of etched

408


tracks one can get nanovalves that open and close at a preset temperature. This can be used for intelligent in-vivo drug delivery (Fink and Chadderton, 2005). The replacement of the classical capacitor plates by ion track-made grids increases the capacity considerably. Polyimide is a very suitable polymer for this purpose, as its dielectric coefficient is only marginally frequency dependent, even at the highest frequencies (Fink, 2004).

14.3

Radiation chemistry of polymers

During the irradiation process, the energy from the radiation source – gamma ray, electron beam or X-ray – is transferred to the processed material, which results in a variety of chemical reactions that alter the molecular structure of the material. Free radicals and ions are produced when polymer is exposed to ionizing radiation (Woods and Pikaev, 1994). The energy required to break one chemical bond is much lower than the energy transferred by the electron beam. The electron beam knocks off electrons when it interacts with polymer (P) and generates free radicals and ions (see Fig. 14.3). These active species are responsible for initiating chemical reactions, leading to modifications like crosslinking (formation of network structure), degradation (chain scission and molecular weight reduction), grafting (polymerizing a monomer and grafting onto the base polymer chain) and curing (simultaneous polymerization and crosslinking). In addition to this, most of the polymers show gas evolution (Chapiro, 1962). In the radiation chemistry of polymers, generally the yield of the products tends to saturate with increasing dose, i.e., the yields of crosslinking, main chain scission, gas evolution, double bond and free radicals decrease with increase of dose (Seguchi et al., 1999). The hydrogen detachment is the primary chemical process in polymer radiolysis, and then the crosslinking, double bond and other products are formed through the reactions of active species such as free radicals. For polymers composed of hydrocarbons, the main gas is hydrogen and the minor gases are CH4, C2H6, C3H8, C2H4, CO2 and CO. In glassy and partially crystalline polymers, the gases generated during irradiation get trapped and may give rise to strains which produce

P

(P+)• + e

Ionization and excitation (a) +

P +e

P

Ionization

(b) P

P• Excitation (c)

14.3 Interaction of polymer with ionizing radiation.


409

cracks in the polymer. These strains can be released by heat treatment of the irradiated samples (Chapiro, 1962). For linear polyethylene, the mechanism of radiolysis can be represented as shown in Fig. 14.4. One hydrogen molecule produces one crosslink or one double bond. Yield of H2 decreases with increase of dose. It is well known that polymers containing double bonds or a phenyl group in the chain have high radiation resistance and the gas evolution is much reduced when compared to saturated polymer. For example in polyisoprene, the H2 yield is only 14% of that of polyethylene, where it is estimated that one double bond reduces the hydrogen evolution to 14% of that of polyethylene (Seguchi, 2001). Gases are formed as a result of atom or side chain abstraction. The nature of the gas closely reflects the structure of the macromolecules. It can be seen from Table 14.3 that the chemical nature of the gases closely corresponds to that of the side group in the repeating unit. Polymers generally undergo structural changes during irradiation. Depending on their chemical structures, some polymers crosslink while others degrade. A polymer with the following structure will be degraded by irradiation if it has no hydrogen at the α position (R1 ≠ H, R2 ≠ H) or crosslink when it contains at least one hydrogen at this position (R1 and/or R2 = H) (see Fig.14.5) (Woods and Pikaev, 1994). For example, crosslinking will occur in polyethylene whereas polypropylene, polymethylmethacrylate, PTFE, etc., in which the α hydrogens in the repeating units are substituted, undergo degradation. The G-value parameter is used to quantify the chemical yield resulting from radiation. It is defined as the chemical yield of radiation in the number of molecules reacted per 100 eV of absorbed energy. G(S) is the number of polymer chain scissions per 100 eV absorbed, and G(X) is the number of polymer crosslink sites per 100 eV absorbed. Materials with a G(S)/G(X) ratio smaller than 1 are favoured for crosslinking. Materials with a G(S)/G(X) ratio greater than 1 tend to undergo degradation. Materials whose G(X) and G(S) values are low are more resistant towards radiation. Gvalues for crosslinking and chain scission for some polymers are given in Table 14.4 (Cleland et al., 2003). More than 50 years ago, polyethylene was the first polymer to be recognized

—H2C—CH2—CH—CH2— —H2C—CH2—CH—CH2— + H2 —H2C—CH2—CH2—CH2—

irradiation

crosslink —H2C—CH

CH—CH2— + H2 double bond

14.4 Mechanism of radiolysis.

410

Advances in polymer processing Table 14.3 Gas evolution during irradiation Polymer

Repeating unit in the polymer chain

Polyethylene

—CH2—CH2—

Polypropylene

—CH2—CH— CH3

Polyisobutylene

Gas formed

H2

H2, CH4

CH3 —C—CH2—

CH4

CH3

Poly(methyl acrylate)

—CH—CH2— H3C

Poly(methyl methacrylate)

H2, CO, CO2, CH4

C O

O O

CH3

C

O

H2, CO, CO2, CH4

—CH2—C— CH3

Poly(vinyl chloride)

—CH—CH2—

HCl

Cl

R1 CH2

C n

R2

14.5 Vinyl polymer.

as crosslinkable by irradiation (Charlesby, 1960) and it is still the most important commercial polymer because of the great variety of products that can be made out of it. Polyethylene has an attractive G(S)/G(X) ratio because its G(S) values are about half as high as its G(X) value. Natural rubber and polybutadiene have a very favourable ratio because of their very low G(S) and high G(X) values respectively. Polystyrene has a favourable ratio, but its G(S) and G(X) values are both extremely low because of the stability of aromatic compounds. The ratios for polytetrafluoroethylene and polyisobutylene are extremely unfavourable for crosslinking. The factors


411

Table 14.4 G-values for crosslinking and chain scission Polymeric material

Crosslinking G(X)

Scission G(S)

Ratio G(S)/G(X)

Low density polyethylene High density polyethylene Polyvinylidene fluoride Polymethylmethacrylate Polymethylacrylate Nylon 6 Nylon 6,6 Polyvinylacetate Atactic polypropylene Isotactic polypropylene Polystyrene Natural rubber Polybutadiene Polytetrafluoroethylene Polyisobutylene Cellulose

0.8–1.1 0.5–1.1 1.00 0.5 0.5 0.67 0.50 0.30 0.27 0.16 0.019–0.051 1.05 5.3 0.1–0.3 0.5 low

0.4–0.5 0.4–0.5 0.30 0.77 0.04 0.68 0.70 0.07 0.22 0.24 0.0094–0.019 0.1–0.2 0.53 3.0–5.0 5 11

0.47 0.56 0.30 1.54 0.07 1.01 1.40 0.23 0.81 1.50 0.41 0.14 0.10 20 10 High

Source: Reprinted from NIMB, vol. 208, Cleland M.R, Parks L.A, Cheng S, ‘Applications of radiation processing of materials’, page 68, Copyright (2003), with permission from Elsevier.

affecting the G-value are chemical structure, radiation dose, atmosphere and temperature (Cleland et al., 2003). For a given polymer, G(X) and G(S) both change with radiation conditions such as the absorbed dose, temperature and environment. G(S) for a polymer generally increases more than G(X) with increasing dose. Therefore when G(X) of a polymer is much greater than G(S), the molecular weight would continuously increase due to continuous crosslinking, but the molecular weight will level off because G(S) will increase faster. When G(X) is greater but not much greater than G(S), G(S) will eventually become equal to G(X) and the molecular weight will show a turning point, with the overall reaction changing from crosslinking to degradation. Continuous degradation will occur when G(S) is greater than G(X). Since G(S) for a polymer increases more than G(X) with increasing dose, it is possible that upon reaching a certain dose level, crosslinking-dominated polymers like polyethylene and natural rubber latex will change to degradation dominated (Cleland et al., 2003). Examples of polymers predominantly undergoing crosslinking and scission are given in Table 14.5 (Charlesby, 1960, 1977; Silverman, 1977, 1981). Polymers with benzene rings in the chain tend to have high radiation stability. Polystyrene and polyethylene terephthalate (PET) are outstanding examples of radiation-resistant materials, while polypropylene, polytetrafluoroethylene (PTFE) and cellulose cannot tolerate very high doses.

412

Advances in polymer processing Table 14.5 List of some polymers undergoing crosslinking and scission Mainly crosslinking

Mainly scission

Polyethylene Polyacrylates Polysiloxanes Polyamides Polystyrene Polyvinyl alcohol Polyhydroxy ethyl methacrylate Polybutadiene Phenol formaldehyde resin Polyurethanes Natural rubber Ethylene–propylene–diene rubber

Polyisobutylene Polymethacrylates Polymethacrylamides Polyvinylidene chloride Polytetrafluoroethylene Chitin Chitosan Cellulose Polymethylstyrene Polypropylene ether

In general, radiation degradation is greater with gamma rays or X-rays in comparison to high-power electron beams, because lower dose rates and longer treatment times allow more oxidative reactions to occur. Recent studies have shown that PTFE, which is inherently of the scissioning type, can be crosslinked by proper choice of irradiation conditions. This can be achieved by irradiation at elevated temperature (>340°C), i.e. above the melting temperature of PTFE under inert atmosphere (Lappan et al., 2000; Tabata et al., 2001). For some polymers, elevated temperatures may increase the mobility of polymer chains and make them more favourable to crosslinking. Oxygen in air usually assists degradation through a peroxide radical mechanism, so an oxygen-free atmosphere would also be more favourable for crosslinking (Sarma, 2005). Another example is polystyrene that crosslinks at ambient temperature and undergoes chain scission above the glass transition temperature. However, experimental and environmental conditions influence the balance between radiation-induced crosslinking and scission reactions for most polymers. Irradiation in the presence of air, particularly at low dose rates and with thin samples, can result in predominantly oxidative scission reactions even for polymers of the crosslinking type (Clough and Gillen, 1990).

14.3.1 Dose requirements in irradiation process The absorbed dose (D) is the amount of radiation energy absorbed per unit mass of material. It is defined in terms of kilograys (kGy). Energy per unit mass, i.e. the specific energy, required for radiation-induced chemical reactions is proportional to the absorbed dose. According to the definition of the common dose unit, one kilogray (kGy) equals the absorption of 1 kilojoule (kJ) per kilogram of the material. Therefore the specific energy (E) required in kilojoules per kilogram (kJ/kg) is just


413

equal to the dose in kilograys: E = D (kJ/kg)

14.1

The specific energy is related to the molecular weight and the G-value (G) of the reaction through the following expressions: E = 6.022 × 1023(100/G) × 1.602 × 10–19/103 = 9.65 × 10 /G 3

= 9.65 × 106/G(mol. wt)

14.2

(kJ/mole)

14.3

(kJ/kg)

14.4

The factors used in this derivation are 6.022 ×10 molecules/mole, (100/G) electron volts per molecule, 1.602 × 10–19 joules/electron volt, 103 joules/kJ and 103/(mol. wt) moles/kg. By combining Eqns 14.1 and 14.4, the following expression for the dose is obtained (Charlesby, 1960, 1977; Silverman, 1977, 1981): 23

D = 9.65 × 106/G (mol. wt) (kGy)

14.5

G-values for many reactions are in the range from 0.1 to 10, with 1 being a typical value for crosslinking pure polymers. For example, if the molecular weight were 100 000 and the G-value were about 1, the dose needed to convert all of the molecules in the material would be about 100 kGy. This dose would be sufficient for many crosslinking applications. On the other hand, if the molecular weight were 100 and the G-value were as high as 10, the dose would be about 10 000 kGy. Such a high dose would be impractical because of the high energy requirement and the high processing cost. The doses used in commercial applications range from 0.1 kGy to 100 kGy. Even higher doses are required for some specific applications. The lowest doses are for processes like polymerization and curing of coatings. Typical doses required for the different radiation processes are given in Table 14.6.

Table 14.6 Dose requirements for various radiation processes Radiation process

Dose required (kGy)

Polymerization of monomers Modification of polymers Degradation of cellulosic materials Degradation of scrap PTFE Crosslinking Curing of coating and adhesives Curing of composites Grafting

10–30 10–50 50–250 500–1500 50–300 10–30 100–200 100°C). Exposure of polymer to high temperature may cause degradation in the polymer and thus affect the life of the product. Chemically crosslinked polymers do degrade when in operation due to oxidative degradation reactions in the presence of unconsumed vulcanizing agents like peroxides, oxygen in air and elevated continuous operating temperatures. Also the crosslinking agent gets absorbed partially by the fillers and on heating this generates additional free radical sites besides reacting with C–H and C–C, causing degradation. In EB crosslinking, since a crosslinking agent and heat energy are not required, degradation can be prevented. A higher degree of crosslinking up to 75–80% is obtained in electron beam crosslinking compared to 50– 55% in chemical crosslinking (Tikku, 2005). The degree of crosslinking is proportional to the radiation dose. The dose rate of the reaction can be varied widely and thus the reaction can be better controlled. Variation of the degree of crosslinking within a product can be achieved in the case of electron beam by a suitable choice of the electron energy, by multiple irradiations with different electron energies and/or by using metallic masks, which absorb the electrons or reduce their energy before they reach the product surface. Hence, it is possible to manufacture products with different material properties from the same starting material (Charlesby, 1977). Thus the unique advantage of radiation crosslinking is that crosslinked structures with improved mechanical and chemical properties can be prepared from thermoplastics and elastomers without any chemical agent and heat. Advantages of radiation-crosslinked plastics for innovative applications like packaging include improved impact resistance, higher thermal stability, good chemical compatibility towards aggressive chemicals, and higher barrier properties towards liquid and organic vapours. In the field of automotive parts manufacture, radiation-crosslinked polyamides 6 and 66 are mainly used due to their advantages such as a higher service temperature (long-term stability, improved heat distortion temperature) and improved chemical resistance (Rouif, 2005).

14.4.2 Degradation Radiation technology has been beneficially utilized to improve processing properties and/or materials compatibility parameters through degradation which involves chain scission, chain branching and/or partial oxidation of the macromolecular materials. Some polymers that undergo degradation upon EB processing include polytetrafluoroethylene (PTFE), polypropylene (PP) and cellulose. In degradation, chain scission occurs, which is the opposite of

418


crosslinking. It involves the rupturing of C–C bonds. Crosslinking increases the average molecular weight, whereas the scission process reduces the average molecular weight. During scission, the polymeric radicals formed will interact with the abstraction of hydrogen atoms and result in the formation of a double bond. Figure 14.8(a) illustrates chain scission in polymethyl methacrylate. In solution, scission proceeds in a different manner. During irradiation, solvent free radicals (X•) are formed, which interact with the polymer to produce polymer free radicals. In the presence of oxygen these free radicals form peroxy species, which on decomposition form smaller molecules. The oxidative degradation of the polymers is shown in Fig. 14.8(b). Although degradation usually brings about deterioration in mechanical properties of polymers, in some cases good applications have been found. Recently much attention has been paid to application of radiation processing technology for degradation, for reasons such as the ability to carry out the process at room temperature, reliability of process control, large-scale application, and economic competitiveness to alternatives. Both gamma and electron beam can be applied for the degradation process.

14.4.3 Curing Curing is a combination of polymerization and crosslinking of monomers and oligomers (polymers with low molecular weight) initiated by radiation. Electron beam curing has emerged as an affordable and efficient non-thermal process which utilizes high-energy electrons and/or X-rays at controlled doses to polymerize and crosslink polymeric material. Traditional thermal curing has many disadvantages, for example it requires long cure time and high energy consumption, creates residual thermal stresses, CH3 —CH2—C—CH2—

CH2

CH3 irradiation

—CH2—C• + CH2

COOCH3

—CH2—C

COOCH3

+ H3C—

COOCH3

(a)

X • + —HC—CH2—

—C•—CH2— + HX

O2 —C•—CH2 + O2

—C•—CH2—

decomposition

(b)

14.8 Mechanism of degradation.

—C

O + O•—CH2—


419

produces volatile toxic by-products and requires expensive tooling for high cure temperatures. The advantages of EB curing technology over the conventional autoclave (thermal) curing process include faster cure time leading to significant improvements in throughput, low-temperature curing which reduces internal stress, ability to utilize low-cost tooling materials, reduced volatile emission, reduced energy consumption and control over curing energy-absorption profile, capacity to cure larger parts and better process control. Several studies have shown manufacturing cost savings of 10–40% compared to thermal processing (Goodman and Palmese, 2003). This is because electron beam processing, as well as the shorter curing time, cures the material at ambient temperature, allowing tools to be made from very low-cost materials such as foam or wood. Savings are marked for complexshaped components such as engine inlets and cryogenic tanks. Another advantage of electron beam processing is its ability to combine several different resin systems in the same curing cycle. This may not be possible with thermal curing because the different systems may need different temperatures or cure times. This is not a problem for electron beam curing because the beam of the electrons is rastered across the part being cured and the dose given to each area of the component can be pre-programmed. EB cured materials have been shown to possess excellent mechanical properties, high glass transition temperature (Tg) and low void content. These materials show excellent property retention after thermal/cryogenic cycling (Farmer et al., 1998). The main potential applications are in the automotive and aerospace industries (Berejka and Eberle, 2002; Saunders et al., 1995). A range of electron beam curable epoxy resins is now commercially available and they have been successfully used in a large number of applications. EB curing of epoxy resins can be carried out either through a free radical mechanism where no initiator is required or through a cationic mechanism where initiator is needed. These materials generally offer low shrinkage on curing (2–3%), a glass transition temperature of up to 400°C, very low water absorption (less than 2%) and an extended storage lifetime at ambient temperature. They all cure in minutes to produce composites with high performance mechanical properties. Composite manufacturing is one of the major areas of applications of radiation curing (Czvikovszky, 2003).

14.4.4 Grafting Radiation-initiated grafting is a powerful method for surface modification of polymer materials (Dworjanyn and Garnett, 1992). Radiation-induced grafting can be performed using different radiation sources. The graft copolymer will have slightly different characteristics depending on the type of radiation used (Svarfvar, 1997). The radiations used are gamma, UV, X-rays and electron beam. Among the various forms of radiations, electron beam radiation has

420


useful advantages. It is powerful and easily attainable, with high penetration depth, a wide range of dose rates, high efficiency and safe operation. A large number of free radicals are formed during irradiation in the material without the use of chemical initiators. These radicals can undergo reaction with another monomer to produce macromolecular chains that are covalently bonded to the irradiated polymer. The dose requirement for grafting is usually less than 10 kGy (Cleland et al., 2003). Grafting has been successfully carried out onto polymer films, membranes, fibres and powder. The surface properties of polymer can be modified by graft copolymerization with different monomers. Hydrophilic properties can be imparted to hydrophobic polymers. The chemically inert surface of low density polyethylene (LDPE) film and sheets were grafted with polyhydroxy ethyl methacrylate (HEMA) through electron beam irradiation. The hydrophilicity of poly-HEMA-g-LDPE is remarkably improved compared to that of ungrafted LDPE (Morshedian et al., 2005). There are numerous reports describing the grafting of polymers such as acrylates onto fibres and fabrics. Most frequently the grafting treatment renders the fibre more hydrophilic, which enhances dyeability. A wide range of fibres have been beneficially grafted, including PP, nylon and wool (Nho et al., 1999). Grafting of monomers to wool improves properties such as permanent press, wrinkle recovery, abrasion resistance and dyeing. 4-Vinyl pyridine copolymer has significantly improved the dyeing characteristics of backbone wool (Garnett, 1979). Biocompatibility of various polymers can be improved through grafting. Ion exchange membranes and fuel cell/battery separator films are made by grafting styrene onto porous polyvinylidene fluoride membranes (Holmberg et al., 1996). Radiation grafting onto cellulose and starch has been carried out to enhance thermal stability, microbial resistance and water absorbance. Grafting of monomers such as acrylamide onto starch has produced super-absorbents for water. Radiation grafting also imparts conductivity to the matrix (Bhattacharya, 2000). This is the unique method of grafting conducting matrix onto the insulating one. It can exhibit several potential applications such as EMI shielding, conducting coating and antistatic agents. Polypyrrole has been successfully grafted onto Teflon (Bhattacharya et al., 1994). The two types of radiation-induced grafting are free-radical grafting and ionic grafting. Free-radical grafting Free-radical grafting proceeds in three different ways: pre-irradiation, peroxidation and mutual irradiation. •

Pre-irradiation: In the pre-irradiation technique, polymer backbone (P) is first irradiated in vacuum or in the presence of inert gas to produce free radicals. The irradiated polymer is then treated with the monomer


•

•

421

(M), in liquid or vapour state or as a solution in a suitable solvent (see Fig. 14.9(a)). Peroxidation: In the peroxidation grafting method, the polymer is subjected to EB radiation in the presence of oxygen or air. The result is the formation of hydroperoxides or diperoxides, depending on the nature of the backbone polymer and irradiation conditions (see Fig. 14.9(b)). The peroxy products are treated with the monomers at higher temperature, when the peroxides undergo decomposition to radicals, which initiates grafting. The advantage of this technique is that the intermediate peroxy products can be stored for a long period before performing the grafting step. With cellulose and polyethylene, it has been observed that the peroxide is stable enough to initiate copolymerization with styrene several years after its original formation (Garnett, 1979). Mutual irradiation: In this case both monomer and polymer backbone are irradiated simultaneously (see Fig. 14.9(c)).

Ionic grafting Radiation grafting can also proceed through an ionic mode, with the ions formed through high-energy irradiation. Ionic grafting may be of two different types: cationic or anionic. The polymer is irradiated to form the polymeric ion, and then reacted with the monomer to form the grafted copolymer. If the polymer system is extremely dry, it is possible to carry out radiation grafting at room temperature or below by cationic or anionic mechanisms. Thus monomers such as styrene, acrylonitrile, methyl styrene, vinyl-n-butyl ether and isobutylene have been radiation grafted by ionic processes to backbone polymers like polyethylene and polytetrafluoroethylene. Higher reaction rates P

P + O2

P• + M (a)

PM

P—O—O—H or P—O—O—P

P—O—O—H

P—O• + OH•

P—O—O—P

2P—O•

P—O• + OH• or P—O• + M

P—O—M

(b) P+M

P• + M• (c)

P—M

14.9 Mechanism of free-radical grafting.

422


can be attained in ionic grafting when compared with corresponding freeradical grafting (Garnett, 1979; Bhattacharya and Misra, 2004). Thus, lower radiation doses are sufficient to bring about grafting. The probable mechanism for radiation-induced cationic grafting of monomer CHR==CH2 to polyethylene is shown in Fig. 14.10(a). An alternative cationic grafting mechanism can proceed through monomer radical cation, which subsequently forms a dimer. Charge localization in the dimer occurs in such a way that the dimer radical cation then reacts with the radical produced by the irradiation of the polymer (see Fig. 14.10(b)). An analogous mechanism involving an anion as the initiator can be proposed for anionic grafting (Garnett, 1979; Bhattacharya and Misra, 2004). Advantages of EB-initiated grafting over chemical grafting Radiation-induced grafting differs from chemical initiation in many aspects. In the radiation technique an initiator is not required and free-radical formation is on the backbone polymer/monomer, whereas in the chemical method, free radical forms first onto the initiator and then is transferred to the monomer/ polymer backbone. Unlike the chemical initiation method, the radiationinduced process is free from contamination, so that the purity of the processed products may be maintained. Chemical initiation is limited by the concentration of the initiator, and it may be difficult to determine the accurate concentration of the initiator in pure form (Bhattacharya and Misra, 2004). Chemical initiation

irradiation

CH2—CH—R + H2C

CH—R (a)

irradiation

CHR

CH2

CHR

CH2 + [CHR

[CHR

CH2]

CH2]

CHR—CH2—CH2—CHR

+ CHR—CH2—CH2—CHR

(b)

14.10 Ionic grafting.

CHR—CH2—CH2—CHR


423

often brings about problems arising from local heating of the initiator, which is absent in the formation of free-radical sites by radiation. The radiation processing, however, is temperature independent, which is therefore considered as a zero activation energy process for initiation. The efficiency of these two processes has been compared, and it is estimated that the same number of initiating radicals are produced in unit time with a radiation dose of 1 rad or a chemical initiator, such as benzyl peroxide, at a concentration of 0.1 M. Due to the large penetrating power of higher-energy radiation, methods using radiation initiation provide the opportunity to carry out grafting at different depths of the base polymer matrix. Moreover, the molecular weight of the products can be better regulated in radiation techniques, and these are also capable of initiation in solid substrates. It should be known whether the polymer is stable in the radiation range of interest.

14.5

Industrial applications

Fifty years of research and development in polymer radiation chemistry have led to a number of commercial applications. Wire and cables The crosslinking of insulation on electrical wires and cables was one of the first practical applications of radiation processing. Polymers used in this application include polyethylene, polyvinylchloride, ethylene–propylene rubber, polyvinylidene fluoride and ethylene–tetrafluoroethylene copolymer. Property improvements obtained by irradiation include increased tolerance to hightemperature environments and overloaded conductors, fire retardation, increased abrasion resistance and tensile strength, reduction in cold flow, and increased resistance to solvents and corrosive chemicals as well as other important characteristics (Bennet, 1979). Irradiated wires are commonly used in automobiles, military vehicles, aircraft, spacecraft and many other applications where high performance is required. Large amounts of flame-retardant radiation-crosslinked wires and cables are incorporated in automobile harnesses, for safety and to reduce the total weight and increase the efficiency of the automobile (Chmielewski et al., 2005). Polyurethane covering the outside jackets of sensor cables for antilock brake systems is also radiation-crosslinked to improve the resistance against hot water (Makkuchi, 1999). Heat-shrinkable products One of the largest applications of EB crosslinking is in the manufacture of heat-shrinkable products because of their higher and uniform degree of crosslinking, ease of handling, excellent heat shrink/expansion properties

424


and fast process with minimum scrap. RAYCHEM has pioneered this technology and is one of the largest users of electron beam crosslinking (Cook, 1990; Cook and Halperin, 1998). Thin-walled plastic tubing and plastic films are crosslinked to obtain an interesting phenomenon called the memory effect from the crosslinked network. Radiation crosslinking fixes or stabilizes the original dimensions of the tubing or films. When the material is heated above the melting point of the crystalline regions that exist in the polymer, it becomes elastic and can be expanded to at least twice its original dimensions. When cooled, it maintains the expanded dimensions but retains the memory of its original dimensions. The material retains its stretched size and shape until the next time it is heated. When heated again, the crosslinks that were created in the sample’s smaller unstretched state pull the material back into its original shape, causing it to contract to the original dimensions. Crosslinking at room temperature without melting of crystallites in the material is the advantage of using ionizing radiation. It cannot be easily achieved with conventional chemical techniques, as these require heating of the material to induce crosslinking, with consequent melting of crystals; upon cooling, crystallinity is reduced. Polyethylene is commonly used for this application (Ota, 1981). The heat shrinkability of the material increases with an increase in the radiation dose (Mishra et al., 2008). The dose required for the manufacture of heat-shrinkable products is in the range 100–250 kGy. Many commercial products have been developed. Some examples are encapsulation for electronic components, bundles of electrical wires, exterior telephone cable connectors, etc. The applications for films are mainly in the food industry, where heatshrinkable wrapping material is used to make attractive sealed packages. Polymer blends Irradiation is found to be very useful in the processing of polymer blends which often undergo physical phase separation due to incompatibility of the components. In a number of blend systems, irradiation has been used to induce crosslinking of one or more of the components, and/or formation of crosslinks between the different phases, resulting in the improvement of physical properties (Zhang and Xu, 1993; Spenadel, 1979). EB irradiation of a physical blend as it emerges from an extruder can be used to lock in a particular desired morphology of one crosslinkable phase that is dispersed in a non-crosslinkable phase (Clough, 2001). When this material is used for subsequent mouldings, the non-crosslinked phase flows, while the crosslinked phase largely maintains its shape. This procedure allows significant control over the properties of the final material made from the blend (Van Gisbergen et al., 1989; Van Gisbergen and Meyer, 1991; Van Gisbergen and Overbergh, 1992).


425

Emulsion polymerization Another application of radiation-induced crosslinking is the formation of polymeric spheres and particles having diameters in the micron and submicron range (McHerron and Wilkes, 1991; Ye et al., 2002). This is accomplished by gamma or EB irradiation of an emulsion containing one or more monomers, usually multifunctional. A variety of applications are possible for such materials, including chromatographic or catalytic supports and latexes. Micro-emulsion made from such processes may have application in nanotechnology (Xu et al., 1999). PTC (positive temperature coefficient) materials The PTC (positive temperature coefficient) effect is an interesting phenomenon exhibited by some conductive polymer composites. The main feature of PTC materials is that upon heating, the conductive system shows a sharp resistivity increase near the melting region of the semicrystalline polymer matrix. Conductive polymer composites typically consist of a polymeric matrix into which carbonaceous filler is incorporated. For example, polyethylene and ethylene–vinyl acetate copolymers (EVA) filled with conductive fillers such as carbon black, can be crosslinked by irradiation to obtain the PTC effect. The PTC effect is sometimes followed by a negative temperature coefficient (NTC) effect, namely a resistivity decrease. The most common explanation for the PTC effect is that as the melting temperature is approached, conductive pathways are broken because of the volume expansion of the matrix. Above the melting temperature, carbon black particles cluster because of their tendency to agglomerate. This consequently results in the formation of new pathways and thus a conductivity increase, giving rise to the NTC effect. Crosslinking can reduce and even eliminate any NTC effect that would be observed after the melting point of the polymer matrix if the polymer has not been crosslinked. Crosslinking may inhibit the rearrangement of filler particles that would otherwise take place in the softened polymer and so it imparts good electrical reproducibility in heating/cooling cycles. PTC intensity increases with an increasing degree of crosslinking (Lee et al., 2002). The importance of PTC materials has been well understood in the electrical and electronic industries. Especially, self-regulating heaters have become the most popular form of electric heat tracing products because this technology eliminates the possibility of heater burnout due to an inability to dissipate internally generated heat, which is often the cause of heater failure. PTC materials can also be used as current limiters, over-current protectors, microswitches and sensors and other devices (Hall, 1999; Huang et al., 2004).

426


Hydrogels Hydrogels are polymeric materials which exhibit the ability to swell in water, i.e. they retain a significant fraction of water within themselves but do not dissolve in water. In medical applications, polymers based on hydrogel biomaterials are largely used. Hydrogels have excellent biocompatibility and can be used for wound dressing, controlled release of drugs and enzyme support. Some polymers that can form hydrogels are polyvinyl alcohol, polyacrylamide, polyvinylpyrollidone, polyethyleneoxide, polyhydroxyl ethyl methacrylate (HEMA) and methyl cellulose. Hydrogels can be prepared by ionizing radiation (Rosiak, 1991; Rosiak and Olejniczak, 1993; Razzak et al., 1999). On irradiation of polymers, radicals are formed which can combine to form hydrogels. If the radicals are favourably positioned, new covalent bonds between the polymer chains are formed and thus the hydrogel turns into an insoluble fraction depending upon the amounts of these new bonds formed. On further irradiation, the amount of the gel increases, although a part of the macromolecules may still be unbound (sol). In the case of hydrogel formation, a three-dimensional network is developed, which exhibits higher molecular weight (Bhattacharya, 2000). Hydrogels prepared by a radiation technique are preferred because the network can be readily controlled and the prepared materials can be sterilized. Most of the products are based on poly(HEMA) hydrogel. Cellulose and its derivatives generally undergo degradation by high-energy radiation. However, radiation crosslinking was observed in highly concentrated solution (paste-like condition) of carboxymethyl cellulose (CMC) (Fei et al., 2000). The major advantage of this crosslinked hydrogel is its biodegradability. Crosslinked CMC is suitable for healthcare products such as surgical operation mats for prevention of bedsores. Crosslinked CMC hydrogel is applied as a coolant to keep freshness of vegetables and fish at low temperatures. Elastomers – vulcanization of natural rubber latex As early as the 1950s, rubbery polymers were found to be crosslinkable when exposed to high-energy radiation. Natural rubber latex is an aqueous dispersion of cis-1,4-polyisoprene particles whose size is usually of the order of several hundred metres. When the latex is irradiated, the rubber molecules in the latex particles are crosslinked to form a kind of pre-vulcanized latex, which is processed by dipping or extrusion to make rubber products such as gloves and pipes. Conventional crosslinking or vulcanization is carried out by heating in the presence of sulphur. A small amount of a toxic substance, nitrosamine, formed during vulcanization, remains in the product. In radiation vulcanization of rubber latex, the products are extremely safe due to the absence of nitrosamine.


427

During the manufacture of automobile tyres, some of the tyre components are irradiated to obtain partial crosslinking before the tyre is assembled. This stabilizes their thickness during the final thermal curing process. It also prevents the steel belt from migrating through its supporting rubber layer, thus a higher quality tyre with more uniform thickness and better balance can be obtained. This allows the tyre to be made thinner to save material and to reduce cost. A thinner tyre also generates less frictional heating on the road. Typical materials are isoprene and diene elastomers. Dose requirements are in the range of 30–50 kGy (Hunt and Alliger, 1979; Bradley, 1984). Table 14.8 gives a comparison of properties of EB-crosslinked and chemically crosslinked elastomeric cables of size 1 × 150 mm2 (Tikku, 2005). Particle size reduction A well-known example of application for degradation is making fine polytetrafluoroethylene (PTFE) powders. It is an inherently chain-scissioning polymer and has a high G-value for scission. EB irradiation of PTFE in air results in the production of micro-powders having lower molecular weight (Lunkwitz et al., 1989; Neuberg et al., 1999). PTFE becomes more brittle on degradation. This effect is used to convert scrap PTFE into fine particles of micron size. The unirradiated scrap is too tough, pasty and slippery to grind, but the irradiated material can be ground easily. The EB treatment of PTFE is now a large-scale industry in which PTFE micro-powders are commonly processed by scissioning of the polymer by irradiation; the polymer can then be easily ground for use as additives in ink and lubricants. The powdered PTFE has additional functional groups that are not present in the original PTFE, e.g. the carboxylic acid group, if irradiation is carried out in the presence of oxygen or air. The polar functional groups such as the carboxylic acid group in PTFE can help improve the compatibility of PTFE with other polymers (Lappan et al., 1999). Melt flow index (MFI) Another effect of degradation is the increase in the melt flow rate of the polymer. Polymers can be intentionally degraded by irradiation in air to Table 14.8 EB-crosslinked vs. chemically crosslinked elastomeric cables Tests

EB crosslinked

Chemically crosslinked

Operating temperature (°C) Minimum bending radius (mm) Radiation resistance (Mrad) Oil resistance (60°C diesel oil) (% swell) Mass of cable (kg/km)

125 62 >200 15 1365

90 80 >100 20 1970

428


improve their processibility. Irradiation of polypropylene (PP) under inert atmosphere causes a combination of chain scission and long-chain branching and improves the melt strength. This allows different moulding options for PP and allows easier conversion of PP into fibres. This technology has been adopted in the commercial processing of PP items (Berejka, 1995). Irradiated polymers can be blended with unirradiated polymers to adjust the melt flow and improve the processibility (Tervoort et al., 2000). Composite materials EB curing of polymer matrix composites (PMC) is an emerging technology for the manufacture of a variety of structures and components for aerospace and automotive applications (Saunders et al., 1995; Berejka and Eberle, 2002). Medium to high energy beams are used in the preparation of advanced composite materials. Doses are typically in the range of 100–200 kGy. Conventionally these composites were cured by thermal means which require ovens or autoclaves operating at high temperature. A number of benefits have been identified for EB curing of PMC. EB curing of PMC is a considerably more energy efficient process compared to thermal curing (Saunders et al., 1997). Electron beam curing can reduce residual internal stresses in the microstructure of composite parts. The stresses are induced during cooling and can develop within layers, due to the low linear expansion of the fibres compared to the resin, or between layers. Such stresses can limit composite part design. However, electron beam curing reduces such stresses because the processing temperature can be ambient or even lower. This is very beneficial for producing cryogenic tanks, which must hold fluids at temperatures of –200 to –255°C. Conventional thermal curing is carried out at 150 to 200°C, and this large difference can cause large stresses, which are eliminated using the electron beam technique. Interface is an important factor, especially in the case of injection mouldable composites of short fibres and thermoplastics, and this interface can be greatly improved through ionizing radiation (Czvikovszky, 2003). Large carbon fibre reinforced building elements for aerospace applications are cured at room temperature with radiation. By this technique, the deformations observed in heat-cured samples are avoided. Large electron accelerators operating with energies ranging from 10 MeV to 20 MeV are required in order to fully penetrate such pieces. The aerospace industry is developing EB technology to make specific products like cryogenic fuel tanks, improved canopy frames for jet aircraft, and all-composite military aircraft (Saunders et al., 2000). Electron beam curing could also be used to repair all the composite components on the aircraft, minimizing downtime and thereby saving money. The use of radiation techniques to process or produce nanostructured materials has been exploited in the production of nanocomposite materials.


429

The addition of inorganic nanoparticles to polymers can enhance conductivity and mechanical toughness, and this is useful for application in areas such as organic batteries, microelectronics, non-linear optics and sensors (Friedrich, 2005). Recently, electron beam-cured layered-silicate and spherical silica epoxy nanocomposites have been developed (Chen and Anderson, 2007). Polymer-based nanocomposite has also been developed by the radiationinduced grafting of inorganic particles onto polymer backbone. Polypropylene/ polyhedral oligomeric silsesquioxane (PP/POSS) nanocomposites were prepared by in-situ radiation-induced grafting of POSS on to PP (Choi et al., 2008). Wood polymer composites with greatly enhanced mechanical properties and moisture repellence can be made by radiation curing of wood impregnated with monomers. Production of wood polymer composite is achieved by impregnation of acrylate, styrene, vinyl acetate or other crosslinkable monomer in wood followed by polymerization using gamma or EB to get a product with enhanced hardness, tensile strength, scratch resistance, moisture repellency and resistance to biodegradation. The process takes the advantage of the fact that wood has high porosity and has up to 80% voids. Commercial products utilizing wood polymer composites include flooring and tabletops (Czvikovszky, 1992). Coatings and adhesives Solvent-free coatings and adhesives are cured (polymerized) by treatment with low energy electron beams (Menezes, 1990). These materials are combinations of oligomers and monomers, which control the viscosity before curing. Volatile solvents are not needed and curing occurs without loss of material. Typical oligomers are acrylated urethane polyester, acrylated epoxies and polyether. A typical multifunctional monomer is trimethylolpropane triacrylate (TMPTA). Dose requirements are relatively low, in the range of 10–30 kGy (Nablo and Tripp, 1977). Polymer recycling The major problems associated with polymer recycling are incompatibility of different polymer types leading to separation of phases, and the degradation of products in the polymer mixture brought about by environmental factors (Chmielewski et al., 2005). Ionizing radiation offers a unique solution to the problem of recycling due to its ability to induce crosslinking or scission of a wide range of polymers without introducing any chemical initiators and without dissolving the sample, thus avoiding phase separation. Beneficial effects of radiation recycling are enhancement in mechanical properties of recovered materials/blends through crosslinking or through surface modification

430


of different phases being combined, and decomposition of polymers, particularly through chain scission, leading to recovery of low molecular weight mixtures (Czvikovszky and Hargitai, 1999). Micro-device production (lithography) Another promising processing technology is the production of microelectronic circuits, micro-machines and ultra-small devices (Pethrick, 1991). Radiationcrosslinkable negative photoresist polymers and radiation-degradable positive photoresist polymers are effectively used in microelectronics. Lithography functions through radiation-induced chemistry which brings about a change in the solubility of the polymer molecules in the exposed regions. Both negative and positive photoresists are used; a negative resist becomes less soluble and a positive resist becomes more soluble upon exposure. In either case, a subsequent solvent development step removes the more soluble material, leaving a patterned polymer surface. These photoresists applied in the manufacture of integrated circuits make it possible to compile more than a million transistors in a few cm3 volume (Czvikovszky, 2003). The conventional lithography process used for making computer chips involves patterning with light of a thin polymer layer that is spin-coated on a silicon wafer, but this photolithography is reaching a limitation due to diffraction effects. Membrane technology (ion track membranes) One of the most exciting developments in ion irradiation of polymers is the creation of microporous and nanoporous membranes having uniform sized pores. Ion tracks formed during heavy ion irradiation on polymers can be etched chemically to get membranes with uniform pores. Track-etched membranes are finding increasing application in areas like microfiltration, ultrafiltration, diffusion, pervapouration, medical science, etc. For many of these applications, porous membranes with uniform pore size and density are required. There are many methods to prepare microporous and ultraporous membranes, such as phase inversion, sintering, track etching, leaching, etc. Membranes prepared by methods other than track etching have disadvantages such as low tensile strength, non-uniform pore density, large pore length, etc. Homogeneity in pore size can be obtained in track-etched membranes. Polymers that are widely used for ion track membranes are polyethylene terephthalate (PET) and polycarbonate (PC). By bombarding PET or PC films with swift heavy ions (e.g. Ar+, N+ or Xe+), latent radiation damage linear tracks are created through the samples. The track-etched membranes thus obtained are used as template materials for the synthesis of micro- and nanostructures. The template base method consists of filling the pores with desired materials. The material deposition


431

into the pores is usually performed by electrodeposition or by chemical polymerization (Ferain and Legras, 2003) or by electroless plating (Piraux et al., 1999). The micro- or nanomaterials that are produced in this way take the form of wires or tubules. Magnetic, conducting and superconducting nanowires and nanotubules in array or isolated mode possessing special properties have been manufactured in this way (Apel, 2001). Track-etched membranes with pore diameters in the range 0.2–1.3 µm have been successfully used as templates for preparation of polypyrrole nanotubules (Starosta, 2006). A number of researchers have succeeded in grafting hydrogels into the pores of ion track membranes. For example, poly(N-isopropylacrylamide) has been grafted onto the nanopores of track-etched membranes of PET (Shtanko et al., 1999; Alem et al., 2008). These membranes also find applications in the biomedical field in filtering bacteria of specific dimensions. Normally, in filter applications, the microholes get clogged. The clogging can be removed by expanding these microholes in order to use them again. Such a possibility has been demonstrated by the work of Reber et al. (1995) in which hydrogel was coated on the inner walls of the microholes. The hydrogel expands at elevated temperature of 80°C, which allows control of the size of microholes. The use of micropore membranes as a controlled drug delivery system has also been demonstrated (Rao et al., 2003). Normally, the medicine is applied to a wound or infected area at regular intervals. The advantage of using these micropore membranes is that the medicine can be provided continuously to the infected area through the pores at a controlled rate.

14.6

Conclusions

Unlike the conventional chemical methods of polymer processing, radiation processing has many advantages which have led researchers to adopt a comprehensive approach towards the successful utilization of radiation sources as an alternative technology for polymer processing. Since then remarkable advancements have been made. The last 50 years have witnessed an exponential growth in radiation processing. At present radiation processing has emerged as an economical, effective and competitive method for developing highquality, high-performance polymeric materials suitable for space, defence, automobile, medical and other sectors. The radiation-induced chemical reactions such as crosslinking, degradation, curing and grafting have been beneficially utilized in industry for the production of special polymeric products like heat-shrinkable polymers, rubber tyres, composites, etc. Among the various radiation-induced modifications, electron beam processing of polymers has gained special importance due to its advantages over other radiation sources and chemical methods. More than 1300 electron accelerators in the range of 200 keV to 10 MeV are in use worldwide for polymer processing. Among

432


the different radiation processing technologies, a few are still in the developing stage and a few in the laboratory stage. The processing technologies having enormous commercial potential which are currently in the developing stage include micro-electronic circuit production, natural polymer processing, production of nanomaterials and nanocomposites. Radiation processing applications involving ion beam treatment of polymers offer exciting prospects for commercialization and are under active investigation in many research laboratories. A major field of potential future application is track-etched membranes. These can serve as templates for making various micro- and nanostructures. With the continuing rapid advances in radiation processing technology, many novel products can be expected in the future.

14.7


Makuuchi K, Nakayama H (1983), ‘Radiation processing of polymer latex’, Progress in Organic Coating, 11, 241–265 Radiation Processing of Polysaccharides, International Atomic Energy Agency, IAEA, TECDOC-1422, November 2004 Singh A, Bahari K (2001), in Utracki L A (ed.), Polymer Blends Handbook, Kluwer Academic Publishers, Dordrecht Knobel T M (1996), ‘Electron beam facilities operation: Contract and dedicated composite curing, Electron Beam Curing of Composites Workshop, Oak Ridge, TN Knolle W, Trautman C (eds) (1999), Ionising Radiation and Polymers, NIMB, p. 151

14.8

References

Alem H, Duwez A S, Lussis P, Lipnik P, Jonas A M, Demoustier-Champagne S (2008), ‘Microstructure and thermo-responsive behavior of poly(N-isopropylacrylamide) brushes grafted in nanopores of track-etched membranes’, J Membrane Sci, 308(1/2), 75–86 doi:10.1016/j.memsci.2007.09.036 Apel P (2001), ‘Track etching technique in membrane technology’, Radiat Meas, 34(1/ 6), 559–566 doi:10.1016/S1350-4487(01)00228-1 Apel P, Schulz A, Spohr R, Trautmann C, Vutsadakis V (1998), ‘Track size and track structure in polymer irradiated by heavy ions’, NIMB, 146(1/4), 468–474 doi:10.1016/ S0168-583X(98)00445-5 Bennet Jr E W (1979), ‘Applications of irradiation to industrial wire and cables’, Radiat Phys Chem, 14(3/6), 947–951 doi:10.1016/0146-5724(79)90132-8 Berejka A J (1995), ‘Irradiation processing in the ‘90’s: Energy savings and environmental benefits’, Radiat Phys Chem, 46(4/6), 429–437 doi:10.1016/0969-806X(95)00188-4 Berejka A J, Eberle C (2002), ‘Elctron beam curing of composites in North America’, Radiat Phys Chem, 63(3/6), 551–556 doi:10.1016/S0969-806X(01)00553-9 Bhattacharya A (2000), ‘Radiation and industrial polymers’, Prog Polym Sci, 25(3/10), 371–401 doi:10.1016/S0079-6700(00)00009-5


433

Bhattacharya A, Misra B N (2004), ‘Grafting: a versatile means to modify polymers. Techniques, factors and applications’, Prog Polym Sci, 29(8/12), 767–814 doi:10.1016/ j.progpolymsci.2004.05.002 Bhattacharya A, De A, Bhattacharyya S N (1994), ‘Preparation of polypyrrole composite with acrylic acid-grafted tetrafluoroethylene–hexafluoropropylene (Teflon-FEP) copolymer’, Syn Met, 65(1), 35–38 doi:10.1016/0379-6779(94)90290-9 Bhowmick A K, Vijayabaskar V (2005), ‘Electron beam curing of elastomers’, in Industrial Radiation Processing, IANCAS Bulletin IV, no. 2, 81–90 Bradley R (1984), Radiation Technology Handbook, Marcel Dekker, New York Chapiro A (1962), Radiation Chemistry of Polymeric Systems, Wiley, London Charlesby A (ed.) (1960), Atomic Radiation and Polymers, Pergamon, New York Charlesby A (1977), ‘Use of high energy radiation for crosslinking and degradation’, Radiat Phys Chem, 9(1/3), 17–29 doi:10.1016/0146-5724(77)90070-X Charlesby A (1991), in Clegg D W, Collyer A A (eds), Irradiation Effects on polymers, Elsevier, New York Charlesby A, Lawler J P (1980), ‘Grafting of acrylic acid on to polyethylene using radiation as initiator’, Radiat Phys Chem, 15(5/6), 595–602 doi:10.1016/01465724(80)90200-9 Chen C, Anderson D P (2007), ‘E-beam-cured layered-silicate and spherical silica epoxy nanocomposites’, J Appl Polym Sci, 106(3/6), 2132–2139 doi:10.1002/app.26915 Chmielewski A G (2006), ‘World-wide developments in the field of radiation processing of materials in the dawn of 21st century’, Proceedings, Nukleonika, 51 (Supplement 1), 53–59 Chmielewski A G, Haji-Saeid M, Shamshad A (2005), ‘Progress in radiation processing of polymers’, NIMB, 236(1/4), 44–54 doi:10.1016/j.nimb.2005.03.247 Choi J-H, Jung C-H, Kim D-K, Ganesan R (2008), ‘Radiation-induced grafting of inorganic particles onto polymer backbone: A new method to design polymer-based nanocomposite’, NIMB, 266(1/6), 203–206 doi:10.1016/j.nimb.2007.11.012 Cleland M R, Parks L A, Cheng S (2003), ‘Applications of radiation processing of materials’, NIMB, 208, 66–73 doi:10.1016/S0168-583X(03)00655-4 Clough R L (2001), ‘High energy radiation and polymers: A review of commercial processes and emerging applications’, NIMB, 185(1/4), 8–33 doi:10.1016/S0168583X(01)00966-1 Clough R L, Gillen K T (1990), in Klemchuk P P, Pospisil J (eds), Oxidation Inhibition in Organic Materials, Vol II, CRC Press, Boca Raton, FL, p. 191 Clough R L, Shalaby S W (eds) (1991), Radiation Effects on Polymers, American Chemical Society, Washington, DC Clough R L, Shalaby S W (eds) (1996), Irradiation of Polymers: Fundamentals and Technological Applications, American Chemical Society, Washington, DC Cook P (1990), ‘Impact and benefit of radiation technology’, Int J Radiat Appl Instr, Part C, 35(1/3), 7–8 doi:10.1016/1359-0197(90)90047-L Cook P, Halperin R (1998), ‘Arthur Charlesby – His impact on creating a new industry’, Radiat Phys Chem, 51(1/6), 7–8 doi:10.1016/S0969-806X(97)00251-X Czvikovszky T (1992), in Singh A, Silverman J (eds), Radiation Processing of Polymers, Oxford University Press, New York Czvikovszky T (2003), ‘Expected and unexpected achievements and trends in radiation processing of polymers’, Radiat Phys Chem, 67(3/4), 437–440 doi:10.1016/S0969806X(03)00081-1 Czvikovszky T, Hargitai H (1999), ‘Compatibilisation of recycled polymers through

434


radiation treatment’, Radiat Phys Chem, 55(5/6), 727–730 doi:10.1016/S0969806X(99)00220-0 Drobny J G (2005), ‘Electron beam processing of elastomers’, Rubber World, 1 July Dworjanyn P, Garnett J L (1992), in Singh A, Silverman J (eds), Radiation Processing of Polymers, Carl Hanser Verlag, Munich Farmer J D, Janke C J, Lopata V J (1998), 43rd Int SAMPE Symp, 43, 1639–1646 Fei B, Wach R, Mitomo H, Yoshii F, Kume T (2000), ‘Hydrogel of biodegradable derivatives 1. Radiation induced crosslinking of CMC’, J Appl Polym Sci, 78(2/14), 278–283 doi:10.1002/1097-4628(20001010)78:23.0.CO;2-9 Ferain E, Legras R (2003), ‘Track-etch templates designed for micro- and nanofabrication’, NIMB, 208, 115–122 doi:10.1016/S0168-583X(03)00637-2 Fink D (ed.) (2004), Transport Processes in Ion Irradiated Polymers and Applications, Springer Series in Materials Science, Vol 65, Springer-Verlag, Berlin, Heidelberg New York Fink D, Chadderton L T (2005), ‘Ion-solid interaction: status and perspectives’, Appl Phys Instr, Braz J Phys 35, 3b doi: 10.1590/S0103-97332005000500003 Fink D, Klett R, Mathis C, Vacik J, Hnatowicz V (1995), ‘Doping of nuclear tracks with fullerene solution’, Radiat Meas 25 (1/4), 85–88 doi:10.1016/1350-4487(95)00040-L Fink D, Petrov A V, Rao V, Wilhelm M, Demyanov S, Szimkowiak P, Behar M, Alegaonkar P S, Chadderton L T (2003), ‘Production parameters for the formation of metallic nanotubules in etched tracks’, Radiat Meas, 36(1/6), 751–755 doi:10.1016/S13504487(03)00240-3 Fink D, Alegaonkar P S, Petrov A V, Wilhelm M, Szimkowiak P, Behar M, Sinha D, Fahrner W R, Hoppe K, Chadderton L T (2005), ‘High energy ion beam irradiation of polymers for electronic applications’, NIMB, 236(1/4), 11–20 doi:10.1016/ j.nimb.2005.03.243 Friedrich K, Zhang Z, Fakirov S (2005), Polymer Composites – from Nano-to MacroScale, Springer-Verlag, Berlin, Heideberg, New York Garnett J L (1979), ‘Grafting’, Radiat Phys Chem, 14(1/2), 79–99 doi:10.1016/01465724(79)90014-1 Goodman D L, Palmese G R (2003), ‘Curing and bonding of composites using electron beam processing’, in Vasile C, Kulshreshtha A K (eds), Handbook of Polymer Blends and Composites, Vol 1, Rapra Technology, Shrewsbury, UK Hall T J (1999), ‘Electrical devices having improved PTC polymeric composition’, US Patent 5,880,668 Holmberg S, Lehtinen T, Näsman J, Ostrovskii D, Paronen M, Serimaa R, Sundholm F, Sundholm G, Torell L, Torkkeli M (1996), ‘Stucture and properties of sulfonated [(poly vinylidene fluoride)-g-styrene] porous membranes’, J Mat Chem, 6, 1309– 1317 doi: 10.1039/JM9960601309 Huang S-J, Lee J-K, Ha C-S (2004), ‘Polymeric positive temperature coefficient materials: dynamic curing effect’, Colloid Polym Sci, 282, 575–582 doi:10.1007/s00396-0030980-4 Hunt J D, Alliger G (1979), ‘Rubber – application of radiation to tire manufacture’, Radiat Phys Chem, 14(1/2), 39–53 doi:10.1016/0146-5724(79)90010-4 Ivanov V S (1992), Radiation Chemistry of Polymers, VSP Publishers, Utrecht Kanjilal D (2001), ‘Swift heavy ion-induced modification and track formation in materials’, Current Science, 80(12/12), 1560–1566 Lappan U, Geissler U, Lunkwitz K (1999), ‘Modification of polytetrafluoroethylene by electron beam irradiation in various atmospheres’, NIMB, 151(1/4), 222–226 doi:10.1016/ S0168-583X(99)00115-9


435

Lappan U, Geissler U, Lunkwitz K (2000), ‘Changes in the chemical structure of polytetrafluoroethylene induced by electron beam irradiation in the molten state’, Radiat Phys Chem, 59(3/6), 317–322 doi:10.1016/S0969-806X(00)00269-3 Lee G J, Han M G, Chung S C, Suh K D, Im E S (2002), ‘Effect of crosslinking on the PTC stability of carbon black filled HDPE/ethylene–ethylacrylate copolymer blend system’, Polym Eng Sci, 42, 81740–81747 Lunkwitz K, Brink H J, Handte D, Ferse A (1989), ‘The radiation degradation of polytetrafluoroethylene resulting in low molecular and functionalized perfluorinated compounds’, Int J Radiat Appl Instr, Part C, 33(6/6), 523–532 doi:10.1016/13590197(89)90309-3 Makkuchi K (1999), in Role of Radiation Processing in Materials Science Applications, KACST, Riyadh, Saudi Arabia McHerron D C, Wilkes G L (1991), in Clough R L, Shalaby S W (eds), Radiation Effects on Polymers, American Chemical Society, Washington, DC Menezes T J (1990), ‘Developments in electron beam curing’, Int J Radiat Appl Instr, Part C, 35(1/3), 52–58 doi:10.1016/1359-0197(90)90056-N Mishra J K, Chang Y W, Lee B C, Ryu S H, (2008), ‘Mechanical properties and heat shrinkability of electron beam crosslinked polyethylene–octene copolymer’, Radiat Phys Chem, 77(5/6), 675–679 doi:10.1016/j.radphyschem.2007.12.004 Morshedian J, Mirzataheri M, Bagheri R, Moghadam M, (2005), ‘Solventless surface modification of LDPE through electron beam radiation grafting of P-hydroxy ethyl methacrylate’, Iranian Polymer Journal, 14(2), 139–145 Nablo S V, Tripp E P (1977), ‘Low energy electron process applications’, Radiat Phys Chem, 9(1/3), 325–352 doi:10.1016/0168-583X(85)90759-1 Neuberg N W, Poszmik G, Sui M (1999), ‘Method of providing friable polytetrafluoroethylene products’, US Patent 5,891,573 Nho Y C, Chen J, Jin J H (1999), ‘Grafting polymerization of styrene onto preirradiated polypropylene fabric’, Radiat Phys Chem, 54(3/6), 317–322 doi:10.1016/S0969806X(98)00277-1 Ota S (1981), ‘Current status of irradiated heat-shrinkable tubing in Japan’, Radiat Phys Chem, 18(1/2), 81–87 doi:10.1016/0146-5724(81)90066-2 Pethrick R A (1991), in Clegg D W, Collyer A A (eds), Irradiation Effects on Polymers, Elsevier, New York Piraux L, Dubois S, Duvail J L, Radulescu A, Demoustier-Champagne S, Ferain E, Legras R (1999), ‘Fabrication and properties of organic and metal nanocylinders in nanoporous membranes’, J Mater Res, 14, 3042–3050 doi: 10.1557/JMR.1999.0408 Rao V, Amar J V, Avasthi D K, Narayana C R (2003), ‘Etched ion track polymer membranes for sustained drug delivery’, Radiat Meas, 36(1/6), 585–589 doi:10.1016/S13504487(03)00206-3 Razzak M T, Zainuddin, Erizal, Dewi S P, Lely H, Taty E, Sukirno (1999), ‘The characterization of dressing component materials and radiation formation of PVA– PVP hydrogel’, Radiat Phys Chem, 55(2/6), 153–165 doi:10.1016/S0969806X(98)00320-X Reber N, Omichi H, Spohr R, Tamada M, Wolf A, Yoshida M (1995), ‘Thermal switching of grafted single ion tracks’, NIMB, 105(1/4), 275–277 doi:10.1016/0168583X(95)00578-1 Rosiak J M (1991), in Clough R L, Shalaby S W (eds), Radiation Effects on Polymers, American Chemical Society, Washington, DC Rosiak J M, Olejniczak J (1993), ‘Medical applications of radiation formed hydrogels’, Radiat Phys Chem, 42(4/6), 903–906 doi:10.1016/0969-806X(93)90398-E

436


Rouif S (2005), ‘Radiation cross-linked polymers: Recent developments and new applications’, NIMB, 236(1/4), 68–72 doi:10.1016/j.nimb.2005.03.252 Sarma K S S (2005), ‘Electron beam technology in industrial radiation processing’, in Industrial Radiation Processing, IANCAS Bulletin IV, no. 2, 128–134 Saunders C B, Lopata V J, Kremers W, Chung M, Singh A, Kerluke D R (1995), ‘Electron curing of fibre reinforced composites: An industrial application for high energy accelerators’, Radiat Phys Chem, 46(4/6), 991–994 doi:10.1016/0969-806X(95) 00307-J Saunders C B, Lopata V J, Kremers W (1997), ‘Electron curing of composite structures for space applications’ at Electron Beam Curing of Composites Workshop, Oak Ridge, TN Saunders C B, Lopata V, Barnard J, Stepanik T (2000), ‘Electron beam curing – taking good ideas to the manufacturing floor’, Radiat Phys Chem, 57(3/6), 441–445 doi:10.1016/ S0969-806X(99)00411-9 Seguchi T (2001), ‘Mechanisms and kinetics of hydrogen yield from polymers by irradiation’, NIMB, 185, 43–49 Seguchi T, Kudoh H, Sugimoto M, Hama Y (1999), ‘Ion beam irradiation effect on polymers. LET dependence on the chemical reactions and change of mechanical properties’, NIMB, 151(1/4), 154–160 doi:10.1016/S0168-583X(99)00132-9 Shtanko N I, Kabanov V Y, Apel P Y, Yoshida M (1999), ‘The use of radiation-induced graft polymerization for modification of polymer track membranes’, NIMB, 151(1/4), 416–422 doi:10.1016/S0168-583X(99)00108-1 Silverman J (1977), ‘Basic concepts of radiation processing’, Radiat Phys Chem, 9(1/3), 1–15 doi:10.1016/0168-583X(85)90768-2 Silverman J (1981), ‘Radiation processing: the industrial applications of radiation chemistry’, J Chem Educ, 58(2/12), 168–173 Singh A, Silverman J (eds) (1992), Radiation Processing of Polymers, Carl Hanser Verlag, Munich Spenadel L (1979), ‘Radiation crosslinking of polymer blends’, Radiat Phys Chem, 14(3/ 6), 683–697 doi:10.1016/0146-5724(79)90104-3 Starosta W, Buczkowski M, Sartowska B, Wawszczak B (2006), ‘Studies on templatesynthesized polypyrrole nanostructures’, Nukleonika, 51 (Supplement 1), 35–S39 Svarfvar B (1997), ‘Electron beam radiation graft modification of preformed polymer architecture’, in Al-Malalika S (ed.), Reactive Modifiers for Polymers, Springer, New York Tabata Y, Ikeda S, Oshima A (2001), ‘Radiation-induced crosslinking and grafting of polytetrafluoroethylene’, NIMB, 185(1/4), 169–174 doi:10.1016/S0168-583X(01) 00763-7 Tervoort T, Visjager J, Graf B, Smith P (2000), ‘Melt processable polytetrafluoroethylene’, Macromolecules, 33, 6440 Tikku V K (2005), ‘Radiation processing in polymer/rubber’, in Industrial Radiation Processing, IANCAS Bulletin IV, no. 2, 91–97 Toimil Molares M E, Brötz J, Buschmann V, Dobrev D, Neumann R, Scholz I U, Schuchert R, Trautmann C, Vetter J (2001), ‘Etched heavy ion tracks in polycarbonate as template for copper nanowires’, NIMB, 185(1/4), 192–197 doi:10.1016/S0168-583X(01) 00755-8 Van Gisbergen J G M, Meyer H E H (1991), ‘Influence of electron beam irradiation on the microrheology of incompatible polymer blends: Thread break-up and coalescence’, J Rheol, 35(1/8), 63–87 doi:10.1122/1.550209


437

Van Gisbergen J G M, Overbergh N (1992), in Singh A, Silverman J (eds), Radiation Processing of Polymers, Oxford University Press, New York Van Gisbergen J G M, Meyer H E H, Lemstra P J (1989), ‘Structured polymer blends: 2. Processing of polypropylene/EDPM blends: controlled rheology and morphology fixation via electron beam irradiation’, Polymer, 30(12/12), 2153–2157 doi:10.1016/00323861(89)90241-3 Woo L, Sandford C L (2002), ‘Comparison of electron beam radiation with gamma processing for medical packaging materials’, Radiat Phys Chem, 63, 845–850 doi:10.1016/S0969-806X(01)00664-8 Woods R J, Pikaev A K (1994), Applied Radiation Chemistry: Radiation Processing, Wiley, New York Xu X, Ge X, Ye Q, Zhang Z, Zuo J, Niu A, Zhang M (1999), ‘Growth of polymer nano particles in microemulsion polymerization initiated with gamma rays’, Radiat Phys Chem, 54(3/6), 279–283 doi:10.1016/S0969-806X(98)00254-0 Ye Q, He W, Ge X, Jia H, Liu H, Zhang Z (2002), ‘Formation of monodisperse polyacrylamide particles by radiation induced dispersion polymerization: I. Synthesis and polymerization kinetics’, J Appl Polym Sci, 86(10/14), 2567–2573 doi: 10.1002/ app.11170 Zhang H, Xu J (1993), ‘The modification of the flexibility of radiation crosslinked PE by blending PE with EVA and CPE’, Radiat Phys Chem, 42(1/3), 117–119 doi:10.1016/ 0969-806X(93)90217-I

15 Novel processing additives for extrusion and injection of polymers O L K U L I K O V and K H O R N U N G, Universität der Bundeswehr München, Germany and M H W A G N E R TU Berlin, Germany

Abstract: This chapter presents experimental results and tentative explanations for the origin of sharkskin and of slip in extrusion of polyolefins with narrow molecular weight distribution as well as plausible mechanisms of sharkskin suppression by the use of viscoelastic polymer processing additives (PPA). Various types of conventional PPAs are reviewed briefly and then novel PPAs are proposed. The novel PPAs are composed of polyethylene glycol (PEG) with molecular weights from 1000 to 10 000 Da, esters of oxoacids of boron or phosphorus and optionally a thickening agent. In contrast to conventional PPAs made from fluorinated polymers and siloxanes, the novel PPAs are hydrophilic and they can be made from cheap and FDA-compliant components. Key words: polyolefins, sharkskin melt fracture, polymer processing additive; polyethylene glycol (PEG), viscoelastic fluid.

15.1

Introduction

The most important processing operations for thermoplastic polymers are extrusion and injection moulding. Processing of neat polymers by extrusion and moulding is rarely possible. Instead, it is common practice to ‘formulate’ compositions containing polymer and a variety of ingredients in relatively small, but critical amounts. These ingredients may be categorized into two classes, namely polymer modifiers and polymer processing additives. In technical literature the abbreviation PPA is often used for polymer processing additive. Polymer modifiers are chemical substances added to the main polymer in order to change its chemical and physical properties. Illustrative examples of polymer modifiers are antiblocking agents, anti-fog agents, antimicrobials, antioxidants, antistatic agents, blowing agents, coupling agents, colourants, dispersion aids, heat stabilizers, light stabilizers, flame retardants, plasticizers, impact modifiers, nucleating agents, reinforcing and non-reinforcing fillers, etc. On the other hand, polymer processing additives facilitate the processing of polymers. Foremost among these additives are lubricants to reduce extrusion pressure and eliminate extrusion defects, agents to eliminate gel-streaking and pinstriping, and release agents, which prevent sticking of the thermoplastic 438

Novel processing additives for extrusion and injection

439

polymer to machine surfaces such as extruder screws, extrusion dies, rolls, injection moulds, and the like. Some of the polymer modifiers simultaneously are polymer processing additives, e.g. zinc stearate, which is a heat stabilizer as well as a lubricant: for details see Lutz and Grossman (2000), Zweifel (2000) and Flick (2001). Polyolefins represent roughly 60% of all the thermoplastic polymers produced and sold in the world. For example, in 2005 global demand for major plastics was represented by 23% for polypropylene (PP), 17% for high density polyethylene (HDPE), 11% for low density PE (LDPE), and 11% for linear low density PE (LLDPE). The use of polyolefins continues to increase due to competitive costs and low environmental impact in comparison with other polymers. Global PE demand is projected to increase from 54 million tonnes in 2002 to 87 million tonnes in 2010. The present revolution in plastics production is the development of metallocene catalysts. Polyolefins with narrow molecular weight distribution made by use of metallocene catalysts are tougher, stronger and cleaner than plastics made with conventional catalysts. Products manufactured with metallocene PE offer downgauging opportunities to reduce costs and environmental pollution. Analysts forecast growth of about 30% per year for metallocene-PE resins, from 1 million tonnes in 2000 to 17 million tonnes in 2010. It is a major problem for the industry that the resins with narrow molecular weight distribution cannot be processed by extrusion without polymer processing additives.

15.2

Flow instabilities, sharkskin, slip

In processing of highly viscous melts of polyolefins, and in particular of mLLDPE with narrow molecular weight distribution, two major problems appear. Firstly, the molten polymer near the die wall degrades because of long exposure to high temperatures. Secondly, smooth extrudates can only be obtained up to a certain output rate. Beyond that, surface irregularities begin to appear. The irregularities such as ‘sharkskin’, ‘stick-slip’, and ‘gross melt fracture’, collectively known in the art as ‘melt fracture’, limit production rates in commercial applications. To suppress sharkskin instability and to achieve higher output rates, a polymer processing additive is typically added to the thermoplastic resin prior to extrusion. ‘Melt fracture’ can be illustrated by pictures of the extrudate appearance and a characteristic curve of extrusion pressure versus extrusion rate. The ‘flow curve’, which is a plot of apparent shear stress versus apparent shear rate, is commonly used for rheological characterization of molten polymers. However, we discuss here the origin of flow instability phenomena, which may include friction and fracture. The use of reduced quantities like apparent shear stress and apparent shear rate in the characterization of friction and fracture would be confusing. Therefore we present characteristic curves of

440


the directly measured quantities, i.e. the pressure (P) at the die entrance versus the averaged extrudate velocity (V), to characterize the flow in a circular die. The average extrudate velocity (V) is derived from the volumetric flow rate Q by V=

4Q πd 2

15.1

where d is the diameter of the die. Below we will use also a term ‘extrusion rate’ instead of V. If needed, the apparent shear stress τa can be calculated from the reported pressure P by

τa = 4L d

15.2

where L is the length of the die, and the apparent shear rate γ˙ a is given by 32 Q 8V γ˙ a = = d πd 3

15.3

We made extrusion with dies from metal and a transparent die from glass with a ram extruder. In Fig. 15.1, the characteristic flow curve is presented 200

200 Reference, 5.2 mm/s Rubber E41, 208 mm/s 3 mm orifice

Pressure, bar

100

100

80

80

60

60

40

40

20

20

10

10 2

10 100 Averaged extrudate velocity, mm/s

1000

15.1 Characteristic curves for LLDPE ‘LL1201 XV’ at 165°C, i.e. plots of extrusion pressure vs. averaged extrudate velocity for a metal die 3 × 16 mm (diameter × length) without coating (Reference, solid line), for the same die coated by silicon rubber E41 (dotted line), and for a sharp diaphragm (dashed line) of 3 mm diameter. Critical extrusion velocities for the onset of melt fracture are marked by crosses in circles, and are given in the legend (upper left corner).


441

for extrusion of a neat LLDPE (LL1201 XV from ExxonMobile Chemicals) through a clean metal die 3 × 16 mm (diameter × length) by a solid line. At low average product velocity, i.e. low extrusion rate, the extrudate has a glossy smooth surface as presented in Fig. 15.2a. Surface irregularity, i.e. sharkskin, appears at higher rate of extrusion, and its onset is marked by a cross at the characteristic curve in Fig. 15.1. From our experimental data, we can distinguish two stages of sharkskin instability. The first one has a ‘wavy’ appearance: see Fig. 15.2b as well as Fig. 15.3a at larger magnification and using a glass die 6 × 32 mm. Dhori et al. (1997) explained this stage in terms of ‘common line motion’ and ‘dewetting–rewetting’. The second one is an ‘eruptive’ stage, which manifests itself in narrow ridges with sharp slopes and wide valleys between the ridges. The appearance of ‘eruptive’ sharkskin observed by use of the metal die is presented in Fig. 15.2c. With the transparent glass die having a sharp rim, we detect two modes of the eruptive sharkskin. We think that one corresponds to detachment of the melt only from the die face (see Fig. 15.3b), while the other is caused by detachment also from the inner die surface near the die exit (see Fig. 15.3d). The latter shows deeper fracturing of the extrudate. The two modes of sharkskin can alternate as shown in Fig. 15.3c. At higher product velocity an abrupt drop in the pressure gradient occurs (Fig. 15.1), which is attributed to a stick-slip transition at the die wall. It first shows up as a sudden change of the flow, i.e. oscillations of

15.2 Appearances of the extrudate produced with a clean metal die 3 × 16 mm at 165°C: (a) glossy smooth surface; (b) ‘wavy’ appearance of sharkskin; (c) eruptive stage of sharkskin; (d) ‘stick-slip’ transitions and a bamboo-like appearance of the extrudate; (e) ‘super flow’, i.e. slip of the extrudate inside the die; (f) crumbs of polymer at the surface of the extrudate; such crumbs appear during surface cavitation inside the die; (g) volume cavitation of the extrudate inside the die; (h) ‘gross melt fracture’.

442


15.3 Appearances of the extrudate produced with a clean glass die 6 × 32 mm at 165°C (extrusion direction is downwards): (a) onset of ‘wavy’ sharkskin; a white line marks the die exit; (b) ‘eruptive’ sharkskin in the low-velocity mode caused by adhesion failure at the die face; (c) the two sharkskin modes of (b) and (d) alternate; (d) eruptive sharkskin in the low-velocity mode caused by adhesion failure both at the die exit and the die inside. It leads to higher ridges; cracks of cohesion failure inside the die are clearly visible and are marked by arrows; (e) transition from stick to slip; a crack of cohesion failure from the die rim is marked by an arrow; (f) transition from stick to slip; a ridge originates at the die followed by a smooth area which is marked by an arrow; (g) onset of surface cavitation inside the die (voids are marked by an arrow); (h) onset of volume cavitation inside the die (voids marked by an arrow).

the pressure gradient and a bamboo-like appearance of the extrudate: see Figu. 15.2d. The appearance of the extrudate at the moment of transition of the boundary condition from slip to stick is presented in Fig. 15.3e, while the moment of the ‘stick-slip’ transition is presented in Fig. 15.3f. Next follows a stage of continuous slip: see Fig. 15.2e. We observe ‘surface cavitation’ at higher rate of extrusion when the extrudate slips: see Fig. 15.2f and Fig. 15.3g. At the stage of continuous slip, the extrudate may separate from the inside of the die in macroscopic areas and drive outside crumbs of polymer between the extrudate and the die surface. We see such crumbs in Fig. 15.2f at the surface of the extrudate. The following stage corresponds to ‘volume cavitation’: see Fig. 15.2g and Fig. 15.3h. Deep craters are visible at the surface of the extrudate. Finaly, the last stage is ‘gross-melt fracture’ presented in Fig. 15.2h. We can detect a sharp drop in the slope of the characteristic curve at the moment of onset of ‘gross-melt fracture’. The characteristic curve of extrusion by use of a sharp metal diaphragm of diameter 3 mm is shown in Fig. 15.1 by a dashed line for comparison. With the use of the sharp diaphragm the onset of sharkskin instability is at lower extrusion rate in


443

comparison with any tubular die, while the ‘stick-slip’ instability is not observed. We detect the sharp drop in the slope of the characteristic curve at the onset of ‘gross-melt fracture’ for the orifice die at about the same extrusion rate as for the tubular dies. Sharkskin instability in blown film production has been known since the end of the 1950s when LLDPE resins were developed. Numerous suggestions for the origin and mechanism of sharkskin instability have been proposed; see, e.g., the following reviews: Bagley et al. (1958), Dennison (1967), Boudreaux and Cuculo (1977–78), Larson (1992), Leonov and Prokunin (1994), Piau and Agassant (1996), Person and Denn (1997), Graham (1999), Denn (2001), and Hatzikiriakos and Migler (2005). The origin of sharkskin is still debated. One widely accepted explanation of sharkskin proposed by Cogswell (1977) focuses on rapid acceleration and stretching of the extrudate surface after the extrusion die. In accordance with ideas of Hill et al. (1990) based on the apparent relation between adhesive failure and melt fracture, and of Howells and Benbow (1962), we believe that sharkskin instability is caused by swelling of the extrudate at the die exit, flowing of the molten polymer around the die rim, stretching of its surface layer outside the die, and periodic failures in adhesion of the molten polymer at the die rim. A sketch of this mechanism is presented in Fig. 15.4. Due to stretching and extrudate swelling, the surface layer accumulates elastic energy and releases it during the act of adhesion failure. The adhesion failure propagates along the boundary as a crack. At the die rim, the crack deviates from the die wall into the inside of the melt and produces a seed crack, i.e., a notch: see a sketch in Fig. 15.5a. The seed crack grows, ruptures the surface layer and

Die rim Melt sticks Die Velocity profiles

Strain Stress Melt

Extrudate swelling

15.4 Sketch of extrusion of neat polymer through a clean die with sharp edge. Acceleration of the surface layer of the extrudate produces additional strain and thereby additional stress.

444

Advances in polymer processing Solid wall

Solid wall

Polymer

Polymer

(a)

(b)

Solid wall Polymer (c)

Elastic layer

Solid wall Polymer (d)

15.5 Sketch presenting unstable development of a crack in the direction towards the source of elastic energy: (a) the polymer detaches from a clean solid wall; a crack deviates to the source of elastic energy (inside the polymer); (b) the polymer slips along the solid wall; a crack separates the polymer from the solid wall and deviates inside the polymer; (c), (d) a thin elastic layer (which is more elastic than the polymer) coats the solid wall. Cracks deviate in the opposite direction, i.e. towards the inside of the elastic layer, but the polymer stays intact.

creates a valley at the surface of the extrudate. Upstream of the crack, the molten PE decelerates, flows along the die rim, and forms another ridge that will detach from the die in the next act of adhesion failure. As mentioned above, the cracks of adhesion failure can separate the melt from the die face (the first mode of the eruptive sharkskin) or from the die face and the die inside (the second mode of the eruptive sharkskin). More details about the ‘eruptive’ mode of sharkskin are given in Kulikov et al. (2007). Slip of polymer melt along a solid boundary attracts much attention from researchers because of its importance for practical applications: see, e.g., Lupton and Regester (1965), Vinogradov and Ivanova (1968), Vinogradov et al. (1972), Chernyak and Leonov (1986), Kalika and Denn (1987), Leonov (1990), Brochard and de Gennes (1992), Migler et al. (1993), Stewart (1993), Adjari et al. (1994), Piau and El Kissi (1994), Mhetar and Archer (1998), Yarin and Graham (1998), Hill (1998), Black (2000) and Dubbeldam and Molenaar (2003). The idea that plastics can show any sort of flow discontinuity is usually not in the focus of rheology. In fact, even in the determination of viscosity (the principal material parameter) any flow discontinuity is usually ignored. Existing theoretical models propose that during slip some layer of low viscosity material appears at the phase boundary caused by disentanglement of molecules. This low-viscosity region does not lead to true interfacial slip, i.e. a velocity discontinuity. Slip occurs not only along a rigid surface but also between immiscible polymers. An anomalous low viscosity of a blend


445

made of two immiscible polymeric melts is well known, e.g. see Zhao and Macosko (2002). In general, it appears that systems which are more incompatible show a larger viscosity reduction. To account for the anomalous low viscosity of blends, interfacial slip was proposed by Lin (1979). A mechanism is based on polymer chain dynamics, was proposed by Furukawa (1989) and Brochard-Wyart (1990) to explain apparent slip at the polymer– polymer melt interfaces. The theories predict a lower entanglement density near interfaces and thus a region with low viscosity near the interfaces. An alternative point of view may be to consider frictional sliding along an interface between two rapidly deforming solids, which is a basic problem of mechanics that arises in a variety of contexts including sliding of machinery parts, machining of materials (e.g. cutting), failure of the reinforced composites (e.g. fibre pullout) and earthquake dynamics (fault rupture): see, e.g., Persson (2000), Granick et al. (2003) and Coker et al. (2005). The problem of steady sliding under classical Coulomb friction is unstable, and leads to self-healing pulses of slip propagation: see Adams (1998), Caroli (2000) and Brener et al. (2005). It is widely accepted that self-healing pulses of slip occur during earthquakes: see, e.g., Heaton (1990) and Nielsen and Madariaga (2003). Weertman (1980) theoretically concluded that a self-healing pulse of slip propagates along the frictional interface between dissimilar elastic solids. In a numerical study Andrews and Ben-Zion (1997) and Shi and Ben-Zion (2006) examined wrinkle-like propagation of slip pulses on a fault between dissimilar elastic solids. The calculations show that propagation of slip pulses occurs in only one direction (referred to below as the positive direction) which is the direction of slip in the more compliant medium. Displacement is larger in the softer medium, which is displaced away from the fault during the passage of the slip pulse. Lambros and Rosakis (1995a, 1995b) as well as Xia et al. (2005) recently performed sliding experiments along a bimaterial interface for several loading configurations and obtained asymmetric bilateral ruptures. In all cases, the rupture fronts in the positive direction had stable properties. Uenishi and Rossmanith (2003) recently detected by optical means a macroscopic interface separation between similar materials during propagation of a Rayleigh pulse. Polymer melts show rubber-like behaviour when deformed rapidly, and we can expect to find similarities between the slip of polymer melts and the slip of rubbers along a solid boundary. Schallamach (1971) noticed that the lateral force required to drag a glass lens across a rubber slab is reduced when detachment waves are formed. The Schallamach waves detach the rubber from the hard surface, and relative motion between the two surfaces occurs only in the regions where contact has temporarily been lost. The effect is similar to a propagating wrinkle in a carpet. A review of the subject of rubber friction was presented by Roberts (1992). Comninou and Dundurs (1978a, 1978b) as well as Gerde and Marder (2001) demonstrated theoretically

446


that a dynamic wave involving separation could propagate along an interface when two elastic media or an elastic body in contact with a rigid surface are compressed and simultaneously sheared. Separation between the sliding objects is actually very small and can be in the range of a few atomic scales. Gerde and Marder (2001) proposed that separation waves or self-healing cracks, resembling bumps on a rug, run along the interface, causing one solid elastic body to slip over the other. Every crack in its propagation separates one elastic body from the other and results in a small relative displacement. The propagation of many cracks manifests itself in continuous slip. A flux of elastic energy is going to the tip of every crack to break the contact bonds between the bodies. At its tail, the banks of the crack collapse and the bonds recover, at least partially. Low adhesion (weak contact bonding) between the contacting bodies would promote slip. In the referred model of slip, the fault propagates in the direction of bulk motion of an elastic body along the rigid surface. Recent experimental observations by Rubio and Galeano (1994) as well as by Baumberger et al. (2003) on the frictional motion of sheared gels (aqueous gelatin) sliding against a PMMA surface and a glass surface also indicate the existence of inhomogeneous modes of wet sliding at low driving velocity. When a block of gel is driven at low velocity along a plate of glass, self-healing slip pulses originate at the trailing edge of the block and propagate to its leading edge along the interface at a velocity of about 8 mm/s, which is much slower than the (transverse) sound velocity of 2 m/s. Resticking occurs quasi-discontinuously by deceleration of slip. When the block of gel is driven at velocities above some threshold (125 µm/s) resticking no longer occurs, and sliding becomes stationary. Adhesion persists while sliding, which excludes the idea that sliding occurs on top of a pure water layer. Opening of the network/glass contact is also excluded, in contrast to Schallamach waves in rubber or to the ‘brittle’ pulses studied by Gerde and Marder (2001). In model experiments we investigated extrusion of a clay paste inside a transparent die (Kulikov and Hornung, 2002). Extrusion of the paste through a tubular die is unstable to spurt for the flow controlled by pressure. In contrast to extrusion of molten polymer, the spurt of clay paste is caused not by a stick-slip transition but by acceleration of slip due to a macroscopic detachment of the extrudate from the inside of the die. In the development of the spurt instability, first we observe quickly fluctuating voids of surface cavitation near the die exit. The voids grow in number and size as the extrusion rate increases. High optical contrast of the voids indicates that the voids are gas bubbles. Macroscopic detachment of the extrudate happens when the voids merge. At that very moment friction drops as the area of contact between the die and the extrudate is diminished. We also observed surface cavitation during extrusion of molten LLDPE in a glass die: see details in Kulikov and Hornung (2004). From these observations of detachments of the


447

10 mm

extrudate from the inside of the die, we might consider local detachment of material from the solid wall (the voids of detachment) or shear cracks as a plausible mechanism of slip. It is known from the literature (see, e.g., Hatzikiriakos and Migler, 2005), and we also observed in our own experiments, that coating of the die with siloxanes or fluorinated polymers, having a low surface energy, may postpone sharkskin onset, but sharkskin reappears at higher extrusion rates even if the extrudate slips inside the die. We used a siloxane fluid AK 2000 from Wacker Chemie to observe periodic stick-slip transitions that happen at a low rate of extrusion as well as two onsets of sharkskin. One onset of sharkskin corresponds to the no-slip boundary condition and the other to slip of the extrudate. Periodic stick-slip transitions at the low extrusion rate start before the sharkskin onset at the no-slip boundary condition and can be recognized by a sudden drop of extrusion pressure at constant extrusion rate. In Fig. 15.6a the shape of the extrudate is disturbed in the middle of the picture by the stick-slip transition. In Fig. 15.6b sharkskin is visible and originated at the no-slip condition (bottom of the picture), while it is suppressed when the extrudate starts to slip (top of the picture). Figure 15.6c represents the case when sharkskin originates both at the no-slip condition (bottom of the picture) and at the slip condition (top of the picture). In Fig. 15.6d sharkskin of the extrudate is suppressed because of slip at high rate of extrusion. The die exit is a natural source of any cracks, as it is a point where the boundary condition of polymer flow suddenly changes. A tentative explanation of the stick-slip transition at a low rate of extrusion is that it could be caused by propagation of shear cracks from the die exit in the upstream direction. As such shear cracks are unstable to deviation into the inside of the melt, we

8 mm/s

11 mm/s (a)

53 mm/s (b)

120 mm/s (c)

(d)

15.6 Extrudate appearances for a steel die 6 × 32 mm coated with silicone fluid AK 2000: (a) an untypical case of early appearance of stick-slip transition at 8 mm/s without sharkskin; (b) the extrudate slips inside the die without sharkskin; sharkskin is present at the lower part of the extrudate which has been produced some moments ago when there was ‘no-slip’ inside the die; (c) onset of sharkskin at the slip boundary condition (upper part); (d) ‘superflow’ at high extrusion rates.

448


observe the onset of sharkskin also during slip. Sharkskin is caused by deviation of the shear crack into the inside of the melt: see the sketch in Fig. 15.5b. At high rates of extrusion, the stick-slip transition could be caused by propagation of self-healing shear cracks that originate from the die entrance and propagate in the downstream direction. If we follow the model of the self-healing shear crack, there is a compressive stress near the tip of the crack, and the cracks do not deviate into the inside of the elastic body when they propagate downstream. Therefore we see stable extrusion even at conditions of high stress outside the die. The onset of sharkskin at the slip condition at low rates of extrusion as well as suppression of the sharkskin at high rates of extrusion are arguments in support of our tentative explanations of sharkskin and slip given above.

15.3

Conventional processing additives

15.3.1 Fluoropolymers In the early 1960s, DuPont Canada accidentally discovered that fluorinated polymers added in a small amount to LLDPE work as slip agents and polymer processing additives to postpone the onset of the sharkskin instability. Fluorinated polymers are still in use as polymer processing additives, e.g. Viton from DuPont, Dynamar from 3M, Kynar from Atofina, and Tecnoflon from Solvay Solexis: see, e.g., Hatzikiriakos and Migler (2005). Such additives are typically employed at a concentration of about 250 to 3000 parts per million (ppm) based on the weight of the thermoplastic material. The main problem, but not the only one, arising in the commercial use of fluorinated polymer as polymer processing additive is a tendency for plate-out of decomposed fluorinated polymer on the extruder screw and/or the die lips (i.e. die build-up). The problem is often severe, requiring shutdown of the equipment and extensive clean-ups. Fluorinated polymers are extremely hydrophobic materials, and incorporating them into polyolefins makes the polymeric material even more hydrophobic. Hydrophobicity is useful in some applications such as the manufacture of fishing lines and nets, but it is not a desirable feature in the use of polyolefins for packaging. With the use of fluorinated additives in blown film production, white dust accumulates at rollers of the production equipment and between film layers. This dust consists of small particles of the fluorinated polymers, which are separated from the polymeric matrix. Fluorine-containing gas can be generated in recycling of polymer products produced with the use of fluorinated polymer processing additive. Such gas contributes to the destruction of the ozone layer of our planet and is potentially carcinogenic. Fluorinated polymer processing additives have higher affinity to metal in comparison with the matrix polymer, and during the extrusion of polymer/


449

PPA blends they deposit at the boundary and displace the matrix polymer there. Schematically, extrusion with the use of polymer processing additives is shown in Fig. 15.7. An important parameter in the use of PPAs is the conditioning time, i.e. the time for the additives to coat metal walls which are in the contact with molten polymer. Fluorinated polymers used as PPAs normally have a high viscosity at processing temperatures. Kinetics of the conditioning process by conventional fluorinated PPA were investigated by Kharchenko et al. (2003). Viscous fluorinated PPA in a polymer matrix does not actively migrate to the boundary of a channel of constant width but coats the wall only if inclusions of PPA in the polymer are in close proximity to the die wall at the entrance of a narrower channel. Once at the boundary, the droplet of the additive deforms and slowly flows to the die exit as a streak. Coating of the wall is more effective if the inclusions are of larger size, but most of the fluorinated PPA stays trapped inside the matrix polymer and does not deposit at the boundary. The conditioning time can be shortened and the additives can be extracted from the polymer matrix in a more efficient way if the viscosity of the additives is essentially less than the viscosity of the high viscosity component, according to Joseph and Renardy (1993) and Joseph (1997). This lubricating technique, called ‘core-annular flow’, is very interesting from a practical and scientific point of view. In a number of cases it was successfully applied for pipeline transport of very viscous oil. Water (the low-viscosity liquid) forms a lubricant layer between the pipe wall and the high-viscosity core. Unfortunately, lowering of the viscosity of fluorinated PPAs generally results

Coating

Die rim

Melt slips Die Velocity profiles

Strain Stress Melt

Extrudate swelling

15.7 Sketch of extrusion of a PPA/polymer blend through a die with a sharp edge. PPA deposits at the die wall, displaces the polymer and produces a thin layer that separates the polymer from the die. The melt slips inside the die if a lubricating PPA is used.

450


in poorer efficiency of the PPAs. Another important issue in the use of hydrophobic PPAs with low viscosity is PPA migration to the surface of products. This migration lowers adhesion of the surface to inks, labels and paints. Welding of a polymeric film produced with such additives is also problematic. To compare conventional fluorinated PPAs with other PPAs, we produced blends of Viton/LLDPE, Kynar/LLDPE and Dynamar/LLDPE. We used a screw-extruder and a die 2 × 60 mm with a 50º entrance cone that reduces from 8 mm diameter at the entrance to the 2 mm circular die. The die was made from steel grade 34CrAlNi7 (1.8550) and nitrided to harden its surface. These and further extrusions were made at average extrudate velocity of about 40 mm/s and at a temperature of 185ºC. First, we extruded 1 kg of the PPA/LLDPE blend, and then we purged the screw extruder with neat LLDPE. The lubrication chart, i.e. percentage of pressure reduction as a function of time since startup of extrusion, for the use of Viton at a concentration of 1000 ppm is presented in Fig. 15.8. The use of the Dynamar/LLDPE blend gives similar results but slightly higher lubrication in comparison with the use of the Viton/LLDPE blend. Actually, lubrication inside the die is slightly higher than the values presented in Fig. 15.8 and following. The Baymod

Geniomer

PEG8K + 2PPDI 60

60

Lubrication, %

Viton

Conditioning

Purging

1000 ppm

40

40

20

20

0

0 0

100

200 Time, min

300

400

15.8 Lubrication charts showing the conditioning of the extruder with the use of polymer processing additives at a concentration of 1000 ppm, followed by purging of the extruder with neat LLDPE. Average extrudate velocity of about 40 mm/s. PPAs used: Viton from DuPont (solid line); Baymod from LanXess (dashed line); Geniomer from Wacker Chemie (dotted line); polyurethane/urea elastomer based on PEG 8000 (dash-dotted line).


451

pressure drop at the entrance of the die has to be taken into account and has to be subtracted from the pressure measured. In our measurements with a diaphragm of 2 mm diameter having the same cone at the entrance, the pressure drop is 20 bar while the reference pressure drop for the use of the 2 × 60 mm die and neat LLDPE taken at the same extrusion rate is 398 bar. The real lubrication values can be derived from the raw data presented in the figures. In extrusion trials with the conventional PPAs based on fluorinated polymers we detected that they accumulate inside the extruder: see Fig. 15.8 as an example. To clean up the extruder from Viton and Dynamar we had to purge it with a blend of LLDPE and mica. The die was heated to 450ºC in an electrical furnace to burn off all organics. We observed also an accumulation of static electrical charge at the extrudate when Dynamar and Viton were used as PPA.

15.3.2 Siloxanes Processing additives based on silicone (siloxane) fluids have been effectively employed for many years to improve melt flow and release in plastics moulding and extrusion: see, e.g., Hauenstein and Romenesko (1996). Liquid organosilanol polymer processing additive Ucarsil PA is available from Union Carbide Corporation. The use of silicone fluids can be complicated by the difficulty of adding the material on-line and a tendency of fluids with low surface energy to cause screw slip in some equipment designs. High molecular weight silicone fluids as well as fluorinated PPAs are hydrophobic and have low electrical conductivity. Their presence in a polymeric film lowers printability and adhesion of the product to labels as well as causing accumulation of static electrical charge at the product surface. Dow Corning has designed its new series of additives MB50-010 Masterbatch as dry, solid pellets that can be blended with the bulk polymer to work as lubricant and release agent during moulding or extrusion. Based on ultra-high molecular weight siloxane gum (with viscosity of about 15–20 Pa s), these materials form a dispersion within the polymer melt, controlling the mobility of the siloxane and virtually eliminating screw slip. High-viscosity silicone gum in pellet form for use as PPA is also available from Wacker Chemie under the trade name of Genioplast: see Geck et al. (2005). ‘Tospearl 240’ from GE Silicons is a microfine spherical silicone resin powder for plastic films used as an antiblocking agent to impart lubricity and improve extrusion rate. The use of ultra-high molecular weight siloxane gum leads to improvement in melt flow and suppression of melt fracture at relatively high content of the additives, which makes the method prohibitively expensive for use in processing of commodities. Both fluorinated polymers and siloxanes are materials with low surface energy, and so far it is widely believed in the industry that only materials with low surface energy can be used as lubricants. Based on this conception, the

452


development of polymer processing additives has been in stagnation since the 1960s.

15.3.3 Stearates Stearates, such as calcium and zinc stearates, are present in many commercial PE resins at a level of about 1000 ppm as heat stabilizers and lubricants. Increasing the additive level typically results in greater pressure reduction and some delay of sharkskin instability: see, e.g., Achilleos et al. (2002).

15.3.4 Polyalkylene oxides, polyesters and their derivatives The use of high molecular weight polyethylene glycol (MW 10 000 Da, preferably from 25 000 to 50 000 Da) as PPA was proposed by Tikuisis et al. (2005). According to Duchesne and Bryce (1989), the use of lower molecular weight PEG does not provide satisfactory results. Dover Chemical announced recently its new Doverlube FL-599, an ester of polyethylene glycol with a fatty acid, for the use as a polymer processing additive for HIPS, PS, PE, PP, ABS and PVC. It also acts as a clarifier in PP, and it can be incorporated into a purging compound to clean extruders between colour changes and to reduce black specks. It was shown by Hong et al. (1999) that addition of small amounts of hyperbranched dendritic polyesters (HBP) significantly improves fibre extrusion and film-blowing processibility of LLDPE. Sharkskin elimination and substantial lubrication were achieved upon addition of about 0.5% of HBP. Liu and Li (2005) and Chen et al. (2007) describe the use of PEG and PEG/diatomite binary additives in improving the processability of polyethylene. Recently esters of boric acid and low molecular weight PEG were proposed by Sato (2004) for use as release agents for metal moulds. These esters have a ratio of PEG molecules to boron atoms above 3 and they are liquid at room temperature. In our experiments, the use of PEG with molecular weights below 1000 Da cured with boric acid as an additive does not suppress sharkskin in extrusion of LLDPE.

15.3.5 Combination of fluorinated PPAs with polyalkylene oxides The conditioning time would be shortened if a combination of the conventional fluorinated PPAs with polyalkylene oxides is used. The polyalkylene oxides work as solvents for thermally degradated polyolefins, which accumulate at the surface of the processing equipment. The use of polyalkylene oxides helps to displace decomposed polyolefins and decomposed fluorinated processing additives from the inside of the processing equipment. There are


453

many publications and patents relating to the use of a combination of fluorocarbon polymers and polyalkylene oxides in extrusion of polyolefins: see, e.g., Duchesne and Bryce (1989), Chiu et al. (1996), Chapman and Oriani (2003) and Woods (2004). Fluorinated PPAs blended with PEG are available from two companies: 3M and Atofina. We made an extrusion trial with the use of Kynar from Atofina, which is a blend of some fluorinated PPA with PEG, at a concentration of 2000 ppm. No static electrical charge of the extrudate was observed in this experiment. We found also that the use of Kynar results in a higher lubrication in comparison with the use of Dynamar and Viton.

15.3.6 Combination of fluorinated PPAs with powders The efficiency of fluorinated processing additives is higher if they are combined with fine powders of boron nitrite (BN) or delaminated clay: see, e.g., Muliawan et al. (2005) and Hatzikiriakos et al. (2005).

15.3.7 Other polymer blends Sharkskin of the PP-type resins with narrow molecular weight distributions was suppressed by the use of adhesive resins (EVA, maleated PP, maleated PER, styrene/maleic anhydride copolymer, etc.) with good adhesion to metal in amounts from about 3 to 5%: see Fujiyama and Inata (2002). The authors confirmed the idea of Ramamurthy (1986) that improved adhesion would delay sharkskin onset. Wang et al. (2008) recently proposed the use of hyperbranched polyethylene (HBPE) as PPA and demonstrated that at a concentration above 3% by weight, the presence of HBPE in m-LLDPE significantly reduced apparent viscosity and postponed the onset of sharkskin.

15.4

Novel processing additives

As discussed above, fracturing of extrudate may happen not only in extrusion of molten polymers but also in extrusion of pastes: see, e.g., Domanti and Bridgwater (2000). We used a transparent die and investigated extrusion of a clay paste in the case of a flow controlled by pressure, see Kulikov and Hornung (2001). We observed periodic detachment of the extrudate from the inside the die surface and seed cracks that appear inside the die near its exit. The seed cracks grow when the extrudate exits the die and produce structures resembling sharkskin in extrusion of molten polymers. To suppress fracturing in extrusion of the clay paste we proposed using a rigid core that extended beyond the die so that the extrudate slips along the core. Sliding friction of the extrudate on that core generates back-pressure at the die exit. The backpressure keeps the extrudate in contact with the inside wall of the die and

454


heals the seed cracks, so that a smooth extrudate can be produced. Geometrical solutions of the sharkskin problem in extrusion of molten LLDPE have also been proposed by Kurtz (1982). Inspired by the geometrical solutions of the sharkskin problem and observations of detachment of the extrudate from the inside of the die as well as of cavitation voids in extrusion of clay pastes, we tested thick rubber coatings to delay the onset of sharkskin in extrusion of molten LLDPE: see Kulikov and Hornung (2004). The idea was to prevent detachment of the extrudate from the die surface during extrusion and thereby to suppress seed cracks. The use of thick rubber inserts gave very good results and suppressed sharkskin completely, but soon we discovered that a thin (0.1 µm) coating works as efficiently as thick ones do. A tentative explanation for this is that the rubber coating suppresses seed cracks in the extrudate by keeping shear cracks, or the cracks of detachment that separate molten polymer from the die surface, from deviating into the molten polymer. Indeed, the crack separating two elastic bodies is unstable to deviate into the more elastic body. Rubber is more elastic in comparison with the melt, and the cracks do not deviate into the molten polymer, see the sketch in Fig. 15.5c. Therefore, no seed crack is produced and the sharkskin instability is suppressed. Similarly, healing of seed cracks is used in industry to reduce shattering of glass bottles and glass windows. So, we elude the sharkskin problem (fracturing of molten polymer) by eliminating the seed cracks. We discovered also that an elastic coating works as an efficient lubricant. To demonstrate lubrication by a thin elastic coating, we extruded a neat LLDPE through a metal die coated by silicon rubber Elastosil E-41 from Wacker Chemie at a temperature of 165°C. The characteristic flow curve is presented in Fig. 15.1 by a dotted line. We see that the extrusion pressure for a given rate of extrusion is much lower in the case of the coated die. The appearance of the extrudate by the use of the rubber coating is presented in Fig. 15.9, and can be compared with the extrudate appearance in Fig. 15.2. The onset of surface defects is marked by a cross at the flow curve in Fig. 15.1 at an extrusion velocity of 208 mm/s, and the appearance of the extrudate at this extrusion rate is presented in Fig. 15.9f. It is interesting to note that in the case of the rubber coating, the extrusion pressure at extrusion velocities above 210 mm/s is slightly higher than in the case of a clean die. A tentative explanation for this is the instability of shear cracks in the downstream direction along the rubber coating. Indeed, the extrudate from the coated die shows a rougher fracturing in comparison with the extrudate from the clean die at extrusion velocities above 250 mm/s: see Figs 15.9g and Fig. 15.9h. A viscoelastic material can be characterized by its storage modulus G1 and loss modulus G2. Molten polymers are viscoelastic liquids, i.e. they flow under load like a viscous liquid but in experiencing high shear rate they manifest properties of an elastic solid body. To characterize elasticity we use


455

15.9 Extrudate appearances for the case of a 6 × 32 mm die coated with silicon rubber E-41 from Wacker Chemie (extrusion temperature 165°C). Extrusion rates were chosen to be close to the extrusion rates shown in Fig. 15.2. (a)–(e) Extrudate with smooth glossy surface; (f) onset of surface defects; (g) volume cavitation of the extrudate inside the die; (h) ‘gross melt fracture’.

an ‘elasticity factor’ G1/G2 at a defined frequency f. The elasticity factor is the inverse of the ratio G2/G1, which is the loss tangent (tan α). The complex viscosity is defined as the complex Young’s modulus G* divided by 2πf. The use of the ratio G1/G2 instead of tan (α) is more convenient when presenting data for complex viscosity and elasticity in the same plot. The elasticity factor G1/G2 and the complex viscosity of LLDPE LL1201 XV as a function of frequency f are presented in Fig. 15.10 at temperatures of 140 and 200°C. The LLDPE used is elastic at frequencies above 10 Hz (G1/G2 >1) and viscous (G1/G2 < 1) at lower frequencies for the temperature of 165°C. Fracturing of molten LLDPE in the case of the sharkskin instability happens on 165°C with characteristic frequencies above 10 Hz. To suppress the fracture we have to produce an elastic coating on the inside of the extrusion equipment. Such a coating can be made by use of viscoelastic polymer processing additives that are more elastic in comparison with molten LLDPE at characteristic frequencies of the sharkskin instability.

15.4.1 Thermoplastic urethane/urea elastomers Thermoplastic urethane (TPU)/urea elastomers were proposed for use as polymer processing additives by Kulikov (2005). Choosing TPU/urea elastomers as PPAs is a challenging task, as possible combinations of hard and soft segments are numberless. In contrast to fluorinated polymers and

456


2

10

1 0.8

8

G 1 /G 2

Complex viscosity, | G*|/ω, kPa s

20

140°C

6

200°C

4 0.2

0.1

2 0.1

1 Frequency, Hz

10

15.10 Elasticity factor G1/G2 and complex viscosity of LLDPE ‘LL1201 XV’ as a function of frequency at temperatures of 140°C (solid lines) and 200°C (dotted lines).

siloxanes, TPU/urea elastomers are materials with a high surface energy. TPU/urea elastomers are immiscible with PE and are characterized by strong affinity to metals. If TPU/Urea elastomers are used that have low viscosity at processing temperatures, the conditioning time may be shortened in comparison with fluorinated PPAs. The elastomers made with long flexible soft chains and aromatic diisocyanates show high elasticity and good efficiency in lubrication and suppression of sharkskin. With the use of aliphatic diisocyanates and polyesters as long flexible segments of the TPU/urea elastomer, it is possible to produce TPU/urea elastomers that are essentially viscous at frequencies above 10 Hz in comparison with the matrix polymer at processing temperature. When elastomers of this type are used as PPAs, they may substantially delay sharkskin onset according to the proposal of Fujiyama and Inata (2002) to use adhesive PPAs. We believe that viscous adhesives slow down detachment cracks, which separate molten polymer from the coated die. A lubrication chart for the use of the ‘viscous’ elastomer Baymod from LanXess blended with LLDPE at a concentration of 1000 ppm is presented in Fig. 15.8. Baymod is made from aliphatic diisocyanates and high molecular weight polyester as a long segment, and it is essentially viscous (G1/G2 < 0.4) in comparison with molten LLDPE at frequencies above 10 Hz at 165°C. In comparison with conventional PPAs (Viton), suppression of sharkskin with Baymod starts at a lower level of apparent lubrication (< 10%). TPU/urea elastomers made from siloxanes as soft segment are much more elastic in comparison with molten LLDPE at high frequencies. This type of


457

elastomer is available commercially for use as PPAs from Wacker Chemie under the trade name of Geniomer. To compare them with other PPAs, we made extrusion of a Geniomer/LLDPE blend through a clean steel die. A lubrication chart, a chart showing percentage of pressure reduction as a function of time since startup of extrusion, for the use of Geniomer at a concentration of 1000 ppm is shown in Fig. 15.8. The sharkskin suppression starts when lubrication is above 20%. We see that in contrast to conventional fluorinated PPAs, Geniomer and Baymod do not accumulate inside the extruder and can be washed out by purging processing equipment with neat LLDPE. Polyethylene glycols (PEG) resemble silanols in molecular structure. Molecules of PEG are soft, but in contrast to siloxanes they are hydrophilic. Thermoplastic urethane/urea elastomers made from PEGs have been known since the end of the 1950s: see Windemuth et al. (1960). Elastomers made from PEG with molecular weights above 2000 Da are hydrophilic and soluble in water. They show high elasticity and can be used as lubricating PPAs. We produced thermoplastic elastomer from PEG 8000 and PPDI diisocyanate (p-phenylene diisocyanate) from DuPont and used it as PPA. A lubrication chart for the use of the PPDI/PEG elastomer at a concentration of 1000 ppm is shown in Fig. 15.8.

15.4.2 Viscoelastic silanols and glycols cured by borates, phosphates and phosphites Blending a silanol fluid with borates turns it into a viscoelastic substance resembling a rubber or even a solid material for short-time mechanical loading, but it slowly sags under its own weight as a very viscous fluid. Silanols cured by borates are used to produce a famous toy with the trade name ‘Silly Putty’. As opposed to thermoplastic polymers that consist of long linear molecules, Silly Putty has silanol molecules bonded by boron atoms into a three-dimensional network. Covalent bonding with boron is weak and unstable in the presence of water molecules. The bonds spontaneously break and reappear after some time. Therefore the rubber-like three-dimensional network slowly flows under continuous load. The silanols cured by borates can be used as PPAs. Phosphoric acid is used to enhance adhesion of the PPA. To compare a silanol fluid cured by borates with other PPAs, we extruded a blend of cured silanol and LLDPE through a clean steel die. The cured silanol was produced by blending 9 g of a Q1-3563 silanol fluid from Dow Corning, 1 g of phosphoric acid and 0.25 g of borax (Na2B4O7). A lubrication chart for the use of this additive at a concentration of 1000 ppm is shown in Fig. 15.11. Silanols cured by borates show about the same level of lubrication but are much cheaper in comparison with siloxane-based TPUE Geniomer. It is easy to vary the elasticity of the cured silanol fluid by employing different combinations of high and low molecular weight silanols and by varying the

458

Advances in polymer processing Silly Putty PEG1.5K + PA

PEG1.5K + BA PEG1.5K + SO + PA 60

60

Lubrication, %

Conditioning

Purging

1000 ppm

40

40

20

20

0

0 0

100

200 Time, min

300

400

15.11 Lubrication charts showing the conditioning of the extruder with the use of polymer processing additives at a concentration of 1000 ppm, followed by purging of the extruder with neat LLDPE. Average extrudate velocity of about 40 mm/s. PPAs used: silanol cured with boric acid (BA) and phosphoric acid (PA) – ‘Silly Putty’ (solid line); PEG 1500 cured with BA (dashed line); PEG 1500 cured with PA (dotted line); PEG 1500 blended with sorbitol (SO) and cured with PA (dash-dotted line).

amount of borates and phosphoric acid. The impact of elasticity of the lubricant on the sliding friction and the ability to suppress melt fracture was investigated by Kulikov et al. (2007). The ability to delay the onset of sharkskin and to lubricate increases with increasing elasticity of the cured silanols. PEG with molecular weights from 1000 to 10 000 Da readily reacts with boric acid and borates and produces a viscoelastic material with a threedimensional network of covalent bonding similar to Silly Putty. PEG of such molecular weights cured by borates is solid at room temperature and can be used as a polymer processing additive. Crumbs or powder of the PEG cured by borates can be directly mixed with pellets of LLDPE at room temperature at concentrations from 500 to 2000 ppm. In general, at processing temperature the cured PEG is less elastic and much less viscous than Silly Putty but shows higher lubrication. A lubrication chart for the use of PEG 1500 cured by boric acid (10% by weight of the blend) at a concentration of 1000 ppm is shown in Fig. 15.11. It is recommended to dissolve boric acid in PEG under heating to 140°C while the ratio of boron atoms to PEG molecules is from 1 to 5. Lubrication levels (maximum lubrication achieved) are measured and presented in Fig. 15.12 vs. the ratio of boron atoms to PEG molecules for PEG 6000 and PEG 1500 used in the blends. We believe that cured PEG


459

Lubrication, %

1000 ppm 40

40

30

30

PEG6K + BA PEG1.5K + BA

20

20 0.1

1 [Boron]/ PEG]

10

15.12 Lubrication level vs. ratio of boron atoms to PEG molecules. Polymer processing additives were produced by blending PEG 6000 (solid line) or PEG 1500 (dashed line) with boric acid (BA).

actively migrates to the die surface and forms a lubrication layer. The products made from PEG with molecular weights from 1000 to 4000 show a more homogeneous distribution in the polymer matrix than the product made from PEG with molecular weights from 6000 to 10 000. PEG also reacts with phosphoric acid and organophosphates. A blend of PEG with phosphoric acid shows high adhesion to metal and good lubrication, and it suppresses sharkskin in extrusion of LLDPE. A lubrication chart for the use of PEG 1500 blended with phosphoric acid (2.5% by weight) at a concentration of 1000 ppm is shown in Fig. 15.11. Sugar alcohols like xylitol, sorbitol and mannitol can be used to enhance lubrication of the PPA based on PEG. A lubrication chart for the use of PEG 1500 blended with sorbitol (2% by weight) and phosphoric acid (2.5% by weight) at a concentration of 1000 ppm is shown in Fig. 15.11. Lubrication levels (maximum lubrication achieved) are measured and presented in Fig. 15.13 vs. the ratio of phosphorus atoms to PEG molecules for PEG 1500 used in the blends. It is recommended to dissolve phosphoric acid and sugar alcohols in PEG under heating to 160°C while the ratio of phosphorus atoms to PEG molecules is from 0.1 to 0.5. The blend of PEG with phosphoric acid is hydrophilic and in the open air it turns into a sticky paste. The blend of PEG with boric acid also attracts moisture but the powder made from such blend does not lose free-flowing performance. Organophosphites are used in industry as an antioxidant agent as they react with oxygen and reduce certain oxides while the organophosphite is

460


Lubrication, %

1000 ppm 40

40

30

30

20

20

PEG1.5K + PA

10

PEG1K + OP 0.01

0.1 [Phosphorus]/[PEG]

10

1

15.13 Lubrication level vs. ratio of phosphorus atoms to PEG molecules. Polymer processing additives were produced by blending PEG 1500 (solid line) with phosphoric acid (PA) or PEG 1000 (dashed line) with organophosphite (OP).

converted to a phosphate ester. We dissolved diorganophosphite ‘Hostanox PAR 62’ from Clariant (MW 605 Da) in PEG 1000 at a temperature of about 190°C and used it as PPA. Lubrication levels (maximum lubrication achieved) are measured and presented in Fig. 15.13 vs. the ratio of phosphorus atoms to PEG molecules in the blends. At high concentrations of the diorganophosphite the blend turns into a viscous and sticky paste and starts to reduce the rate of extrusion. Therefore it is recommended to keep the ratio of phosphorus atoms to PEG molecules in the range from 0.1 to 0.5 or to add the components to LLDPE separately. Lubrication charts with the use of PEG 6000 with 4% by weight of the organophosphite as PPA at a concentration 1000 ppm are presented in Fig. 15.14 for a short-time extrusion, i.e. 1 kg of PPA/LLDPE extruded, the extruder being purged by neat LLDPE, and for a long-time extrusion. In experiments with the blends of PEG 1000 and PEG 6000 with organophosphite we saw that the blends used as PPA may show a high level of lubrication but after a relatively long induction time. Organophosphates are characterized by higher affinity to metal surfaces in comparison to organophosphites. Therefore our observations can be tentatively explained by slow conversion inside an extruder of organophosphites to organophosphates that accumulate inside the extruder and interact with PEG. Also we detected higher lubrication if the organophosphite/PEG blend was exposed to temperatures above 200°C in the open air. Heating the blend to such high temperatures helps to convert some part of organophosphite to organophosphate. As mentioned above, in our experiments we used LLDPE ‘LL1201 XV’ having an antioxidant agent in its composition. It was observed that neat

Novel processing additives for extrusion and injection PEG6K–F PEG6K + OP (4%)

PEG6K PEG6K + OP (4%)

60

60 Conditioning

Lubrication, %

461

Purging

1000 ppm

50

50

40

40

30

30

20

20

10

10

0

0 0

100

200 Time, min

300

400

15.14 Lubrication charts showing the conditioning of the extruder with the use of polymer processing additives at a concentration of 1000 ppm, followed by purging of the extruder with neat LLDPE. Average extrudate velocity of about 40 mm/s. PPAs used: neat PEG 6000 and a fresh die (solid line); neat PEG 6000 and a die already exposed for 10 hours to extrusion of LLDPE (dashed line); PEG 6000 blended with organophosphite (4%) in short-time extrusion (dotted line) and in long-time extrusion (dash-dotted line).

PEG with molecular weights from 6000 to 10 000 Da blended with LL1201 at a concentration from 250 to 1000 ppm resulted in a high lubrication level but after a relatively long induction interval. We detected also in short-time extrusions of the PEG/LLDPE blend that the maximum lubrication level achieved is higher if the die was exposed to a long-time extrusion of LL1201 in comparison with the case when a fresh die was used. Lubrication charts for the use of neat PEG 6000 at 1000 ppm with a fresh die, i.e. cleaned by heating and water, and a die exposed for 10 hours to extrusion of LL1201 XV are presented in Fig. 15.14. These observations can be tentatively explained by interaction of PEG and the organophosphite that is present in LL1201. We see that extrusion of an industrial grade of LLDPE that comprises organophosphite as an antioxidant can be improved simply by the use of neat PEG with molecular weights from 6000 to 10 000 Da.

15.4.3

Gels for use as PPAs

Bones slide along each other in joints of our skeleton with a friction coefficient as low as 0.002. For comparison, the friction coefficient of smooth steel

462


against ice and snow lubricated by a layer of water is about 0.05. Taking a close look at a human hip-joint we see that bones are coated by elastic cartilages, and the gap between the cartilages is filled by a synovial fluid. The synovial fluid lubricates better than water, showing that the low friction coefficient observed cannot result from a hydrodynamic mechanism of lubrication: see Herman (2007). The synovial fluid is an elastic gel and a high molecular weight (225 kDa) glycoprotein, lubricin, plays an important role there. The lack of lubricin and the change of the synovial fluid into a Newtonian fluid were observed in patients with rheumatoid arthritis, see Sokoloff (1978) and Pieterse (2006). If we take nature as an expert in sliding friction we can accept that the use of elastic gels as lubricants might be a good choice.

15.4.4 Gels based on PEGs for use as PPA Industry widely uses greases for lubrication of ball-bearings and for parts sliding along each other. The greases are blends of a lubricant with a thickening agent. Silica fume, soap or diurethane polymer are examples of thickening agents. Blends of PEG having molecular weights from 1000 to 10 000 Da with a thickening agent can be used as PPA. Blends of PEG with colloidal particles having sizes from 1 to 1000 nm or blends of PEG with high molecular weight polymers that are soluble in PEG may show high elasticity and viscosity if the concentration of the thickening agent is high enough. The elasticity factor G1/G2 and the viscosity of the blends PEG 1000 and PEG 10 000 with silica fume were measured at a temperature of 165°C. Complex viscosities of the blends (measured at a frequency of 0.1 Hz) as a function of the silica content are presented in Fig. 15.15, and the corresponding elasticity factors G1/G2 (measured at a frequency of 2.15 Hz) are presented in Fig. 15.16. We see that the blends turn into high viscosity gel-like materials if the content of silica fume is above 7.5%. Blends of PEG 1000 and silica fume (the latter in proportions of 0.025%, 0.1%, 1.0% and 10.0%) were prepared and used as PPA at a concentration of 2000 ppm for LLDPE. 1 kg of the PPE/LLDPE blends was extruded, lubrication levels were measured and their maximum values are presented in Fig. 15.17. We observe neither lubrication nor suppression of sharkskin when blends of PEG 200 or PEG 600 with silica fume are used instead of the PEGs with molecular weight from 1000 to 10 000. Moreover, these low molecular weight PEGs lubricate the extruder in its feeding zone. As the result of this undesirable lubrication, the extrusion rate drops greatly (at least 30%). In contrast to low molecular weight PEGs, when PPAs based on PEG with molecular weights from 1000 to 10 000 Da are used at a concentration from 1000 to 5000 ppm, the extrusion rate will be a few per cent higher at the same rotation speed of the screw as compared to the rate without additive. Not only silica fume but other mineral powders

Complex viscosity, |G*|/ω , kPa s


463

104

102

100 PEG1K + AS PEG10K + AS 10–2 6

7 8 Silica fume (AS), wt%

9

15.15 Complex viscosity of blends of PEG with silica fume at 165°C versus silica concentration: PEG 1000 (solid line); PEG 10000 (dashed line).

PEG1K + AS PEG10K + AS

G 1 /G 2

10

1

0.1 6

7 8 Silica fume (AS), wt%

9

15.16 Elasticity factor G1/G2 of blends of PEG with silica fume at 165°C versus silica concentration: PEG 1000 (solid line); PEG 10000 (dashed line).

having particles with size below 1000 nm can be used as a thickening agent. For example, if a colloidal powder of titanium oxide from Degussa is used as a thickening agent, the lubrication efficiency is not much different compared with the use of silica fume. Polyethylene oxide (PO) of high molecular weight (MW > 5 MDa) was used as a thickening agent for PEG. Blends of PEG 1000 with PO (0.01%,

464


60

Lubrication, %

2000 ppm

40

40

20 0.01

20 0.1 1 Silica fume, wt%

10

15.17 Lubrication level of polymer processing additives made by blending PEG 1000 with different weight fractions of silica fume as thickening agent.

Lubrication, %

2000 ppm 40

40

30

30

20 0.01

20 0.1 1 Polyethylene oxide, wt%

10

15.18 Lubrication level of polymer processing additives made by blending PEG 1000 with different weight fractions of polyethylene oxide (PO) as thickening agent.

0.065%, 0.1% and 1.0%) at a concentration of 2000 ppm were used as PPAs. Lubrication levels (maximum lubrication achieved) for each blend are measured and presented in Fig. 15.18 vs. concentration of PO. The blend of PEG 1000 with 0.1% of PO shows a much higher level of lubrication in comparison with the neat PEG and the blends of PEG 1000 containing 0.065% or 1.0% of PO by weight. In a similar way, the lubrication chart that corresponds to the blend of PEG 1000 with 1.0% of silica fume in Fig. 15.17 shows a higher level of lubrication in comparison with neat PEG and the blends of PEG


465

1000 having 0.1 and 10% of silica fume by weight. These facts can tentatively be explained by the transfer of the thickening agent together with liquid PPA to the die wall, and by the accumulation of the thickening agent near the surface. The blend is too viscous at high concentration of the thickening agent, and a transfer of the blend to the wall is slow. Therefore, there exists an optimum content of the thickening agent that can be seen in Fig. 15.17 and 15.18. We used the blend of PEG 1000 with 0.1% of polyethylene oxide (PO) as PPA at concentrations of 1000, 2000, 3500, 5000; and 10 000 ppm. The blends show clear threshold behaviour. There is no sharkskin suppression at a concentration of 1000 ppm while we observe good lubrication and suppression of sharkskin at a concentration of 2000 ppm. When the blend is used at a concentration of 1% the extrusion rate drops about 15%, so the blend can be recommended for the use as PPA only at relatively high concentrations from 2000 to 5000 ppm. This is not a problem as PEG 1000 is cheap, and the use of PEGs with molecular weights from 1000 to 6500 Da is known in industry at concentrations from 0.1 to 10 wt% to improve hot sealability and printability of PE films, see Wolinski (1965). Silica fume as well as polyethylene oxide as thickening agents improve the lubrication of the PPAs based on neat PEG with molecular weights from 1000 to 4000 or the esters of this PEG and oxoacids of boron and phosphorus. Lubrication charts for the use of the blends PEG 1500 (at a concentration of 2000 ppm) cured by borates and containing silica fume (6%) or PO (6%) as PPAs are presented in Fig. 15.19. The viscosity of the PEGs increases over-proportionally as their molecular weight grows from 6000 to 10 000 Da and the use of silica fume or PO as a thickening agent is not efficient in this range of molecular weights.

15.4.5 Gels based on blends of PEG with diisocyanates Blending of PEG 10 000 with silica fume gives only a slight improvement in comparison with neat PEG 10 000 for use as PPA. Much greatr improvement can be obtained if some type of diisocyanate is used to thicken PEG 10 000. Blending of polyisocyanates and PEGs can be done at room temperature without a complicated procedure to extract water from PEG that is common in the manufacture of thermoplastic polyurethane/polyurea elastomers. After mixing, the blend can be heated above the melting temperature of PEG and dosed in a liquid form as PPA, or it can be pelletized for gravimetric dosing and blended with LLDPE. Polyisocyanates can be combined with other thickening agents. We blended powders of PPDI diisocyanate and PEG 10 000 at room temperature (0.16% and 0.8%), heated the blend to a temperature of about 120°C and added the product to LLDPE in a concentration of 250 ppm. Lubrication charts for the use of the blends PEG 10 000 containing 0.16% by weight of PPDI (designated PEG10K+ppdi) and 0.8% by weight

466

Advances in polymer processing PEG1.5K

PEG1.5K + BM

PEG1.5K + BM + AS

PEG1.5K + BM + PO

60

60

Lubrication, %

Conditioning

Purging

2000 ppm

50

50

40

40

30

30

20

20

10

10

0 0

100

200 Time, min

300

0 400

15.19 Lubrication charts showing the conditioning of the extruder with the use of polymer processing additives at a concentration of 2000 ppm, followed by purging of the extruder with neat LLDPE. Average extrudate velocity of about 40 mm/s. PPAs used: neat PEG 1500 (solid line); PEG 1500 cured with borates (dashed line); PEG 1500 cured with borates and with addition of 6% of silica fume (AS, dotted line); PEG 1500 cured with borates and with addition of 6% of polyethylene oxide (PO, dash-dotted line).

(designated PEG10K+PPDI) as well as for neat PEG 10 000 (PEG 10K) are presented in Fig. 15.20. We made extrusion of about 5 kg of PPA/LLDPE blend with a PPA concentration of 1000 ppm consisting of PEG 10 000 with 0.16% of diisocyanate PPDI to see the lubrication dynamics (Fig. 15.21). The level of lubrication rises slowly during 2–3 hours of extrusion. We may speculate in a similar way as above that the thickening agent based on isocyanate accumulates at the die wall and creates a layer of an elastic gel that works as a lubricant.

15.4.6 Gels based on blends of PEG with silanols Silanols quickly react with borates when mixed with a solution of borates in PEG. Therefore, one can dissolve borates in PEG and add silanols to the solution. Otherwise we can blend the components at room temperature, and then heat them to the melting temperature of PEG. Silanols and PEG react with borates, and the cured silanol as the minor component is dispersed inside the PEG. At processing temperatures, the dispersion of the cured silanol in PEG works as an efficient viscosity enhancer, and the blend can be

Novel processing additives for extrusion and injection PEG10K

PEG10K + ppdi

PEG10K + PPDI

60

60 Conditioning

Lubrication, %

467

Purging

250 ppm

50

50

40

40

30

30

20

20

10

10

0

0 0

100

200 Time, min

300

400

15.20 Lubrication charts showing the conditioning of the extruder with the use of polymer processing additives at a concentration of 250 ppm, followed by purging of the extruder with neat LLDPE. Average extrudate velocity of about 40 mm/s. PPAs used: neat PEG 10 000 (solid line); PEG 10000 + ppdi cured by 0.16% by weight of diisocyanate (dashed line); PEG 10 000 + PPDI cured by 0.8% by weight of diisocyanate (dotted line). 60

60

Lubrication, %

1000 ppm 50

50

40

40

30

30

20

20

PEG10K + ppdi 10

10

PEG1.5K + Silly Putty PEG1.5K + SO + PA + AS

0

0 0

100

200 Time, min

300

400

15.21 Lubrication charts showing the conditioning of the extruder with the use of polymer processing additives at a concentration of 1000 ppm, followed by purging of the extruder with neat LLDPE. Average extrudate velocity of about 40 mm/s. PPAs used: PEG 10 000 + ppdi cured by 0.16% by weight of diisocyanate (solid line); PEG 1500 cured by boric acid with a silanol fluid (dashed line); PEG 1500 with sorbitol and silica fume cured by phosphoric acid (dotted line).

468


used as a PPA. The lubricating efficiency of the PPA increases as the content of silanol in the blend increases from 1 to 10 wt% of the PEG, but it drops at higher concentrations of silanol. A tentative explanation may be a covalent bonding of PEG and silanols by boron. Due to the bonding, the less viscous PEG migrates to the wall of the die and pulls the more viscous cured silanol with it. With low viscosity PEG as a carrier, a larger amount of the cured silanol can be extracted from the polymer matrix to the die wall in comparison with the use of only cured silanols as PPA. If the concentration of silanol in the blend is too high, the blend has a high viscosity and will not actively migrate to the wall, so the lubricating efficiency decreases. In viscosity measurements of the PEG–silanol blends by use of a rotational rheometer, the cured silanol separates in time from the PEG and coats the metal surfaces with a thin layer of high viscosity (about 2000 Pa s). In processing of the corresponding PPA/LLDPE blend, the cured silanol as the material of higher viscosity will accumulate at the wall, while the PEG as the material of low viscosity will flow along the wall, and both of these components contribute to lubrication. A lubrication chart for the use of blends comprising PEG 1500 (81%), boric acid (9%) and a Q1-3563 silanol fluid from Dow Corning (10%) at a concentration of 1000 ppm as PPA for a long-time extrusion is presented in Fig. 15.21.

15.4.7 Gels based on blends of PEG with other thickening agents A fine dispersion of polymers that are not dissolvable in PEG can also be used as PPA. For example, high molecular weight polymers that are soluble in water can be blended with PEG in water, and then the PPA would be prepared by removing the water. We used a high molecular weight polyacrylamide (MW 11 MDa, content in the blend 6%) and PEG 1500. The blend shows a definite improvement in lubrication in comparison with neat PEG 1500. Following Cody et al. (1999) some stearates may be used as thickening agents for PEG. We used sodium stearate (1%) as a thickening agent for PEG 1500 and also detected an improvement in lubrication. A monoester of polyethylene glycol with a fatty acid is used in industry as a hydrophilic additive (antifog agent): see, e.g., Birnbrich et al. (2004), and as a compatibilizer for polymer–wood composites: see, e.g., Drabeck et al. (2005). We may expect also that such PEG derivatives may be used in PPA composition as thickening agents. Several thickening agents can be combined in the composition of novel PPAs. A lubrication chart for the use of a blend of PEG 1500 with phosphoric acid (12.5%), silica fume (1%) and sorbitol (4%) at a concentration of 1000 ppm as PPA is presented in Fig. 15.21. We see that such a complex blend used as PPA manifests a short induction time and a high lubrication level.


15.5

469

Interaction of polymer processing additives with polymer modifiers

Thermoplastic polymers are often compounded with mineral particles with size above 1000 nm: mineral pigments, antiblock agents (talc or diatomaceous earth) and fillers. Fine particles like delaminated clay or BN may improve the efficiency of fluorinated PPAs, but mineral particles of larger size reduce their efficiency. Mineral fillers absorb polymer processing additives and abrade the layer of PPA from the metal surface. This is a problem for the use of conventional PPAs, as fluorinated polymers are expensive and the use of higher amounts to compensate for the absorption and abrasion by mineral fillers and polymer modifiers is costly. Treating mineral particles by silicone fluids or glycols helps to keep a reasonable amount of the fluorinated polymers in the blend. In contrast to conventional fluorinated PPA, the novel PPAs presented here are cheap and can be used at higher concentrations to compensate absorption of them by mineral particles and abrasion of the lubrication layer. They can be used in amounts from 200 to 10 000 ppm for processing of commodities and polymeric composites by extrusion, especially those composites containing natural fibres and clay. In contrast to PPAs based on fluorinated polymers, the novel PPAs are hydrophilic. Metal soaps like zinc or calcium stearates that are normally present in industrial grades of polyolefins also compete with fluorinated polymer processing additives. These soaps are soluble in PEG and work as thickening agents, and therefore do not limit the use of PPAs based on PEG. Organophosphites (esters of phosphorous acid) are used in industry as antioxidant agents for polyolefins, and organophosphates (esters of phosphoric acid) are in use as flame retardants. Organophosphates and organophosphites are soluble in PEG and under heating the blend undergoes a transformation by transesterification into high molecular weight esters with release of volatile alcohols. We found that a combination of PEG of moderate molecular weight with an organophosphite (an antioxidant agent) can be used as PPA, and the processing of some industrial grades of LLDPE already comprising such an antioxidant can be improved simply by addition of PEG in concentrations from 200 to 2000 ppm: see Section 15.4.2 above. The novel PPAs reduce apparent viscosity and suppress sharkskin in extrusion of LLDPE with narrow molecular weight distribution. The efficiency of the novel PPA is lower for blends of LLDPE/LDPE with a wide distribution of molecular weight, and a higher concentration of the PPA has to be used to suppress sharkskin. PEG slightly dissolves in PE and we may tentatively explain the reduction in the PPA efficiency by its interaction with low molecular weight components of the LDPE/LLDPE blend. Further systematic experimental investigations are required to understand the interactions of the novel PPAs with other polymer modifiers and with various polyolefins.

470

15.6


Future trends in polymer processing additives

It seems that nature produces elastic lubricating gels not by secreting high molecular weight molecules of lubricin and hyalurona (6–10 MDa), but by assembling them in vivo from blocks of relatively low molecular weight (about 7 kDa) with enzymes: see, e.g., Archer et al. (1999). We may expect as a future trend the development of ‘smart’ lubricants that turn into viscoelastic fluids after blending of components inside the processing equipment. Combinations of various thickening agents for PEGs with molecular weights from 1000 to 10 000 can also form an interesting and promising direction of research.

15.7

Conclusions

Nature uses elastic gels as lubricants in joints of bones, and demonstrates excellent results in lowering of sliding friction. It is remarkable that most designs of nature are based on sliding, see Vogel (2000). Human technologies differ very much from natural ones. Owing to the historical development of technology, fluorinated polymers dominate the use as external lubricants and polymer processing additives for polyolefins. The PPAs made from fluorinated polymers are inherently costly, but the main problem in their use is that they are not friendly to the environment. Production and recycling of polymer blends made by use of fluorinated PPAs creates fluorine-containing gases. Such gases contribute to the destruction of the ozone layer of our planet, and they are potentially carcinogenic. Taking from nature the idea of hydrophilic gel-like lubricants, we developed novel polymer processing additives that do not contain fluorine. The novel PPAs are composed of polyethylene glycol (PEG) with molecular weights from 1000 to 10 000 Da, esters of oxoacids of boron or phosphorus, and optionally a thickening agent. The thickening agent enlarges the level of apparent lubrication, especially if PEG with molecular weight from 1000 to 4000 Da is the base of the PPA. It has been shown already that the use of PEG with molecular weight from 1000 to 6500 Da as an additive to PE improves printability and hot welding of the produced PE films. So we may expect to improve also hot welding and printability of LLDPE films made by the use of novel PPAs. The novel PPAs show better lubrication efficiency for LLDPE with narrow molecular weight distribution in comparison with conventional (fluorinated) PPAs. As opposed to fluorinated PPAs, the novel PPAs can be produced by blending FDA-compliant chemicals at room temperature. Together with our experimental results, we presented here tentative explanations for the origin of sharkskin and of slip as well as plausible mechanisms of sharkskin suppression by the use of viscoelastic PPAs.


15.8

471


Information on conventional polymer processing additives is available on the Internet, e.g. at http://www.polymerprocessing.com and http:// www.specialchem4polymers.com. Specific information about these PPAs is available directly from the producers: • • • • • • • • • • •

Dynamar from 3M (http://www.3m.com) Viton from DuPont (http://www.dupontelastomers.com) Kynar from Arkema (http://www.arkema-inc.com) Tecnoflon from Solvay Solexis (http://www.solvaysolexis.com) Ucarsil PA-1 from Union Carbide Corporation (http:// www.unioncarbide.com) Doverlube from Dover Chemical (http://www.doverchem.com/) Geniomer and Genioplast from Wacker Chemie (http://www.wacker.com) Silicon Masterbatch from Dow Corning Silicones (http:// www.dowcorning.com) Tospearl from GE Silicons (http://www.gesilicones.com) Advalube and Paraloid from Rohm and Haas (http://www.rohmhaas.com) Michel XO PA from M Michel and Co. (http://www.mmichel.com).

The following scientific journals publish regular articles about the use of PPAs: Additives for Polymers, Plastics Additives & Compounding from Elsevier, and Journal of Vinyl and Additive Technology from Wiley InterScience.

15.9

Acknowledgements

We are grateful to Ludwig Kassecker for technical support and Tobias Himmel for rheological measurements. We thank the following companies for contributing polymers and chemicals: ExxonMobil Chemical, Dow Corning, Wacker Chemie, Bayer, Huntsman, DuPont, Borealis, Clariant, Degussa, 3M, Atofina, Huhtamaki, LanXess and Lyondell. Financial support by the German Science Foundation (Deutsche Forschungsgemeinschaft) is gratefully acknowledged.

15.10 References Achilleos E C, Georgiou G, Hatzikiriakos S G (2002), ‘Role of processing aids in the extrusion of molten polymers’, J. Vinyl and Additive Techn., 8: 11–24 Adams G G (1998), ‘Steady sliding of two elastic half-spaces with friction reduction due to interface stick-slip’, J. Appl. Mech., 65: 470–475 Adjari A, Brochard-Wyart F, de Gennes P-G, Leibler L, Viovy J-L, Rubinstein M (1994), ‘Slippage of an entangled polymer melt on a grafted surface’, Physica A, 204: 17–39 Andrews D J, Ben-Zion Y (1997), ‘Wrinkle-like slip pulse on a fault between different materials’, J. Geophys. Res., 102: 553–571

472


Archer C W, Caterson B, Benjamin M, Ralphs J R (eds) (1999), Biology of the Synovial Joint, Informa Health Care, New York Bagley E B, Cabot I M, West D C (1958), ‘Discontinuity in flow curve of polyethylene’, J Appl. Phys., 29: 109–110 Baumberger T, Caroli C, Ronsin O (2003), ‘Self-healing slip pulses and the friction of gelatin gels’, Eur. Phys. J., E11: 85–93 Birnbrich R C P, Wild C, adurschel P (2004), Hydrophilic additive, US Pat. Nn. 6,699,922 Black W B (2000), Wall slip and boundary effects in polymer shear flow, Thesis (PhD), University of Wisconsin-Madison. Available at http://grahamgroup.che.wisc.edu/pub/ black_dissert.pdf Boudreaux E Jr, Cuculo J A (1977–78), ‘Polymer flow instability: A review and analysis’, J. Macro Sci., C16(l): 39–77 Brener E A, Malinin S V, Marchenko V I (2005), ‘Fracture and friction: Stick-slip motion’, Eur. Phys. J., E 17: 101–113 Brochard F, de Gennes P G (1992), ‘Shear dependent slippage at a polymer/solid interface’, Langmuir, 8: 3033–3037 Brochard-Wyart F (1990), ‘Slippage at the interface between two slightly incompatible polymers’, C. R. Acad. Sci., Ser. II: Mec., Phys., Chim., Sci., Terr. Univers., 310: 1169–1173 Caroli C (2000), ‘Slip pulses at a sheared frictional viscoelastic/nondeformable interface’, Phys. Rev., E62: 1729–1737 Chapman G R Jr, Oriani S R (2003), Process aid for melt processible polymers, US Pat. Appl. No. 20030236357 A1 Chen J, Liu X, Li H (2007), ‘Improvement in processability of metallocene polyethylene by ultrasound and binary processing aid’, J. Appl. Polym. Sci., 103: 1927–1935 Chernyak Yu B, Leonov A I (1986), ‘On the theory of the adhesive friction of elastomers’, Wear, 108: 105–138 Chiu R, Taylor J W, Cooke D L, Goyal S K, Oswin R E (1996), Melt fracture elimination in film production, US Pat. No. 5,550,193 Cody S L, Hoy M R, Roche E J, Walter E J (1999), Soft gelatin pharmaceutical dosage form, US Pat. No. 5,916,590 Cogswell F N (1977), ‘Stretching flow instabilities at the exits of extrusion dies’, J. NonNewtonian Fluid Mech., 2: 37 Coker D, Lykotrafitis G, Needleman A, Rosakis A J (2005), ‘Frictional sliding modes along an interface between identical elastic plates subject to shear impact loading’, J. Mech. Phys. Solids, 53: 884–922 Comninou M, Dundurs J (1978a), ‘Can solids slide without slipping?’, Int. J. Solids Struct., 14: 251–260 Comninou M, Dundurs J (1978b), ‘Elastic interface waves and sliding between two solids’, J. Appl. Mech., 45:325–330 Denn M M (2001), ‘Extrusion instabilities and wall slip’, Annu. Rev. Fluid Mech., 33: 265–287 Dennison M T (1967), ‘Flow instability in polymer melts: a review’, Plastics Polymers, 35:803–808 Dhori P K, Jeyaseelan R S, Giacomin A J, Slattery J C (1997), ‘Common line motion III: Implications in polymer extrusion’, J. Non-Newtonian Fluid Mech., 71: 231–243 Domanti A T J, Bridgwater J (2000), ‘Surface fracture in axisymmetric paste extrusion. An experimental study’, Trans Ichem, E. V 78, Part A: 68–78 Drabeck G W Jr, Bravo J, DiPierro M, Andrews A C, McKinney J M, Hollo B, Chundury


473

D (2005), Polymer-wood composites and additive systems therefor, US Pat. No. 6,942,829 Dubbeldam J and Molenaar J (2003), ‘Dynamics of the spurt instability in polymer extrusion’, J. Non-Newtonian Fluid Mech., 112, 217–235 Duchesne D J and Bryce V (1989), Extrudable thermoplastic hydrocarbon polymer composition, US Pat. No. 4,855,360 Flick E W (Ed.) (2001), Plastics Additives: An Industrial Guide, Knovel Corporation, N.Y. Fujiyama M and Inata H (2002), ‘Melt Fracture Behavior of Polypropylene-Type Resins with Narrow Molecular Weight Distribution. II. Suppression of sharkskin by addition of adhesive resins’, J. of Appl. Polym. Sci., 84, 2120–2127 Furukawa H (1989), ‘Sliding along the interface of strongly segregated polymer melts’, Phys. Rev. A, 40, 6403–6406 Geck M, Jerschow P, Staiger G, Fuhrmann O (2005), Pelletized organopolysiloxane material, US Pat. Appl. No. 20050004296 A1 Gerde E, Marder M (2001), ‘Friction and fracture’, Nature, 413: 285–288 Graham M D (1999), ‘The sharkskin instability of polymer melt flows’, Chaos, 9(1): 154–163 Granick S, Lee H, Zhu Y (2003), ‘Slippery questions of stick when fluid flows past surfaces’, Nature Materials, 2: 221–227 Hatzikiriakos S, Migler K (2005), Polymer Processing Instabilities: Control and Understanding, Marcel Dekker, New York Hatzikiriakos S G, Rathod N, Muliawan E B (2005), ‘The effect of nanoclays on the processability of polyolefins’, Polym. Eng. Sci., 45: 1098–1107 Hauenstein D E, Romenesko D J (1996), Method of modifying polyolefin with diorganopolysiloxane process aid, European Patent No. EP0722981 Heaton T H (1990), ‘Evidence for and implications of self-healing pulses of slip in earthquake rupture’, Phys. Earth Planet Int., 64: 1–20 Herman I P (ed.) (2007), Physics of the Human Body, Springer, Berlin Hill D A (1998), ‘Wall slip in polymer melts: A pseudo-chemical model’, J. Rheol., 42: 581–601 Hill D A, Hasegava T, Denn M M (1990), ‘On the apparent relation between adhesive failure and melt fracture’, J. Rheol., 34: 891 Hong Y, Cooper-White J J, Mackay M E et al. (1999), ‘A novel processing aid for polymer extrusion: rheology and processing of polyethylene and hyperbranched polymer blends’, J. Rheol., 43:781–793 Howells E R, Benbow J (1962), ‘Flow defects in polymer melts’, Trans. Plastics Inst., 30: 240–253 Joseph D D (1997), ‘Steep wave fronts on extrudates of polymer melts and solutions: lubrication layers and boundary lubrication’, J. Non-Newtonian Fluid Mech., 70(3): 187–203 Joseph D D, Renardy Y Y (1993), Fundamentals of Two-Fluid Dynamics, Part II: Lubricated Transport, Drops and Miscible Liquids, Springer-Verlag, New York Kalika D S, Denn M M (1987), ‘Wall slip and extrudate distortion in linear low-density polyethylene’, J. Rheol., 31: 815–834 Kharchenko S B, McGuiggan P M, Migler K B (2003), ‘Flow induced coating of fluoropolymer additives: Development of frustrated total internal reflection imaging’, J. Rheol., 47: 1523–1545 Kulikov O (2005), ‘Novel processing aids for extrusion of polyethylene’, J. Vinyl and Additive Techn., 11, 127–131

474


Kulikov O, Hornung K (2001), ‘A simple geometrical solution to the surface fracturing problem in extrusion processes’, J. Non-Newtonian Fluid Mech., 98, 107–115 Kulikov O, Hornung K (2002), ‘Wall detachment and high rate surface defects during extrusion of clay’, J. Non-Newtonian Fluid Mech., 107, 133 Kulikov O, Hornung K (2004), ‘A simple way to suppress surface defects in the processing of polyethylene’, J. of Non-Newtonian Fluid Mech., 124: 103 Kulikov O, Wagner M, Hornung K (2007), ‘Silanols cured by borates as lubricants in extrusion of LLDPE. Impact of elasticity of the lubricant on sliding friction’, Rheol. Acta, 46 741–754 Kurtz S J (1982), Methods for reducing melt fracture during extrusion of a molten narrow molecular weight distribution linear ethylene copolymer, US Pat. No. 4,348,349 Lambros J, Rosakis A J (1995a), ‘Development of a dynamic decohesion criterion for subsonic fracture of the interface between two dissimilar materials’, Proc. R. Soc. Lond., A451: 711–736 Lambros J, Rosakis A J (1995b), ‘Shear dominated transonic interfacial crack growth in a bimaterial – I. Experimental observations’, J. Mech. Phys. Solids., 43: 169–188 Larson R G (1992), ‘Instabilities in visco elastic flows’, Rheol. Acta, 31: 213–263 Leonov A I (1990), ‘On the dependence of friction force on sliding velocity in the theory of adhesive friction of elastomers’, Wear, 141: 137–145 Leonov A I, Prokunin A N (1994), Nonlinear Phenomena in Flows of Viscoelastic Polymer Fluids (1st edn). Chapman & Hall, London Lin C C (1979), ‘A mathematical model for viscosity in capillary extrusion of two– component polyblends’, Polym. J. (Tokyo), 11: 185–192 Liu X, Li H (2005), ‘Effect of diatomite/polyethylene glycol binary processing aid on the melt fracture and the rheology of polyethylenes’, Polym. Eng. Sci., 45: 898–903 Lupton J M, Regester J W (1965), ‘Melt flow of polyethylene at high rates’, Polym. Eng. Sci., 5: 235–245 Lutz J T Jr, Grossman R F (2000), Polymer Modifiers and Additives, JL Enterprises, Bensalem, PA Mhetar V, Archer L A (1998), ‘Slip in entangled polymer solutions’, Macromolecules, 31: 6639–6649 Migler K B et al. (1993), ‘Slip transition of a polymer melt under shear-Stress’, Phys. Rev. Lett:, 70: 287–290 Muliawan E B, Rathod N, Hatzikiriakos S G, Sentmanat M (2005), ‘Boron nitride and fluoropolymer combinations: Interactions and their performance as processing aids’, Polym. Eng. Sci., 45: 669–677 Nielsen S, Madariaga R (2003), ‘On the self-healing fracture mode’, Bull. Seismol. Soc. Am., 93: 2375–2388 Person T J, Denn M M (1997), ‘The effect of die materials and pressure-dependent slip on the extrusion of linear low-density polyethylene’, J. Rheol., 41: 249–265 Persson B N J (2000), Sliding Friction: Physical Principles and Applications, Springer, Heidelberg Piau J M, Agassant J F (eds) (1996), Rheology for Polymer Melt Processing, Elsevier, Amsterdam Piau J M, El Kissi N (1994), ‘Measurement and modelling of friction in polymer melts during macroscopic slip at the wall’, J. Non-Newtonian Fluid Mech., 54: 121–142 Pieterse N (2006), Development of a dynamic hip joint simulation model, Thesis (MSc), University of Pretoria. Available from http://upetd.up.ac.za/thesis/available/etd-03152006142935/unrestricted/00dissertation.pdf (Accessed 7 November 2007)


475

Ramamurthy A V (1986), ‘Wall slip in viscous fluids and influence of materials of construction’, J. Rheol., 30: 337–357 Roberts A D (1992), ‘A guide to estimating the friction of rubber’, Rubb. Chem. Technol., 65: 673–686 Rubio M, Galeano J (1994), ‘Propagative slipping modes in a spring-block model’, Phys. Rev., E50: 1000–1004 Sato S (2004), Release agent for metallic mold, US Pat. Appl. No. 2040083925 A1 Schallamach A (1971), ‘How does rubber slide?’, Wear, 17: 301–312 Shi Zh, Ben-Zion Y (2006), ‘Dynamic rupture on a bimaterial interface governed by slipweakening friction’, Geophys. J. Int., 165: 469–484 Sokoloff L (ed.) (1978), The Joints and Synovial Fluid, Vol I, Academic Press, New York Stewart C W (1993), ‘Wall slip in the extrusion of linear polyolefins’, J. Rheol., 37(3): 499–513 Tikuisis T, Arnould G, Bayley J, Chisholm P S, Marshall S (2005), High molecular weight polyethylene glycol as polymer process aids, US Pat. Appl. No. 20050070644 A1 Uenishi K, Rossmanith H P (2003), ‘Optical methods in fault dynamics’, Opt. Laser. Eng., 40: 325–339 Vinogradov G V, Ivanova L I (1968), ‘Wall slippage and elastic turbulence of polymers in the rubbery state’, Rheol. Acta, 7: 243–254 Vinogradov G V, Insarova N I, Boiko B B, Borisenkova E K (1972), ‘Critical regimes of shear in linear polymers’, Polym. Eng. Sci., 12: 323–334 Vogel S (2000), Cats’ Paws and Catapults: Mechanical Worlds of Nature and People, W.W. Norton, New York Wang J, Kontopoulou M, Yea Z, Subramanian R, Zhu S (2008), ‘Chain-topology-controlled hyperbranched polyethylene as effective polymer processing aid (PPA) for extrusion of a metallocene linear-low-density polyethylene (mLLDPE)’, J. Rheol., 52: 243–260 Weertman J J (1980), ‘Unstable slippage across a fault that separates elastic media of different elastic constants’, J. Geophys. Res., 85: 1455–1461 Windemuth E, Schnell H, Bayer O (1960), High molecular weight polyether urethane polymers, US Pat. No. 2,948,691 Wolinski L E (1965), Polyethylene resin containing a solid polyethylene glycol., US Pat. No. 3,222,314 Woods S S (2004), Melt processable thermoplastic polymer composition employing a polymer processing additive containing a fluorothermoplastic copolymer, US Pat. No. 6,734,252 Xia K, Rosakis A J, Kanamori H, Rice J R (2005), ‘Laboratory earthquakes along inhomogeneous faults: directionality and supershear’, Science, 308: 681–684 Yarin A, Graham M D (1998), ‘A model for slip at polymer/solid interfaces’, J. Rheol., 42: 1491–1504 Zhao R, Macosko C W (2002), ‘Slip at polymer–polymer interfaces: Rheological measurements on coextruded multilayers’, J. Rheol., 46: 145–167 Zweifel H (2000), Plastic Additives Handbook (5th edn), Hanser Publishers, Munich

16 Processing of macro, micro and nanocomposites Y- L L U, Y- P W U and L - Q Z H A N G, Beijing University of Chemical Technology, China

Abstract: This chapter surveys some recent researches on two main classes of processing methods for the preparation of rubber-based nanocomposites: (a) directly compounding nanoparticles into the matrix; and (b) in-situ generation of the nanodispersion phase during vulcanization through in-situ radical polymerization of metallic salts of unsaturated carboxylic acid. The characteristics of microstructure and properties of the corresponding rubber nanocomposites related to processing conditions are reviewed. The microstructural evolutions of rubber-based nanocomposites, including rubber filled with carbon black or silica and rubber/layered silica nanocomposites during storage and the curing process, and their influence on properties of the nanocomposites, are also introduced. Key words: melt blending, latex compounding, in-situ generation, rubber, nanocomposites.

16.1

Background of rubber reinforcement

As far as the mechanical properties (tensile strength, stiffness, abrasion resistance, fatigue resistance, etc.) are concerned, most elastomers would be of no practical value if they were not reinforced. For many years, fine nanosized particles of carbon black, and more recently silica, have been incorporated into rubber compounds at about 20–25% volume concentration to give major improvements in the physical properties. As for the outstanding reinforcement of carbon black and silica in comparison with fillers of larger particles, some simple general criteria have been commonly accepted. They emphasize the importance of particle size, structure and surface activity, but the reinforcing effect of carbon black and silica is attributed primarily to their very small particle size, of the order of nanometers rather than micrometers. The concept that the size of the reinforcing particle is the primary factor in rubber reinforcement, and that such materials were nanocomposites, was first introduced around 2000.1,2 Hamed stated: ‘Sufficiently small hard domains give good reinforcement, even when matrix/domain bonding is poor, e.g., with graphitized or fluorinated carbon black. However, if the domain size is greater than about a micron, reinforcement is absent or minimal, regardless of domain shape, and even if bonding between the matrix and domains is quite strong. Small hard domains (of the order of 0.1 µm) are necessary and 479

480


sufficient for substantial reinforcement.’2 Actually, a ‘particle size factor’ implies a ‘surface activity factor’ because when the particle size is less than 100 nm, atoms located at the particle surface account for a large proportion of the total number of atoms in the particle. Generally, excellent reinforcement of filler requires a fine dispersion of nanoparticles, and a certain degree of interfacial interaction between filler and rubber. In this chapter, we will discuss preparation of rubber-based nanocomposites by polymer processing and the forming mechanisms of nanostructure, and review the structure and properties of various rubber-based nanocomposites.

16.2

Melt blending method

16.2.1 Nanoparticles/rubber nanocomposites The commonl employed processing method of direct mechanical mixing of the filler particles into rubber may be termed a melt blending technique. During the mixing process, the filler dispersion process undergoes a number of stages: (a) incorporation of the powder; (b) ‘wetting’ the powder (i.e., the penetration of the rubber into the agglomerate); (c) deagglomeration (i.e., breaking up the agglomerate into aggregate or the smallest constituent element); and (d) randomization of the particles throughout the volume.3 Among these, the wetting step involves the surface energy and compatibility between filler and rubber, and deagglomeration is generally the rate-controlling step. The clusters of particles held together by cohesive forces (agglomerates) are successively broken apart by hydrodynamic stresses imposed on the external surfaces of the deforming matrix. The separating force is proportional to the viscosity of the matrix, and also depends on the shape of the particle and the orientation of agglomerates. Since elastomers have higher viscosity than plastics, the nanoparticles are easy to disperse in the rubber matrix in comparison with mechanical mixing of powders into molten plastics. The degree of filler dispersion in the rubber matrix can be measured by the strain dependence of the elastic modulus, G′, which has been termed the ‘Payne effect’. In practice, due to the difference in the surface energy between filler and rubber, the dispersed filler aggregates in the polymer matrix have a tendency to agglomerate, especially at high loading, leading to chain-like filler structures or clusters. That is, the well-dispersed fillers can flocculate to form more developed agglomerates or a filler network during storage, especially during vulcanization.4,5 This flocculation is mainly determined by its kinetics, which is related to the diffusion constant of the filler aggregates in the polymer matrix and the mean distance between aggregates. The diffusion constant is controlled by temperature, the viscosity of the polymer matrix, and the effective diameter of the aggregates, and the interaggregate distance depends on the


481

filler loading, the structure parameter, i.e., CDBP number, and the surface area of the filler. For the same filler–polymer system, the filler networking can be manipulated by changing the diffusion constant through processing of the compounds. According to the above analysis, the final degree of filler dispersion in the rubber matrix is related to both the initial degree of filler dispersion during the mixing process and filler flocculation during storage and vulcanization. In order to improve the filler dispersion and depress filler networking,6 applied approaches could be divided into thermodynamic and kinetic approaches. •

•

Thermodynamic approaches: (a) increase the filler–polymer interaction and compatibility by filler surface modification and/or by using coupling agents; (b) reduce the difference in surface characteristics, especially in surface energy, between polymer and filler by filler surface modification. Kinetic approaches: (a) improve the initial filler dispersion in the compound; (b) increase the mean surface-to-surface distance between aggregates by changing the filler morphology; (c) increase the bound rubber content to increase the effective aggregate size and the viscosity of the polymer matrix to hinder reagglomeration of dispersed nanoparticles; (d) introduce a small quantity of crosslinks between polymer molecules to increase the effective molecular weight and hence the viscosity of the polymer matrix to impede reagglomeration; (e) reduce unnecessary scorch time for compound processing and increase the cure rate to lock filler aggregates in place before a developed filler network is formed.

In practice, thermodynamic and kinetic factors are affected mutually and two examples are presented in the following. Silica/rubber nanocomposites with the coupling agent modification Due to strong filler–filler interaction and weak filler–rubber interaction, it is difficult to disperse silica in rubber and to attain strong adhesion at the interface relative to carbon black. To improve silica dispersion and reduce agglomeration of silica particles, modification of silica with a bifunctional silane during the mixing operation is introduced,6 which is also called in-situ modification. The most widely used silane coupling agent is bis-(3(triethoxysilyl)-propyl)-tetrasulfide (TESPT).7–12 Relative to the case without TESPT, the dispersion of silica particles was greatly improved after TESPT modification (Fig. 16.1),13 and correspondingly filler networking can be effectively depressed via reduction of the specific component of the silica surface energy with surface modification and increased bound rubber to impede the reagglomeration, which can be demonstrated by the weaker Payne effect (Fig. 16.2). As a result, the compound exhibits significantly improved processability, and the vulcanizate shows better dynamic properties and

482


1 µm

1 µm

(a)

(b)

16.1 TEM photographs of silica/SSBR vulcanizates (a) without TESPT and (b) with TESPT.

2500

2000 50%

G1/kPa

60°C, 1Hz without TESPT

1500

with TESPT

1000

500

0 0.1

1

10

100

ε /%

16.2 Curves of storage modulus vs. strain of silica/SSBR compounds (a) without TESPT and (b) with TESPT.

improved abrasion resistance. For the commercialization of the Michelin ‘green tire‘ in 1992, the key techniques are the application of silica as the principal filler and a bifunctional silane coupling agent in tread compounds. Besides silica, some other inorganic nanoparticles have been developed and applied to prepare rubber nanocomposites by melt blending combined


483

with in-situ modification, such as nano-magnesium hydroxide (Mg(OH)2), nanoscaled calcium carbonate (CaCO3),7–11 etc. For example, nano-Mg(OH)2 endowed the EPDM compounds with both better mechanical properties and fire resistance than micro-Mg(OH)2. Surface modification of nano-Mg(OH)2 particles substantially improves the properties of nanocomposites, but had little effect on fire resistance.14 Functionalized blacks – carbon–silica dual phase filler (CSDPF)/rubber nanocomposites In general, interaction between different surfaces is lower than that between surfaces in the same category. To further improve the dispersion of carbon black and silica in rubber matrix, in response to ever more demanding requirements from the tire industry (the tradeoff between wear resistance, rolling resistance and wet skid resistance), Cabot Corp. developed CSDPF. There are two groups of CSDPFs, namely CRX 2000 and 4000 (Fig. 16.3), classified by their distribution of silica, silica-surface coverage and silicon content. CSDPF 4000 has a higher silica content and much higher silicasurface coverage relative to CSDPF 2000. The silica of CSDPF 2000 is distributed throughout the aggregates, while silica is concentrated at the aggregate surface of CSDPF 4000. When incorporated in hydrocarbon polymers, both CSDPFs are characterized by higher filler–polymer interaction in relation to a physical blend of carbon black and silica, and lower filler– filler interaction compared to both traditional fillers. Evidence of the higher filler–polymer interaction of the silica-containing fillers in hydrocarbon polymers can be obtained by comparing the adsorption energies of polymer

A Section A-A A

CSDPF 2000

CSDPF 4000

16.3 General views of CSDPF 2000 and CSDPF 4000.14

484


analogs on the filler surfaces, measured by IGC and bound rubber contents.15 The higher surface activity on the carbon domains is attributed to the change in surface microstructure of the carbon black phase. Upon doping with a foreign substance in the carbon phase, more surface defects in the graphitic crystal lattice and/or smaller crystallite dimensions are formed, leading to more edges of carbon basal planes, which were verified by scanning tunneling microscopic (STM) investigation.15 The edges of basal planes and crystal defects have been proposed as the active centers for rubber adsorption. The lower filler–filler interaction of CSDPF 2000 and 4000 can be demonstrated by the ‘Payne effect’.16 Although from the chemical composition point of view, both CSDPF 2000 and 4000 are between carbon black and silica, what is actually observed here is that the two new fillers give a smaller Payne effect (Fig. 16.4) than carbon black and silica. After TESPT modification, the Payne effects of both CSDPF 2000 and 4000 further reduce. This unique behavior is readily explained by the low filler–filler interaction due to their hybrid surfaces. From the view of global performance of different tires, the CSDPF 4000 is recommended for passenger tires17 and the CSDPF 2000 for truck tire tread formulations.18

16.2.2 Fibrillar silicate/rubber composites The dominant fibrillar silicate (FS) is attapulgite (AT) or palygorskite, and FS has the structural formula Mg5[Al]Si8O20(HO)2(OH2)4•H2O. FS is composed of many fibrillar nano-single crystals, 100–3000 nm in length and 10–30 nm in diameter, which compactly arrange in parallel to form into crystal bundles, and then these crystal bundles agglomerate into particles having diameters of 5–50 µm (Fig. 16.5(a).19 When purified AT particles are dispersed in water by ultrasonic vibration or strong stir, they are dramatically and rapidly separated into lots of nanofibrils about 30 nm in diameter and less than 1500 nm in length (Fig. 16.5(b)). Figure 16.6(a) and (b) shows the typical microstructure of AT/rubber composites prepared by a traditional mechanical mixing technique, combined with an in-situ modifying method. This clearly illustrates that almost all AT particulates are separated into numerous fibrils of nano-scale diameter in rubber matrices (styrene butadiene rubber (SBR) and nitrile rubber (NBR) are only taken as two examples). The aspect ratio of nanofibrils is roughly estimated to be 5:30. By comparison, Fig. 16.6(c) presents the dispersion morphology of AT/SBR composites prepared by directly blending without the in-situ modification. Obviously, the dispersion of AT in SBR is fairly poor and some large particulates are observed. It is a very important characteristic that the interaction between single crystals in AT is extremely small due to a similar line–line contact. Furthermore, there are a lot of interstices between these geometrically agglomerated nano-

Processing of macro, micro and nanocomposites 12.0

485

OESSBR/BR 70/30, Filler: equal volume 70°C, 10 Hz

G’, MPa

10.0 Silica 8.0

6.0

N234

4.0

CSDPF 4210

2.0

CSDPF 4210/ TESPT 2.7 phr

Strain, %

0.0 0.1 12.0

1 G’, MPa

10

100

NR, Filler: 50 phr 70°C, 10 Hz

10.0

8.0 Silica 6.0 N234 4.0 CSDPF 2125 2.0

CSDPF 2125/ TESPT 1.5 phr

Strain, %

0.0 0.1

1

10

100

16.4 Strain dependence of G ′ of vulcanizates filled with a variety of fillers.14

scale single crystals. These characteristics will facilitate the decohesion (separation) of AT micro-agglomerates into smaller-scale crystal bundles and single crystals in water or by surface chemical modification. The dispersion of AT in rubber matrix may obey the erosion mechanism, i.e., during mechanical mixing, the silane coupling agent first modifies the outer layer of AT dispersion, which weakens the interactions between fibril crystals and disjoints the fibril crystals located in the outer layer. Consequently, fibril crystals are exfoliated from dispersed AT particles from the outer layer to the inner until the AT particles completely disappear.

486


1 µm

(a)

(b)

16.5 Typical microstructures of AT.

Table 16.1 summarizes mechanical properties of various rubber-based short nanofibril silicate nanocomposites made by a traditional mechanical mixing technique, combined with an in-situ modifying method. It strongly reveals that nanofibril silicates can endow various rubbers with excellent performance, such as SBR, NBR, NR, EPDM, CNBR, HNBR and CR. It should be pointed out that the excellent properties of AT/rubber nanocomposites are not only related to the nanodispersion of AT in rubber matrix, but also attributed to the strong interfacial interaction between nanofibril silicate and rubber, which benefits from chemical bonding of the silane coupling agent with those active hydroxyl groups located on the surface of fibril silicate crystals.20 In addition, the processing of composites filled with AT nanofibers was much better than with compounds containing micron-sized short fibers. This nanocomposite appears to be a good replacement for traditional short-fiber/ rubber composites.

16.2.3 Rubber/layered silicate nanocomposites The melt blending method might be the most widely used method for preparing rubber/clay nanocomposites (RCNs) from the viewpoint of the present rubber compounding praxis. The related R&D work has profited from some general rules deduced for thermoplastic-based nanocomposites.21 Some main factors determining the microstructures and final performances of RCNs prepared with melt blending have been disclosed, as described below. The organic modification of clay For successfully preparing RCNs with the melt blending method, the clay must be organically modified to reduce its hydrophilicity and facilitate the


1 µm

487

1 µm (b)

(a)

1 µm

(c)

16.6 TEM photographs of (a) AT/SBR nanocomposites, (b) AT/NBR nanocomposites, and (c) AT/SBR microcomposites by melt blending without in-situ modification.

488

Properties Shore A hardness

SBR 45

AT/SBR NR 73

41

AT/NR 74

NBR 54

AT/NBR CNBR AT/CNBR EPDM AT/ EPDM HNBR AT/HNBR CR 83

61

72

70

78

57

80

51

AT/CR 83

Stress at 100% strain (MPa)

0.8

7.6

1.0

9.1

1.2

12.1

2.3

8.3

1.4

7.0

1.3

28.6

1.1

13.8

Tensile strength (MPa)

2.3

14.7

29.3

24.5

3.2

22.7

9.7

22.7

4.7

10.7

6.1

30.4

6.6

14.6

Elongation at break (%) Tear strength (kN/m)

556 11.6

344 60.8

660 30.3

390 65.4

*The AT loading level is 40 g/100 g rubber.

308 14.3

290 58.8

412 22.9

237 48.5

356 18.1

177 36.1

270 13.9

109 26.4

509

113

no data no data


Table 16.1 Mechanical properties of palygorskite/rubber nanocomposites*


489

intercalation of rubber chains into the gallery of clay layers. Alkyl ammonium salts are the most widely used surfactant for preparation of organically modified clay (OMC). Kim et al. investigated the influence of alkyl length of the surfactant on the spatial distribution of OMC in NBR matrix.22 They prepared a series of OMC modified by octylamine (C8-MMT), dodecylamine (C12MMT) and octadecylamine (C18-MMT), respectively. TEM observation, as shown in Fig. 5 in their report, showed that the dispersion state obviously improved (i.e., the dispersion dimension reduced but distribution homogeneity increased) with increasing alkyl length. In another publication by the same authors on the same RCN systems,23 WAXD patterns (as shown in Fig. 2 in that report) suggested there were exfoliated structures in C12-MMT/NBR and C18-MMT/NBR systems; nevertheless there was intercalated structure in the C8-MMT/NBR one. Zheng et al. studied the influence of the length and dosage of alkyl ammonium salts on the microstructure of EPDM/OMC nanocomposites.24 It was found that the intercalated structures could be obtained in the EPDM filled with OMC modified by octadecyltrimethylammonium chloride (MMT-C18) and the intercalation extent increased with surfactant dosage increasing, whereas the EPDM chains could not be intercalated into the gallery of OMC modified by hexadecyltrimethylammonium salt (MMTC16). A study of the SBR/OMC system conducted by Schön et al., also confirmed the above viewpoint that longer alkyl chains and larger dosage of the modifier would facilitate the formation of a better dispersion state.25 Features of rubber compatibilizers or coupling agents Similar to the case of thermoplastic-based clay nanocomposites, it is easy to obtain good dispersion of clay particles in the rubber matrix with high polarity, for example NBR22,23,26 and HNBR.27 Polar rubber could be used as the compatibilizer to improve the dispersion state of OMC as well as the interaction with non-polar rubber. For instance, epoxidized natural rubber (ENR)28 and maleic anhydride grafted EPDM (EPDM-g-MAH)24,29,30 were used for NR/ OMC and EPDM/OMC, respectively. Kim et al. also disclosed that adding silane coupling agent could enhance the dispersion of OMC (C18-MMT) in NBR.31 Melt compounding conditions Both Schön et al.25 and Gatos et al.30 drew the same conclusion that high shear mixing force could improve dispersion of OMC in rubber matrix from their researches on SBR/OMC and EPDM/OMC, respectively. Schön disclosed that compounding on an open roll mill was better than on an internal mixer. In contrast, Gatos drew the reverse conclusion. Gatos also disclosed that a high compounding temperature might improve the dispersion of clay. The

490


better dispersion of clay resulted in an obviously improved reinforcing effect of OMC, both modulus and strength of the nanocomposites increasing. Vaia and Giannelis established the thermodynamic principles for polymer melt intercalation OMC according to their studies on polystyrene-type polymerbased clay nanocomposites.32,33 The free energy change (∆F) upon intercalation was separated into an energy change (∆E) due to new intermolecular interactions and combinational entropy change (∆S) associated with conformational change of the polymer (∆Spolymer) and the modifier’s aliphatic chains (∆Schain), which could be expressed by the following two equations: ∆F = ∆E – T∆S

16.1

∆S ≈ ∆Schain + ∆Spolymer

16.2

For non-polar polymer, ∆E caused by the change of intercalated structures is small, so that the intercalation should be determined mainly by ∆S. According to theoretical calculations32 and experimental validation,33 there is a critical gallery height hc (~2.4 nm, with corresponding basal spacing ~3.4 nm) for the compounding system containing clay modified by octadecylammonium salt (initial basal spacing ~2.3 nm) and an arbitrary aliphatic polymer (nonpolar), below which the penalty for polymer confinement is compensated for by entropy gains of the tethered surfactant chains associated with layer separation, and overall entropy change is near zero. For non-polar polymer having weak interactions with surfactant chains and silicate surfaces, ∆F for intercalation of polymer into the silicate gallery is always below zero if h < hc, meaning that the intercalated structure with gallery height less than hc should be a thermodynamically favorable and stable structure. In some nonpolar rubber, however, the intercalation structures with gallery height far larger than hc (i.e., 2.4 nm) were obtained by melt blending. Figure 16.7 shows the WAXD pattern and TEM images of IIR/OMC nanocomposites.34 Liang et al. assumed that this phenomenon should be mainly attributed to the features of rubber being different from those of thermoplastic. In comparison with common thermoplastic polymer, rubber has such a larger molecular weight that the viscosity and shear stress for rubber compound during melt blending are quite high. Rubber chains are oriented by shear stress to a larger extent, especially at the gap region between two rolls, which can decrease the entropy of chains before their intercalation. Therefore, ∆Spolymer related to polymer confinement would be reduced during melt intercalation, so that hc for rubber in this dynamic state may be far larger than hc (~2.4 nm) estimated based on thermoplastic. These analyses can also explain the influence of mixing shear magnitude on the dispersion of OMC. There have also been some studies revealing that the microstructures would greatly change during the curing process; and the curative system, the


491

d (nm) 108765 4

3

2

1

Intensity (counts)

1000 800 600 400 5.87 nm 200

2.94 nm

0 1

2

3

4

100 nm

5 6 2θ (°)

7

8

9 10

500 nm

16.7 WAXD pattern (top) and TEM images (bottom) of IIR/OMC binary model compound (100/10 wt). The asterisk indicates (001) peak for OMC dispersed in IIR matrix. Copied from Ref.34

organic intercalant type, and processing parameters (temperature and pressure) might be the main factors influencing this change. The researches on this aspect will be reviewed in detail in Section 16.5 rather than here. One of the main advantages in performance for RCNs prepared by melt blending is high reinforcement with quite low filler loading. Table 16.2 summarizes the reinforcement levels of OMC in some rubbers. Very interestingly, the RCNs always exhibit very high values of the elongation at break. It has been suggested that the nanodispersed clay pellets could facilitate the energy dissipation and retard or slow the cracks growing during tension. Another advantage of RCNs is high gas barrier performance. Figure 16.8 shows the relationship between OMC loading and nitrogen gas permeability

492


Table 16.2 Optimal reinforcement level of OMC in different rubbers reported in references Rubbers

Compatibilizer

OMC type*

OMC loading


Elongation at break (%)

Ref.

NR

NA

10 phr

~19.0

>1000

[35]

NR ENR

NA NA

10 phr 10 phr

15.0 21.2

>700 767

[36] [37]

NR

ENR

2 phr

~28.0

>1200

[28,38]

SBR

NA

10 phr

~14.5

>1300

[39]

SBR EPDM EPDM EPDM EPDM NBR HNBR IIR IIR

NA NA NA MAH-g-EPDM MAH-g-EPDM NA NA NA NA

MMT-ODA I.30P MMT-ODA MMT-ODA I.30P I.28E OMC Nanocor I.28E OMC Nanocor MMT-ODA MMT-ODA MMT-ODA MMT-ODA MMT-ODA MMT-ODA MMT-ODA MMT-ODA MMT-ODA I.30P

10 phr 10 phr 15 phr 15 phr 5 phr 8 wt% 10 phr 20 phr 10 phr

~5.0 ~18.0 15.5 22.2 ~12.0 ~14.0 ~32.0 ~15.0 ~20.0

~450 ~570 470 620 NA >1500 >700 NA 760

[40] [40] [24] [24] [29] [22] [27] [41] [42]

* MMT = montmorillonite; ODA = octadecylamine.

1.00

Pc / Pp

0.95

0.90

0.85

0.80

0.75 0

2

4 6 Loading of clay (vol%)

8

16.8 Influence of clay loading on the relative permeability of IIRCNs prepared by melt blending. PC and PP are permeability of the composite and pure polymer, respectively. 0 vol% represents pure IIR vulcanizate. Redrawn according to data from Ref.41


493

of IIRCN.41 It can be seen that addition of nanoclay could definitely improve the gas barrier properties of IIR-based composites, but not to a great extent. The study on NBR (acrylonitrile content ~34%)/OMC showed that addition of only 15 phr OMC can decrease air permeability of NBR composites about 80%.43 This difference in the degree of improvement of gas barrier properties upon nanoclay layers should be attributed to the difference in the dispersion state of clay in non-polar (i.e., IIR) and polar (i.e., NBR) rubbers. Lu et al. prepared a series of highly filled RCNs with the melt blending process.44 It was found that highly filled RCNs (up to ~60 wt% OMC) have intercalated structures, and the dispersion homogeneity of clay layers is improved with increased loading of OMC. Highly filled RCNs exhibited extremely high modulus and greatly improved gas barrier properties, as shown in Figs 16.9 and 16.10. For the above advantages of OMC, some researchers suggested that OMC could be a promising substitute for conventional reinforcing fillers (i.e., carbon black and silica) for rubbers.45,28,36

16.3

Latex compounding method

16.3.1 Layered silicate/rubber nanocomposites Up to now, technologies for preparing rubber–clay nanocomposites have included mainly rubber melt or solution intercalation of organoclay and the latex route using pristine clay.46 Compared with the melt or solution method, the approach of co-coagulating rubber latex and clay aqueous suspension,46 where pristine clay (non-organoclay) is employed, is promising for industrialization due to the low cost of pristine clay, the simplicity of the preparation process, and superior cost/performance ratio. Figure 16.11 shows TEM micrographs of four rubber–clay nanocomposites containing 20 phr clay, NR-clay, SBR-clay, NBR-clay, and CNBR-clay. In Fig. 16.11, the dark lines are the intersections of the silicate layers. From Fig. 16.11(a)–(d), there are both individual layers and stacking silicate layers with a thickness of about 10–30 nm, and the dispersion of clay in each of the four rubber matrixes is excellent. Therefore, the co-coagulation technique is effective and applicable to rubbers with a latex form. The XRD experimental results demonstrated that the rubber–clay nanocomposites prepared by co-coagulation are a kind of partly exfoliated structure, in which the rubber molecules ‘separate’ the clay into either individual layers or just silicate layer aggregates of nanometer thickness without the intercalation of rubber molecules into clay galleries. It is important to note that the non-exfoliated layer aggregates without the intercalation of rubber molecules in rubber–clay nanocomposites prepared by co-coagulation are different from the intercalated layer aggregates in polymer–clay nanocomposites

494


prepared by melt blending. The reason is that the mechanism for forming nanocomposite structures by the latex route is totally different from that by melt compounding. Consequently, the nanocomposites prepared by the melt compounding method bear both intercalated and exfoliated silicate layers simultaneously. In contrast, the non-exfoliated layer aggregates in rubber– clay nanocomposites prepared by co-coagulation are formed by the reaggregation of exfoliated clay layers during the co-coagulating process. Therefore, it is according to the unique nanocompounding mechanism that

RCN’s E / Neat rubber’s E ’

120

25°C

100 80 60 40

100°C

20 –70°C 0 0

10

20 30 40 OMC Content (wt%) (a)

50

60

350 25°C


300 250 200 150 100

100°C 50 –70°C 0 0

10

20 30 40 OMC Content (wt%) (b)

50

60

16.9 Influence of OMC content on storage modulus (E ′) ratio of RCNs to neat rubbers under different temperatures: (a) EPDMCNs; (b) SBRCNs; (c) ECOCNs. Copied from Ref.44


495

1200 25°C


1000

800 600 400 100°C 200 –70°C 0 0

10

20 30 40 OMC Content (wt%) (c)

50

60

16.9 (Cont’d) 1.1 1.0

EPDMCN

0.9

SBRCN ECOCN

0.8

Pc / Pp

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0

20

40

60 80 100 Loadig of OMC (phr)

120

140

160

16.10 Influence of OMC loading on N2 gas permeability of RCNs. 0 phr means neat cured rubber sample. Pc and Pp represent the gas permeability of composites and neat rubber, respectively. Copied from Ref.44

this kind of structure prepared by co-coagulation is named a ‘separated’ structure. In addition, the concept of ‘separated’ structure can also explain the difference from the partly exfoliated structure prepared by melt blending. The formation mechanism of ‘separated’ structures can be illustrated by the schematic of the mixing and co-coagulating process presented in Fig. 16.12. At the stage of mixing, the rubber latex particles were mixed with the

496


200 nm

200 nm

(a)

(b)

200 nm

200 nm (c)

(d)

16.11 TEM micrographs of four rubber–clay nanocomposites: (a) NRclay; (b) SBR-clay; (c) NBR-clay; (d) CNBR-clay.

clay aqueous suspension, in which clay was dispersed into individual silicate layers. After adding a flocculant, the flocculant coagulated the rubber latex and the silicate layers simultaneously, but the rubber macromolecules did not exactly intercalate into the galleries of clay. This resulted mainly from the competition between separation of rubber latex particles and reaggregation of single silicate layers upon addition of flocculant. Since rubber latex particles are composed of several molecules, the existence of latex particles between the galleries of silicate layers in the water medium should result in completely separated (exfoliated) silicate layers. However, cations of flocculant cause

Clay aggregates

Flocculant cation

Adding flocculant

Clay

After co-coagulation

Rubber macromolecule

Exfoliated layer

16.12 Schematic illustration of the mixing and co-coagulating process.


Latex particle

497

498


separated silicate layers to reaggregate so that the rubber latex particles between the silicate layers may be expelled. As a result, there are some nonexfoliated layers in the nanocomposites. In the meantime, due to the fact that the amount of latex is more than that of silicate layers and the latex particles agglomerate rapidly, the reaggregation of silicate layers is evidently obstructed to some extent by the agglomerated latex particles around the silicate layers. Consequently, the size of aggregates of silicate layers is at the nanometer level, and the nanocomposites thus obtained contain both the exfoliated silicate layers and non-exfoliated (not intercalated) aggregates of nanometer thickness in the rubber matrix. According to the above nanocompounding mechanism, the factors affecting the final dispersion level of nanocomposites include mainly the size of rubber latex particles, the ratio of rubber latex to clay suspension, and the speed of co-coagulating. It can be expected that, based on the present investigations, the smaller latex particles, the more latex content, and the faster speed of cocoagulating rubber latex and clay layers will provide nanocomposites with fewer non-exfoliated layer aggregates, and even completely exfoliated nanocomposites. Up to now, the completely exfoliated structure could only be observed in SBR/MMT and styrene–isoprene–butadiene rubber/MMT nanocomposites containing less than 4 wt% MMT prepared by in-situ living anionic polymerization.47,48 The mechanical properties of three rubber–clay nanocomposites – SBRclay, NR-clay and CNBR-clay – are listed in Table 16.3. Compared to the corresponding conventional rubber–clay composites containing the equivalent amount of clay (20 phr), all three nanocomposites exhibit substantially higher 300% stress, shore A hardness, tensile strength and tear strength. The largely increased reinforcement and the tear resistance of the nanocomposites should be ascribed to the dispersed structure of clay at the nano level, the high aspect ratio and the planar orientation of the silicate layers.49 The gas permeabilities of gum SBR vulcanizate, SBR-clay nanocomposites (SBR-clay NC), conventional SBR-clay composites (SBR-clay MC) and SBR filled with carbon black (SBR-N330) are presented in Fig. 16.13. In Table 16.3 Mechanical properties of rubber–clay nanocomposites (NC samples) and conventional rubber–clay composites (MC samples) with 20 phr clay Sample

Stress at 300% strain (MPa) Tensile strength (MPa) Elongation at break (%) Shore A hardness Tear strength/(kN/m)

SBR–clay

NR–clay

CNBR–clay

MC

NC

MC

NC

MC

NC

2.1 2.4 400 52 16.5

7.4 14.5 548 60 45.3

2.7 11.6 568 41 22.8

12.3 26.8 644 54 44.1

5.2 9.0 444 60 24.4

– 18.0 228 82 46.5


499

Permeability (m2 · Pa–1· s–1)

8 7 6

SBR-N330

5 SBR-clay MC 4 3 SBR-clay NC 2 0

5

10 15 Volume fraction (%)

20

25

16.13 Effect of filler volume fraction on gas permeability.

Fig. 16.13, the nitrogen permeabilities reduce with the increase of the amount of filler, and SBR-clay nanocomposites have the best gas barrier property among the three classes of composites. Compared with the gum SBR vulcanizate, the nitrogen permeability of SBR-clay nanocomposites with 1.96, 7.40 and 13.8 vol% clay reduces by 27.3%, 54.1% and 61%, respectively. The reason is that the silicate layers having large aspect ratio and planar orientation lead to the great increase of the diffusion distance by creating a much more tortuous path for the diffusing gas.

16.3.2 Cabot elastomeric composite To increase the mixing efficiency of carbon black in NR, and to improve the material properties, Cabot Corp. developed carbon black/NR masterbatch (also called Cabot Elastomeric Composite – CEC) prepared by latex mixing (see Fig. 16.14).50 During the preparation process, the carbon black slurry is prepared by finely dispersing carbon black in water mechanically without any surfactant, and the slurry is injected into the mixer at very high speeds and mixes continuously with the NR latex stream. Under highly energetic and turbulent conditions, the mixing and coagulation of polymer with filler is completed mechanically at room temperature in less than 0.1 second, without the aid of chemicals. This essentially reduces the adsorption of nonrubber substances on the filler surface, thus polymer–filler interaction can be better preserved, leading to a better abrasion resistance. For the carbon blacks that are easily dispersed, the abrasion resistance of CEC is comparable to that of dry mixed counterparts at lower loading. For carbon blacks with high surface area and/or low structure that have poor dispersibility, CEC gives advantages over dry mixing in abrasion resistance, especially at higher filler loading, due to the powerful ability of the CEC process to achieve excellent filler dispersion.

500

Advances in polymer processing Carbon black

Continuous process

Water

Mechanical dispersion

NR Iatex

Liquid mixing/coagulating

Filter slurry

Dewatering

Oil/other chemicals (Optional)

Drying

CEC (NR/Carbon black) (Optioal – NR/carbon blak/oil/other chemicals)

16.14 CEC process.

Technically, the grades of carbon blacks used for rubber reinforcement can be greatly expanded via CEC technology. Traditionally, since carbon blacks with surface area higher than 160 m2/g and CDBP lower than 60 mL/ 100 g cannot be dispersed by dry mixing, they are not considered rubber grades. With CEC technology, however, over the practical range of loading, carbon blacks with surface area as high as 260 m2/g and CDBP as low as 40 mL/100 g can be dispersed in the polymer matrix with excellent dispersion. This is a great advantage of CEC technology for rubber reinforcement as these carbon blacks may impart some unusual properties to the filled rubber compounds. For example, for carbon blacks with high surface area and/or low structure, the tear strength can reach a very high level which has never been obtained with conventional compounds. The tradeoff among the tensile strength, elongation at break and hardness can also be considerably improved. The dispersion of carbon black in CEC is excellent and distribution is uniform relative to the dry-mixed compound (Fig. 16.15). In fact, the carbon black dispersion and distribution is fully completed at the very early stages of the CEC process. Due to the minor mechanical energy input during the following processing, the mechanical breakdown of polymer molecules is not significant. Consequently, compared with dry-mixed compounds, the tear and tensile strengths of the vulcaniztes are higher, the hysteresis and heat build-up are lower, and the flex fatigue life is substantially longer (Fig. 16.16).


X 1000 CEC-N234 50 phr/Oil 5 phr

501

X 1000 Dry mixing-N234 50 phr/Oil 5 phr

16.15 TEM images of CEC and dry-mixed compounds. 100% Flex life (+90% average) Hysteresis* (–10% to –15%) Tear strength (+15%) Tensile strength (+20%) Heat build-up* (–10°C) CEC

Dry-mixed

*Lower is better.

16.16 Comparison of vulcanizate properties between CEC and drymixed compounds.

16.4

In-situ polymerization of metallic salts of unsaturated carboxylic acid (MSUCA) during peroxide curing of elastomers

The metallic salts of unsaturated carboxylic acid (MSUCA) were originally used as coagents for peroxide curing of rubbers. They not only increase the crosslinking efficiency of the vulcanization process and the crosslink density, but also enhance the bonding strength between rubber compounds and metal materials as well.51 Crosslinking polybutadiene rubber (BR) with zinc dimethacrylate (ZDMA) results in a very hard rubber with high resilience, which has been used for many years in forming rubber cores for two-piece

502


golf balls.52 In the late 1980s, the dramatic reinforcement of hydrogenated nitrile–butadiene rubber (HNBR) by ZDMA was discovered,53–55 which has attracted significant interest from researchers. Now the addition of MSUCA as reactive fillers to rubber is becoming a new approach to achieve sustainable reinforcement. Many studies have revealed that MSUCA-reinforced rubbers are a kind of nanocomposite, and there are fine particles of diameter 10–30 nm in the elastomer matrix.56–63 It is believed that such nanostructures are generated from the in-situ polymerization of MSUCA, and they impart novel properties to this kind of composite. In this section, some current recognitions, viewpoints and research results about MSUCA/rubber composites are reviewed and summarized, including (i) processing of MSUCA/rubber nanocomposites (NCs) and mechanism of nanodispersion generation; (ii) performance features of MSUCA-reinforced rubber NCs and their advantages; and (iii) application of in-situ polymerization of MSUCA in the preparation of functional elastomer composites.

16.4.1 Processing of MSUCA/rubber nanocomposites (NCs) and mechanism of nanodispersion generation There are two processing methods for preparation of MSUCA/Rubber NCs. One is that MSUCA (powders of micro-level in general) is directly compounded into a rubber.64 The other is that, through the neutralization of a metal oxide or metal hydroxide and methacrylic acid (MMA) or acrylic acid (AA), a MSUCA is prepared in situ in rubber during mixing. The corresponding reports revealed that the reinforcing effect of the latter is better than that of the former. However, this in-situ processing method might corrode metal parts of the mixing machine due to the acidity of carboxylic acid. Nomura et al.57 first investigated the course of nanodispersion of ZDMA in HNBR during peroxide curing, and brought out the point that in-situ radical polymerization of ZDMA introduced generation of nanodispersions, which has been confirmed by many other studies58–63 and is well accepted now. The course of nanodispersion formation during peroxide curing of the MSUCA/rubber composite is schematically displayed in Fig. 16.17. In the uncured MSUCA/rubber compound, there are small amounts of MSUCA monomers diffused in the rubber matrix, the in-situ free-radical polymerization of which would be initiated during the process of curing the compound with peroxide. The products of in-situ polymerization, poly-MSUCA (P-MSUCA), are incompatible with the rubber matrix, so that they aggregate to form nanodispersion. With the in-situ polymerization, the ‘dissolving balance’ of MSUCA monomer in the matrix rubber is broken constantly, causing MSUCA monomers to continually diffuse from MSUCA particles into the matrix rubber to take part in the in-situ polymerization. The above procedure is


503

MSUCA particles (micro level) Rubber matrix Dissolved MSUCA monomer P-MSUCA Nano-particles

In-situ polymerization of MSUCA initiated by peroxide radical

MSUCA further dissolved in rubber

16.17 Schematic representation of nanodispersion of MSUCA in rubber matrix induced by their in-situ polymerization during peroxide curing.

repeated continuously during the curing of the compound with peroxide, and as a result, the majority of the original MSUCA particles dispersed in the compound disappear, and many nano-granular dispersions of 10–20 nm diameter are formed after the compound is cured. It has been disclosed that the nanodispersion of MSUCA through their insitu polymerization could be achieved in various elastomer matrixes.57–63 TEM morphologies of various MSUCA/rubber nanocomposites are displayed in Fig. 16.18. This clearly indicates that there are significant amounts of nanolevel dispersion, the darker phase in various elastomer matrixes. According to the principle of TEM imaging, it can be judged that the darker phases in the TEM photographs should represent the Zn-rich regions and be aggregates of poly-ZDMA. The dispersion of these nano phases is extremely homogeneous, compared to the polymer-based composites directly filled with nanofillers. On the other hand, some micron-level dispersions remain in all cured composites, as shown in Fig. 16.19, which are assumed to be unreacted micron-level ZDMA particles.63 Substantial residual ZDMA particles of micrometer level in the cured composite become stress concentrators during the tensile course, resulting in the destruction of the composite at low elongation and the decrease in final tensile strength of the nanocomposites. For instance, the amount and dimensions of microdispersion in XA/POE are far larger than those of in SR634/POE, so that both tensile strength and elongation at break of XA/POE are obviously less than those of SR634/POE. According to present studies, there are three factors to determine the conversion of ZDMA (i.e., the residual amount of unreacted ZDMA particles in the cured composites), and these are discussed below under the following headings.

504


100 nm

100 nm

100 nm (a)

(b)

(c)

100 nm

100 nm

100 nm (d)

(e)

(f)

16.18 TEM morphology of ZDMA (SR634)/elastomer (30/100) cured composites: (a) SR634/SBR, (b) SR634/POE, (c) SR634/EPM, (d) SR634/EPDM, (e) SR634/NBR, (f) SR634/HNBR. Copied from Ref.63 SR634 denotes a commercial ZDMA product having a trademark of Saret 634, which was produced by US Sartomer Co. POE is the abbreviation of poly olefin elastomer, which has the trademark of Engage and is produced by Dupont-Dow Co.

The ratio of the initiator (peroxide) to the monomer Saito et al. investigated the polymerization behavior of ZDMA in NBR and HNBR, and found that the conversion increased with increasing peroxide concentration, but increased little at more than 90% conversion; the molecular weight of P-ZDMA is inversely proportional to [peroxide]1/2 at low peroxide concentration, but depends only slightly on the peroxide concentration at high peroxide levels and high conversion of ZDMA.56

Processing of macro, micro and nanocomposites (a)

505 (b)

(c)

(d)

(e)

(f)

(g)

16.19 SEM morphology of ZDMA/rubber (30/100) cured composites: (a) SR634/POE, (b) XA/POE, (c) SR634/EPM, (d) XA/EPM, (e) SR634/ SBR, (f) SR634/NBR, (g) SR634/HNBR. Copied from Ref.63 XA denotes the other type ZDMA produced by Xian Organic Chemical Technology Plant, P.R. China.

506


The reactivity of rubber with radicals There is competition between the in-situ reaction of ZDMA and crosslinking of the matrix elastomer, because both kinds of reaction need to be initiated by radicals produced by heat decomposition of peroxide. Lu et al. developed an infrared spectroscopic method to trace the course of in-situ polymerization of ZDMA in POE.60 The integral intensity of the 831 cm–1 band, i.e. the inplane rocking mode of ==CH2, was used to determine the residual amount of ZDMA in composites cured at 165°C for different times (as shown in Fig. 16.20), through which the course of in-situ polymerization of ZDMA in POE was traced, and the dynamic curve was described (as shown in Fig. 16.21). By comparing this course with the curing behavior of POE, it was found that the in-situ polymerization of ZDMA almost completes at the beginning stage of curing of POE and the substantial crosslinking reaction starts subsequently. This result suggested that, in the saturated elastomer matrix, the most peroxide radical should attack the ZDMA monomer to initiate its in-situ polymerization at the beginning stage of curing, because the reactivity of saturated polymer chains with the radical is far lower than that of ZDMA. As a result, the in-situ reactions of ZDMA would be completed more thoroughly, and the residual amount of unreacted ZDMA microparticles would be less. On the other hand, in the unsaturated rubber matrix, the reactivity of ZDMA with radical would be comparable to that of rubber chains, and then the in-situ reactions of ZDMA would be affected to a large extent by curing reactions, and the conversion might be less. For example, the amount of

1651 831

720

20 min 5 min 1 min 0 min

1800

1600

1400 1200 1000 Wave Number, cm–1

800

600

16.20 FTIR spectrum of ZDMA/POE/D25 (30/100/6) cured at 165°C for different times: 0 min; 1 min; 5 min; 20 min. Copied from Ref.60


507

1.0

0.6

0.5

0.745

0.8

0.6

C t′

A830/A720

0.4

0.3 0.4 0.2 0.2

4.1 min

0.1 0.0 0

2

4

6

8 10 12 14 Curing time, min (a)

16

18

0.0 20

100 90 80

T90 = 20 min

60 50 40

T10 = 4 min

Torque, dN.m

70

30 20 10 0 0

5

10

15

20 25 30 Curing time, min (b)

35

40

45

16.21 Comparison between the course of (a) in-situ polymerization of ZDMA in POE and (b) peroxide curing. Redrawn according to data from Ref.60 In graph (a), the 720 cm–1 band associated with the deformation mode of —(CH2)n— (n > 5) of POE was used as a reference band for normalization. The relative conversation of ZDMA at a curing time C ′t was calculated from C ′t = (A0 – At)/(A0 – A20), where A0, At and A20 are the normalized areas of 831 cm–1 at 0 min, t min and 20 min (T90), respectively.

residual unreacted ZDMA microparticles in cured ZDMA/SBR composites (Fig. 16.19(e)) is very large.63 During the peroxide curing process of MSUCA/rubber compound, the grafted P-MSUCA could also be formed through a terminate reaction between

508


the poly-MSUCA radical and the rubber chain radical, or a transfer reaction of the poly-MSUCA radical to the rubber chain. Therefore, the grafted PMSUCA fraction and molecular weight (MW) of P-MSUCA are also greatly affected by the reactivity of matrix rubber with the radical. The study conducted by Saito et al. disclosed that the number of grafted poly-ZDMA on HNBR slightly increases up to 80% conversion and sharply increases above 80%; nevertheless the number of grafted poly-ZDMA on NBR is considerable (0.397 fraction in total poly-ZDMA) when the conversion of ZDMA is only 57.4%, and is 4.6 times that of HNBR at 80% conversion.56 They also found that the MW of poly-ZDMA in NBR is about one-fourth that in HNBR. Ikeda et al. studied the copolymerization of ZDMA and perfluoroalkyl acrylates (RsFA) in different solvents58 and their copolymerization in various amounts of DMF in the presence of HNBR to simulate the in-situ copolymerization.65 Their results also revealed that the MW and graft ratio of poly-ZDMA depend on the double bond content of the HNBR. In summary, poly-MSUCA generated by in-situ polymerization in high-saturated rubber would have a low fraction of grafted structure but high MW. For this reason, the dimension of the nanodispersion (i.e. aggregates of poly-ZDMA) in elastomers with higher saturation such as POE, EPM, EPDM and HNBR is somewhat larger than that in elastomers with lower saturation such as SBR and NBR (see Fig. 16.18). The dispersion state of ZDMA in the compound before in-situ reaction In the course of nanodispersion of MSUCA induced by their in-situ polymerization during peroxide curing, the diffusion of MSUCA monomer from their particles to the rubber matrix is a control step, which is greatly affected by the contact area between MSUCA particles and the matrix polymer. Therefore, the dispersed level of MSUCA particles in the compound is an essential factor for the morphology of the MSUCA/elastomer composite. Fine dispersion of MSUCA particles and high contact area would facilitate the in-situ polymerization of MSUCA and lead to higher conversion and smaller amounts of residual unreacted micro-MSUCA particles. In the case of MSUCA powders being directly compounded into rubber, there are two factors determining the dispersion level of MSUCA in the uncured compound: the features of the matrix rubber and the original particle size of MSUCA.63 As shown in Figs 16.22 and 16.23, during mechanical mixing a larger number of micro-level ZDMA particles were ground into the smaller particles and even ultra-fine particles whose sizes are nano level, leading to reduction in dimension and amount of micro-level ZDMA particles in EPM, SBR, EPDM, NBR and HNBR. The dispersion of ZDMA particles in the compound is predominantly affected by the rheological features of matrix elastomer during the compounding process. The rheology measurements show that EPDM, NBR, HNBR and SBR exhibit higher apparent shear stress


509

(τw), EPM has medium τw, and POE has less τw, under their corresponding mixing temperatures.63 The higher shear stress benefits the grinding of ZDMA particles during the compounding process. The smaller difference in affinity between MSUCA and matrix rubber (i.e., NBR and HNBR) would lead to a greater amount of monomers dissolved in the rubber, and accelerate the diffusion of monomers from MSUCA particles to the rubber matrix during the process of peroxide curing, resulting in higher conversion of in-situ polymerization of MSUCA. For instance, the original dispersion states of ZDMA particles in the SBR and NBR compounds (Fig. 16.22(i) and (m)) and the reactivity of SBR and NBR with radicals are similar, but the conversion in the cured NBR compound was much higher, and there were fewer residual unreacted ZDMA particles compared to the cured SBR composite. Only in the elastomer with the lower shear stress such as POE did the original dimension and shape of ZDMA particles have a considerable influence on the dispersion state of ZDMA particles (see Fig. 16.22(c) and (d)).

(a) SR634 powders

(b) XA powders

(c) SR634/POE

(d) XA/POE

16.22 SEM morphology of ZDMA powders (SR634 and XA) elastomer (30/100) compounds before curing. Copied from Ref.63

510


(e) SR634/EPM

(f) XA/EPM

(g) SR634/EPDM

(h) XA/EPDM

(i) SR634/NBR

(j) XA/NBR

16.22 Cont’d


(k) SR634/HNBR

(l) XA/HNBR

(m) SR634/SBR

(n) XA/SBR

511

16.22 Cont’d

In the case of in-situ preparation of MSUCA through the neutralization reaction during mixing of the compounds, almost all produced MSUSA are dispersed in the rubber matrix at nano level (as shown in Fig. 16.24), so that the conversion of in-situ polymerization of MSUCA might be higher and there should be little microdispersion in cured composites. Consequently, the MSUCA/rubber nanocomposites prepared in this way usually exhibit higher strength,66,67 compared to the directly compounding method.64

16.4.2 The performance features of MSUCA-reinforced rubber NCs and their advantages In general, the MSUCA/rubber nanocomposites exhibit the mechanical property characteristics of high strength, high hardness and high elongation at break. However, MSUCA has different reinforcing effects for different elastomers. The reinforcement levels obtained with various peroxide-cured elastomers using ZDMA or MMA/ZnO are summarized in Table 16.4. It was suggested

512

Advances in polymer processing 200 nm

200 nm

4 µm

(a) SR634/POE

(b) SR634/EPM

(c) SR634/EPDM

200 nm

200 nm 2000 nm (d) SR634/NBR

(e) SR634/HNBR

(f) XA/NBR

16.23 TEM morphology of ZDMA (SR634)/elastomer (30/100) compounds before curing. Images (a)–(d) were copied from Ref.63

that the degree of reinforcement would depend upon three factors, as also illustrated in Table 16.4. The tensile strength of the HNBR composite is far superior to that obtained with other elastomers, due to its affinity to MMA salts, moderate radical reactivity and microcrystallinity created upon hydrogenation.55 Lu et al. studied the mechanical properties of various ZDMA-reinforced elastomer nanocomposites, and claimed that the saturation and regularity of rubber chains are two essential features to determine the mechanical properties of the composites at room temperature.64 They also proposed a possible model for the microstructure of ZDMA/rubber composites to interpret the mechanism, which is schematically represented in Fig. 16.25. It was expected that there might be multiple interactions between poly-ZDMA and matrix elastomer. In Fig. 16.25(a), the physical adsorption between poly-ZDMA nanoparticles and rubber chains is relatively weak individually, but their sum is great due


513

200 nm

1000 nm

(a) ZnO/MMA/NBR compound

100 nm

2000 nm

(b) ZSC compound (ZnO/MMA/HNBR)

16.24 TEM images of MSUCA/elastomer compound by in-situ neutralization reaction between ZnO and MMA. The equivalent loading of ZDMA in the NBR compound is about 30 phr. HNBR/ZnO/ MMA as grade ZSC2295 was purchased form Zeon Chemicals, Inc. (Japan).

514


Table 16.4 Dependence of reinforcement degree of MMA/ZnO/rubber composites on properties of rubbers Elastomer

Tensile strength of elastomer reinforced with ZnO/MMA or ZDMA (MPa)

Affinity to MMA salt

Moderate radical reactivity

Crystallization ability

HNBR NBR NR POE (Engage 8180) EPM EPDM SBR BR

50–60 20–30 < 30

30–40 > 30 15–20 15–20 < 10

means positive or enhanced reactivity; means negative or no reactivity. Grafted poly-ZDMA-nanoparticles

Poly-ZDMA-nanoparticles

Rubber chains (a)

Rubber chains (b)

2+ Zn2+ Zn Zn Zn2+ Zn2+ 2+ Zn2+ Zn2+ Zn 2+

Rubber chain

Short grafted poly-ZDMA

Zn2+ Zn2+

Ionic cluster (c)

16.25 Schematic representation of the multiple-microstructure model for rubber reinforced by ZDMA: (a) the physical adsorption between poly-ZDMA nanoparticles and rubber chains; (b) the chemical grafting of poly-ZDMA nanoparticles onto rubber chains; (c) the ionic cluster structure through the static electronic attraction. Copied from Ref.64


515

to the extremely high specific area of the nanoparticle. This structure can be considered as the physical crosslinking point for the composite, but it is not thermally stable. As shown in Fig. 16.25(b), if the nanoparticle consists of poly-ZDMA chains grafted onto different rubber chains, a crosslinking structure will form, while if the nanoparticle traps only one grafted poly-ZDMA chain, a nanoparticle chemically suspended onto the rubber chain will form. Both will exhibit much stronger interfacial interaction than structure (a). Some grafted poly-ZDMAs with low molecular weight are not able to form nanoparticles alone because the chains are too short. However, they can form ionic clusters through static electronic attraction, as represented in Fig. 16.25(c). This structure is also the physical crosslinking, and a similar structure exists in the ionic polymer or metallic oxide crosslinked carboxyl rubber.68,69 It was supposed that the major difference between ionic clusters and polyZDMA nanoparticles is that the density of ionic bonds in nanoparticles is much higher than that in ionic clusters. For this reason, Tg for poly-ZDMA is above 300°C and nano-poly-ZDMA particles are thermally stable in common applications of rubber composites, while the ionic clusters disassociate over 100°C.68 For different ZDMA/rubber composites, the fraction for each microstructure is not similar, resulting in different mechanical properties of the composite. For composites with high-saturated or total-saturated matrix rubber, such as EPM, POE, HNBR and EPDM, there is little grafted poly-ZDMA,56,60 so that the major interaction between nanoparticles and rubber is physical absorption. This interaction has an important mechanical feature that rubber chains can slide on the surface of the nanoparticle. Additionally, the covalent crosslinking density is low in this kind of composite. Therefore, this kind of composite readily exhibits stress relaxation at low strain and high elongation at break. If the regularity of the rubber chain is great enough, the tensile induced crystallization will occur at high strain, thus high strength can be obtained at room temperature. The affinity of rubber to ZDMA and the appropriate (but not exorbitant) fraction of grafted poly-ZDMA would enhance the stress transfer and be of benefit in achieving superior reinforcement. ZSC2295, which has a tensile strength of ~60 MPa, is a commercial HNBR compound reinforced with ZnO and MMA (produced by Zeon Chemicals, Inc.). Matrix HNBR such as Zeptol 2020 has a hydrogenation degree of 90%.55 In contrast, ZDMA-reinforced HNBR having high saturation, such as Zeptol 2000 (hydrogenation degree over 98%), only has ~40 MPa tensile strength, when the optimum reinforcement is achieved.64 For composites with low-saturated rubber matrix, such as NBR, SBR and BR, the grafted poly-ZDMA nanoparticles and ionic clusters take the major fraction, and covalent crosslinking densities are also high. The morphology study63 revealed that there are substantial residual ZDMA particles in cured SR634/SBR composites. Therefore, compared with ZDMA/NBR, ZDMA/

516


SBR has more ionic clusters and covalent crosslinking but less grafted polyZDMA nanoparticles. The grafted nanoparticle and covalent crosslinking are the structures with little feature of stress relaxation. Therefore, the macromolecular chain of this kind of composite is more constrained and has higher efficiency of stress transfer, which tends to cause high modulus but low elongation and strength. It was shown in some studies and patents53–55,62,70,71 that using ZDMA to reinforce rubber could enhance the high-temperature performance of rubber composites. Nevertheless, there is other research59 disclosing that the high temperature strength of some rubber/ZDMA composites is much lower than that of carbon black reinforced composites. Figure 16.26 compares tensile strengths at room temperature (~20°C) and at high temperature (~120°C).64 The strengths of these composites at 120°C are not high and not more than 7.5 MPa. Comparing the high-temperature strengths of all composites, those of NBR and HNBR composites hold the first place, that of SBR composite takes the second place, and those of EPM, POE and EPDM are lower. This At room temperature (~20°C) At high temperature (120°C)

40 35


30 25 20 15 10 5

34.8%

18.4%

30.0% 8.84%

10.6%

16.0%

0 SBR

NBR

EPM

POE

HNBR

EPDM

16.26 Comparison between tensile strengths of various elastomers reinforced by 30 phr ZDMA (SR643) at room temperature (~20°C) and those at high temperature (120°C). The percentage above each dark column is the retention rate of tensile strength of each composite at high temperature with respect to that at room temperature. Copied from Ref.64


517

phenomenon implies that the high polarity and unsaturation of the matrix elastomer might benefit the high-temperature properties of ZDMA/rubber composites. It is very interesting that the ratio of the high-temperature strength over the room-temperature strength is basically proportional to the content of double bonds in the matrix rubber. Generally, the strength reduction of a composite at high temperature reflects the thermal instability of its microstructures. Hence the strength retention ratio at high temperature demonstrates the fraction of thermally stable structures existing in the composite. Apparently, composites with low-saturated matrix rubber, such as SBR and NBR, should have more thermally stable structures. As mentioned above, in composites with high-saturated matrix rubber, such as EPM, POE, EPDM and HNBR, the majority of interactions between rubber and nanopoly-ZDMA particles belong to physical adsorption. Under high temperature, this physical absorption is easily disassociated, so that the strength of this kind of composite falls dramatically. Due to the high polarity of HNBR, the strength of physical interactions is strong, so that the absolute high-temperature strength of ZDMA/HNBR is relatively high. On the other hand, the fractions of grafted poly-ZDMA and covalent crosslinking in composites with unsaturated rubber matrix are much larger, and these are thermally stable, so that the strength of this kind of composite falls no more than does that of composites with high-saturated matrix. Apart from zinc dimethacrylate (ZDMA) and zinc diacrylate acid (ZDA), which are usually used for MSUCA-reinforced rubber composites, some other types of MSUCA, such as magnesium methacrylate (MgMA), sodium methacrylate (NaMA) and lithium acrylate (LiA), reinforced elastomer systems were also investigated.72–77 The highest levels of reinforcement achieved in these systems are summarized in Table 16.5. The mechanical properties of elastomers reinforced by these non-zinc types of MSUCA were superior to those of carbon black reinforced rubber vulcanizates. The in-situ preparation Table 16.5 Mechanical properties of various elastomer composites reinforced by non-zinc type of MSUCAs, when their tensile strength reaches the highest level Elastomer type

MSUCA type (optimum loading for strength, phr)

Tensile strength, MPa

Elongation at break, %

Tear strength, kN/m

Ref.

EVMa NBR SBR EVMa CPEa

MgMA (50) Mg(OH)2/MAA (50b) MgO/MAA (30b) NaOH/MMA (50b) LiOH/AA (40b)

22.5 29.2 30.6 ~35 ~28

~350 474 483 ~350 ~150

84 NA NA ~95 NA

[72] [73] [74] [75,76] [77]

a

EVM and CPE are abbreviations of ethylene–vinyl acetate rubber, and chlorinated polyethylene, respectively. b Equivalent loading of MSUCA.

518


and polymerization of LiA also play important roles in the improvement of water-swelling properties of CPE vulcanizates, providing a new approach to the preparation of water-swelling elastomers.77 Very interestingly, the molar ratio of metallic oxide (MO) or hydroxide (MOH) to carboxylic acid (CA) always exhibits a significant influence on the final mechanical properties of the elastomer reinforced with MSUCA generated in-situ through the neutralization reaction, the exact reason for which is still unclear at present. Table 16.6 shows this influence in various compound systems. As far as the tensile strength is concerned, Table 16.6 The influence of molar ratio of metallic oxide (MO) or hydroxide (MOH) to carboxylic acid (CA) on mechanical properties of various elastomers reinforced by in-situ prepared MUSCA. Rubber type

MO or MOH/CA

Influence of molar ratio on mechanical properties

Ref.

HNBR

ZnO/MAA

The best reinforcement over the broadest range of loadings was obtained with a molar ratio of 0.75

[52]

NBR

ZnO/MAA

Increase of molar ratio from 0.25 to 1.5 resulted in increase of modulus and simultaneous decrease of elongation at break, but little change in tensile strength

[66]

EPDM

ZnO/MAA

Hardness and tear strength directly increased with molar ratio increasing from 0.4 to 1.0; the tensile strength reached the highest level in the molar ratio range of 0.7–0.8; the elongation at break decreased rapidly when the molar ratio was over 0.8

[67]

NBR

Mg(OH)2/MAA

The best reinforcement over the 10–50 phr loading range was obtained with a molar ratio of 0.5 (equivalent).

[72]

SBR

MgO/MAA

The tensile strength and elongation at break reached their maximum values at a molar ratio of 0.5 (equivalent), where on the contrary the hardness reached its minimum value. The MgO/MAA ratio of 0.50–0.75 was recommended for a good combination of all the mechanical properties

[74]

EVM

NaOH/MAA

The optimum molar ratio for mechanical properties is 1 (equivalent)

[75]

CPE

LiOH/AA

The tensile strength reached its maximum value at a molar ratio of 1 (equivalent), where on the contrary the elongation at break reached its minimum value

[77]


519

the optimum molar ratio for the zinc carboxylate reinforced rubber composite is always over 0.5, in other words the amount of ZnO is excessive.55,66,67 In the other types of MSUCA-reinforced elastomer composites, however, the best reinforcement is often obtained when MO or MOH and CA are equivalent.73–75,77 Besides excellent mechanical properties, MSUCA/elastomer systems also exhibit some other advantages compared to rubber composites reinforced with conventional fillers such as carbon black and silica. The MSUCA/ elastomer compounds possess good processing ability. Their viscosity did not change with loading of MSUCA, nor did the hardness of the resulting composites increase,55 whereas those of carbon black or silica compounds increased rapidly. This is mainly because the nanoparticles having reinforcement did not exist in the mixing compound, but would be formed in-situ during the curing process of the compound with peroxides. It was also found that zinc carboxylate could provide good adhesion of rubber to the metal substrates. For this application, ZDA showed more effect than ZDMA.78

16.4.3 Application of in-situ polymerization of MSUCA in the preparation of functional elastomer composites The studies on MSUCA (i.e., Zn, Mg, Na and Li) reinforced elastomers suggest that the in-situ reaction of MSUCA in an elastomer matrix during peroxide curing could be an effective way to obtain a fine nano-level dispersion and strong interactions between the dispersion phase and the matrix in polymer composites. Enlightened by these works, some researchers employed this kind of in-situ reaction method to disperse some functional substances at nano level in the elastomer matrix, and thereby prepared some novel functional elastomer composites.58,79–82 Because of the special electronic structure of lanthanides, they can efficiently remedy the weak absorption of lead between 40 and 88 keV, the regular energy region of X-rays used for medical diagnosis. In addition, lanthanides have a much lower toxicity than lead. Thus, polymer composites containing lanthanides suggest a novel and efficient way for overall and efficient X-ray shielding. However, most rare earths are micro-sized particles, and also have poor compatibility with a rubber matrix. As a result, it is difficult to finely and homogeneously disperse the inorganic rare earth compounds into the rubber matrix, especially for a high loading. The poor dispersion and interfacial interaction not only lead to the poor mechanical properties of the composites, but also are assumed to limit the shielding performance of the dispersed phase, probably because of the presence of more local, micro, non-filled rubber parts and the lowering of the probability of interaction between the dispersed phase and the X-rays. Liu et al. used gadolinium acrylate (Gd(AA)3),

520


which is a rare earth complex compound (RECC) having radical reactivity, to prepare rubber-based rare earth composites with an in-situ reaction method.80 Figure 16.27, showing SEM and TEM images of Gd(AA)3/natural rubber compounds (uncured) and vulcanizates, reveals that the dispersion state of the Gd(AA)3 phase was greatly improved by in-situ reactions of Gd(AA)3 during rubber curing with peroxide (see SEM images), and many nano-level substances containing Gd were formed in the cured composites (see TEM images). Figure 16.28 shows that the parameter P, which denotes the X-ray shielding property of the composites, gradually increases with an increase of vulcanization

(a)

(b)

200 nm 1000 nm (c)

(d)

16.27 SEM images of fracture surfaces of Gd(AA)3/NR composites (content of Gd(AA)3 50 wt%): (a) uncured compound; (b) the vulcanizates. TEM images of the vulcanizates: (c) low magnification; (b) high magnification. Images (a) and (b) were copied from Ref.80


0.20

1

25%

2

35%

3

40%

4

50%

4

0.18

P/(mmPb/(g/cm2))

521

3

0.16

2

0.14 0.12

1

0.10 0.08

T0

T50

T90

T100

Curing degree

16.28 The X-ray protection properties of Gd(AA)3/NR composites with different Gd(AA)3 contents varying with curing degree (i.e., T0, T50, T90 and T100). Copied from Ref.80

time. For instance, when the Gd(AA)3 content of the composites is 50 wt%, the shielding property of the composites cured for T100 is 16.7% greater than that of the uncured one. The patterns of interaction between substances and X-rays mainly include the photon effect, the Compton effect, and the electron– positron pair effect. Because the electron–positron pair effect only occurs at high energies (threshold 1.01 MeV, dominating at energies above ca. 10 MeV), the former two effects play a more important role in the absorption of X-ray energy by Gd(AA)3/NR composites with respect to medical X-rays (lower than 88 keV). The distributive dispersion of the shielding phase (substance containing Gd) obviously increased and the Gd(AA)3 dispersed ‘everywhere’ in the rubber matrix after in-situ reaction of Gd(AA)3 during the curing process of matrix rubber with peroxide, which was assumed to result in reducing the non-filled micro-local space and increasing the interaction probability between photons and the shielding phase in the shielding-phasefilled polymer composites. Therefore, the shielding property of the composites is apparently enhanced by in-situ reaction. Liu et al. also applied this kind of in-situ reaction approach to prepare rare-earth complex/rubber composites with luminescent properties.79 They first synthesized a novel europium complex, acrylate (1,10-phenanthroline) bis-(2-thenoyltrifluoroacetonato) europium(III) [Eu(tta)2(aa)(phen)], which combines the excellent fluorescence property of [Eu(tta)3(phen)] and the reactivity of acrylic acid (AA) with radicals. Various amounts of the obtained reactive complex (powder) were mixed with nitrile–butadiene rubber (NBR)

522


and peroxide to form compounds (uncured composites). The compounds were vulcanized to obtain [Eu(tta)2(aa)(phen)]/NBR composites (cured composites). SEM and TEM observations (as shown in Fig. 16.29) disclosed that the dispersion dimension for cured composites is far finer than that for uncured composites. It was assumed that the dispersion phase of cured [Eu(tta)2(aa)(phen)]/NBR composite should be composed of nearly nanosized aggregates of poly[Eu(tta)2(aa)(phen)] and residual [Eu(tta)2(aa)(phen)] particles with reduced dimension. WAXD experiments disclosed that the crystalline degree of [Eu(tta) 2(aa)(phen)] in the composite decreases dramatically, undergoing a crosslinking process, revealing that in-situ reactions

(a)

(c)

(b)

(d)

(e)

16.29 SEM images of the surfaces of the [Eu(tta)2(aa)(phen)]/NBR composite containing 16.17 wt% Eu complex: (a) uncured compound; (b) cured composites. TEM images of [Eu(tta)2(aa)(phen)]/NBR composite containing 16.17 wt% Eu complex: (c) uncured compound; (d) cured composite at low magnification; (e) cured composite at high magnification. Copied from Ref.79


523

(probably including polymerization and grafting) of [Eu(tta)2(aa)(phen)] initiated by peroxide radical should take place during crosslinking of NBR matrix with peroxide. Figure 16.30(a) and (b) shows the emission spectra of uncured and cured [Eu(tta)2(aa)(phen)]/NBR composites with different contents of Eu complex, respectively. It can be seen that cured and uncured composites possess similar

612

450 3

Emission intensity (a.u.)

400

1 2 3 4

4

350

– – – –

2.8% 8.3% 12.5% 16.7%

300 250 2

200

618 150 100

1 590

50 579

625

597

651

0 580

600 620 Wavelength (nm) (a)

640

660

1400 4


1200 3

1000 800

1 2 3 4

– – – –

2.8% 8.3% 12.5% 16.7%

2

600 1 400 200 0 580

600 620 Wavelength (nm) (b)

640

660

16.30 Fluorescent emission spectra (λex = 385 nm) of (a) uncured and (b) cured [Eu(tta)2(aa)(phen)]/NBR with different concentration of [Eu(tta)2(aa)(phen)]. Copied from Ref.79

524


emission spectra, and exhibit the characteristic emission bands assigned to Eu3+. Nevertheless, fluorescent intensities of cured composites are much stronger than those of their uncured counterparts, for instance the intensity at 612 nm of cured composite is about 2–5 times higher than that of uncured composite filled with the same amount of [Eu(tta)2(aa)(phen)]. The emission intensities at 612 nm of uncured and cured composites are summarized and plotted as a function of weight concentration of [Eu(tta)2(aa)(phen)] in Fig. 16.31. The fluorescent emission intensity at 612 nm of cured [Eu(tta)2(aa)(phen)]/NBR finely dispersed composites prepared via an in-situ reaction method is much stronger than that of their uncured counterparts. However, the dependence of emission on Eu3+ content departs from linearity at a considerably lower filling level (see Fig. 16.31). In general, this nonlinear relationship between fluorescent intensity and rare-earth content suggests formation of ionic aggregates in which metal ions are close together and direct energy transfer among them. It was expected that formation of nearly nano-sized poly[Eu(tta)2(aa)(phen)] aggregates in the rubber matrix during the curing process with peroxide should have different effects on the luminescence property of prepared composites: 1. Positive effects: The dispersion level for cured composites is much finer and more uniform than those of uncured ones, meaning a much larger surface area of rare-earth complex particles in cured composites at the same content of Eu3+, which would lead to a higher fraction of Eu3+ in 1400 Uncured composite Cured composite


1200

1000 800 600

400 200

0 0

2

4 6 8 10 12 14 16 Content of Eu-AAPhen (wt%)

18

16.31 A plot of the emission-intensity variation at 612 nm against rare-earth complex content for uncured and cured composites. Copied from Ref.79


525

dispersion particles that are easily irradiated by excitation light and emit fluorescence. Some fraction of poly[Eu(tta)2(aa)(phen)] would probably be grafted onto rubber chains through in-situ reactions. The strong chemical bonds between the Eu3+ complex and matrix rubber would facilitate energy transfer between them, enhancing excitation efficiency. 2. Negative effects: WAXD experiments disclosed that [Eu(tta)2(aa)(phen)] in uncured composites still preserves a crystal state; therefore, Eu3+ ions are strictly fixed in a crystalline lattice and hard to aggregate together. As a result, energy transfer between neighboring Eu3+ ions would hardly occur. While the dispersion particles in cured composites formed by poly[Eu(tta)2(aa)(phen)] have no crystalline phase, some ionic aggregates might form within these poly[Eu(tta)2(aa)(phen)] particles, which results in some excited Eu3+ ions not emitting fluorescence and depresses the emission efficiency of Eu3+. In addition, the high surface area of the dispersion particles increases the possibility of direct contact among them, which might cause energy transfer among Eu3+ complex particles. These positive and negative factors compete with each other to determine the final fluorescence intensity of composites. The results (Figs 16.29 and 16.30) clearly demonstrate that the positive effects caused by formation of fine poly[Eu(tta)2(aa)(phen)] particles through in-situ reactions are far more significant than its negative effects. Ikeda et al. chose ZDMA and fluorine-containing acrylate as the monomers for the in-situ copolymerization in HNBR during peroxide crosslinking, attempting to obtain excellent reinforcement and lower surface friction as well.58,65,82 They studied the copolymerization behaviors of ZDMA and different types of fluorine-containing acrylate (as shown in Fig. 16.32) in O

O

C2H5

CF3—(CF2)7—S—N—(CH2)2—O—C—CH

CH2

O RfSA O CH3 CF3—(CF2)7—(CH2)2—O—C—C

CH2

Rfm O CF3—(CF2)7—(CH2)2—O—C—CH

CH2

Rfa

16.32 Fluorine-containing acrylate78: RfSA: 2-(Nethylperfluorooctanesulfonamide)ethyl acrylate; RfM: 2perfluorooctylethyl methacrylate; RFA: 2-perfluorooctylethyl acrylate.

526


different solvents.82 It was expected that the sulfonamide group in RfSA could enhance the miscibility of RfSA with ZDMA, leading to the composition of the copolymer being almost the same as that of the comonomers at high conversions of both monomers. As a result, in the case of the in-situ copolymerization of ZDMA and RfSA in HNBR, the copolymerizations proceeded without phase separation of RfSA, which is preferable for improvement of the characteristics of the rubber. The resulting HNBR/ZDMA/ RfSA nanocomposites processed superior tensile strength, which was about 35 MPa, and about 60% lower friction than that obtained by in-situ polymerization of ZDMA alone in HNBR. The enhanced strength and low friction were considered to arise from the carboxylic salt in the ZDMA unit and the perfluoro-alkyl group in the RfSA unit, respectively. The surface tension and atomic compositions of this type of nanocomposites with curing course were characterized by the contact angle (CA) test and XPS, respectively.58 The results revealed that the surface compositions were different from their average compositions. The RfSA unit in the copolymer was segregated at the surface of the vulcanizates to minimize the surface tension while increasing curing time. We conclude this section with a short prospect on future researches on MSUCA/rubber composites prepared by the in-situ reaction method. Although much valuable work has been carried out in this field, and thereby some MSUCA/rubber composites with superior strength, novel functional characteristics or both have been obtained, some fundamental problems, such as the exact microstructures of the nanodispersion phase formed by insitu reactions and their interaction states with matrix rubber, and the relationships between the progress of in-situ reactions and microstructures, have not yet been clearly understood. The authors suggest that more attention and efforts on these fundamental problems would lead to new breakthroughs in this field. Flexible design and manufacture of nanostructures could be achieved to some extent.

16.5

Microstructural evolution of rubber-based nanocomposites during storage and the curing process

In order to maximize the modification effects of the nanodispersion phase, the most attention and efforts have been paid to improving the dispersion degree of nano phases and enhancing their interfacial interactions with the matrix polymer in the field of preparation and processing of polymer-based nanocomposites (PNCs). After the process of compounding, the dispersed nano phases would tend to aggregate or flocculate by themselves due to their inherent surface characteristics. In the case of thermoplastic-based nanocomposites, the dispersing microstructures could be ‘frozen’ and stabilized


527

by crystalline and/or amorphous domains of the matrix polymer. On the other hand, the Tg values of rubber and elastomeric materials are far below 0°C, so that the viscosity of the rubber compound is relatively low and the mobility of polymer chains in the compound is quite high under normal storage temperatures (room temperature, RT) or the usual curing temperature (i.e., 140–180°C). As a result, the microstructure of rubber-based nanocomposites (RNCs) could change significantly during the storage and curing process, which should also deserve great attention for preparing successful RNCs. Unfortunately, there have been only a few studies on this aspect, and many problems have not been understood clearly, especially for these new types of RNCs. In this section, some current research results and recognitions on this aspect will be reviewed, including: (i) the microstructural evolution of rubber compound filled with carbon black or silica (i.e., conventional reinforcing fillers) during storage; and (ii) the microstructural evolution of rubber/clay nanocomposites (RCNs) during the curing process.

16.5.1 Microstructural evolution of rubber compounds filled with carbon black or silica during storage Interactions between fillers and matrix rubber Bound rubber (BR) is an old concept in rubber science, and is known to be one of the major factors in carbon black reinforcement. Bound rubber is defined as the fraction of polymer that cannot be extracted from an uncured (carbon black) filled compound by a good solvent of the gum elastomer. Above a sufficient filler level (in the 15–20% weight range), a highly swollen rubber–filler gel remains after all the free rubber has been extracted by the solvent. It was proved subsequently that the concept of bound rubber is also valid in silica-filled compounds. Some researchers believed that the magnitude of bound rubber reflects the strength of interactions between filler and rubber.83 For a given elastomer, the amount of bound rubber at fixed filler content depends on the structural area, structure (or morphology) and surface activity of the filler. Moreover, bound rubber also shows a dependence on the processing conditions of the compound, such as mixing and storage times. Leblanc studied the evolution of bound rubber with storage time (at room temperature, RT) for BR/N330 carbon black binary compounds prepared under different mixing energies, the results of which are shown in Fig. 16.33.84 It can be seen that the bound rubber increases quickly with maturation time until a plateau is reached. In addition, a significant effect of the mixing energy level on the maximum bound rubber was noted, which corresponded well with similar observation on styrene–butadiene rubber compounds.85 With increase of mixing energy, the changing degree of the bound rubber obviously decreases. Through remilling, further evolution of bound rubber

528


Bound rubber (% gum in cpd)

30

25

20

15

0

: Mix 2A (460 MJ/m3)

: Mix 2A remilled

: Mix 2B (1480 MJ/m3)

: Mix 2B remilled

: Mix 2C (2420 MJ/m3)

: Mix 2C remilled

200

400

600 800 1000 1200 1400 Storage time (h)

16.33 Effect of mixing energy and storage time on bound rubber content in a model binary polybutadiene/carbon black compound. Copied from Ref.84

was obtained, which is particularly significant for the two compounds with the lowest mixing energy (mixes 2A and 2B). The authors also proposed the following equation to fit bound rubber storage maturation data for BR of different types: BRt = BR0 + ( BR∞ – BR0 )(1 – e – k

t

)

16.3

where BR0 and BR∝ are, respectively, the initial and stabilized bound rubber fraction, t is the maturation time, and k is a fitting parameter. By nonlinear regression, the parameters of this equation could be obtained, namely a stabilized BR value. They also applied this approach to some data of other carbon-filled rubber systems,86 and thereby the stabilized bound rubber values for various carbon-filled rubber compounds were obtained. By considering that the rubber chains obey Gaussian statistics, Cohen Addad, Touzet and Frebourg found the following relationship for bound rubber:87 BR =

cS p M0 ⋅ ⋅ ε a N Nv A0

Mn

16.4

where M 0 is the average molecular weight of the monomer units, A0 is the average area of the elementary adsorption sites on the filler surface, εa ≈ 1 is a numerical factor accounting for chain stiffness and surface coverage (this factor is still lacking a clear indication at present), M n is the number-average molecular weight of the polymer, Sp is the specific surface area of the filler


529

(m2/g), NAv is the Avogadro number (6.023 × 1023), and c is the filler concentration (g/g of gum polymer). Through application of this equation, Leblanc calculated the A0 values in various carbon-filled rubber compounds with the obtained stabilized bound rubber values; the results are shown in Table 16.7. It can be seen that the calculated A0 values are very close, irrespective of the carbon black specific surface area and the nature of the polymer.88 In addition, Wang et al. demonstrated a drastic decrease in bound rubber with increasing solvent extraction temperature on SBR–carbon black compounds.89 With respect to these results and discussions, Leblanc suggested that the interactions between carbon black and rubber (which result in the formation Table 16.7 Areas of elementary adsorption site on filler surface, through applications of Cohen-Addad’s equation on Leblanc and Hardy’s data (copied from Ref.88 Elastomer type Parameter

Mv (from bale) a

Mv (masticated ) N326b compounds: bound rubber (%)c A0 (nm2)d mean A0 N330° compounds: bound rubber (%)c A0 (nm2)d mean A0 N347f compounds: bound rubber (%)c A0 (nm2)d mean A0 a

NR SMR 5CV

Polybutadiene 98% cis-1,4 NeoCis BR40 EniChem

EPR 42% PP Dutral CO 054 EniChem

EPDM 46% PP; 3.5% ENB Dutral TER 054/E EniChem

580 000

225 000

55 000

65 000

330 000

230 000

65 000

50 000

30.33

23.57

8.06

7.80

81.7–46.5 64.1

43.6–44.6 44.1

35.8–42.3 39.1

43.8–33.7 38.8

32.77

25.22

8.12

8.96

72.8–414 57.1

39.7–40.5 40.1

35.9–41.4 38.2

37.3–28.7 33.0

33.31

27.52

9.62

9.34

77.2–43.9 60.6

38.5–39.4 39.9

31.7–37.4 34.6

38.7–29.7 34.2

In Banbury mixer up to 1470 MJ/m3 mixing energy. N326: Sp = 84 m2/g. c From toluene extraction experiments; extrapolated to infinite storage time at room temperature. d First number is calculated with Mv measured on rubber sample cut from bale; second number is calculated with Mv measured on masticated sample. e N330: Sp = 83 m2/g. f N347: Sp = 90 m2/g. b

530


of bound rubber) have a purely physical nature and there are likely to be topological constraints exerted on chain segments by the appropriate (geometrical) elements on the surface of the filler particles. Based on this hypothesis, Leblanc also proposed the molecular origin of bound rubber. The rubber–filler interaction resulting from topological constraints is schematically displayed in Fig. 16.34. It was expected that the polymer segment of structural units and the filler topological site possibly have the corresponding reciprocal geometry, in the appropriate orientation and at the right time. During mixing, the probability of such favorable events is obviously quite high. Once this topological interaction has taken place, it is quite obvious that, in order to release it, the free portions of the chain must exert on the constrained units not only sufficient stresses but also stresses in the appropriate direction, which would require quite a high energy level to be statistically significant for bound rubber to vanish. Compared to NR and BR, EP-type rubber segments involved in interaction with the carbon black surface have locally higher flexibility, because of the absence of a double bond. As a result, they are less prone to permanent topological interaction with the rigid filler surface. This would explain why EP rubbers exhibit lower bound rubber values. An investigation by Donnet and Custodero on the surface morphology of carbon black particles using an atomic force microscope was also in agreement with the above viewpoint about the feature of carbon black–rubber interactions.90 In essence, not all fractions of the elastomer in the compound are involved in bound rubber, and many experimental results and theories about bound rubber have demonstrated that the highest molecular weight chains do Rigid chain motif blocked by the appropriate site on filler surface

Dangling segment

Surface of carbon black particle

C-C bond half-free rotation

Steps-like structure at edges of broken graphitic plies

Dangling segment

Graphitic layers

16.34 Pictorial view of rubber–filler interaction resulting from topological constraints when an elastomer motif conformation meets locally the surface structure of a carbon black particle. Copied from Ref.88


531

preferentially adsorb on filler particles. From this point of view, Leblanc implied that initially adsorbed short chains would be progressively replaced by larger ones during storage under RT, and gave an acceptable model for expressing the evolution in molecular weight of the bound rubber with storage, as follows: M n ( t ) = M n (0) + [ M n ( ∞ ) – M n (0)](1 – e – bt )

16.5

where M n (0) and M n ( ∞ ) are the number-average molecular weights of bound rubber directly after compound preparation and after an infinitely long storage period, respectively, and b is a kinetic parameter (it was selected as 0.0003 in their calculation). With the above relationship, the evolution in A0 with storage for N330 carbon black-filled BR compound was also calculated through application of a modified Cohen-Addad equation. The result revealed that the interaction site area (A0) increased from 0.785 to 0.916 nm2 over a storage period of 2000 h. Consequently, it was believed that the main reason for bound rubber increasing with storage is slow replacement of short chains initially involved in bound rubber by large ones; and the secondary reason is a slight increase of A0 with respect to the limited variation of this parameter.88 Choi91 investigated the variation of bound rubber content of filled rubber compounds with storage time by using silica-filled and carbon-filled SBR compounds with different silane contents. Bis-(3-(triethoxysilyl)-propyl)tetrasulfide (TESPT) was employed as a silane coupling agent. He prepared two types of filled SBR compounds: the master batch (MB) compounds containing matrix rubber, filler, silane coupling agent, cure activators and antidegradants; and the final mixing (FM) compound that was prepared by mixing the curatives with MB compounds. The results (as shown in Figs 1 and 2 in his report) revealed that the bound rubber contents in both MB and FM decreased with storage time under 30°C for the silica-filled compounds without silane, which was believed to result from the release of some of the bound rubber molecules due to the large difference in polarity between rubber (except for silicone rubber) and uncoated silica. In contrast, for the silicafilled compounds containing silane, the bound rubber content clearly increased with storage time, and the initial increment was especially remarkable. It was expected that the silane would have high reactivity with silica even at the low temperature of 30°C. First, silanol of the silica reacts with the ethoxy group of the silane to form a siloxane bond (primary reaction) and then polysulfide of the silane bonded to the silica surface is dissociated and reacts with the rubber chain (secondary reaction). The bound rubber contents of the silica-filled FM are a little higher than for the respective MB compound. The variations in the behavior of the bound rubber content with storage time for carbon black-filled rubber compounds are quite different from those for silica-filled ones (as shown in Figs 3 and 4 in Choi’s report). When silane is not added or its content is low (0–4.8 phr for MB compounds and 0–2.4 phr

532


for FM ones), the bound rubber content increases slightly with increase of storage time. At higher silane levels the bound rubber increases dramatically following a 10–20 day period of slow increase. It was suggested that the polysulfide linkages of the retained silane in the rubber could be dissociated to form radicals as storage time elapsed and sulfidic radicals can react with rubber molecules. The pendent group terminated by silane residue would react with unreacted functional groups on the carbon black surface. Also unlike in the case of silica-filled rubber, bound rubber contents of the FM compounds filled with carbon black are obviously higher than those for the respective MB ones. It was believed that carbon black could react with sulfur and cure accelerator during mixing, so that ≡CS · radicals or ≡CX (X = SH or residues of a cure accelerator) were formed. Then the radicals on the filler surface could react with rubber chains and this would result in the formation of chemical bonds between the rubber and the filler. Choi’s results also showed a different influence of storage temperature on the variation of the bound rubber content between silica-filled compounds and carbon blackfilled ones, also suggesting the existence of different chemical mechanisms for them. Anyway, the work conducted by Choi suggested that chemical mechanisms, apart from the physical one, would also play very important roles in the evolution of bound rubber for carbon black-filled compounds containing silane coupling agent. Flocculation of fillers and formation of filler–filler networks Because of the surface energy difference between rubber and filler, the filler aggregates will inexorably tend to flocculate during storage and vulcanization of the compound, forming filler–filler networks.92 The Payne effect of filled rubber compounds is manifest as a dependence of the viscoelastic storage modulus on the amplitude of the applied strain at a given condition of temperature and vibration frequency. Above approximately 0.1% strain amplitude, the storage modulus decreases rapidly with increasing amplitude. At sufficiently large strain amplitudes (roughly 20%), the storage modulus approaches a lower bound.93,94 The magnitude of modulus decreasing with strain amplitude is considered as an indirect representation of the strength of filler–filler networks in filled rubber compounds.95 Böhm and Nguyen employed the Payne effect as a measure of filler flocculation to study the factors influencing the flocculation course of carbon black in the unvulcanized compounds.96 Figure 16.35 shows a schematic plot where ∆E′ = ∆E′ (e = 14%) – ∆E′ (e = 0.2%). The course of increase in ∆E′ is regarded as being caused by carbon black flocculation during annealing at evaluated temperature for various rubber compounds, where the molecular weight of matrix rubber, the type and loading of carbon black, and the initial dispersion state were selected as changing parameters.


E ′ (MPa)

8

533

After flocculation

6

4

∆E(t) Before flocculation

∆E(0)

2

0.1

1 10 Strain amplitude (%)

16.35 Schematic representation of the strain amplitude dependence of the dynamic storage modulus E ′ and the effect of carbon black flocculation. Adapted from Ref.96

Gerspacher et al. argued that low strain dynamic measurements could result in the deformation and destruction of the filler network and therefore reduce the sensitivity of this method. They traced the flocculation of carbon black in the compound through recording the electrical resistance of the compound as a function of time and aging temperature, based on the fact that the resistivity of a rubber compound depends largely on the local fluctuation of density of the interpenetrated carbon black/polymer networks.97 Most of the conclusions derived from these two different researches are similar: the lower the molecular weight of the polymer, the smaller the filler and the higher annealing temperature associated with the higher flocculation rate. However, there are some contrary conclusions concerning the following aspects. Study of the Payne effect in annealed compounds suggested that a filler content that is obviously above the percolation concentration, bad initial dispersion of filler and high structure of filler would result in stronger flocculation during annealing.96 On the other hand, the electrical resistance experiments implied that a filler content near the percolation concentration, good initial dispersion and low structure of filler would favor the course of flocculation.97 Besides the above results, the study of Gerspacher et al. also disclosed that the higher bound rubber fluctuation, the higher flocculation of carbon black during annealing, and the addition of oil and curative would significantly affect the course of the flocculation. They also found that the flocculation of carbon black still occurred in the cured compound, though its magnitude was much smaller than in the uncured one and was slowed with increase of crosslinking density. Wang92 suggested that the flocculation would be influenced by two main factors: (i) the diffusion of aggregates due to Brownian motion to form

534


thermodynamically stable agglomerates; and (ii) the mean distance between filler aggregates. For the former factor, the diffusion constant (∆) can be given by the Einstein–Stokes relation as follows:

∆ = kT ⋅

1 6 πηa

16.6

where k is the Boltzmann constant, T the absolute temperature, η the medium viscosity, and a the particular diameter. It can clearly be concluded from the diffusion point of view that higher viscosity of polymer, larger aggregate size (or effective aggregate size) of filler and higher filler structure would be favorable for a reduced rate of flocculation. For the latter factor, Wang and co-workers established the relation of mean distance between aggregate (δagg) with surface area (S), loading and structure of filler, as expressed by the following equations:

δ agg = 6000 ( kφ –1/3 β –1/3 – 1) β 1/43 ρS β=

φeff = 0.181DBPA + 1 φ 1.59

16.7 16.8

where ρ is the density, φ is the volume fraction of the filler, k is a constant which is dependent on the arrangement of aggregates in the polymer matrix, and β is an expansion factor which is the ratio between the effective volume fraction (φeff) and the actual volume fraction (φ) of the filler. It is obvious from Eqn 16.7 that the higher the surface area and loading of the filler, and the shorter the inter-aggregate distance, the easier would be the filler flocculation. Equation 16.8 includes the effect of DBPA, i.e., the higher the structure, the smaller will be the distance. However, in comparison to surface area, the effect of DBPA on the δegg is much smaller. In conclusion, for filled rubber compounds, both storage and remilling before the vulcanization process should be necessary. The storage could enhance the interaction between filler and rubber (bound rubber increasing). The remilling process might also increase the bound rubber content but more importantly can break down the filler–filler networks formed during storage and can improve filler dispersion.

16.5.2 Microstructural evolution of rubber/clay nanocomposites (RCNs) during the curing process For most rubber-based composites, the curing process under high temperature and high pressure is needed after compounding so as to obtain crosslinked nanocomposites. Some researches have disclosed that the microstructures of


535

rubber/clay nanocomposites (RCNs) could be changed considerably during the curing process. It is possible that LeBaron and Pinnavaia98 first reported the great change in the dispersion structure of OMC in elastomer matrix caused by the curing process. They found that crosslinking reactions between poly(dimethylsiloxane) (PDMS) containing terminal hydroxyl groups and tetraethyl orthosilicate (TEOS) in the presence of the OMC afforded elastomeric nanocomposites where the clay nanolayers were exfoliated. However, subsequent researches on this aspect mainly concentrated on RCNs cured with the sulfur vulcanization system. The reported works revealed that there might be two types of microstructural change caused by the curing process, as follows: 1. Change in local microstructures of clay particles: This type of change could be deduced from WAXD patterns of RCNs before and after curing. It can be divided into three sub-classes for this type of change. (a) Further intercalation: The position of the (001) reflection peak of OMC shifted to a lower angle, or the (001) peak disappeared (implying the formation of exfoliated structures or intercalated ones having low order degree). (b) Confinement or collapse: The position of the (001) peak shifted to a higher angle after curing, but it was still smaller than that of pure OMC used in this compound. (c) Deintercalation: The position of the (001) peak was larger than that of the pure OMC after curing. In some cases, even totally inorganic clays could be formed. 2. Change in spatial distribution of clay particles: This kind of change includes the dimensions of dispersed clay particles and the homogeneity of their distribution on a large scale. The information can usually be obtained by using TEM. As far as the reinforcement of OMC as concerned, the homogeneity of their spatial distribution is more important than whether some exfoliated clay layers exist. So TEM observation at low magnification is more suitable for the discovery of this type of microstructural change. Some works have been carried out on the reasons and mechanisms for these microstructural changes, in order to take advantage of this kind of change to tailor the microstructures of RCNs. Results to date suggest that there are two classes of factors influencing microstructural evolution of RCNs during the curing process: chemical and physical factors. The chemical factors include two aspects: the reactions between rubber and curatives at the initial stage of the curing process, and the reactions between curatives and aminetype intercalants within the silicate galleries. Usuki et al. studied the influence of the curing accelerator type on microstructures of crosslinked EPDM/clay nanocomposites (EPDMCNs).99 They found that the silicate layers of the clay were exfoliated and almost dispersed to the monolayers in the cured EPDMCNs when thiuram (TS) or dithiocarbamate (PZ) type vulcanization

536


accelerator was used; whereas the dispersibility of clays in EPDM was insufficient when thiourea (NPV/C), thiazole (M) or sulfonamide (CA) type was used. They assumed that the radicals produced by thermal dissociation of thiuram or dithiocarbamate could combine with carbon atoms in EPDM chains to polarize EPDM molecules, which resulted in the further intercalation of EPDM molecules into the clay galleries through hydrogen bonds between the polar EPDM and the clay surface. Some other studies on NR,35 ENR,37 EPDM100,101 and HNBR102 base clay nanocomposites conducted by KargerKocsis et al. disclosed that the confinement, deintercalation and further intercalation could occur simultaneously during vulcanization of RCNs. They speculated that a Zn complex with sulfur and amine groups of the organic amine-type intercalants could be formed during vulcanization. This Zn complex is intermediate for sulfur vulcanization and would react with rubber. If it migrated into the rubber matrix to take part in the vulcanization, the intercalants would be extracted out of the clay galleries, resulting in confinement and even deintercalation. However, if it caused rubber crosslinking inside the galleries, the rubber molecules would be further inserted into the clay galleries, resulting in further layer separation or delamination/exfoliation. They also disclosed that types of amine intercalants play important roles in the structural changes of clay layers. When primary amine (i.e., octadecylamine, ODA) was used as intercalant, confined and even deintercalated structures were always found in the crosslinked RCNs. In contrast, when quaternary amine, which has less reactivity with curatives, was used, confinement and deintercalation almost did not happen.101,102 Karger-Kocsis et al. used peroxide as a substitute for sulfur curatives to cure HNBR/OMC composites. In contrast to sulfur vulcanization, which in combination with primary amine intercalants produced a confined/deintercalated clay structure, the peroxide curing yielded well-ordered intercalated nanocomposites.103 Very recently, Lu et al. reported the dramatic impact of curing temperature on the microstructure of sulfurcured IIRCNs.42 It can be deduced from Fig. 16.36(a) that deintercalation occurred significantly only when curing temperature was above a certain level (140°C in this case). They discussed the mechanism for the deintercalation according to the WAXD patterns of the mixture containing OMC and sulfur curatives after thermal treatments, as shown in Fig. 16.37. The results show that the reactions between curatives and amine intercalants occurring at high temperature resulted only in layer separation and delamination rather than in deintercalation in the absence of rubber. The high temperature and high pressure during vulcanization were considered as two physical factors influencing the microstructural changes of RCNs. As mentioned in Section 16.2.3, some thermodynamically unstable intercalated structures could be obtained by melt compounding. At room temperature, these structures could remain due to the high viscosity of the compound. At high temperature, however, they would spontaneously transform


537

Uncured IIRCN 8765 4

3

d (nm) 2

1

3.24 nm (001) 1 µm

3000

IIRCN110 3.61nm (002) 2500

Uncured IIRCN 1.05nm (003)

(001) 5.20 nm

1 µm IIRCN140

Intensity (counts)

2000

1500

(002) 2.80nm (003) (004) 1.02 1.31nm nm 5.03 nm IIRCN110 1 µm 1.29 nm

IIRCN160

IIRCN140

5.08 nm 1000

1.30 nm 1 µm 500

4.86 nm

IIRCN160 IIRCN180

1.30 nm

IIRCN180

1 µm

0 1

2

3

4

5 6 2θ (°) (a)

7

8

9 10

(b)

16.36 (a) WAXD patterns and (b) TEM images of IIRCNs (IIR:OMC = 100:10) cured at different temperatures. WAXD curves were shifted vertically for clarity. Copied from Ref.42

Advances in polymer processing 450 2.2 nm 400 350

OMC

3.55 nm(001)

Intensity (a.u.)

300 250

1.79 nm (002)

OMC + CA

6.89 nm(001)

200 150

OMC + CA–110°C × 1h

4.33 nm(001)

OMC + CA–160°C × 1h

100 4.53 nm(001)

50

OMC + CA–180°C × 1h

0 1

2

3

4

5

6

7

8

9

10

2θ (°)

16.37 Influence of treating conditions on WAXD patterns of the mixture of OMC and curing agents (CA). The WAXD patterns of pure OMC and untreated mixture of OMC and CA were also plotted for a comparison. The mixture contained: OMC 10 parts, zinc oxide (ZnO) 5 parts, stearic acid (SA) 2 parts, tetramethyl thiuram disulfide (accelerator TMTD) 1.0 parts, 2-mercapto benzothiazole (accelerator M) 0.5 parts, sulfur (S) 1.8 parts, N-phenyl-α-naphthylamine (antioxidant A) 1.0 parts. Copied from Ref.42 d (nm) 876 5

4

3

2

1

900 800 700

Intensity (counts)

538

600

3.30 nm

2.05 nm

500

160°C × 1h

3.10nm

400

120°C × 1h

5.88 nm

300 200

80°C × 1h

100

Untreated

0 1

2

3

4

5 6 2θ (°)

7

8

9

10

16.38 WAXD patterns of untreated IIRCN and IIRCN (IIR:OMC = 100:10) thermally treated at different temperatures and at AP for 1 h. The asterisks indicate (001) peak for OMC dispersed in IIR matrix. The dotted line indicates the location of the silicate (001) reflection of pure OMC. The curves were shifted vertically for clarity. Copied from Ref.34


539

d (nm) 876 5 4

3

2

1

1500 1350 3.06 nm 1200

Intensity (counts)

1050 3.07 nm

2.07 nm

900 1.49nm

750 600

3.11nm

160°C × 15MPa × 1h 160°C × 5.0MPa × 1h

450 3.30 nm

300

160°C × 4.5MPa × 1h

2.05nm

150

160°C × AP × 1h

0 1

2

3

4

5

6

7

8

9

10

2θ (°)

16.39 Influence of pressure on WAXD patterns of IIRCN treated at 160°C for 1 h. The asterisks indicate (001) peak for OMC dispersed in IIR matrix. The dotted line indicates the location of the silicate (001) reflection of pure OMC. Copied from Ref.34

to thermodynamically stable structures, as shown in Fig. 16.38.34 It was also found that high pressure could enhance and accelerate this transformation, as shown in Fig. 16.39. More importantly, high pressure was discovered to be the critical factor causing the aggregation of originally well-dispersed OMC particles.104 As shown in Fig. 16.40, the dimensions of dispersed OMC particles clearly increased and the homogeneity of their spatial distribution reduced dramatically after high-pressure vulcanization, compared to uncured compounds. In contrast, the spatial dispersion state would be greatly improved after vulcanization under atmospheric pressure (AP), as shown in Fig. 16.41. The dispersion state of OMC plays an extremely important role in determining the properties of RCNs. Table 16.8 shows that not only mechanical but also gas-barrier properties of IIRCN cured under 3 MPa pressure are clearly superior to those of IIRCN cured under 15 MPa pressure.105 When Lu et al. studied the effects of heat and pressure on microstructures of IIRCNs prepared by solution intercalation, the vaporization effect of residual solvents on the microstructures was disclosed.106 Figure 16.42 (WAXD patterns) and Figure 16.43 (TEM images) reveal that the vaporization of residual solvents occurring during high-temperature treatment under AP could expand

540


2500 4.28 nm

IIRCN cured at 15MPa × 180°C

2000 1500 1000 2.08 nm

500

1µm

1.35nm

IIRCN cured at 15 MPa × 150°C

0 2500

Uncured IIRCH

2000 1500

Intensity (cps)

1000 500

6.65nm 3.72nm

0 2500 2000

1µm Uncured IIRCN

SBRCN cured at 15MPa × 150°C 5.05nm

1500 1000 2.12nm

500

1.39nm

0 2500

Uncured SBRCN

1µm SBRCN cured at 15 MPa × 150°C

2000 1500 1000 500

6.43nm 3.27nm 2.08nm

0

1µm 2θ (°) (a)

Uncured SBRCN (b)

16.40 (a) WAXD patterns and (b) TEM images of cured and uncured IIRCN and SBRCN prepared by melting intercalation. Copied from Ref.104

and delaminate the clay galleries, resulting in great improvement in spatial distribution. Enlightened by these results, a new method to prepare RCNs by using a melt-blending process was also developed. A certain amount of an organic solvent and OMC were first mixed to obtained pre-swelled OMC (PSOMC). The PSOMC was incorporated into the rubber by melt blending. The obtained RCNs have similar microstructural features to that of S-IIRCN.

IIRCN cured at AP × 180°C

2500 2000 1500

Intensity (cps)

500

4.45 nm

0 2500

SBRCN cured at AP × 150°C

IIRCN cured at AP × 180°C

2000 1500 4.52 nm

1000 500

2.18 nm

0 1

2

3

4

5 6 2θ (°) (a)

7

8

9

10 SBRCN cured at AP × 150°C (b)


1000

16.41 (a) WAXD patterns and (b) TEM images of IIRCN and SBRCN cured at atmospheric pressure. The vertical scales of the WAXD patterns are the same as for Fig. 40(a) in order to allow comparison. Copied from Ref.104

541

542

Advances in polymer processing Table 16.8 Comparison of properties between IIRCNs cured at 15 MPa pressure and 3 MPa pressure Curing pressure (MPa) Property

15

3

Shore A hardness (deg) Elastic modulus, G′ (MPa) Tensile strength (MPa) Elongation at break (%) Tear strength (kN/m) Relative permeability (Pc/Pp)*

43 4.9 14.8 657 19 0.70

47 8.0 20.0 631 23 0.54

*Pc and Pp are permeability of the composite and pure polymer, respectively. d (nm) 9876 5 4

3

1800

2

1

3.50 nm

1600 3.70 nm

Untreated

1200

1.81nm

3.63nm

AP × 80°C × 1h

1000 1.73nm

3.39nm

800

I001

Intensity (counts)

1400

600

AP × 100°C × 1h

1.68nm

AP × 120°C × 1h

3.26nm

400

AP × 140°C × 1h 3.13 nm

200

AP × 160°C × 1h

0 1

2

3

4

5

6

7

8

9

10

2θ (°)

16.42 WAXD patterns of untreated IIRCN and S-IIRCNs treated at different temperatures and at AP for 1 h. The asterisks indicate the (001) peak of OMC dispersed in the IIR matrix. The curves are shifted vertically for clarity. Copied from Ref.106

During the curing process, the solvents within the silicate galleries were vaporized, which caused exfoliation and prevented aggregation of the silicate layers. As a result, RCNs prepared by this method have better dispersion morphology, and therefore their mechanical and gas-barrier properties are better than those of RCNs prepared by traditional melt blending. For instance, the tensile strength and gas-barrier property of IIRCN (100/5) prepared by


543

100nm

500nm

(a)

500nm

100nm

(b)

16.43 TEM photographs of S-IIRCN treated in conditions of (a) AP at 80°C for 1 h, and (b) AP at 160°C for 1 h. Left: low magnification; right: high magnification. Copied from Ref.106

this new method are ~75% and 15% higher than those of the counterpart prepared by melt blending, respectively. It was expected that the spatial aggregation of clay particles in a rubber matrix caused by high pressure should happen mainly within the initial

544


period of curing, when most of the rubber chains are in a flowing state. Therefore, increasing the curing rate by elevating the temperature might be a feasible means to reduce the negative effect of high pressure. Lu et al. researched the influence of curing temperature on microstructures and properties of RCNs.42 The results show that the reactions during the initial curing period result in further intercalation of rubber chains into clay galleries and improve the spatial dispersion of clay particles. High pressure leads to the aggregation of clay layers to form larger clay agglomerates. Increasing the curing rate by elevating the temperature can reduce this aggregation. However, exorbitant curing temperatures (i.e., 160–180°C for this system) also strongly accelerate the reactions between amine intercalants with CAs, resulting in deadsorption of intercalants with the formation of larger amounts of inorganic clays and retarding further intercalation of IIR chains into silicate galleries induced by the curing reactions. There is an optimum curing temperature range to obtain the best dispersion, i.e., around 140°C for the researched IIRCN system (as shown in Fig. 16.36). Because the crosslinking density, filler dispersion state and filler–rubber interactions are strongly influenced by curing temperature, the gas-barrier and mechanical performances of IIRCN cured under different temperatures are quite different (as shown in Figs 16.44 and 16.45). In accordance with the most recent results, some suggestions for preparing successful RCNs may be made, as follows. (a) If commercial OMC (the

1.0

P c /P p

0.9

0.8

0.7

0.6

0.5 110

120

130 140 150 160 Curing temperature (°C)

170

180

16.44 Impact of the curing temperature on the gas barrier property of cured IIRCN (IIR:OMC = 100:10). Pc and Pp represent the gas permeability of the composite and pure matrix rubber, respectively. Copied from Ref.42


545

22 A

B

C

800

20

760

16

740

14

720

12 10

700

8

680

6

Elongation at break (%)


780 18

660 110

120

130 140 150 160 170 Vulcanization temperature (°C)

180

16.45 Changes of tensile strength and elongation at break of IIRCN (IIR:OMC = 100:10) with vulcanization temperature. Copied from Ref.42

intercalants are usually organic amine) is used, the vulcanization system and curing parameters (i.e., curing temperature and pressure) must be carefully designed. (b) The specific OMC for preparing RCNs should be developed, in which the intercalants do not react with CAs (i.e., quaternary amine intercalants are better than primary one) or interact with clay layers very strongly.

16.6

References

1. Zhang L Q, Wu Y P, Wang Y Q, et al., ‘Nanoreinforcing and nanocompounding technique of rubber’, China Synthetic Rubber Industry, 2000 23(2) 71–77. 2. Hamed G R, ‘Reinforcement of rubber’, Rubber Chem Technol, 2000 73(3) 524– 33. 3. Tadmor Z and Gogos C G, Principles of Polymer Processing, Second Edition, New York, Wiley Interscience, 2006. 4. Lin C J, Hergenrother W L, Alexanian E, Böhm G G A, ‘On the filler flocculation in silica-filled rubbers. Part I. Quantifying and tracking the filler flocculation and polymer–filler interactions in the unvulcanized rubber compounds’, Rubber Chem Technol, 2002 75(5) 865–90. 5. Lin C J, Hogan T E, Hergenrother W L, ‘On the filler flocculation in silica-filled rubbers. Part II. Filler flocculation and polymer–filler interaction’, Rubber Chem Technol, 2004 77(1) 90–114. 6. Wang M J, ‘Effect of polymer–filler and filler–filler interactions on dynamic properties of filled vulcanizates’, Rubber Chem Technol, 1998 71(3) 520–89. 7. Wolff S, ‘Optimization of silane–silica OTR compounds. Part 1. Variations of mixing temperature and time during the modification of silica with bis(3triethoxysilylpropyl) tetrasulfide’, Rubber Chem Technol, 1982 55(4) 967–89.

546


8. Goerl U, Hunsche A, Mueller A, Koban H G, ‘Investigations into the silica/silane reaction system’, Rubber Chem Technol, 1997 70(4) 608–23. 9. Ten Brinke J W, Debnath S C, Reuvekamp L A E M, Noordermeer J W M, ‘Mechanistic aspects of the role of coupling agents in silica-rubber composites’, Comp Sci Technol, 2003 63(8) 1165–74. 10. Choi S S, ‘Strain and anisotropic effects on migration behaviors of antiozonants in carbon black-filled natural rubber vulcanizates’, Polymer Testing, 2002 21(7) 741–44. 11. Hasse A, Klockmann O, Wehmeier A, Luginsland H D, ‘Influence of the amount of di- and polysulfane silanes on the crosslinking density of silica-filled rubber compounds’, Kautschuk Gummi Kunststoffe, 2002 55(5) 236–43. 12. Sae-oui P, Thepsuwan U, Hatthapanit K, ‘Effect of curing system on reinforcing efficiency of silane coupling agent’, Polymer Testing, 2004 23(4) 397–403. 13. Wu Y P, Zhao Q S, Zhao S H, Zhang L Q, ‘The influence of in situ modification of silica on filler network and dynamic mechanical properties of silica-filled solution styrene–butadiene rubber’, J Appl Polym Sci, 2008 108(1) 112–18. 14. Zhang Q, Tian M, Wu Y P, Lin G, Zhang L Q, ‘Effect of Particle Size on the Properties of Mg(OH)2-filled rubber composites’, J Appl Poly Sci, 2004 94(6) 2341–46. 15. Wang M J, Tu H, Murphy L and Mahmud K, ‘Carbon/silica dual phase filler, a new generation reinforcing agent for rubber, Part VIII. Surface characterization by IGC’, Rubber Chem Technol 2000 73(4) 666–77. 16. Wang M J, Mahmud K, Murphy L, Patterson W J, ‘Carbon/silica dual phase filler, a new generation reinforcing agent for rubber, Part I. Characterization’, Kautschuk. Gummi Kunststoffe, 1998 51(5) 348–50, 353–60. 17. Wang M J, Kutsovsky Y, Zhang P, Murphy L J, Laube S, Mahmud K, ‘New generation carbon-silica dual phase filler, Part I. Characterization and application to passenger tire’, Rubber Chem Technol, 2002 75(2) 247–63. 18. Wang M J, Zhang P, Mahmud K, ‘Carbon/silica dual phase filler, a new generation reinforcing agent for rubber, Part IX. Application to truck tread compound’, Rubber Chem Technol, 2001 74(1) 124–37. 19. Tian M, Liang W L, Rao G Y, Zhang L Q, Guo C, ‘Surface modification of fibrillar silicate and its reinforcing mechanism on FS/rubber composites’, Comp Sci Technol, 2005 65(7–8) 1129–38. 20. Tian M, Lu Y L, Liang W L, Cheng L J, Zhang L Q, ‘Structure and properties of novel fibril silicate/rubber nanocomposites’, Polym J, 2006 38(11) 1105–13. 21. Karger-Kocsis J, ‘Thermoset rubber/layered silicate nanocomposites. Status and future trends’, Polym Eng Sci, 2004 44(6) 1083–93. 22. Kim J, Oh T, Lee D, ‘Morphology and rheological properties of nanocomposites based on nitrile rubber and organophilic layered silicates’, Polym Int, 2003 52(7) 1203–08. 23. Kim J, Oh T, Lee D, ‘Preparation and characteristics of nitrile rubber (NBR) nanocomposites based on organophilic layered clay’, Polym Int, 2003 52(7) 1058–63. 24. Zheng H, Zhang Y, Peng Z, Zhang Y, ‘Influence of the clay modification and compatibilizer on the structure and mechanical properties of ethylene–propylene– diene rubber/montmorillonite composites’, J Appl Polym Sci, 2004 92(1) 638–46. 25. Schön F, Thomann R, Cronski W, ‘Shear controlled morphology of rubber/organoclay nanocomposites and dynamic mechanical analysis’, Macromol Symp, 2002 189 (Polymer Science Insights) 105–10.


547

26. Nah C, Ryu H J, Kim W D, Chang Y W, ‘Preparation and properties of acrylonitrile– butadiene copolymer hybrid nanocomposites with organoclays’, Polym Int, 2003 52(8) 1359–64. 27. Gatos K G, Sawanis N S, Apostolov A A, Thomann R, Karger-Kocsis J, ‘Nanocomposite formation in hydrogenated nitrile rubber (HNBR)/organomontmorillonite as a function of the intercalant type’, Macromol Mater Eng, 2004 289(12) 1079–86. 28. Teh P L, Mohd Ishak Z A, Hashim A S, Karger-Kocsis J, Ishiaku U S, ‘On the potential of organoclay with respect to conventional fillers (carbon black, silica) for epoxidized natural rubber compatibilized natural rubber vulcanizates’, J Appl Polym Sci, 2004 94(6) 2438–45. 29. Li W, Huang D, Ahmadi S J, ‘Preparation and properties of ethylene–propylene– diene rubber/organomontmorillonite nanocomposites’, J Appl Polym Sci, 2004 94(2) 440–45. 30. Gatos K G, Thomann R, Karger-Kocsis J, ‘Characteristics of ethylene propylene diene monomer rubber/organoclay nanocomposites resulting from different processing conditions and formulations’, Polym Int, 2001 53(11) 1191–97. 31. Kim J T, Lee D Y, Oh T S, Lee D H, ‘Characteristics of nitrile–butadiene rubber layered silicate nanocomposites with silica coupling agent’, J Appl Polym Sci, 2003 89(10) 2633–40. 32. Vaia R A, Giannelis E P, ‘Lattice model of polymer melt intercalation in organicallymodified layered silicates’, Macromolecules, 1997 30(25) 7990–99. 33. Vaia R A, Giannelis E P, ‘Polymer melt intercalation in organically-modified layered silicates: model predictions and experiment’, Macromolecules, 1997 30(25) 8000–09. 34. Liang Y R, Ma J, Lu Y L, Wu Y P, Zhang L Q, Mai Y W, ‘Effects of heat and press on intercalation structures of isobutylene–isoprene rubber/clay nanocomposites. I. Prepared by melt blending’, J Polym Sci: Part B: Polym Phys, 2005 43(19) 2653–64. 35. Varghese S, Karger-Kocsis J, ‘Melt-compounded natural rubber nanocomposites with pristine and organophilic layered silicates of natural and synthetic origin’, J Appl Polym Sci, 2004 91(2) 813–19. 36. Arroyo M, López-Manchado M A, Herrero B, ‘Organo-montmorillonite as substitute of carbon black in natural rubber compounds’, Polymer, 2003 44(8) 2447–53. 37. Varghese S, Karger-Kocsis J, Gatos K G, ‘Melt compounded epoxidized natural rubber/layered silicate nanocomposites: structure–properties relationships’, Polymer, 2003 44(14) 3977–83. 38. Teh P L, Mohd Ishak Z A, Hashim A S, Karger-Kocsis J, Ishiaku U S, ‘Physical properties of natural rubber/organoclay nanocomposites compatibilized with epoxidized natural rubber’, J Appl Polym Sci, 2006 100(2) 1083–92. 39. Mousa A, Karger-Kocsis J, ‘Rheological and thermodynamical behavior of styrene/ butadiene rubber–organoclay nanocomposites’, Macromol Mater Eng, 2001 286(4) 260–66. 40. Wu Y P, Ma Y, Wang Y Q, Zhang L Q, ‘Effects of characteristics of rubber, mixing and vulcanization on the structure and properties of rubber/clay nanocomposites by melt blending’, Macromol Mater Eng, 2004 289(10) 890–94. 41. Liang Y R, Wang Y Q, Wu Y P, Lu Y L, Zhang H F, Zhang L Q, ‘Preparation and properties of isobutylene–isoprene rubber (IIR)/clay nanocomposites’, Polym Testing, 2005 24(1) 12–17.

548 42.

43.

44.

45.

46. 47.

48.

49.

50.

51. 52. 53.

54. 55. 56.

57. 58.

59. 60.

Advances in polymer processing Lu Y L, Li Z, Mao L X, Li Y, Wu Y P, Liang Y R, Zhang L Q, ‘Impact of curing temperature on microstructures and properties of isobutylene–isoprene rubber clay nanocomposites’, J Appl Polym Sci, 2008 110(2) 1034–42. Nah C, Ryu H J, Kim W D, Choi S S, ‘Barrier property of clay/acrylonitrile– butadiene copolymer nanocompsoites’, Polym for Adv Technol, 2002 13(9) 649– 52. Lu Y L, Li Z, Yu Z Z, Tian Ming, Zhang L Q, Mai Y M, ‘Microstructure and properties of highly filled rubber/clay nanocomposites prepared by melt blending’, Comp Sci Technol, 2007 67(14) 2903–13. Joly S, Garnaud G, Ollitrault R, Bokobza L, Mark J E, ‘Organically modified layered silicates as reinforcing fillers for natural rubber’, Chem Mater, 2002 14(10) 4202–08. Zhang L Q, Wang Y Z, Yu D S, Wang Y Q, Sun Z H, Chinese Patent CN 98 101496.8, 1998. Zhang Z J, Zhang L N, Li Y, Xu H D, ‘New fabricate of styrene–butadiene rubber/ montmorillonite nanocomposites by anionic polymerization’, Polymer 2005 46(1) 129–36. Zhang Z J, Zhang L N, Li Y, ‘Effect of the addition of toluene on the structure and properties of styrene–isoprene–butadiene rubber/montmorillonite nanocomposites’, Macromol Mater Eng, 2005 290(5) 430–37. Wu Y P, Jia Q X, Yu D S, Zhang L Q, ‘Structure and properties of nitrile rubber (NBR)-clay nanocomposites by co-coagulating NBR latex and clay aqueous suspension’, J Appl Polym Sci, 2003 89(14) 3855–58. Wang M J, Wang T, Shell J, Mahmud K, ‘NR/carbon black masterbatch produced with continuous liquid phase mixing’, Kautschuk Gummi Kunststoffe, 2002 55(7– 8) 388–96. Costin R, Nagel W, Ekwall R, ‘New metallic coagents for curing elastomers’, Rubber Chem Technol, 1991 64(2) 152–61. Martin F S, Melvin T, Pieronl J K, US Patent 4,266,772, 1981. Tocuchet P, Rodriguez G, Gatza P E, Butler D P, Crawford D, Teets A R, et al., ‘Rubber compound for tracked vehicle track pads’, US Patent 4,843,114, 27 June, 1989. Medalia A, Alesi A L, Mead J L, ‘Pattern abrasion and other mechanisms of wear of tank track pads’, Rubber Chem Technol, 1992 65(1) 154–75. Klingender R C, Oyama M, Satio Y, ‘High-strength compound of highly saturated nitrile and its applications’, Rubber World, 1990 202(3) 26–31. Saito Y, Nishimura K, Asada M, Toyoda A, ‘Polymerization behavior of zinc methacrylate study of zinc methacrylate/rubber/peroxide compounds, Part 2’, J Jpn Rubber Soc, 1994 67 867–72. Nomura A, Takano J, Toyoda A, Saito T, ‘Structural analysis of high strength HNBR/ZDMA composites’, J Jpn Rubber Soc, 1993 66 830–38. Ikeda T, Yamada B, Tsuji M, Sakurai S, ‘In situ copolymerization behavior of zinc dimethacrylate and 2-(N-ethylperfluoro-octanesulphonamido) ethyl acrylate in hydrogenated nitrile–butadiene rubber during peroxide crosslinking’, Polym Int, 1999 48 446–54. Wu S M, Lu Y L, Gu L, Zhang L Q, ‘In-situ polymerized nano-PZDMA/POE composite’, China Rubber Ind, 2002 49 709–15. Lu Y L, Liu L, Shen D Y, Yang C, Zhang L Q, ‘Infrared study on in situ polymerization of zinc dimethacrylate in poly(α-octylene-co-ethylene) elastomer’, Polym Int, 2004 53 802–08.

Processing of macro, micro and nanocomposites 61. 62. 63.

64. 65.

66.

67. 68. 69. 70.

71. 72.

73. 74.

75. 76.

77.

78. 79.

80.

549

Zhao Y, Lu Y L, Liu L, Feng Y X, Zhang L Q, ‘ZDMA/NBR nano-micro composite I’, China Rubber Synth Ind, 2001 24 350–53. Zhao Y, Lu Y L, Liu L, Feng Y X, Zhang L Q, ‘ZDMA/NBR nano-micro composite II’, China Rubber Synth Ind, 2002 25 35–38. Lu Y L, Liu L, Yang C, Tian M, Zhang L Q, ‘The morphology of zinc dimethacrylate reinforced elastomers investigated by SEM and TEM’, Eur Polym J, 2005 41 577– 88. Lu Y L, Liu L, Ming Tian, Geng H P, Zhang L Q, ‘Study on mechanical properties of elastomers reinforced by zinc dimethacrylate’, Eur Polym J, 2005 41 589–98. Ikeda T, Yamada B, ‘Simulation of the in situ copolymerization of zinc methacrylate and 2-(N-ethylperfluoro-octanesulphonamido)ethyl acrylate in hydrogenated nitrile– butadiene rubber’, Polym Int, 1999 48 367–72. Yuan X H, Peng Z L, Zhang Y, Zhang Y X, ‘In situ preparation of zinc salts of unsaturated carboxylic acids to reinforce NBR’, J Appl Polym Sci, 2000 77 2740– 48. Peng Z L, Liang X, Zhang Y X, Zhang Y, ‘Reinforcement of EPDM by in situ prepared zinc dimethacrylate’, J Appl Polym Sci, 2002 84 1339–45. Sato K, ‘Ionic crosslinking of carboxylated SBR’, Rubb Chem Technol, 1983 56 943–58. Eisenberg A, Hird B, Moore R B, ‘A new muliplet-cluster model for the morphology of random ionomers’, Macromolecules, 1990 23(18/26) 4098–4107. Granatowicz D S, Morris M T, Pilkington M V, Tompkin G R, ‘High modulus belt composition and belts made therewith’, European Patent 0864607 A1, 16 September, 1998. Keisaku Y, Kiyoshi I, Junichi K, ‘Rubber composition’, European Patent 0589701 A1, 30 March, 1994. Du A H, Peng Z L, Zhang Y, Zhang Y X, ‘Polymerization conversion and structure of magnesium methacrylate in ethylene–vinyl acetate rubber vulcanizates’, J Appl Polym Sci, 2004 93 2379–84. Yuan X H, Zhang Y, Peng Z L, Zhang Y X, ‘In situ preparation of magnesium methacrylate to reinforce NBR’, J Appl Polym Sci, 2002 84 1403–08. Yin D H, Zhang Y, Zhang Y X, Peng Z L, Fan Y Z, Sun K, ‘Reinforcement of peroxide-cured styrene–butadiene rubber vulcanizates by methacrylic acid and magnesium oxide’, J Appl Polym Sci, 2002 85 2667–76. Du A H, Peng Z L, Zhang Y, Zhang Y X, ‘Properties of EVM vulcanizates reinforced by in situ prepared sodium methacrylate’, J Appl Polym Sci, 2003 89 2192–2200. Du A H, Peng Z L, Zhang Y, Zhang Y H, ‘Fracture morphology and mechanical properties of ethylene/vinyl acetate rubber vulcanizates reinforced by in situ prepared sodium methacrylate’, J Polym Sci, Part B: Polym Phys, 2004 42 1715–24. Ren W T, Peng Z L, Zhang Y, Zhang Y X, ‘Water-swelling elastomer prepared by in situ formed lithium acrylate in chlorinated polyethylene’, J Appl Polym Sci, 2004 92 1804–12. Costin R, Ekwall R, Naget W, ‘Metallic coagents for curing elastomers’, Rubber World, 1992 August 23–30. Liu L, Lu Y-L, He L, Zhang W, Yang C, Liu Y-D, Zhang L-Q, Jin R-G, ‘Novel europium-complex/nitrile–butadiene rubber composites’, Adv Funct Mater, 2005 15 309. Liu L, He L, Yang C, Zhang W, Jin R-G, Zhang L-Q, ‘In situ reaction and radiation protection properties of Gd(AA)3/NR composites, Macromol Rapid Commun, 2004 25 1197–1202.

550 81.

82. 83. 84. 85.

86.

87.

88. 89. 90. 91.

92. 93. 94. 95.

96.

97.

98. 99. 100.

101.

Advances in polymer processing Liu L, Zhang X J, Jing R G, Yang C, He L, Zhang W, Zhang L Q, ‘Studies on the preparation of Sm(AA)3/NBR composites by in situ reaction method and their fluorescent properties’, Chem J Chinese Univ, 2004 25 570–74. Ikeda T, Yamada B, ‘Copolymerization of zinc methacrylate and perfluoroalkyl acrylates in different solvents’, J Appl Polym Sci, 1999 71 1499–1504. Dannenberg E M, ‘Bound rubber and carbon black reinforcement’, Rubber Chem Technol, 1986 59 512–20. Leblanc J L, Stragliati B, ‘An extraction kinetics method to study the morphology of carbon black filled rubber compounds’, J Appl Polym Sci, 1997 63 956–70. Leblanc J L, Evo C, Lionnet R, ‘Composite design experiments to study the relationships between the mixing behavior and rheological properties of SBR compounds’, Kautschuk Gummi Kunststaffe, 1994 40 401. Leblanc J L, Staelraeve A, ‘Studying the storage maturation of freshly mixed rubber compounds and its effects on processing properties’, J Appl Polym Sci, 1994 53 1025–35. (a) Cohen-Addad J P, Touzet S, ‘Poly(dimethylsiloxane)–silica mixtures: intermediate states of adsorption and swelling properties’, Polymer 1993 34(16) 3490-98; (b) Cohen-Addad J P, Frebourg P, ‘Gel-like behaviour of polybutadiene/carbon black mixtures: n.m.r. and swelling properties’, Polymer 1996 37(19) 4235–42. Leblanc J L, ‘A molecular explanation for the origin of bound rubber in carbon black filled rubber compounds’, J Appl Polym Sci, 1997 66 2257–68. Wang M J, Wolff S, Tan E H, Rubber Chem Technol, 1993 66 178. Donnet J B, Custodero E, Carbon, 1992 30 813. Choi S-S, ‘Influence of storage time and temperature and silane coupling agent on bound rubber formation in filled styrene–butadiene rubber compounds’, Polym Testing, 2002 21 201–08. Wang M J, ‘Effect of polymer–filler and filler–filler interactions on dynamic properties of filled vulcanizates’, Rubber Chem Technol, 1998 71 520. Fletcher W P, Gent A N, ‘Non-linearity in the dynamic properties of vulcanized rubber compounds’, Trans Inst Rubber Ind, 1953 29 266–80. Payne A R, ‘The dynamic properties of carbon black-loaded natural rubber vulcanizates, Part I’. J Appl Polym Sci, 1962 6 53–57. Lion A, Kardelky C, Haupt P, ‘On the frequency and amplitude dependence of the Payne effect: theory and experiments’, Rubber Chem Technol, 2003 76 533–47. Böhm G G A, Nguyen M N, ‘Flocculation of carbon black in filled rubber compounds. I. Flocculation occurring in unvulcanized compounds during annealing at elevated temeperatures’, J Appl Polym Sci, 1995 55 1041–50. Gerspacher M, Nikel L, Yang H H, O’Farell C O, Worth F, Schwartz G A, Aires B, ‘Flocculation in carbon black filled rubber compounds’, Kautschuk Gummi Kunststoffe, 2002 55(11) 596–604. LeBaron P C, Pinnavaia T J, ‘Clay nanolayer reinforcement of a silicone elastomer’, Chem Mater, 2001 13 3760–65. Usuki A, Tukigas A, Kato M, ‘Preparation and properties of EPDM–clay hybrids’, Polymer, 2002 43(8/26) 2185–89. Gatos K G, Thomann R, Karger-Kocsis J, ‘Characteristics of ethylene propylene diene monomer rubber/organoclay nanocomposites resulting from different processing conditions and formulations’, Polym Int, 2004 53 1191–97. Gatos K G, Karger-Kocsis J, ‘Effects of primary and quaternary amine intercalants


551

on the organoclay dispersion in a sulfur-cured EPDM rubber’, Polymer, 2005 46(9/26) 3069–76. 102. Gatos K G, Sawanis N S, Apostolov A A, Thomann R, Karger-Kocsis J, ‘Nanocomposite formation in hydrogenated nitrile rubber (HNBR)/organomontmorillonite as a function of the intercalant type’, Macromol Mater Eng, 2004 289 1079–86. 103. Gatos K G, Százdi L, Pukánszky B, Karger-Kocsis J, ‘Controlling the deintercalation in hydrogenated nitrile rubber (HNBR)/organo-montmorillonite nanocomposites by curing with peroxide’, Macromol Rapid Commun, 2005 26 915–19. 104. Liang Y-R, Lu Y-L, Wu Y-P, Yong M, Zhang L-Q, ‘Pressure, the critical factor governing final microstructures of cured rubber/clay nanocomposites’, Macromol Rapid Commun, 2005 26 926–31. 105. Lu Y L, Liang Y R, Wu Y P, Zhang L Q, ‘The determining effect of the pressure applied in vulcanization on microstructures and properties of IIR/clay nanocomposites’, National Conference for Polymer Science of China Symposium, Beijing, 2005, C-O-1630/p260. 106. Lu Y L, Liang Y R, Wu Y P, Zhang L Q, ‘Effects of heat and pressure on microstructures of isobutylene–isoprene rubber (IIR)/clay nanocomposites’, Macromol Mater Eng, 2006 291 27–36.

17 Micromolding of polymers D Y A O, Georgia Institute of Technology, USA

Abstract: In the past three decades, much effort has been made worldwide in the adaptation of traditional molding techniques into microfabrication, resulting in the emerging micromolding technology. This recent trend in technological development is primarily driven by the increasing demand for net-shape microcomponents in the fast growing biomedical, biochemical, electronics, and telecommunication industries. Micromolding processes are replication-based processes, involving feature-transferring mechanisms, e.g., injection, compression, casting and intrusion, for replicating the master pattern on a mold to a plastic material. As in conventional molding, the material in micromolding is deformed in its liquid or semi-liquid state inside a mold and subsequently solidifies to lock the geometry. With micromolding techniques, microsized parts or parts containing microfeatures can be fabricated in a cost-effective way with a feature-transfer resolution down to tens of nanometers, and a variety of engineering materials including polymers, metallic alloys, and ceramics can be processed. This chapter provides an overview on the state of the art in the micromolding technology, focusing on common micromolding techniques, particularly injection- and embossing-based micromolding processes. Due to the so-called size effects, some proven designs and processes in conventional molding may fail in micromolding, and thus rethinking is often needed. Hence, the focus of this chapter is placed on describing the differences between these new micromolding processes and their macrocounterparts and the rationales for the necessary modifications and improvements. Key words: micromolding, polymer microfabrication, injection molding, hot embossing.

17.1

Introduction

Miniaturization is becoming a consistent trend in technological development, particularly in the electronics, optics, telecommunication and biomedical fields. In the microelectromechanical systems (MEMS) industry only, a twofold increase in market size from 2005 to 2009 has been predicted (Nexus, 2005). It is well known that a smaller device uses less material and consumes less energy. More importantly, smaller devices can perform better. For example, a smaller actuator appears more agile and can access useful surface forces, such as surface tension and electrostatic forces. By utilizing the scaling benefits, innovative microdevices with better performance and functionality than their macrocounterparts can be designed. At present, most microdevices and microsystems are made of silicon and 552


553

its derivatives and fabricated using lithographic methods; however, the use of polymers for microfabrication is growing rapidly. The advantages of using polymeric materials as a replacement for silicon and its derivatives are evident considering their versatile properties and mass-production capabilities. There are more than several thousand different types of polymeric materials available for designers to choose. Their properties are versatile, and different mechanical, optical and electrical functions can be achieved with polymeric materials. Opposite material properties, e.g., insulative versus conductive, hydrophilic versus hydrophobic, and opaque versus transparent, may be achieved with appropriate design and selection of polymers. The recent developments in dopable semiconductive polymers, piezoelectric polymers, and other functional polymers, not only allow polymers to be used as a substitute for silicon in electronics and actuators, but also open up new applications that could not be realized before. According to Rotting et al. (2002), the technologies used in polymer microfabrication can be divided into two groups: direct techniques, where each device is manufactured separately, and replication techniques, where a large number of parts are produced by replicating a single master. While direct methods such as lithography, laser ablation and micromilling have been used mainly for prototyping purposes, most research and development activities in polymer microfabrication have been focused on replication methods. The three most promising replication methods are microinjection molding, hot embossing, and casting (Heckele and Schomburg, 2004; Mekaru et al., 2004; Rotting et al., 2002). Because they were adapted from conventional polymer molding processes, these replication methods are also referred to as micromolding in the literature (Heckele and Schomburg, 2004). These micromolding processes are able to deliver a replication resolution down to 10 nanometers (Gross, 2006; Heckele and Schomburg, 2004). This is indeed quite surprising, considering the relatively large size of the polymer molecule. It is generally believed that the macromolecule with a similar size as that of the feature tends to adapt to the form of the mold. So far, micromolded parts have shown great commercial potential for a variety of applications, including microfluidics, diffractive optics, LCD display panels, sensors, actuators, allpolymer electronics, and many others (Becker and Locascio, 2002; Gates et al., 2005; Giboz et al., 2007; Heckele and Schomburg, 2004; Rotting et al., 2002). Although most micromolding papers were published in the past ten years, some earlier work dates back to the 1970s (Bartolini et al., 1970; Gale et al., 1978; Hannan et al., 1973; Ulrich et al., 1972). While initial investigations were directed toward testifying the ability of conventional molding processes for producing microstructured surfaces, the recent endeavors are more focused on process development and optimization related to processing, tooling and materials. It should be acknowledged that micromolding benefits greatly

554


from the broad knowledge base supported by the amount of expertise in the traditional polymer processing industry, which has led to standardized process sequences, a high level of automation, and short cycle times, as well as computer aided engineering (Bauer et al., 1999). However, a general consensus has also been reached in the micromolding community that rethinking is often required in order to successfully adapt a macroscale molding process for microscale applications. Because of the so-called size effects, the consideration of process setup, tooling, material structuring, and simulation in micromolding is quite different from that in conventional molding. Such size effects may be related to the change of rheological properties of the polymer liquid at confined geometry (Yao and Kim, 2002a) or may be simply caused by the different influences of different physical phenomena in a different size domain. For example, diffusion and surface effects represent slow and weak processes in macroscale but may dominate in microscale. Therefore, from materials to machines, from tooling to process setup, and from molding to part handling, micromolding inherently presents a unique set of technical problems which must be distinctively addressed for successful applications. In their classical polymer processing textbook, Tadmor and Gogos (1979, 2006) defined molding and casting as the same category of processes which comprise all the different ways for stuffing molds with thermoplastics or thermosetting polymers. The major deformation mechanism in classical molding is shear flow. In micromolding, the definition of molding needs to be relaxed. The geometry involved in microparts is typically more threedimensional, and different deformation mechanisms including shear, compression and elongation may be involved during processing. In fact, some newly developed micromolding techniques are hybrid in nature. For example, hot embossing of a microshell pattern with a rubber pad as a soft countertool involves film stretching (Nagarajan and Yao, 2007), the major deformation mode used in traditional stretch forming processes, e.g., thermoforming. Another example is the hybrid embossing and intrusion process (Pan et al., 2004), where a polymeric liquid is compression molded and undergoes die flow in an open channel/cavity. Also, the material in micromolding is often deformed in a semisolid or semiliquid state where complex rheological behaviors are involved. For example, hot embossing is sometimes operated at a temperature only slightly above the glass transition temperature for an amorphous polymer and, for a semicrystalline polymer, the processing temperature can be slightly below the melting temperature. At such a meso-state, the polymer is in reality not a simple liquid. Furthermore, the plastication process may not be thermally induced as used for thermoplastic polymers. In some cases, radiation-based curing methods are more useful. Particularly, UV-curable resins are widely used in micromolding. In addition, other softening and/or solidification processes,


555

including solvent-assisted plastication (Khang and Lee, 2000; Wang et al., 2005) and isothermal softening and solidification of a slowly crystallizing polymer (Yao et al., 2007a), have also been reported in micromolding applications. To better understand the scope of micromolding, one may consider the three different types of microstructures commonly concerned in micromolding: surface microstructures, shell microstructures, and discrete microparts, as shown in Fig. 17.1. For surface microstructures, the characteristic size of the microstructure is much smaller than the substrate thickness. During patterning, surface microstructures undergo localized deformation at the surface of the substrate. In contrast, shell microstructures experience deformation over dimensions larger than the film thickness, and the patterned film is marked by shell-type geometry with a relatively uniform shell thickness. In addition, discrete 3-D microparts (Fig. 17.1c) can be produced by micromolding techniques, particularly with microinjection molding. These microparts have an overall dimension smaller than a millimeter and a weight of a few milligrams or smaller. It should be noted that different manufacturing techniques may be needed to produce these different microstructures. The micromolding community has now reached a general agreement regarding the definition of micromolding: micromolding involves molding of microparts and parts with microstructures or even nanostructures. Note that this definition does not exclude the inclusion of submicrofeatures and nanofeatures with a characteristic

(a)

(b)

(c)

17.1 Three different types of microstructures typically involved in micromolding: (a) surface microstructures, (b) shell microstructures, (c) discrete microparts.

556


feature size smaller than a micron. Some other authors (Ruprecht et al., 2002; Whiteside et al., 2003) also proposed that the scope of micromolding should further include precision molding of parts with dimensional tolerances in the micrometer range but without dimensional limit. This chapter provides an overview on the state of the art in micromolding technology, highlighting common micromolding techniques, particularly injection- and embossing-based micromolding processes. Since micromolding presents a new set of problems regarding materials, machinery, tooling and processing, the focus of this chapter is placed on describing the differences between micromolding processes and their macrocounterparts and the rationales for the necessary modifications and improvements.

17.2

Classification of micromolding processes

Micromolding has been a multidisciplinary field; the development of this field has been contributed by researchers with diverse backgrounds in polymer engineering, physics, chemistry, electronics, and biomedical and biochemical engineering. This diversity has allowed different expertise from different areas to be cross-fertilized on the same platform of micromolding, thus enabling the development of innovative processes. However, because of the multidisciplinary nature, different terminologies have been communicated by different researchers. One objective of this chapter is thus to classify existing micromolding processes based on their common features from different perspectives. Understanding of a group of similar processes would facilitate the development of new processes in a similar cluster. As in conventional molding processes, the micromolding material undergoes coupled mechanical and thermal influences in liquid, rubbery and solid states. The resulting thermomechanical history determines the structure and the state of stresses and thus the properties and performance of the micromolded part. As in conventional molding, micromolding mainly uses three mechanisms for applying the molding forces, namely injection, compression and casting. In the injection mode, the process is called microinjection molding, similar to conventional injection molding, but with modified tooling and process setup. The hot embossing process, operated in the compression mode, is basically a hot-mold compression molding process. It is used not only for fabricating microparts and microstructured surfaces, but also as a lithographic step for patterning resist-coated silicon/glass wafers. The latter case is often referred to as imprinting, e.g., nanoimprinting, because the patterned aspect ratio is very small (less than one). In the casting mode, low viscosity resins, e.g. monomers for poly(dimethylsiloxane) (PDMS), are used. Because of the diminishing gravitational forces in microscale, microcasting is often assisted by vacuum or pressurized gases. Besides the three major mechanisms, micromolding also uses microscale surface forces to deform the molding


557

material. One widely used surface force is the surface tension force. For example, casting can be facilitated by the capillary effect. More interestingly, surface tension has been used as reflow force for creating smoother surfaces, e.g. smoother microchannels (Shiu et al., 2008), and curved surfaces, e.g. microlens arrays (Pan et al., 2004; Yang et al., 2004). From materials perspectives, the solidification mechanism of the molding polymer notably affects the design and setup of the molding process. Micromolding processes may thus be categorized based on different solidification mechanisms. The three major solidification mechanisms in conventional molding, namely thermoplastic (either by vitrification for amorphous polymers or by crystallization for semicrystalline polymers), reactive (with monomers or oligomers), and thermosetting (with prepolymers), also dominate in micromolding. However, other mechanisms that cannot be realized in conventional molding may now work for micromolding. Given the small amount of polymer needed by a micropart or microstructured film and thus the easy removal of the solvent by drying or coagulation, polymer solutions can become suitable materials for selective micromolding applications. For example, a polymer solution may be spin cast onto a microstructured surface to form a patterned film. A thin solid polymer film can also be plasticized by absorbing solvent vapors and then re-solidified by extracting the solvent (Khang and Lee, 2000; Wang et al., 2005). With regard to the types of thermal control, micromolding can be grouped into variothermal processes and constant-temperature processes. For thermoplastic micromolding, it is an established fact that a high mold temperature close to the softening temperature of the polymer is required, particularly for molding high-aspect-ratio microstructures (Despa et al., 1999; Wimberger-Friedl, 2000; Yao and Kim, 2002b). This thermal cycling may be eliminated by employing a non-thermoplastic plastication method, e.g. a solvent vapor-assisted method. Yao et al. (2007a) recently showed that constanttemperature embossing can be achieved for slowly crystallizing polymers by embossing the polymer at its amorphous phase and subsequently crystallizing the polymer at the same mold temperature. For reactive and thermosetting polymers, constant-temperature or even room-temperature micromolding can be performed by employing a non-thermal curing method, e.g., by UV radiation. In conventional molding, the mold is typically made of a steel material. Although a hard and durable mold material is desired in micromolding, fabrication of a metallic micromold with well-defined microstructures or even nanostructures is costly and challenging. Therefore, silicon-based lithographic materials are sometimes directly used as tooling materials in micromolding. However, silicon is rather brittle and thus may only be used for prototyping purposes. To produce multiple molds, the silicon structures can be replicated by casting to a relatively softer but tougher material, e.g. acrylic, epoxy, and PDMS (Malek et al., 2007). These soft tools are not hard

558


and durable enough for microinjection molding but are well suitable for low force processes, e.g. casting. In particular, PDMS has been widely used as a tooling material for soft molding (Michel et al., 2001; Xia and Whitesides, 1998). In terms of these different tooling materials, micromolding processes may be separated into two groups: hard mold processes and soft mold processes. It should be further noted that production volume is an important consideration in process selection. Therefore, it may be reasonable to divide micromolding processes into three categories: prototyping, batch, and volume. From prototyping to volume production, there is an increasing demand for more durable tools and more reliable processes.

17.3

Microinjection molding

The typical stages involved in microinjection molding are mold closing, injection, holding, cooling, plastication, mold opening, and part ejection. Note that some of these stages, such as cooling and plastication, may occur concurrently. After the mold closes, a ram or piston is used to inject a plasticized material into a sealed mold, preferably air vacuumed before injection. The speed control mode is then switched to a pressure control mode. The pressure controlled holding stage lasts until the gate freezes. The machine then plasticates a prescribed amount of new material for preparation of the next shot. At the same time, cooling continues until a set ejection temperature is reached. The mold is then opened, and the molded part is ejected. Figure 17.2 only illustrates molding of surface microstructures. The technology for molding microparts is considered to be similar. Like the

(a)

(b)

(c)

(d)

17.2 Typical stages involved in microinjection molding: (a) mold closing, (b) injection and holding, (c) cooling and plastication, (d) mold opening and part ejection.


559

substrate for hosting surface microstructures, runners and gates are used to connect the microcavities. The runners are typically much larger than an individual microcavity and are separated from the molded microparts by degating. Note that the molding stages involved in microinjection molding are almost the same as in conventional injection molding. However, due to the presence of microscale or even nanoscale features, different considerations on machine and process setup are needed in order to achieve the necessary feature replication.

17.3.1 Microinjection molding machines As compared to a bulkier injection molding machine used in conventional molding, the following characteristics are desired for a microinjection molding machine: (1) accurate metering or dosing, (2) small shot size, (3) high injection rate, (4) short response time, (5) small but accurate clamping force, and (6) excellent stability and repeatability. Microinjection molding machines typically employ servo motors with precision ball bearings to achieve accurate movement of the injection plunger. There are mainly four types of injection units used: (a) reciprocating type, (b) screw-plunger type, (c) screw-plunger-plunger type, and (d) plunger type. For a relatively large shot size, e.g., around 5 grams, a reciprocating injection unit works well. This type of injection unit can be used in molding of microstructured parts with a relative large substrate or a plurality of microparts at a single shot. At smaller shot sizes, a separate plunger for dosing and injection can be used. In this case, the screw is only used for plastication. However, it is difficult to accurately control the dose when a single plunger is used both for dosing and for injection. An improved design employs two separate plungers, one for dosing and one for injection. This screw-plunger-plunger type of injection unit is currently adopted in some well-known commercial microinjection molding machines, e.g., Battenfeld Microsystem 50. For prototype microinjection molding, singleplunger screwless machines can be used. During each shot, a premeasured amount of material is charged into the plunger chamber, softened by conduction heating (often assisted by compression and spreading), and then injected into the mold. Efforts were also made to develop new plastication and injection mechanisms for microinjection molding, e.g., ultrasonic plastication (Michaeli et al., 2004) and impact injection molding (Nian and Yang, 2005). For more information on microinjection molding machines, the reader is referred to recent articles by Chang et al. (2007) and Giboz et al. (2007).

17.3.2 Rapid thermal cycling of molds Macrosized parts with low aspect ratio microstructures, e.g. compact disks, can be molded using conventional molding processes without significant

560


modification of the tooling and the process. As the aspect ratio increases, both mold filling and feature ejection become more difficult. One particular obstacle in microinjection molding comes from the difficulty in filling highaspect-ratio microstructures. A conventional mold temperature significantly below the softening temperature of the material would cause a premature freezing problem; the polymer melt would prematurely solidify before the full feature depth could be filled. This molding difficulty can be reduced by increasing the mold temperature. This, however, may result in substantially increased or intolerably long cycle time. To resolve the conflict, one needs a mold with a rapid heating and cooling capability. An elevated mold temperature close to or above the polymer softening temperature is used for mold filling, and a cold mold temperature is used for cooling. Different names for the technology have been coined in the literature, e.g. variothermal processes (Michaeli and Spennemann, 2000; Schinkothe and Walther, 2000; Weber and Ehrfeld, 1999), rapid thermal response molding (Yao et al., 2006), dynamic mold temperature control (Chen et al., 2006a), etc., but all deal with rapidly heating and cooling the mold. Since repeatedly heating and cooling a relatively massive mold mass takes considerable time and energy, a means for rapidly heating only the mold surface prior to the injection stage is desirable. To facilitate this, the thermal mass at the mold surface needs to be reduced. This can be achieved by thermally insulating the surface layer from the bulk of the mold by a layer of thermal insulation. At the same time, thermal mismatching between the surface portion and the remaining mold should be minimized. Although a considerable amount of work (e.g., Kim and Suh, 1986; Jansen and Flaman, 1994; Yao and Kim, 2002c) in mold rapid heating dealt with a multilayer mold design involving heterogeneous materials, recent developments (Chen et al., 2006b; Yao et al., 2006) have been more focused on the use of a single metallic material for mold construction. In particular, Xu et al. (2001) proposed a single metallic mold design with a low thermal inertia by employing air pockets inside the mold. These air pockets function as thermal insulations, and thus a separate thermal barrier layer made of a solid dielectric material can be eliminated. The methods used for rapid mold heating basically fall into three categories: (1) electrical resistive heating, (2) convective heating, and (3) radiation heating. When electrical resistive heating is used, an electrical current is directed to flow at the mold surface. Confinement of the current density at the surface of a bulky metallic mold can be achieved by known methods such as induction heating and proximity heating. In the case of induction heating, an electrical coil passing high-frequency current is placed near the mold surface to induce eddy current. Because of the so-called skin effect (Brown, 1947) at high frequency, the Joule heating is confined at the mold surface. At present, the induction heating method is probably the most popular method for rapid


561

mold heating in microinjection molding (Chen et al., 2006b; Michaeli and Spennemann, 2000; Schinkothe and Walther, 2000; Weber and Ehrfeld, 1999). The presence of the external coil as used in induction heating can be eliminated using the proximity heating method (Yao et al., 2006). In proximity heating, the facing mold halves serve as a coil. The elimination of the external coil allows this method to be useful for in situ mold temperature control after the mold is closed. Convective mold heating with oil is an established practice in conventional injection molding. Convective heating is usually much slower than surface resistive heating because convective heating is energy limited while the power input in resistive heating can be easily changed. In conventional molding, electrical cartridge heaters are also used for mold heating. Like oil heating, cartridge heating is a slow heating method. However, the performance of these heating methods may be improved in microinjection molding applications, given the smaller size of the mold. Smart engineering designs may be developed for enhancing the productivity of these methods for heating smaller molds. Radiation heating is typically not used in conventional injection molding but represents a common practice in some other polymer processes such as thermoforming. In microinjection molding, infrared radiation has been used for heating relatively small mold inserts (Chang and Hwang, 2006; Yu et al., 2007). For this application, infrared lamps and a transparent window are incorporated in the mold design. The above description covers the basic methods used in rapid mold heating. Compared with rapid heating, rapid mold cooling is relatively easier to achieve as long as a low thermal mass is presented. When the energy from heating is confined at the mold surface, the total energy to be taken away by cooling is minimal. Therefore, a rapid heating capability of the mold typically also infers a rapid cooling capability. For such a mold, rapid cooling can be simply achieved in a conventional way, e.g., by circulating water in the mold base. When air pockets or conformal air channels are used as thermal insulation near the mold surface, the cooling medium can be directed into these air voids during cooling, thus enhancing the cooling performance (Yao et al., 2006).

17.3.3 Processing strategies for microinjection molding Microinjection molding frequently involves a large variation in flow thickness, particularly for microstructured articles, where microstructures are placed on the surface of a considerably thick substrate. This causes race tracking among different flow fronts with different flow thicknesses. As a result, the material fills the thick section first, and the flow may hesitate at the entrance to the microstructures, as shown in Fig. 17.3. If the mold should be cold and the hesitation time should be longer than the freezing time of the polymer, short shot of microstructures would occur.

562


Hesitation

17.3 Race tracking of flow fronts caused by a vast difference in flow thicknesses.

One feasible processing strategy for alleviating the premature freezing problem is to increase the injection rate. The higher injection speed reduces the contact time between the polymer melt and the mold. Additionally, the amount of viscous dissipation increases at the higher injection speed, thus further offsetting the unwanted cooling effect. According to numerical simulations (Zhao et al., 2003) and experimental measurements (Whiteside et al., 2003), the shear rate in high-speed microinjection molding could reach as high as 106 s–1, more than an order of magnitude higher than that used in conventional injection molding. Such high shear rates actually exceed the endurance limit of most thermoplastic polymers. However, molecular characterization of microinjection molded polyoxymethylene at such high shear rates did not show a significant reduction in molecular weight (Whiteside et al., 2003). This finding is somewhat surprising, but may be attributed to the rapid cooling effects occurring as the material fills the cavity. It should be noted that the rheological properties of polymer melts including the endurance limit are typically characterized at a much longer time scale. Further, the measured shear rate is an apparent shear rate; the actual shear rate could be significantly lower due to the possible presence of wall slip at an extremely high injection speed. There are some limitations associated with the high speed injection method. First, due to the inertia of the plunger, there is a response or delay time in the speed buildup process. In microinjection molding, this response time could be an important portion in the entire injection stage. Second, accurate control of the switchover from speed control to pressure control is difficult for a small shot size, again because of the inertia effect of the ram at a high speed. This could result in inconsistent part quality. When microstructured thick substrates are molded, the microstructures are actually filled not during the speed-controlled injection stage but during the pressure-controlled holding stage. In the entire speed-controlled injection stage, the material mainly fills the substrate. Therefore, if designed appropriately, a pressure control scheme starting from the beginning of the injection stage may be advantageous in microinjection molding applications. A commercially available process called


563

X-melt, developed by Engel Machinery (Herlihy, 2007), uses the energy stored in the polymer melt to force the polymer melt into the mold cavity. The process sequence is shown in Fig. 17.4. Simply put, the polymer melt is compressed to build up the pressure. When the shutoff valve is opened, the statically pressurized polymer melt expands into the mold cavity, driven by the energy stored in the melt. Due to this characteristic, the process may be named expansion injection molding, but one should not confuse this process with expansion processes for foaming, involving internal gas pressure in the polymer melt. The main advantage of this method over the high speed method is that no injection ram is used during the injection stage and thus the inertia effect is minimal. Further, the highest injection pressure is attained at the beginning of the injection stage. This allows the polymer to flow at the highest speed at the beginning of the injection stage, thus most effectively suppressing the premature freezing problem. Note that in the speed control scheme, the injection pressure is near zero at the startup. Because of the limited compressibility of the polymer melt, a possible limitation of the expansion molding process is the large cushion material used and thus an increased potential for thermal degradation. The above discussion is rooted in the use of a cold mold in microinjection molding. When a mold with a rapid heating and cooling capability is available and the mold can be rapidly heated to the polymer softening temperature prior to the injection stage, the premature freezing problem can be eliminated. When a near-isothermal molding condition is used during filling, Yao and Kim (2002d) proposed a different injection strategy. Rather than increasing the injection speed, they proposed using a low speed filling strategy in hot mold filling. At lower speeds, lower pressure is involved, and the molecular Shutoff valve

(a)

(b)

(c)

17.4 Stages involved in expansion injection molding: (a) dosing, (b) compression, (c) expansion injection.

564


orientations can be reduced. Further, filling of cavities with different characteristic thicknesses becomes more scalable. This helps reduce hesitation and race tracking in the mold cavity.

17.4

Hot embossing

The standard hot embossing process is essentially an open-die compression molding process, involving several sequential steps, as shown in Fig. 17.5. First, a preheated thermoplastic film is placed between two heated mold platens with temperature above the softening temperature of the polymer. The elevated mold temperature is considered necessary for pattern transfer because a cold mold would result in premature freezing of the polymer. Next, the polymer film is pressurized and shaped by closure of the mold and the microfeatures on one or both of the platens transfer to the polymer film. Finally, the entire embossing, including the polymer and the mold, is cooled below the polymer softening temperature, and the platens are separated for the removal of the embossed film. Due to its simplicity in tooling and process setup, hot embossing has been widely used in polymer micro/nanofabrication. Another advantage over microinjection molding is the low stress developed in the process. In compression molding, the polymer melt flows a much shorter distance than in injection molding. The reduction in flow stresses helps improve the dimensional accuracy and stability of the molded part, and at the same time protects the tool from damage caused by large strains. As a result, relatively brittle and/or soft materials, e.g. glass, silicon, or even rubber, may be used as mold materials. While enjoying the benefits from its simple process and tooling setup, the hot embossing process is subjected to some processing difficulties. Because of the open-die setup, significant lateral flow may occur when high embossing pressure is needed for embossing high-aspect-ratio microstructures, resulting in a large reduction of the substrate thickness. The actual thickness of the

Heating

Cold Stamp

Heated

Cooling

Heated

Cooling

Film Heating

(a)

(b)

(c)

Cold

(d)

17.5 Sequential stages involved in hot embossing: (a) preheating, (b) embossing, (c) holding and cooling, (d) ejection.


565

embossed film typically ranges from 20 µm to 200 µm, depending on the process parameters, the polymer material, and the geometry of the mold (Worgull et al., 2006). When the target thickness is larger than this range, high embossing pressure is difficult to build up. For this reason, thicker parts with high-aspect-ratio microfeatures are often produced by microinjection molding. Furthermore, the standard hot embossing process is designed for replicating surface features onto polymer substrates. Shell patterns and discrete features (e.g., microgears, waveguides, micro-through-holes, among others) are difficult to fabricate. Some embossing-based process variants have been developed for addressing the above limitations and enhancing the performance and productivity of the standard hot embossing process. A brief account of these new developments is given in the following sections.

17.4.1 Efficient thermal cycling Thermal cycling of mold temperature is a built-in capability in hot embossing. Unlike injection molding, hot embossing mainly relies on a hot mold for feature transfer. Other strategies such as high speed injection and expansion molding as used in microinjection molding are difficult to implement in hot embossing. As a result, the productivity of the hot embossing process hinges largely on the efficiency of the thermal cycling. The three main types of heating approaches, namely resistivity heating, convective heating and radiation heating, as used in microinjection molding have also been investigated in hot embossing and have greatly enhanced productivity. In particular, Kimberling et al. (2006) implemented the proximity heating method in hot embossing and reduced the standard cycle time of several minutes or above to below 10 s. Additionally, due to the simplified tool design, some other methods that cannot work in injection molding may now become feasible in hot embossing. A fluid-based embossing process has been developed by Chang and Yang (2005), where a fluid (steam, gas and oil) is used both as a pressure medium for uniform pressing and as a heating and cooling medium for the polymer film. The cycle time in this fluid-based system is approximately 30 s or shorter. Ultrasonic heating has also been introduced to hot embossing for cycle time reduction (Liu and Dung, 2005; Mekaru et al., 2007). With ultrasonic heating, the surface layer of the embossing polymer can be rapidly heated in less than 10 s. The limitation of this method, however, is due to the difficulty in replicating concaved features and further due to the extremely small area that can be heated (Chang and Yang, 2005). All the above methods for rapid thermal cycling involve heating and cooling a single embossing mold. Yao et al. (2007b) have investigated the use of two stations, one hot and one cold, for rapid thermal cycling. This embossing technique does not rely on complex design of the mold insert in

566


achieving a low thermal inertia, but rather is developed as a processing strategy for enhancing productivity. As illustrated in Fig. 17.6, two upper mold bases are employed; one is maintained at a constant hot temperature and the other at a constant cold temperature. During the embossing stage, the hot base is used as the backing for the stamp. When the embossing stage finishes, the backing switches to the cold base. With this tool setup strategy, the heating and cooling stages are decoupled, and hence rapid thermal cycling is achieved. The attachment and separation of the stamp and the two stations can be achieved in a non-mechanical manner, e.g. using vacuum force. The cycle time using this two-station approach is in the order of 10 s.

17.4.2 Constant-temperature embossing The thermal cycling problem may also be tackled from materials perspectives. In theory, any materials (not necessarily a thermoplastic) that can be softened and solidified during processing can be used in embossing. If softening and solidification could occur at the same temperature, constant-temperature embossing would become possible. The recently developed solvent-assisted embossing method (Khang and Lee, 2000; Wang et al., 2005) may be considered as a specific method falling into this category. Since diffusion is an efficient process at micrometer sizes, solvents including supercritical fluids can rapidly diffuse into the thin polymer layer on the timescale of the process. The absorbed solvent works as a plasticizer and allows the polymer to soften. After the solvent is removed, the polymer is hardened again. In this case, the polymer is processed under a constant mold temperature. Constant-temperature embossing can also be achieved through chemical curing of monomers or oligomers. Among such processes, the most noteworthy process is the UVcuring embossing process (Gates et al., 2005; Heckele and Schomburg, 2004) in which a UV curable resin is cured by UV radiations. Rather than using solvents or a chemical curing process, Yao et al. (2007a) have investigated the feasibility of utilizing the unique property of slowly crystallizing polymers for achieving constant-temperature embossing. Crystallizable amorphous polymer films were embossed above the polymer’s glass transition temperature for pattern transfer and subsequently crystallized for solidification at the

Start Hot

Embossing

Cooling

Cold Stamp Polymer

17.6 Process sequence in two-station embossing.

Demolding


567

same temperature. The total cycle time with this processing strategy was found to be in the order of the half-time of crystallization of the polymer. Solid-state forming is a standard method used by the metal forming industry. Polymer is subjected to a large amount of elastic recovery if forged at a temperature below its softening temperature. Existing investigations (Xu et al., 2000; Yao and Nagarajan, 2004) showed that the dimensional recovery of the forged polymer microstructures is significantly affected by process conditions, as well as properties of the material used. However, it is worth mentioning that the method can be used for devices with relatively low dimensional accuracy requirement, such as scaffolds for tissue engineering and heat convective surfaces for electrical packaging. This process may be better controlled if capable prediction capabilities and process optimization schemes can be developed and utilized. When imprinting microchannels on a Teflon substrate, Yao and Nagarajan (2004) found that a combination of high forging speed and proper dwell time can be used effectively to reduce the amount of dimensional recovery. Recently, interest in microfeature embossing using solid-state forming is growing. Superplastic materials, including amorphous metals, in place of polymeric materials have been investigated by different researchers (Bohm et al., 2001; Yeh et al., 2006).

17.4.3 Embossing of discrete microfeatures At the moment, precise 3-D microparts are mainly produced using microinjection molding. The molding results, however, are often compromised because of the complex tool setup and the high amount of stresses introduced to the part during injection molding. It is thus advantageous to use hot embossing, a low-stress process with a simpler tool and process setup, for precision fabrication. Heckele and Durand (2001) developed a technique for producing throughholes by hot embossing. They used a substrate with two layers of different materials. After embossing, the tool features protruded through the upper layer into the lower layer. After removal of the lower layer, through-holes were left on the upper layer. Werner (2005) described a process involving identical top and bottom mold halves, both containing pins, whose top surfaces are attached to each other upon mold closure. By this process, through-holes with only a thin residual layer remaining can be embossed. Mazzeo et al. (2007) developed a tool set for punching thin plastic films. They were able to emboss holes as small as 500 µm in diameter. The methods described above all involve a through-thickness action for fabricating through-holes. Nagarajan et al. (2007) prototyped a hybrid punching and embossing process for through-thickness embossing of 3-D parts. The embossing tool included a punching head and to-be-replicated features in the socket behind the punching head. The built-in punching head allowed for a through-thickness action and

568


provided a close-die environment for embossing pressure buildup. The method was used to successfully emboss multichannel millimeter waveguides each of which weighed approximately 0.5 g. Kuduva-Raman-Thanumoorthy and Yao (2008) further developed this process for through-thickness embossing of true microparts. They used an embossing stamp with through-thickness microcavities and performed a through-thickness embossing step for fabrication of discrete microparts (Fig. 17.7), about 1 mg each part. After embossing, the embossed parts are attached to thin residual films of a thickness less than 10 µm on both sides. The residual films are then mechanically detached from the microparts during ejection.

17.4.4 Embossing of shell patterns Polymer thin films patterned with microstructures at a characteristic size greater than the film thickness are difficult to fabricate using standard hot embossing. Such shell microstructures experience deformation over dimensions larger than the film thickness, and the patterned film is marked by shell-type geometry with little change in the film thickness during the patterning process (Dreuth and Heiden, 1999). To emboss shell patterns, a pair of matching dies is needed. Such a matching pair requires extremely precise alignment of the mating surfaces. Small misalignment at a micrometer level could cause tool failure. Furthermore, the tiny space between the matching surfaces may be easily jammed by the polymer, resulting in ejection difficulty and, in worse cases, damage of the microstructures on the tool. Figure 17.8 shows the reported methods for patterning shell-type microgeometries. A modified hot embossing method involving a soft countertool was discussed by Dreuth and Heiden (1999). In this method, the second die is replaced by a cushion material, as shown in Fig. 17.8b. The cushion material and the polymer film are both softened by heat during

Heated platen

Cooled platen

Microparts Heated platen

Startup

Cooled platen Embossing

Cooling and demolding

17.7 Steps involved in through-thickness embossing of discrete microparts.


569

Microstructured

Microstructured

Polymer film

Polymer film Microstructured

Sacrificial cushion

(a)

(b)

Mold Pressure

Polymer film

Se

Poly

mer

film

Vacuum Heated embossing roll

Pressurized (c)

(d)

17.8 Different techniques for fabricating shell microstructures: (a) hot embossing using a pair of microstructured dies, (b) hot embossing using a microstructured die and a countertool made of a deformable polymer, (c) micro thermoforming, (d) roller embossing.

embossing. After cooling of the two materials, the cushion material is sacrificed, e.g. dissolved by a solvent, to recover the structured polymer film. The advantage of this process is that alignment of mating surfaces is not needed. Ikeuchi and Ikuta (2005) used this method for fabricating polymer membrane microchannels. Paraffin was used as a cushion material, and membrane microchannels were fabricated with two polymer films: poly(methyl methacrylate) film and poly(lactic acid) film. There are some issues involved in the use of a sacrificial countertool. The thermoplastic cushion material needs to be softened together with the embossing film, and therefore the two materials need to have a similar softening temperature. Moreover, the cushion material should have appropriate deformability. To address these issues, a rubber-assisted embossing process using rubber as a soft countertool was

570


recently investigated by Nagarajan and Yao (2007). The rubber countertool further helps with part release during ejection. There also exist continuous processes for patterning/texturing thin polymer films. These processes are based on roll-to-roll setups, involving a heated embossing roll and a pressure roll (Fig. 17.8d) for transferring the surface textures on the embossing roll to a continuous film. The surface texture on the film modifies the feel and appearance of the film and is therefore desired in many end-use applications, including packaging, diapers, raincoats, and disposable goods. However, in order to use a roll embossing process for precision film structuring, accurate and spatial control of the roll surface temperature is needed (Michaeli et al., 2005). Efforts have also been made to scale down conventional thermoforming techniques for precision structuring of polymeric film (Heckele and Schomburg, 2004; Truckenmuller et al., 2002). Truckenmuller et al. (2002) used a microthermoforming process to pattern 25-µm thick PS film with 125-µm deep and 250-µm wide microchannels. During this microthermoforming process, the film was held between a structured mold insert and a flat mold platen and then thermoformed under the action of pressurized gas. Although this method was successfully employed to thermoform microfluidic analysis chips, it is not suitable for achieving a high aspect ratio of the channel (Heckele and Schomburg, 2004). Further, uniform film thickness is difficult to reach.

17.5

Micromold making

Many methods have been proposed for making stamps/molds for micromolding applications. These can be classified into three main types: (1) lithographic methods, (2) replication methods, and (3) material ablation or removal methods. Lithography is a well-established technique, particularly for patterning microstructured silicon wafer. Such silicon tools have been used widely in embossing/imprinting-based processes, but mostly for prototyping applications. High-aspect-ratio features can be produced on a resist polymer using X-ray lithography and deep UV lithography. Deep reactive etching can be used for creating high-aspect-ratio features on silicon wafers. If only shallow features are needed, lithography can also be used for patterning steel and other more isotropic material. For example, in the mold making industry, for many years a wax coating and acid etching process has been used for creating shallow microtextures on mold surfaces. The microstructures on a soft or brittle substrate made of polymer resist, silicon or glass can be transferred to a more durable metallic material by the replication method. Different coatingbased techniques, including electroplating or electroforming, physical vapor deposition and chemical vapor deposition, can be used for feature transfer. In particular, the well-known LiGA process is based on lithography and


571

electroforming. Replication has also been used for transferring the pattern from silicon to a polymeric material, and the patterned polymeric material is used as a prototyping mold. With multiple soft molds created from a silicon template, more parts can be replicated. However, these soft tools are mostly useful for low pressure and low temperature processes. The soft molding method based on PDMS is one such process. Studies have also been performed on micromold making using hot embossing with a thermoplastic polymer (Dirckx et al., 2007). A recent review on micromolds made of non-conventional materials is given by Malek et al. (2007). For making a durable metallic micromold, material ablation or removal methods are more frequently used. Microscale material removal can be performed by mechanical micromachining processes (e.g., micromilling), electrical discharge machining, electrochemical machining, and high energy beam machining processes (e.g., with ultrashort pulse laser, focused ion beam, electron beam, plasma, etc.). A comparison between different processes usable for metallic mold insert fabrication has been given by Giboz et al. (2007). Similar to conventional molding, smooth surface finishes are highly desirable for a micromold. The surface finish of the structures produced by material removal processes is typically rough, as compared with the size of the feature. In conventional mold making, burs and unexpected asperities can be reduced by various no-contact polishing processes, e.g. ultrasonic polishing, electrochemical erosion, etc. These traditional methods may be adapted to micromold polishing, but so far there has been little effort reported. Surface smoothness may also be improved by taking advantage of some natural processes, e.g. surface tension caused reflow. For instance, Shiu et al. (2008) created a pattern with a smooth surface finish by embossing a polymer only halfway into a mold with microchannels. The polymer pattern is then used as a stamp for embossing smooth microchannels on a second material. Alternatively, the smooth pattern on the polymer template can be transferred to a metallic material by a coating-based replication method as described earlier. In the case of a lens array pattern, different processes (Chen et al., 2008; Jiang et al., 2007; Pan et al., 2004) can be used to create a smooth surface. As in conventional molding, draft angles are also desired in demolding. According to a review by Heckel and Schomburg (2004), most problems in micromolding are not caused by mold filling, but by demolding. It is indeed a difficult task to eject microfeatures with aspect ratio greater than 10 if no draft angle and/or lubrication agents are employed. Some techniques are available for incorporating draft angles to a straight wall, including inclined lithography (de Campo and Greiner, 2007; Turner et al., 2003), under-dosing lithography (Yang and Young, 2004), and femtosecond laser machining. In addition to the draft angle, the mold surface often needs to be treated with fluorinated coating material to lower its surface energy (Nezuka et al., 2008), thus improving feature release during ejection.

572

17.6


Emerging topics in micromolding

Although the field of micromolding has been under development for more than 30 years, there is still active research and development going on in this field. Particular areas of interest include: (1) structural and morphological study, (2) process control and monitoring and (3) modeling and simulation. Since these areas are still under active research and to some extent still need to be defined, the following only touches upon some general considerations. Processing–structure–property relations are of great importance in polymer processing. It is well known that the hierarchical development in structure of conventionally molded parts is spatially distributed, with local morphology highly dependent on the local thermomechanical history. The structure and properties of conventionally molded parts are typically examined by microscopic methods such as optical microscopy, scanning electron microscopy (SEM), transmission electron microscopy (TEM), thermal analysis such as differential scanning calorimetry (DSC) and dynamic mechanical analysis (DMA), mechanical testing methods including tensile, compression, bending and impact tests, and optical inspections, e.g. birefringence measurement. To measure spatially varied structure and properties, molded parts are often microtomed into thin sections of micrometer thickness. Although fundamentally similar, the morphology in micromolded parts could be significantly different from that in conventionally molded parts due to the different thermomechanical history involved, particularly under the geometrical constraint of a miniaturized mold cavity. For instance, the cross-sectional morphology of microinjection molded PP was found to be quite different from that of a conventionally molded PP; the near-amorphous skin was found to be relatively thinner and the variation in spherulite size was smaller. In order to determine the local variation of morphology and properties, some suitable microscopic, thermal, physical and mechanical testing methods need to be identified and adapted to the characterization of micromolded parts. Polarized optical microscopy and SEM, the two most widely used microscopic methods for conventionally molded parts, have also been reported to be useful for micromolded parts (Ito et al., 2007; Whiteside et al., 2004; Zhang and Lu, 2008). For highresolution microscopic characterization, atomic force microscopy (AFM) and TEM are capable tools. If a blend or a composite material is involved, high-resolution spectroscopic methods such as X-ray photoelectron spectroscopy, energy dispersive spectroscopy, static secondary ion mass spectroscopy and static light scattering, though not reported in micromolding applications, are excellent tools for characterizing variation of composition in a small dimension. Blends and composites are fundamentally interesting in micromolding because their phase or morphological structure may have a characteristic size close to the part feature size. The resulting structural distribution in this case is dependent on the feature size, but, if designed


573

appropriately, innovative products may be created. The phase separation micromolding process (Vogelaar et al., 2005) and the templated phase migration process (Shang et al., 2007) proposed recently are two specific examples. Compared with morphological characterization, thermal analysis and mechanical testing can become more difficult because these tests typically require the use of a large sample size. However, the aforementioned analytical characterization method, if used and interpreted appropriately, may be useful for indirect characterization of thermal and mechanical properties. In particular, AFM-based methods including nano indention were found to be capable of measuring mechanical properties of extremely delicate features (Greenway et al., 2003; Whiteside et al., 2004). Process control and monitoring is also difficult in micromolding. Measurement of pressure and temperature distribution in a microcavity requires an extremely high resolution. Furthermore, the sensor placed in the microcavity needs to be so small that it does not interfere with the flow field and the thermal field. It is conjectured that wall slip may exist in the microchannel flow; however, measurement of the velocity at the surface of a microchannel would indeed be a challenge. So far, there have been only limited efforts addressing these aspects (Ono et al., 2007; Whiteside et al., 2004, 2005a, 2005b). However, new developments may come along in the area of miniature sensors based on the MEMS technology. Inadequacy of the standard simulation packages in micromolding has been revealed by many authors (e.g., Ilinca et al., 2004; Kemmann and Weber, 2000; Kim and Turng, 2006; Piotter et al., 2002; Worgull and Heckele, 2004; Wu and Wu, 2007; Yang et al., 2006; Yao and Kim, 2002a; Yu et al., 2002). The insufficiency is considered to be caused by the inappropriate use of macroscale material properties in micromolding simulation, as well as by the neglect of other micro/nanoscale influences. Therefore, rheological characterization of polymeric liquids in confined geometries would be an important research topic. Simulation schemes for micromolding with justification of micro/nanoscale influences are highly desired. One major difficulty is the characterization and modeling of the complex rheological behavior of the material in micromolding. For instance, an amorphous polymer typically deforms at a temperature only slightly above its Tg in hot embossing. The inclusion of an accurate viscoelastic model considering nonlinearity is often a challenge for high Weissenberg number and high Deborah number flows. The problem is further complicated by the plausible size effects at microscale, particularly when the characteristic size scale of the material is comparable to that of the molding geometry. This could be scarce for a pure polymer, but would likely occur for composite or blend materials.

574

17.7


References

Bartolini R, Hannan W, Karlsons D and Lurie M (1970), ‘Embossed hologram motion pictures for television playback’, Appl Opt, 9, 2283–2290. Bauer H D, Ehrfeld W, Paatzsch T, Smaglinski I and Weber L (1999), ‘Advanced micromolding of optical components’, SPIE, 3878, 261–270. Becker H and Locascio L E (2002), ‘Polymer microfluidic devices’, Talanta, 56, 267– 287. Bohm J, Schubert A, Otto T and Burkhardt T (2001), ‘Micro-metalforming with silicon dies’, Microsyst Technol, 7, 191–195. Brown G H (1947), Theory and Application of Radio-Frequency Heating, Van Nostrand, New York. Chang J H and Yang S Y (2005), ‘Development of fluid-based heating and pressing systems for micro hot embossing’, Microsyst Technol, 11, 396–403. Chang P-C and Hwang S-J (2006), ‘Experimental investigation of infrared rapid surface heating for injection molding’, J Appl Polym Sci, 102, 3704–3713. Chang P-C, Hwang S-J, Lee H-H and Huang D-Y (2007), ‘Development of an externaltype microinjection molding module for thermoplastic polymer’, J Mater Proc Technol, 184, 163–172. Chen S C, Jong W R and Chang J A (2006a), ‘Dynamic mold surface temperature control using induction heating and its effects on the surface appearance of weld line’, J Appl Polym Sci, 101, 1174–1180. Chen S C, Jong W R, Chang Y J, et al. (2006b), ‘Rapid mold temperature variation for assisting the microinjection of high aspect ratio micro-feature parts using induction heating technology’, J Micromech Microeng, 16, 1783–1791. Chen Y, Yi A Y, Yao D, Klocke F and Pongs G (2008), ‘A reflow process for glass microlens arrays fabrication by use of precision compression molding’, J Micromech Microeng, Article Number: 055022. de Campo A and Greiner C (2007), ‘SU-8: a photoresist for high-aspect-ratio and 3-D submicron lithography’, J Micromech Microeng, 17, R81–R95. Despa M S, Kelly K W and Collier J R (1999), ‘Injection molding of polymeric LIGA HARMs’, Microsyst Technol, 6, 60–66. Dirckx M, Mazzeo A D and Hardt D E (2007), ‘Production of micro-molding tooling by hot embossing’, Proc ASME International Manufacturing Science and Engineering Conf, 141–150. Dreuth H and Heiden C (1999), ‘Thermoplastic structuring of thin polymer films’, Sensors and Actuators, 78, 198–204. Gale M T, Kane J and Knop K (1978), ‘ZOD images: Embossable surface-relief structures for color and black-and-white reproduction’, J Appl Photogr Eng, 4, 41. Gates B D, Xu Q, Stewart M, Ryan D, Willson C G and Whitesides G M (2005), ‘New approaches to nanofabrication: molding, printing, and other techniques’, Chem Rev, 105, 1171–1196. Giboz J, Copponnex T and Mele P (2007), ‘Microinjection molding of thermoplastic polymers: a review’, J Micromech Microeng, 17, R96–R109. Greenway G R, Allan P S and Hornsby P R (2003), ‘The characterisation and physical testing of micro-mouldings’, SPE ANTEC Proc, 1995–1999. Gross G L W (2006), ‘The production of nanostructures by mechanical forming’, J Phys D: Appl Phys, 39, R363–386. Hannan W J, Flory R E, Lurie M and Ryan R J (1973), ‘HoloTape: a low-cost prerecorded


575

television system using holographic storage’, J Soc Motion Pict Telev Eng, 82, 905– 915. Heckele M and Durand A (2001), ‘Microstructured through-holes in plastic films by hot embossing’, Proc 2001 Euspen’s 2nd Int Conf, 196–198. Heckele M and Schomburg W K (2004), ‘Review on micro molding of thermoplastic polymers’, J Micromech Microeng, 14, R1–R14. Herlihy G (2007), ‘X-melt – a precision technology for micro-moulding and thin-wall parts’, in Coates P D, Polymer Process Engineering 07 – Enhanced Polymer Processing, University of Bradford, Bradford, UK. Ikeuchi M and Ikuta K (2005), ‘Development of membrane micro embossing (MeME) process for self-supporting polymer membrane microchannel’, Proc IEEE Int Conf on Micro Electro Mechanical Systems (MEMS), 133–136. Ilinca F, Hetu J F and Derdouri A (2004), ‘Numerical simulation of the filling stage in the micro-injection molding process’, SPE ANTEC Proc, 636–640. Ito H, Kazama K and Kikutani T (2007), ‘Effects of process conditions on surface replication and higher-order structure formation in micromolding’, Macromol Symp, 249–250, 625–634. Jansen K M B and Flaman A A M (1994), ‘Construction of fast-response heating elements for injection molding applications’, Polym Eng Sci, 34, 894–897. Jiang L-T, Huang T-C, Chiu C-R, Chang C-Y and Yang S-Y (2007), ‘Fabrication of plastic microlens arrays using hybrid extrusion rolling embossing with a metallic cylinder mold fabricated using dry film resist’, Optics Express, 15, 12088–12094. Kemmann O and Weber L (2000), ‘Simulation of micro injection molding processes’, SPE ANTEC Proc, 576–580. Khang D Y and Lee H H (2000), ‘Room temperature embossing lithography’, Appl Phys Lett, 76, 870–872. Kim B H and Suh N P (1986), ‘Low thermal inertia molding’, Polym Plast Technol Eng, 25, 73–93. Kim S W and Turng L S (2006), ‘Three-dimensional numerical simulation of injection molding filling of optical lens and multiscale geometry using finite element method’, Polym Eng Sci, 46, 1263–1274. Kimberling T E, Liu W, Kim B H and Yao D (2006), ‘Rapid hot embossing of polymer microfeatures’, Microsyst Technol, 12, 730–735. Kuduva-Raman-Thanumoorthy R and Yao D (2008), ‘Hot embossing of discrete microparts’, SPE ANTEC Proc, 2487–2491. Liu S J and Dung Y T (2005), ‘Ultrasonic vibration hot embossing – A novel technique for molding plastic microstructure’, Int Polymer Processing, 20, 449–452. Malek C K, Coudevylle J-R, Jeannot J-C and Duffait R (2007), ‘Revisiting micro hotembossing with moulds in non-conventional materials’, Microsyst Technol, 13, 475– 481. Mazzeo A D, Dirckx M and Hardt D E (2007), ‘Single-step through-hole punching by hot embossing’, SPE ANTEC Proc, 2977–2981. Mekaru H, Yamada T, Yan S and Hattori T (2004), ‘Microfabrication by hot embossing and injection molding at LASTI’, Microsyst Technol, 10, 682–688. Mekaru H, Goto H and Takahashi M (2007), ‘Development of ultrasonic micro hot embossing technology’, Microelectronic Engineering, 84, 1282–1287. Michaeli E H W, Fink B and Blomer P (2005), ‘Dynamic control of roll temperature’, Kunststoffe-Plast Europe, 95, 51–53. Michaeli W and Spennemann A (2000), ‘Injection moulding microstructured functional surfaces’, Kunststoffe Plast Europe, 90, 52.

576


Michaeli W, Spennemann A and Gartner R (2004), ‘New plastification concepts for micro injection moulding’, J Polym Eng, 24, 81–93. Michel B, Bernard A, Bietsch A, Delamarche E, et al. (2001), ‘Printing meets lithography: soft approaches to high-resolution patterning. IBM J Res Dev, 45, 697–719. Nagarajan P and Yao D (2007), ‘Rubber assisted embossing process’, Proc 65th Society of Plastics Engineers Annual Technical Conf, Cincinnati, OH, pp. 2967–2971. Nagarajan P, Yao D, Ellis T S and Azadegan R (2007), ‘Through-thickness embossing process for fabrication of three-dimensional thermoplastic parts’, Polym Eng Sci, 47, 2075–2084. Nexus (2005), Market analysis for MEMS and microsystems III, 2005–2009. Nezuka O, Yao D and Kim B H (2008), ‘Replication of microstructures by roll-to-roll UV-curing embossing’, Polym-Plast Technol Eng, 47(9), 865–873. Nian S C and Yang S Y (2005), ‘Molding of thin sheets using impact micro-injection molding’, Int Polym Proc, 20, 441–448. Ono Y, Whiteside B R, Brown E C, Kobayashi M, Cheng C-C, Jen C-K and Coates P D (2007), ‘Real-time process monitoring of micromoulding using integrated ultrasonic sensors’, Trans Inst Measurement and Control, 29, 383–401. Pan L W, Shen X J and Lin L W (2004), ‘Microplastic lens array fabricated by a hot intrusion process’, IEEE/ASME J Microelectromech Syst, 13, 1063–1071. Piotter V, Mueller K, Plewa K, Ruprecht R and Hausselt J (2002), ‘Performance and simulation of thermoplastic micro injection molding’, Microsyst Technol, 8, 387–390. Rotting O, Ropke W, Becker H and Gartner C (2002), ‘Polymer microfabrication technologies’, Microsyst Technol, 8, 36–32. Ruprecht R, Gietzelt T, Mueller K, Piotter V and Hausselt J (2002), ‘Injection molding of microstructured components from plastics, metals and ceramics’, Microsyst Technol, 8, 351–358. Schinkothe W and Walther T (2000), ‘Reducing cycle times – Alternative mould temperature control for microinjection moulding’, Kunststoffe Plast Europe, 90, 62. Shang Y R, Wei M, Kazmer D, Barry C and Mead J (2007), ‘Experimental validation of numerical simulation for phase decomposition of a binary polymer thin film on a patterned substrate’, SPE ANTEC Proc, 2916–2920. Shiu P P, Knopf G K, Ostojic M and Nikumb S (2008), ‘Rapid fabrication of tooling for microfluidic devices via laser micromachining and hot embossing’, J Micromech Microeng,18, 025012. Tadmor Z and Gogos C G (1979), Principles of Polymer Processing, First Edition, John Wiley & Sons, New York. Tadmor Z and Gogos C G (2006), Principles of Polymer Processing, Second Edition, John Wiley & Sons, New York. Truckenmuller R, Rummler Z, Schaller T and Schomburg W K (2002), ‘Low-cost thermoforming of micro fluidic analysis chips’, J Micromech Microeng, 12, 375–379. Turner R, Desta Y, Kelly K, Zhang J, Geiger E, Cortez S and Mancini D C (2003), ‘Tapered LIGA HARMs’, J Micromech Microeng, 13, 367–372. Ulrich R, Weber H P, Chandross E A, Tomlinson W J and Franke E A (1972), ‘Embossed optical waveguides’, Appl Phys Lett, 20, 213–215. Vogelaar L, Lammertink R G H, Barsema J N, et al. (2005), ‘Phase separation micromolding: a new generic approach for microstructuring various materials’, Small, 1, 645–655. Wang Y, Liu Z M, Han B X, et al. (2005), ‘Compressed-CO2-assisted patterning of polymers’, J Phys Chem B, 109, 12376–12379. Weber L and Ehrfeld W (1999), ‘Micromoulding’, Kunststoffe-Plast Europe, 89, 192.


577

Werner M (2005), ‘Hot embossing of through-holes in cyclo-olefin copolymer’, Diploma Thesis, Technical University of Denmark. Whiteside B R, Martyn M T, Coates P D, Allan P S, Hornsby P R and Greenway G (2003), ‘Micromoulding: process characteristics and product properties’, Plast Rubber Compos, 32, 231–239. Whiteside B R, Martyn M T, Coates P D, Greenway G, Allen P and Hornsby P (2004), ‘Micromoulding: process measurements, product morphology and properties’, Plast Rubber Compos, 33, 11–17. Whiteside B R, Spares R, Howell K, Martyn M T and Coates P D (2005a), ‘Micromoulding: extreme process monitoring and inline product assessment’, Plast Rubber Compos, 34, 380–386. Whiteside B R, Brown E C, Ono Y, Jen C-K and Coates P D (2005b), ‘Real-time ultrasonic diagnosis of polymer degradation and filling incompleteness in micromoulding’, Plast Rubber Compos, 34, 387–392. Wimberger-Friedl R W (2000), ‘Injection molding of sub-µm grating optical elements’, J Injection Molding Technology, 4, 78–83. Worgull M and Heckele M (2004), ‘New aspects of simulation in hot embossing’, Microsyst Technol, 10, 432–437. Worgull M, Hetu J-F and Kabanemi K K (2006), ‘Modeling and optimization of the hot embossing process for micro- and nanocomponent fabrication’, Microsyst Technol, 12, 947–952. Wu C H and Wu S M (2007), ‘Flow analysis of micro-channel in injection molding’, J Polym Eng, 27, 107–127. Xia Y and Whitesides G M (1998), ‘Soft-lithography’, Angew Chem Int Ed Engl, 37, 550–575. Xu J, Locascio L, Gaitan M and Lee C S (2000), ‘Room-temperature imprinting method for plastic microchannel fabrication’, Anal Chem, 72, 1930–1933. Xu X R, Sachs E and Allen S (2001), ‘The design of conformal cooling channels in injection molding tooling’, Polym Eng Sci, 41, 1265–1279. Yang H H, Chao C K, Wei M K and Lin C P (2004), ‘High fill-factor microlens array mold insert fabrication using a thermal reflow process’, J Micromech Microeng, 14, 1197–1204. Yang S P and Young W B (2004), ‘Microinjection molding with LIGA-Like process’, Int Polym Proc, 19, 180–185. Yang W, Chang D and Yang V (2006), ‘True 3D numerical simulation of micro injection molding’, SPE ANTEC Proc, 1270–1274. Yao D and Kim B (2002a), ‘Simulation of the filling process in micro channels for polymeric materials’, J Micromech Microeng, 12, 604–610. Yao D and Kim B (2002b), ‘Injection molding high aspect ratio microfeatures’, J Injection Molding Technology, 6, 11–17. Yao D and Kim B (2002c), ‘Development of rapid heating and cooling systems for injection molding applications’, Polym Eng Sci, 42, 2471–2481. Yao D and Kim B (2002d), ‘Increasing flow length in thin wall injection molding using a rapidly heated mold’, Polym-Plast Technol Eng, 41, 819–832. Yao D and Nagarajan P (2004), ‘Cold forging method for polymer microfabrication’, Polym Eng Sci, 44, 1998–2004. Yao D, Kimberling T E and Kim B (2006), ‘High-frequency proximity heating for injection molding applications’, Polym Eng Sci, 46, 938–945. Yao D, Nagarajan P and Ramasubramani K R T (2007a), ‘Constant-temperature microfeature embossing with slowly crystallizing polymers’, Int Polym Proc, 22, 375–379.

578


Yao D, Nagarajan P, Li L and Yi A Y (2007b), ‘Two-station embossing process for rapid fabrication of surface microstructures on thermoplastic polymers’, Polym Eng Sci, 47, 530–539. Yeh M S, Lin H Y, Lin H T and Chang C B (2006), ‘Superplastic micro-forming with a fine grained Zn-22Al eutectoid alloy using hot embossing technology’, J Mater Proc Technol, 180, 17–22. Yu L, Kuo C G, Lee L J, Keolling K W and Madou M J (2002), ‘Experimental investigation and numerical simulation of injection molding with micro-features’, Polym Eng Sci, 42, 871–886. Yu M C, Young W B and Hsu P M (2007), ‘Micro-injection molding with infrared assisted mold heating system’, Mater Sci Eng. A, 460–461, 288–295. Zhang K F and Lu Z (2008), ‘Analysis of morphology and performance of PP microstructures manufactured by micro injection molding’, Microsyst Technol, 14, 209–214. Zhao J, Mayes R H, Chen G, Chan P S and Xiong Z J (2003), ‘Polymer micomould design and micromoulding process’, Plast Rubber Compos, 32, 240–247.

18 Reactive polymer processing and design of stable micro- and nanostructures A V M A C H A D O and J A C O VA S, University of Minho, Portugal and V B O U N O R - L E G A R E and P C A S S A G N A U, Université de Lyon, France

Abstract: Reactive processing is an alternative and promising method to produce micro- and nanostructured polymeric materials with controlled structure. Using this method it is possible to take advantage of the knowledge on reactive systems (polymerization, modification and blending) that have been conducted during processing. Thus, this chapter shows the potential of this technique to prepare micro- and nanostructured polymeric materials. Key words: reactive processing, copolymers, blends, modification, polymerization, compatibilization, nanostructures.

18.1

Micro- and nanostructures

18.1.1 Introduction Micro- and nanostructured materials are materials with a microstructure that has a characteristic length scale in the order of a few micro/nanometers. The properties of these materials deviate from those of conventional ones. This deviation results from the reduced size and/or dimensionality of the nanometersized structures. An outcome of solid state physics and chemistry is the insight that most properties of solids depend on their microstructure, i.e. the chemical composition, arrangement of the atoms (atomic structure) and the size of the solid in one, two or three dimensions. Thus, changes in the properties of a solid will be noticed if one or several of these parameters are changed. However, if it is not possible to control the structure at atomic level, the desired properties can also be obtained by designing the micro- and nanostructure at the molecular level. Polymeric systems that exhibit unique properties can be directly attributed to the presence of structural entities having dimensions in the micro- and nanometer range. Because of the special contribution of these micro- and nanosized entities, this class of polymeric systems can be collectively designated as micro- and nanostructured polymeric materials. 579

580


Understanding and controlling the mechanisms of phase separation and micro- and nanostructure formation in polymer systems enables the enhancement of the performance of these materials in several applications. For example, co-continuous blends of high- and low-melting-point polymers where the latter is the major component, have increased thermal and mechanical properties (such as toughness, stress at break or high-temperature creep resistance and similar processability) in relation to the original polymers. Copolymers exhibiting nanostructured phases show different optical properties and enhanced mechanical properties when compared with traditional copolymers.

18.1.2 Preparation methods In recent years several strategies have been developed to prepare well-defined and predictable polymer structures. The great issue here is that the desired properties and functions are arrived at not by manipulation of the structure at atomic or molecular level, but by designing larger, nanoscopic building blocks, made of complex fluids (i.e., block copolymers, ion-containing polymers, polymer networks) and often of controlled shape (i.e., micellae, dendrimers, stars, combs, disks). The research for such materials will add a new dimension to the available range of properties and functions in polymers and other materials. The challenge is to go back to the conventional processing– structure–property correlations and develop new principles to create structures of controlled length scales, which would result in novel materials with distinct (and unusual) properties. Until now a lot of research has been performed and as a result various routes leading to these new micro- and nanostructured polymers have been reported. The production of polyolefins with multimodal microstructural distributions in a single metallocene and a single reactor is an attractive method for producing polymers with balanced properties with simpler reactor technology. For example, copolymerization of ethylene and 1-octene carried out with an in situ supported rac-[dimethylsilylbis(methylbenzoindenyl)] zirconium dichloride catalyst produces polyethylene/α-olefin copolymers with broad and bimodal short chain branching distributions and narrow molecular weight distributions [1]. Polymeric microtubules and nanofibrils have been prepared by depositing a solution of the desired polymer within the pores of microporous template membranes [2]. This method typically entails synthesizing the desired material within the pores of a microporous template membrane. The template membranes employed contain cylindrical pores with monodisperse diameters, which extend through the entire thickness of the membrane. The template method has been used to prepare tubular and fibrillar micro- and nanostructures composed of metals, semiconductors, electronically conductive polymers, carbons, and other materials [2].

Reactive polymer processing and design

581

Nanostructured polymers are also accessible by the use of lithographic techniques. Imprinting or embossing is a well-known technique to generate microstructures in hard polymers by pressing a rigid master containing surfacerelief features into a thin thermoplastic polymer film that is then heated close to or, more generally, above Tg [3]. Nanoimprint lithography has the potential of high throughput due to parallel processing, does not require sophisticated tools and allows nanoscale replication for data storage [4, 5]. The quality of the nanoimprinting process depends on a number of experimental parameters like Tg, melt viscosity, adhesion of the polymer to the mold, etc. [6]. Nanoimprint lithography has primarily been used to emboss hard thermoplastic polymers. The micromolding and embossing of elastomers has attracted considerable interest as these materials have found important applications in soft-lithographic techniques such as microcontact printing [7, 8]. The advantage of microcontact printing is the ability to pattern surfaces chemically at the sub-micron level. The fabrication of ~100 nm surface relief features has proven to be much more difficult, as nanoscale structures easily collapse [9, 10]. Suh et al. [11, 12] used capillary force lithography for patterning a polymer film. When a patterned polydimethylsiloxane (PDMS) mold is placed on a spin-coated polymer film and then heated above the glass transition temperature (Tg) of the polymer, the capillarity forces the polymer to melt into the void space of the mold, thus yielding a negative replica when the mold is removed. In forming polymer micro-to-nanostructures by capillary force lithography, dewetting of polymer films was observed when the void space of the mold was not completely filled with the films. Dewetting is also a convenient way of creating ordered micro- and nanostructures. Higgins and Jones [13] investigated the effects of surface topography on polymer dewetting by casting poly(methyl methacrylate) films on glass substrates that are roughened directionally by rubbing. They observed an anisotropic dewetting, the period of which is in accordance with that of the directional rubbing. They showed that the dewetting pattern follows the substrate pattern period, leading to the formation of droplet arrays. The structuring of the topography of polymers by plasma treatment can occur at micrometer and nanometer scales and can influence adhesion, optical and wettability properties of the materials. These topography modifications are of great interest to the biomedical (contact lenses, implants) and optical industries (reflection, absorption) [14]. Electrospinning has been shown to be an effective method for the production of structured polymer fibers with diameters in the range from several micrometers down to tens of nanometers, which are of considerable interest to various kinds of applications [15]. Phase separation of polymers into lamellar structures has been used to generate 50–100 nm thick periodic layers with different refractive indices,

582


which can be achieved from solution casting. Block copolymers comprising two (or more) flexible, chemically incompatible and dissimilar blocks, e.g. poly(styrene) and poly(isoprene), can be microphase separated into a variety of morphologies with nanometer-scale dimensions. This self-assembly process is driven by an unfavorable mixing enthalpy and a small mixing entropy, while the covalent bond between the two blocks prevents macrophase separation. The microphase separated morphology that is formed (spheres, lamellae, inverse spheres and several more complex shapes) depends on the polymers used and on their volume fractions [16]. When the morphology can be controlled and turned into a useful structure, phase separation of block copolymers can be a powerful tool to fabricate nanostructures without additional lithography and processing steps. Solid-state mechanical alloying, wherein the constituent polymers are mixed as solids at cryogenic temperatures, has been successfully used to prepare blends of thermoplastics with nanoscale morphology [17–19]. The dramatically reduced chain mobility in solid-state multicomponent systems effectively prevents phase separation during blending and promotes the formation of nanoscale morphologies. Mechanical alloying generally refers to the high-energy ball milling of two or more dissimilar materials to produce homogeneous alloys at the molecular or atomic level and is responsible for the ongoing development of novel metastable and nanostructured inorganic alloys possessing interesting mechanical, optical, magnetic, and electronic properties [20, 21]. High-energy milling of polymeric materials subjects the blend components to a complex deformation field in which shear, multiaxial extension, fracture, and cold-welding develop concurrently. Recently, it has been demonstrated that noncovalent bonds can be used in constructing block-copolymer-like molecular complexes [22–24] called supramolecules, which serve equally well as building units of nanostructures. Such noncovalent bonds include hydrogen bonding, ionic interactions, coordination complexation, and charge-transfer interactions. These supramolecular routes for block-copolymer-like architectures provide not only a new option to create nanostructured materials, but also a fascinating means to design ‘smart’ materials that respond to external stimuli or conditions [25, 26]. Direct formation or cleavage of supramolecular complexes under a desired condition can thus be a basic strategy to develop materials with several tunable morphologies in the bulk. An alternative and promising method to produce micro- and nanostructured polymeric materials with controlled structure is by reactive extrusion. Using this method it is possible to take advantage of the knowledge on polymerization, modification and blending via reactive extrusion that has been progressively accumulated. Thus, this chapter will show the potential of this technique to prepare micro- and nanostructured polymeric materials.

Reactive polymer processing and design

18.2

583

Reactive extrusion

18.2.1 The role of reactive extrusion One of the significant advantages of the extruder over batch reactors is to guarantee a continuous reactive bulk process, e.g. high viscosity solvent-free reactive systems. Actually, reactive processing combines polymer processing and chemical reaction. Consequently, very specific conditions such as high viscous medium (η ~ 103 Pa.s), high temperatures (T ~ 250°C) and short residence times (t ~ 1 min) can be found in this type of process. Actually, reactive extrusion is now being viewed as an efficient means of continuously polymerizing monomers and/or modifying polymers. Particularly, co-rotating and counter-rotating intermeshing twin screw extruders proved to be a good technical and economical solution for reactive processing of thermoplastic polymers. The numerous possible advantages of using the extruder as a reactor can be described as follows: • • • • • • • •

Fast and continuous process – productivity No solvents necessary – environmentally friendly Good mixing and transport of high-viscosity media No torque limitations for high-viscosity reactive media Chemical modification and/or compounding in a single step Possibility to process complex formulations (filler, plasticizer, etc.) Removal of side products, or monomers, by efficient devolatilization Turning of residence and reaction times by modular screw profile.

However, there are also some drawbacks in using an extruder as a chemical reactor, which are actually the counterparts of the main advantages: •

• • •

Limited number of accessible chemical reactions: – Fast kinetics – very limited residence times (T2Crit)>99%

Tensile bar width (mm)

18.8

3S(M2) 18.7 2S(M2)

18.6

2S(M2) 3S(M2)

18.5 Yes 18.4

No 0

10

20

30

40

50

60

70

80

90 100 110 120 130 140 150 160 170 180 190 200 210 220 Collected Cycle

20.11 Control chart of measured tensile bar width and cycle rejection according to T2 criterion (right axis).


Coolant off

676

18.9


677

manufacturing cells. Third, the immediate diagnosis of the variance and residuals for a defective molding process expedites the correction of the molding process while also enhancing worker knowledge.

20.5

Future trends

20.5.1 Advanced sensing technologies Advances in materials and manufacturing technologies have enabled new types of sensors that are smaller and more robust with greater capabilities. For example, many transducers used to measure the pressure of a melt stream have relied on a deflecting diaphragm, fluid capillary, and associated strain gages to provide a feedback signal. While efforts have been made to reduce the size and increase the performance of sensors based on this technology, these sensors remain on the order of 10 mm in diameter with a response time on the order of 10 ms. By comparison, advanced piezoelectric materials and manufacturing methods have enabled the production of melt pressure sensors on the order of 1 mm in diameter with response times on the order of 10 µs! For this reason, piezoelectric technology has largely supplanted strain gage technology in pressure sensing applications not requiring a high heat environment or very slow measurements. Similar advances are being realized in other sensing applications. For example, thermocouples currently remain well entrenched throughout the industry. Yet, a plethora of newer and potentially better sensors exist including fiber optic temperature sensors, infrared pyrometers, pyroelectric sensors, resistive temperature detectors, semiconductor junction thermometers, and others [26]. A similar abundance of sensing technologies exists for the measurement of other process states such as force, position, velocity, etc. Such advances in sensor capability and miniaturization will provide for more process states to be more accurately measured so as to thereby control and optimize plastics manufacturing processes. Advances in sensing technology are being fueled not only by the sensors themselves, but by our ability to better use the sensed information. The online simulations and multivariate modeling applications described in this chapter are examples of how the sensed information can be better leveraged. Process controllers will increasingly fuse the data from multiple sensor streams with advanced logic to provide reliable estimates of currently unobservable process states such as flow rate, viscosity, density, etc. In parallel, multimode sensors are already available to simultaneously monitor pressure and temperature. It can be expected that such multi-mode sensors will incorporate similar logic and provide an augmented vector of process states, including estimations of unobservable process states.

678


20.5.2 Networked process, quality, and plant controls It was widely recognized by the year 2000 [27] that ‘the web changes everything’. Only recently, however, have plastics processing machinery designs begun to expose their networked future. With all due respect to Fig. 20.2, this supervisory control architecture is in the process of becoming obsolete. Already, embedded microcontrollers running open source operating systems such as RT-Linux have brought about more compact, lower cost, networked processing machinery. At the same time, data transfer between sensors and controllers is increasingly using standard protocols such as CANbus (Controller Area Networks) [28]. From a networked machine, it is a small extension for a remote process server to directly coordinate the networked SISO controllers embodied within a factory’s processing machinery. The benefits of such networked capability are too great to ignore at both the tactical and strategic level. At the tactical level, the networked architecture provides continuous monitoring of all the process states to a factory-wide server with extended analysis and data storage capabilities. As such, the server can provide extensive process diagnostics, semi-infinite data logging for quality assurance purposes, semi-infinite recipe storage, centralized factory-side maintenance scheduling, and many other functions. At a more strategic level, the server–client architecture moves the control from the production floor to a management level. While such centralized control may seem counter-intuitive in operations emphasizing worker empowerment, it is a necessity for lights-out operations in which machine settings should not be changed arbitrarily.

20.6


There is extensive literature available that may be of interest to the reader. Researchers and practitioners alike are encouraged to participate in conferences sponsored by the Society of Plastics Engineers as well as the International Polymer Processing Society, both of which have symposia related to process analysis, instrumentation, and control. There are a few books that may also be of particular enduring interest. For process developers, The Measurement, Instrumentation, and Sensors Handbook provides a very comprehensive and modern treatment of the function and application of most sensors used in the plastics industry [26]. For machinery designers, The Mechanical Systems Design Handbook provides a comprehensive review of the modeling, measurement, and control of manufacturing processes [29]. For process engineers, the book Precision Injection Molding provides a sampling of our current understanding of a broad range of issues, both fundamental and applied, that are important in the molding of high precision components [30] ranging from the nano to the


679

macro scale. For design and tooling engineers, the book Injection Mold Design Engineering [31] provides fundamental analysis and examples of mold design.

20.7

Acknowledgments

This work has been partially supported by the National Science Foundation under award DMI-0428366/0428669, as well as MKS Instruments.

20.8

References

1. K. K. Wang and J. Zhou, ‘Concurrent-engineering approach toward the online adaptive control of injection molding process’, CIRP Annals – Manufacturing Technology, vol. 49, pp. 379–382, 2000. 2. S. Kenig, A. Ben-David, M. Omer, and A. Sadeh, ‘Control of properties in injection molding by neural networks’, Engineering Applications of Artificial Intelligence, vol. 14, pp. 819–823, 2001. 3. C. Lotti and R. E. S. Bretas, ‘Using neural networks to predict injection molded iPP shrinkage’, in Society of Plastics Engineers Annual Technical Conference – ANTEC, San Francisco, CA, 2002. 4. T. Petrova and D. Kazmer, ‘Incorporation of phenomenological models in a hybrid neural network for quality control of injection molding’, Polymer-Plastics Technology & Engineering, vol. 38, pp. 1–18, 1999. 5. J. Chen, M. Savage, and J. Zhu, ‘Development of artificial neural networks-based inprocess flash monitoring (ANN-IPFM) system in injection molding’, Lecture notes in computer science, vol. 4493, p. 1165, 2007. 6. J. Zhang, ‘Fault diagnosis in injection moulding via cavity pressure signals’, International Journal of Production Research, vol. 16, pp. 1–14, 2007. 7. S. L. Mok and C. K. Kwong, ‘Application of artificial neural network and fuzzy logic in a case-based system for initial process parameter setting of injection molding’, Journal of Intelligent Manufacturing, vol. 13, pp. 165–176, 2002. 8. T. Samad, ‘Intelligent control in the process industries: Considerations for future research’, in Proceedings of the IEEE Conference on Decision and Control, Kobe, Japan, 1996, pp. 4512–4513. 9. K. T. Chiang, ‘The optimal process conditions of an injection-molded thermoplastic part with a thin shell feature using grey-fuzzy logic: A case study on machining the PC/ABS cell phone shell ’, Materials and Design, vol. 28, pp. 1851–1860, 2007. 10. H. Zhou, P. Zhao, and W. Feng, ‘An integrated intelligent system for injection molding process determination’, Advances in Polymer Technology, vol. 26, pp. 191– 205, 2008. 11. B. Fan, D. O. Kazmer, W. C. Bushko, R. P. Theriault, and A. J. Poslinski, ‘simulation of injection–compression molding for optical media’, Polymer Engineering & Science, vol. 43, pp. 596–606, 2003. 12. K. K. Wang, ‘Twenty years of CIMP research towards CAE for injection molding’, in International Mechanical Engineering Congress and Expositionm Materials Division, Advances in Computer-Aided Engineering (CAE) of Polymer Processing, Chicago, IL, 1994.

680


13. A. D. Drozdov, ‘Finite viscoelasticity and viscoplasticity of semicrystalline polymers’, Continuum Mechanics and Thermodynamics, vol. 19, pp. 111–132, 2007. 14. H. H. Chiang, ‘Simulation and verification of filling and post-filling stages of the injection-molding process’, PhD Dissertation, Cornell University, Ithaca, NY, 1989. 15. M. L. Williams, R. F. Landel, and J. D. Ferry, Journal of the American Chemical Society, vol. 77, p. 3701, 1955. 16. D. O. Kazmer, R. Nageri, V. Kudchakar, B. Fan, and R. X. Gao, ‘Validation of three on-line flow simulations for injection molding’, Polymer Engineering and Science, vol. 46, pp. 274–288, 2006. 17. K. J. Åström and B. Wittenmark, Adaptive Control, 2nd edn. Reading, MA: AddisonWesley, 1995. 18. Z. Chen, L.-S. Turng, and K.-K. Wang, ‘Adaptive online quality control for injectionmolding by monitoring and controlling mold separation’, Polymer Engineering and Science, vol. 46, pp. 569–580, 2006. 19. G. E. P. Box, W. G. Hunter, and J. S. Hunter, ‘Statistics for Experimenters: An Introduction to Design, Data Analysis, and Model Building’, in Wiley Series in Probability and Mathematical Statisics, 1978, pp. 374–452. 20. J. E. Moyal, ‘Stochastic processes and statistical physics’, Journal of the Royal Statistical Society. Series B (Methodological), vol. 11, pp. 150–210, 1949. 21. C. Zhao, F. Wang, N. Lu, and M. Jia, ‘Stage-based soft-transition multiple PCA modeling and on-line monitoring strategy for batch processes’, Journal of Process Control, vol. 17, pp. 728–741, 2007. 22. S. Wold and M. Josefson, ‘Multivariate calibration of analytical data’, in R.A. Meyers (ed.), Encyclopedia of Analytical Chemistry. Chichester, UK: John Wiley & Sons, pp. 9710–9736, 2000. 23. S. Wold and K. O. L. Sundin, ‘Method for monitoring multivariate processes’, US Patent No. 5,949,678, 1999. 24. F. W. Breyfogle, Implementing Six Sigma: Smarter Solutions Using Statistical Methods. New York: John Wiley, 1999. 25. B. Efron and G. Gong, ‘A leisurely look at the bootstrap, the jackknife, and crossvalidation’, The American Statistician, vol. 37, pp. 36–48, 1983. 26. J. G. Webster, The Measurement, Instrumentation, and Sensors Handbook. Boca Raton, FL: IEEE/CRC Press, 1999. 27. S. Ceri, P. Fraternali, and S. Paraboschi, ‘XML: Current developments and future challenges for the database community‘, Proc. 7th Int. Conf. on Extending Database Technology (EDBT), Springer, LNCS, pp. 3–17, 2000. 28. J. Kaiser and M. A. Livani, ‘Invocation of real-time objects in a CAN bus-system’, IEEE International Symposium on Object-oriented Real-time distributed Computing, May, 1998. 29. O. D. Nwokah and Y. Hurmuzlu, The Mechanical Systems Design Handbook. Boca Raton, FL: IEEE/CRC Press, 2001. 30. J. Greener and R. Wimberger-Friedl, Precision Injection Molding Munich: Carl Hanser, 2005. 31. D. Kazmer, Injection Mold Design Engineering. Munich: Carl Hanser, 2007.

21 Computer modeling and simulation of polymer processing T C L I M, SIM University (UniSIM), Singapore.

Abstract: With the advancement of nanotechnology tools, nanoscale polymers (e.g. polymeric nanofibers) have been developed by various methods such as liquid drawing, phase separation, template drawing, electrospinning, etc. In tandem with any progress of polymer processing, modeling techniques of such processes are required to enable better theoretical understanding on how the fundamental properties of the polymer melts/solution depend on the processing conditions and the processing parameters. Unlike bulk polymers, nanoscale polymers have an extremely high surface area to volume ratio. As a result, bulk properties are no longer applicable due to the size effects. In addition, a nanoscale polymer should be viewed as discrete masses with interlocking chains rather than as a continuum solid, by virtue of its size. The use of molecular mechanics would then take into consideration the discrete mass, as well as the influence of surface properties that give rise to size effects. This chapter introduces the concept of interatomic and intermolecular energy potential functions, computational methods, and size effects, that are applicable for understanding the solid and flow properties of nanoscale polymers. Key words: molecular dynamics, Monte Carlo, potential energy functions, size effects.

21.1

Introduction

Structures and components at nanoscale level exhibit superior properties in comparison to those at bulk level [1]. By ‘nanoscale’, one refers to an entity wherein at least one dimension is of the length scale of 100 nm or lower. In view of advantageous properties at nanoscale level, various structures and components have been fabricated [2–6]. Of specific interest at the present day are nanoscale polymers. Recently, success in processing of nanofibers has been reported [7, 8]. The success in developing of nanofibers with due consideration to their potential application calls for the need to develop modeling techniques for pre-production design analysis in order to minimize prototype production. Due to their length scale, nanocomponents are best understood at the molecular level, hence the need to consider interaction amongst atoms or clusters of atoms. This differs from simulation of bulk properties of polymeric solids, which uses principles of solid mechanics. It also differs significantly 681

682


from computer simulation of bulk polymeric melt flow, which uses rheological properties of the polymer, the processing condition, and the mold geometry. The success of any simulation of the structural properties of a nanomaterial, and its processing, are ultimately governed by the accuracy of the interatomic potential used in the simulation, and the length of the time-step increment of the simulation. The interatomic potentials covered include (a) van der Waals systems, (b) covalent interactions, and (c) ionic interactions. Computer simulations of nanoscale components can be traditionally categorized into two broad groups: (a) Monte Carlo, and (b) molecular dynamics simulations. Calculations are based on parameters from the interatomic potentials.

21.2

Interatomic potentials

To consider the interatomic potentials, one begins with the overall energy, i.e. the Hamiltonian N

N –1

N –2

i =1

i =1

i =1

H = Σ Vi + Σ U 2b ( ri ,i+1 ) + Σ U 3b (θ i ,i+1,i+2 ) N –3

N –1

N

i =1

i =1 j ≥i+3

+ Σ U 4b (τ i ,i+1,i+2,i+3 ) + Σ Σ U N b ( rij )

21.1

which is a summation of the kinetic energy, V, and the potential energy, U, for N number of atoms where r, θ, and τ refer to bond length, bond angle and torsional angle respectively. The kinetic energy Vi =

pi 2m

21.2

where pi and m refer to the momentum of the ith atom and its mass, is applicable for analysis of molecules undergoing rapid deformation and/or vibration. In the case of static loading or quasi-static deformation for modeling of mechanical properties, Vi = 0 is assumed. The interatomic potentials, U, can be broadly categorized into bonded (or adjacent, intramolecular) interactions and non-bonded (which includes non-adjacent, intramolecular and intermolecular) interactions. The bonded interactions can be further classified into two-body (stretching) potential, three-body (bending) potential and fourbody (torsional) potential, as shown in Fig. 21.1. The non-bonded interaction occurs between two atoms not directly bonded within the same molecule as well as between atoms of different molecules. Each of these forms of potentials will be furnished in the subsequent sub-sections. A summary of abridged potential functions is shown in Table 21.1.

Computer modeling and simulation of polymer processing j

j

τijk

rij i

683 l

j

i

θ ijk

Stretching

k

i

k

Bending

Twisting

(a)

2b

2b

2b

3b

3b

4b

2b

2b

3b

3b

4b

4b

Nb

(b)

21.1 (a) Types of potential; (b) example of polymeric chain with N = 6. Table 21.1 Summary of interatomic potential functions Harmonic potentials Two-body

1 2

k s (r ij – r ij ,e ) 2

Three-body

1 2

k θ (θ ijk – θ ijk ,e ) 2

Anharmonic potentials

D{1 – exp[–a(rij – rij,e)]}2 1 2

k θ (cos θ ijk – cos θ ijk , e ) 2 5

Σ

a n cos n τ ijkl

Four-body

Not applicable

kτ

Non-bonded

Not applicable

6    12    4 ε  σ  – σ     r ij   r ij    

n =0

684


21.2.1 Bonded potentials The two-body potential considers the energy stored when two adjacent atoms, bonded to one another, undergo relative displacement. The simplest potential is based on a simple harmonic oscillator consisting of a mass m attached to a rigid wall by a spring of stiffness ks. Hence the load Fs applied in stretching the spring by an amount x can be written as Fs = –ksx

21.3

and 2 Fs = ma = m d 2x dt

21.4

according to Hooke’s Law and Newton’s Second Law respectively. Since the summation of energy, H, is constant 1 2

mv 2 + U = H

21.5

Differentiating with respect to t, with due consideration of Eq. (21.4), leads to [9]: dx  F + d U  = 0 dt  s d x 

21.6

Since (dx/dt) ≠ 0, U= –

∫ F dx s

21.7

Substituting Eq. (21.3) into Eq. (21.7) gives U 2b =

1 2

k s ( rij – rij ,e ) 2

21.8

where rij is the bond length between the ith and jth atoms with subscript e referring to equilibrium bond length, and rij – rij,e = x [10–18]. One major disadvantage of the harmonic potential is the symmetry of potential about the equilibrium, and hence predicting too large a force at long distance (see Fig. 21.2). A more accurate two-body potential was given as long ago as 1929 by Morse [19]: U2b = D{1 – exp [– a (rij – rij,e)]}2

21.9

where D and a, the Morse oscillator parameters, are obtained by curvefitting from experimental results. This potential function has been widely used when large bond stretching is anticipated [20–24]. The three-body potential considers the energy stored when the bond angle, θ, formed by three atoms is altered, i.e. bent, due to relative displacement.


685

U

Harmonic

D

Morse

0

R

r

21.2 The advantage of a convergent function, such as Morse potential, over the divergent function (e.g. harmonic potential) for large bond stretching.

The simple harmonic version of the bending can be written in a similar fashion to Eq. (21.8), i.e. U 3b =

1 2

kθ (θ ijk – θ ijk ,e ) 2

21.10

where kθ refers to the bending stiffness by the angle formed by atoms i, j and k [14–18]. A more frequently applied three-body potential 1 2

kθ(cos θijk – cos θijk,e)2

21.11

is known as the Quadratic Potential in Cosine [10–13, 20–25]. Though giving an almost similar result, Eq. (21.11) gives a lower result from Eq. (21.10) when θ > θ0 and vice versa. In addition, Eq. (21.11) is asymmetric about the equilibrium bond angle, and hence more realistic. The four-body potential describes the energy stored when a bond undergoes torsion about its own axis. The harmonic potential, as furnished in Eqs (21.8) and (21.10), is generally not applicable for twisting of bonds, unless one confines the rotation to its periodic angle. This is because the changes to the potential follow the sinusoidal manner. A direct form for a change in torsion angle of τ is the periodic function [14, 15, 17]: U 4b =

1 2

kτ [1 + s cos ( nτ )]

21.12

where kτ is the torsional stiffness, n is the periodicity of the potential (or order of symmetry) and s is the phase factor (or barrier to rotation). An alternate, and more frequently used, function is [10–13]:

686


U 4b = kτ Σ a n cos n τ

21.13

n=0

where the coefficients an [26] are listed in Table 21.2. Other forms of Eq. (21.13) group the torsional stiffness and coefficients such that [16, 18, 21–23, 25]. 3

U 4b = Σ Cn cos n τ

21.14

n=0

where the coefficients Cn (in kJ/mol) [27–29] are given in Table 21.3.

21.2.2 Non-bonded potentials Non-bonded potentials refer to the ‘weak’ van der Waals interaction between atoms not directly bonded together (i.e., at least separated by three covalent bonds in a backbone chain), as well as between atoms of different molecules. The force between such atoms is strongly repulsive when brought closer together but mildly attractive, with a minimum and followed by decay, when separated. There are two major sets of potential functions used: exponential6 potential and the Lennard–Jones (12-6) potential. The exponential-6 function is written as [16, 23, 30].

U N b = Arij–6 + B exp (– Crij )

21.15

or [15]   6  rij     ζ   σ  6     U N b = 4 ε       21.16  exp ζ  1 – σ   –       ζ – 6   ri j      ζ – 6     where A, B and C are coefficients which depend on the system under consideration and ζ is a scaling factor, with ε and σ being constants (also Table 21.2 Coefficients for an by Ryckaert and Bellemans [26] n

0

1

2

3

4

5

an

1

1.31

–1.414

–0.3297

2.828

–3.3943

Table 21.3 Coefficients for Cn n

0

1

2

3

References

Cn Cn Cn Cn Cn

8.832 8.77 8.3 8.37 0

18.087 –18.4096 –18.4096 –18.4096 –4.4

4.88 0 0 0 0

–31.8 26.78 26.78 26.78 6.4

[25, 27] [18, 21, 28, 29] [22, 28, 29] [23, 28, 29] [16]


687

known as the LJ parameters) comparable to the more frequently applied Lennard–Jones function [16–18, 21–24]:

  σ  12  σ  6  U Nb = 4 ε    –     rij     rij 

21.17

Since the potential decays at long distance, Rigby and Roe [10–13] introduced the truncated LJ function such that  4 ε [(σ / rij ) 12 – (σ / rij ) 6 ]; U Nb =  ; 0

rij ≤ 1.5σ rij ≤ 1.5σ

21.18

Other forms of the LJ function have been attempted by Sun et al. [14] U Nb

6   σ  12 σ   = 4ε   – 2     rij     rij 

21.19

by Noid and Pfeffer [20]  σ  9 3  σ  6  U Nb = 2 ε    –    2  rij     rij  

21.20

and for the case of interaction between molecules   σ 12   σ 6   21.21 U Nb = 24 ε  2 13 –  7   r   r  In addition to the covalent interactions and van der Waals interaction, another type of interatomic potential is the interionic potential (also known as the electrostatic or Coulombic interactions), written as Uc = Σ

i≠ j

±q2 – Aq 2 = 2 πε 0 R 4 πε 0 rij

21.22

whereby ε0 = permittivity of free space R = nearest-neighbor separation rij = interionic distance = aijR and the Madelung constant A= Σ

i≠ j

±1 ai j

21.23

Although this function is extremely simple, it is very difficult to cope with computationally because of its long-range nature. The problem in calculating the lattice sum for the Coulombic potential, i.e. the Madelung constant, is

688


that the sum is conditionally convergent; that is, the result depends on the order in which terms are taken in the summation. The interionic potential is, however, generally absent for the case of polymeric systems, except where charges exist in the polymeric chain group. The simplest and yet reliable results can be obtained when the following assumptions are commonly made [10–13, 17, 20–23, 25]: • •

Groups of CH2 and CH3 are lumped together as beads. Models furnished in Section 21.2 are applied without considering crossterms.

Other simplifying assumptions made are: • •

Neglect of two-body potentials, i.e. assumption of rigid bond lengths, for the case where greater compliance is expected in bending [25]. Neglect of four-body potentials for the case where purely zigzag polymeric chains are stretched, i.e. no torsion of bonds [20].

Consideration of individual atoms in the CH2 and CH3 groups has also been made to investigate its effect on accuracy [16, 31]. It has been found that grouping of CH4 and CF4 as spheres gives better static results, while consideration of four-site bonds for these molecules leads to better dynamic properties in comparison to experimental results [31]. In recent years, crossterms have been considered for attaining greater accuracy [32]. Lists of potential function parameters (such as stiffness, with or without cross-terms) and molecular geometries (such as bond length, bond angle and molecular packing) are given in some references [10–13, 15–18, 20–23, 32].

21.3

Monte Carlo approach

21.3.1 Theory A commonly used computer simulation technique for modeling the structures of materials is the Monte Carlo (MC) approach, so called because it uses random numbers in evaluation of particle displacements during the course of a simulation run. One popular way of implementing a MC simulation is to use the Metropolis method. Starting from an initial configuration, say a random distribution of particles (atoms), one atom i is picked randomly and displaced in a random direction by a random amount, from rim to rin , subject to the maximum displacement being the adjustable parameter δrmax. The change in potential energy of the system, δUmn, resulting from the movement of this atom is then calculated for an assumed form for the interatomic potential. If the atomic displacement is ‘downhill’ in energy terms (δUmn ≤ 0), then the new position is accepted unconditionally. If, however, the move is ‘uphill’ in energy terms (δUmn ≥ 0), then the move is accepted only conditionally, subject to the Boltzmann probability factor, exp(–δUmn/kBT).


689

21.3.2 Applications The MC method was the first simulation technique applied, by Metropolis et al. in 1953, to determine the equilibrium properties of solids and liquids [33]. After a sufficient number of configurations have been sampled, the average value of the energy fluctuations yields the constant volume-specific heat and, in the case of solid, fluctuations in the stress tensor determine the elastic constants [34]. The MC approach has also been extended to bulk polymeric systems, notably by Mattice et al. [35–39]. The Monte Carlo approach by Vao-soongnern et al. [40] simulates the molecular structure of a common polymer – polyethylene – as a nanofiber. As before [35], these polymeric chains are placed on a diamond lattice from which every second site has been removed, giving the second nearest neighbor diamond (2nnd) lattice. Simulation of the polymer fiber consists of two parts: (a) generation of the initial polymer fiber structure, and (b) relaxation of the initial structure to thermodynamic equilibrium. The initial step employs only self-avoiding walks with the excluded volume condition. Next, intramolecular and intermolecular interactions are introduced [35, 36]. The initial structure is then relaxed in order to minimize its potential energy by the Dynamic Monte Carlo technique [36–38]. The equilibrated free-standing thin film structure was obtained from previous work by Doruker and Mattice [39]. The fiber is formed by collapsing this thin film after extending, about three to four times, on one side of the periodic box in the perpendicular direction to the normal axis of the thin film surface. This new box is large enough so that no parent chain interacts with its images. Thus, the periodic boundary seems to apply only in one direction for this step. Two methods for fiber formation were introduced: (a) elongation method, and (b) cut-andpress method. In the former, fiber formation is generated in a similar manner to thin film formation from bulk. In the latter, a new fiber is generated from an existing larger fiber structure. Further details of these two methods are given by Vao-soongnern et al. [40]. It was shown that both methods give equivalent fiber structure when the number of chains and the chain length are the same. Among the parameters simulated are (a) shape of the cross-section (almost circular), (b) radial density profile, bead distribution and segregation of chain ends, (c) local orientation, (d) global equilibrium properties of the chains, and (e) surface energy of the fiber.

21.4

Molecular dynamics approach

Molecular dynamics (MD) provides an alternative to the MC approach. A major weakness of the MC method is its biases in generation of random numbers. Moreover, the basis of the MD technique is to numerically integrate the coupled differential equations of motion for a system of many interacting

690


polymer chains. Since such simulations follow the exact dynamic trajectory of each atom, a wealth of information concerning, for example, the structural, dynamics and mechanical properties of the polymer can be computed from the simulated data.

21.4.1 Theory From Newton’s equation of motion, Eq. (21.4), and the force–potential relation, Eq. (21.7), we have

mi

d 2 ri = – dU 2 d ri dt

21.24

The finite difference method is employed to numerically solve the equations of motion such that positions and velocities at time t lead to those at a later time step, t + δt. The differential equation, such as Eq. (21.24), can be solved by a finite difference approach. One of the most commonly used approximations is the Verlet (1967) algorithm [41], which requires knowledge of the particle position ri(t) and the acceleration ai(t) at time t, as well as the position at the previous time step, ri(t – δt). Performing Taylor expansions for ri(t ± δ t) about ri(t) gives ri(t + δt) = ri(t) + vi(t)δt +

1 2

ai(t)(δt)2 + …

21.25

ri(t – δt) = ri(t) – vi(t)δt +

1 2

ai(t)(δt)2 – …

21.26

Substituting Eq. (21.26) into Eq. (21.25) gives the Verlet algorithm: ri(t + δt) = 2ri(t) – ri(t – δt) + ai(t) (δt)2 + …

21.27

which is accurate to the order of (δt)4. Subtracting Eq. (21.26) from Eq. (21.25), we have the velocity vi ( t ) =

ri ( t + δ t ) – ri ( t – δ t ) 2 δt

which is accurate to the order of (δt)2. The algorithm is as follows: • • • • •

Step Step Step Step Step

1: 2: 3: 4: 5:

Start with r(t) and r(t – δt). Calculate acceleration a(t) from Eq. (21.4). Use Eq. (21.27) to calculate r(t – δt). Calculate v(t) from Eq. (21.28) if desired. Replace r(t – δt) with r(t), and r(t) with r(t + δt).

The advantages are: • •

The advancement of positions is all in one step. It is compact and simple to program.

21.28


• •

691

It is time-reversible. It conserves energy well even with a relatively long time step.

The disadvantages are: • •

The velocities at t can be calculated only after r(t + δt) are known. To start a trajectory, we need to know r(t) and r(t – δt) rather than r(t) and v(t).

As an alternative to Eq. (21.28), the Verlet algorithm was extended to the Velocity Verlet algorithm [42]: v ( t + δ t ) = v( t ) + 12 [ a ( t ) + a ( t + δ t )]

21.29

The algorithm is as follows: • • •

Step 1: Start with r(t) and v(t) and calculate a(t). Step 2: Calculate r(t + δt) using Eq. (21.25). Step 3: Calculate velocities at mid-step using

• •

δt   v  t +  = v( t ) + 1 a ( t ) δ t . 2 2  Step 4: Calculate a(t + δt). Step 5: Complete velocity move using δt   v ( t + δt ) = v  t +  + 1 a( t + δt ) δt . 2 2 

The advantages are: • • • • • •

(a) The kinetic energy at time (t + δt) is available. (b) One can start with positions and velocities at time t. (c) It is numerically stable. (d) It is simple to program. (e) It is time-reversible. (f) It conserves energy well even with relatively long time steps.

The disadvantage of the Velocity Verlet algorithm is the requirement of two velocity steps.

21.4.2 Applications The MD approach has been applied for simulating various physical behaviors of polymeric systems [10–13, 16, 17, 20–24]. Prediction of elastic properties has been attempted [14, 15, 18, 25, 43–45]. The first step is the generation of a basic structural cell. Parameters for a proposed polymeric structure as well as its packing, assumed as a crystal structure, are used as inputs using energy minimization. The unit cell is then distorted by a known amount and

692


the resulting change in potential energy is calculated. The resultant force can be calculated from Eq. (21.7) or from F = – dU dr

21.30

In this direct approach, the calculated force and displacement, leading to stress and strain, gives rise to the modulus, e.g. [18]. Stiffness coefficients can also be obtained based on either. • changes in the total potential energy of the structures with applied small strain increments – hence the ‘energy’ approach [44], or • the volume-average change in the atomic-level stress tensors – hence the ‘stress’ approach. In the energy method, 2 U + U– – 2 U0 Cii = 1 ∂ U = + V0 ∂ε i2 V0 ε i2

21.31

where U = total potential energy V0 = volume εi = component i of strain tensor U+ = minimized energy under tension U– = that under compression U0 = energy of the undeformed structure. In the energy method, only the diagonal terms of the stiffness matrix can be calculated accurately because of significant errors involved in evaluation of second derivatives by the finite difference method. In the force method, 2 σ – σ i– ∂σ i Cij = 1 ∂ U = = i+ V0 ∂ε i ∂ε j 2ε j ∂ε j

21.32

where σi = ith component of the internal stress tensor σi+ = component of the stress tensor under tension σi– = component of the stress tensor under compression. The advantage of the stress method is that only the first derivative need be evaluated numerically since the internal stress can be calculated in the minimization process. The result of the generated 6 × 6 matrix is then compared to the idealistic matrix. Shown here is the 6 × 6 matrix for the isotropic amorphous state [43]:


λ + 2 µ    Cij =     

λ λ + 2µ

symm

λ λ λ + 2µ

0 0 0 µ

0 0 0 0 µ

0 0  0  0 0  µ 

693

21.33

where the Lame constants, λ and µ, can be calculated from the generated matrix as 3 6 λ = 1 Σ Cii – 2 Σ Cii 3 i =1 3 i=4

21.34

6 µ = 1 Σ Cii 3 i=4

21.35

and can be related to the following elastic moduli [44]

21.5

 3λ + 2 µ  E = µ   λ+µ 

21.36

G=µ

21.37

K=λ + 2µ 3

21.38

  ν = 1 λ  2 λ + µ

21.39

Continuum and semi-continuum approaches

Due to their long-established history, continuum theories can be employed for polymer nanofibers if the entire nanofiber structure is large enough to be considered as a continuous medium instead of discrete masses of atoms or beads.

21.5.1 Large deformation Large deformation of nanofibers, such as nanoscale biofilaments, calls for the need to consider energy and/or non-linear properties. Two major considerations [46] are the bending energy (assuming the filament neutral axis being inextensible, hence no stretching) and the stretching force. The bending energy, Ub, for a filament of Young’s modulus E, second moment of

694


area I and filament length L being bent into a radius of curvature Rc, is written as [47]

k U b = EIL2 = f 2 2 Rc

∫

L

0

2

 ∂t  ds  ∂s 

21.40

where t is the tangent vector on an arc length s of the filament, with kf being the flexural rigidity of the filament. The stretching force, F, required to extend a filament of length L by an amount x is given by Boal [46] as F=

3k B T x 2ξ p L

21.41

where kB = Boltzmann constant = 1.38 × 10–23 J/K T = absolute temperature (K) ξp = persistent length. The persistent length, ξp, is a parameter which describes the curviness of a filament, such that ξp >> L refers to a straight filament while ξp >h0 the bulk properties dominate, but as h approaches h0 the effect of the surface becomes significant. Surface stresses have been recognized to play an important role when length scales become comparable with the atomic scale. For example,


695

Cammarata [49] investigated the effect of surface stress on the critical thickness criterion during the growth of a thin film, while Rice and Chuang [50] studied the diffusive growth of cavities where surface effects are important. The model by Miller and Shenoy [48] incorporates surface stresses by using a standard continuum mechanics formulation with elastic surfaces. The model was essentially a specific case of more general continuum formulations including surface effects, as presented by Gurtin and Murdoch [51]. The importance of the surface will, according to the model, depend directly on the characteristic length h0 = S/E. This quantity depends on the material considered, the nature of its free surface and the degree to which surface construction affects the elastic response of the atomic bonds. Structures in bending, such as plane-strain plates and beams, were also analyzed and their result compared with direct atomistic simulation [48]. The significance of the size effect in bending is larger than that in uniaxial tension. In bending, Dc is still about 10% too large for plates as thick as 5 nm, and can be in error by as much as 50% for thinner plates. The extent of this error also depends on the nature of the free surface of the material.

21.6

Conclusions

It is expected that, for nanoscale polymers, the large ratio of cross-sectional perimeter to the cross-sectional area will contribute to the increased stiffness. Critical issues involved are: • • • •

Geometrical arrangement of molecules, such as orientation and packing Proper choice of potential parameters and molecular geometries Choice of time step for saving time without significant compromise to accuracy Determination of surface modulus.

The use of molecular dynamics modeling, with particular attention to the effect of free surface, will provide better understanding of the effects of size, polymeric system and molecular arrangement on the final properties of the nanoscale polymer. Such understanding will aid in performance optimization and cost-cutting in engineering design and development.

21.7

References

1. E.W. Wong, P.E. Sheehan and C.M. Lieber, ‘Nanobeam mechanics: elasticity, strength, and toughness of nanorods and nanotubes’, Science, 277, 1971–1975 (1997). 2. N.C. Seeman, ‘The use of branched DNA for nanoscale fabrication’, Nanotechnology, 2, 149–159 (1991). 3. M. Lieberman, M. Tabet, D. Tahmassebi, J. Zhang and T. Sasaki, ‘Self-assembly approach to protein design’, Nanotechnology, 2, 203–205 (1991). 4. J.A. Dagata, W. Tseng, J. Schneir and R.M. Silver, ‘Nanofabrication and characterization using a scanning tunneling microscope’, Nanotechnology, 4, 194–199 (1993).

696


5. A.F. Jankowski and M.A. Wall, ‘Synthesis and characterization of nanophase face-centeredcubic titanium’, NanoStructured Materials, 7, 89–94 (1996). 6. C. Montemagno and G. Bachand, ‘Constructing nanomechanical devices powered by biomolecular motors’, Nanotechnology, 10, 225–231 (1999). 7. T. Ondarcuhu and C. Joachim, ‘Drawing a single nanofibre over hundreds of microns’, Europhysics Letters, 42, 215–220 (1998). 8. A. Theron, E. Zussman and A.L. Yarin, ‘Electrostatic field-assisted alignment of electrospun nanofibres’, Nanotechnology, 12, 384–390 (2001). 9. A. Hinchliffe, Modelling Molecular Structures, Wiley, Chicheter, UK, 1996. 10. D. Rigby and R.J. Roe, ‘Molecular dynamics simulation of polymer liquid and glass. I. glass transition’, J. Chem. Phys., 87, 7285–7292 (1987). 11. D. Rigby and R.J. Roe, ‘Molecular dynamics simulation of polymer liquid and glass. ii. short range order and orientation correlation’, J. Chem. Phys., 89, 5280–5290 (1988). 12. D. Rigby and R.J. Roe, ‘Molecular dynamics simulation of polymer liquid and glass. 3. chain conformation’, Macromol., 22, 2259–2264 (1989). 13. D. Rigby and R.J. Roe, ‘Molecular dynamics simulation of polymer liquid and glass. 4. free-volume distribution’, Macromol., 23, 5312–5319 (1990). 14. Z. Sun, R.J. Morgan and D.N. Lewis, ‘Calculation of crystalline modulus of syndiotactic polystyrene using molecular modelling’, Polymer, 33, 725–727 (1992). 15. C.F. Fan, T. Cagin, Z.M. Chen and K.A. Smith, ‘Molecular modeling of polycarbonate. 1. force field, static structure, and mechanical properties’, Macromol., 27, 2383–2391 (1994). 16. D.W. Noid, R.E. Tuzun and B.G. Sumpter, ‘On the importance of quantum mechanics for nanotechnology’, Nanotechnology, 8, 119–127 (1997). 17. Y. Jin and R.H. Boyd, ‘Subglass Chain dynamics and relaxation in polyethylene: a molecular dynamics simulation study’, J. Chem. Phys., 108, 9912–9923 (1998). 18. K. Fukui, B.G. Sumpter, M.D. Barnes and D.W. Noid, ‘Molecular dynamics studies of the structure and properties of polymer nano-particles’, Comp. Theo. Polym. Sci., 9, 245– 254 (1999). 19. P.M. Morse,‘Diatomic molecules according to the wave mechanics. II Vibration levels’, Physical Reviews, 34, 57 (1929). 20. D.W. Noid and G.A. Pfeffer, ‘Short time molecular dynamics simulations: stressed polyethylene results’, J. Polym. Sci. B. Polym. Phys., 27, 2321–2335 (1989). 21. B.G. Sumpter, D.W. Noid and B. Wunderlich, ‘Computer experiments on the internal dynamics of crystalline polyethylene: mechanistic details of conformational disorder’, J. Chem. Phys., 93, 6875–6889 (1990). 22. D.W. Noid, B.G. Sumpter and B. Wunderlich, ‘Molecular dynamics simulation of twist motion in polyethylene’, Macromol., 24, 4148–4151 (1991). 23. B.G. Sumpter, D.W. Noid and B. Wunderlich, ‘Computational experiments on the motion and generation of defects in polymer crystals’, Macromol., 25, 7247–7255 (1992). 24. R.E. Tuzun, D.W. Noid and B.G. Sumpter, ‘The dynamics of molecular bearings’, Nanotechnology, 6, 64–74 (1995). 25. D. Brown and J.H.R. Clarke, ‘Molecular dynamics of an amorphous polymer under tension. 1. phenomenology’, Macromol., 24, 2075–2082 (1991). 26. J.P. Ryckaert and A. Bellemans, ‘Molecular dynamics of liquid n-butane near its boiling point’, Chem. Phys. Lett., 30, 123 (1975). 27. D.J. Steele,‘ An ab initio investigation of the torsional potential function of n-butane’, J. Chem. Soc., Faraday Trans. II, 81, 1077 (1985). 28. R.H. Boyd, ‘Method for calculation of the conformation of minimum potential-energy and thermodynamic functions of molecules from empirical valence-force potentials – applications to the cyclophanes’, J. Chem. Phys., 49, 2574 (1968).


697

29. R.A. Sorensen, W.B. Liam and R.H. Boyd, ‘Prediction of polymer crystal structures and properties: a method utilizing simultaneous inter- and intramolecular energy minimization’, Macromol., 21, 194 (1988). 30. D.E. Williams, ‘Nonbonded potential parameters derived from crystalline hydrocarbons’, J. Chem. Phys., 47, 4680–4684 (1967). 31. M. Schoen, C. Hoheisel and O. Beyer, ‘Liquid CH4, liquid CF4 and the partially miscible liquid mixture of CH4/CF4: a molecular dynamics study based on both a spherically symmetric and a four-centre lennard–jones model’, Molecular Phys., 58, 699–709 (1986). 32. K. Tashiro, ‘Molecular dynamics calculation to clarify the relationship between structure and mechanical properties of polymer crystals: the case of orthorhombic polyethylene’, Comp. Theo. Polym. Sci., 11, 357–374 (2001). 33. N.A. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller and E. Teller, J. Chem. Phys., 21, 1087 (1953). 34. M.L. Klein and R.D. Murphy, ‘Elastic constants of solid Ar, Kr and Xe: a Monte Carlo Study’, Phys. Rev. B, 6, 2433, 1972. 35. R.F. Rapold and W.L. Mattice, ‘Introduction of short and long range energies to simulate real chains on the 2nnd lattice’, Macromolecules, 29, 2457–2466 (1996). 36. J.H. Cho and W.L. Mattice, ‘Estimation of long-range interaction in coarse-grained rotational isomeric state polyethylene chains on a high coordination lattice’, Macromolecules, 30, 637–644 (1997). 37. P. Doruker and W.L. Mattice, ‘Reverse mapping of coarse-grained polyethylene chains from the second nearest neighbor diamond lattice to an atomistic model in continuous space’, Macromolecules, 30, 5520–5526 (1997). 38. P. Doruker and W.L. Mattice, ‘Dynamics of bulk polyethylene on a high coordination lattice’, Macromolecular Symposia, 133, 47–70 (1998). 39. P. Doruker and W.L. Mattice, ‘Simulation of polyethylene thin films on a high coordination lattice’, Macromolecules, 31, 1418–1426 (1998). 40. V. Vao-soongnern, P. Doruker and W.L. Mattice, ‘Simulation of an amorphous polyethylene nanofiber on a high coordination lattice’, Macromol. Theory Simul., 9, 1–13 (2000). 41. L. Verlet, ‘Computer “experiments” on classical fluids, I Thermodynamical properties of Lennard-Jones molecules’, Physical Reviews, 159, 98 (1967). 42. W.C. Swope, H.C. Anderson, P.H. Berens and K.R. Wilson, ‘A computer simulation method for the calculation of equilibrium constants for the formation of physical clusters of molecules: application to small water clusters’, J. Chem. Phys., 76, 637 (1982). 43. C.F. Fan and S.L. Hsu, ‘Application of the molecular simulation technique to characterize the structure and propeties of an aromatic polysulfone system. 2. Mechanical and thermal properties’, Macromol, 25, 266–270 (1992). 44. M. Hutnik, A.S. Argon and U.W. Suter, ‘Simulation of elastic response in the glassy polycarbonate of 4,4′-isopropylidenediphenol’, Macromol., 26, 1097–1108 (1993). 45. Y. Sakata, A.P. Unwin, T.M. Nicholson and I.M. Ward, ‘The elastic properties of crystalline syndiotactic polypropylene’, Comp. Theo. Polym. Sci., 7, 175–182 (1997). 46. D. Boal, Mechanics of the Cell, Cambridge University Press, Chapter 2 (2002). 47. L.D. Landau and E.M. Lifshitz, Theory of Elasticity, Pergamon Press, Oxford (1986). 48. R.E. Miller and V.B. Shenoy, ‘Size-dependent elastic properties of nanosized structural elements’, Nanotechnology, 11, 139–147 (2000). 49. R.C. Cammarata, ‘Surface and interface stress effects in thin films’, Prog. Surface Sci., 46, 1–38 (1994). 50. J.R. Rice and T.J. Chuang, ‘Energy variations in diffusive cavity growth’, J. Am. Ceram. Soc., 64, 46–53 (1981). 51. M.E. Gurtin and A.I Murdoch, ‘A continuum theory of elastic material surfaces’, Arch. Ration. Mech. Anal., 57, 291–323 (1975).

22 Joining, machining, finishing and decorating of polymers P K DV Y A R L A G A D D A, Queensland University of Technology, Australia and S B U T D E E and A S U E B S O M R A N, King Mongkut’s University of Technology, Thailand

Abstract: This chapter reviews techniques for joining plastics such as friction stir, laser and resistance welding. It also discusses quality issues such as process control and measuring bond strength. Finally, it considers the role of polishing as a finishing technique and its application in such areas as protection against wear. Key words: joining of plastics, friction stir welding of plastics, laser welding of plastics, resistance welding of plastics, polishing of plastics.

22.1

Joining process and applications

22.1.1 Processes and parameters of joining Processes of joining metallic and non-metallic materials can be categorized in two groups. The first group uses energy based on thermal power such as welding, brazing, soldering, etc., and the second group is based on mechanical tightening such as mechanical fitting and assembly. The parameters of the joining system in each process are different depending on the technique of each joining process. In this section we consider the energy used to generate the power to join systems, such as force, thermal, radiation, etc., as in the examples below. Friction stir welding The process of friction stir welding (FSW) was developed by Mishra (2005) as shown in Fig. 22.1. The process is based on the formation of welds on the microstrure. Process parameters (Elangovan, 2007) for FSW affect the welding efficiency; the base metal used, the tool’s rotational speed, the welding speed and the axial force establish the relationship between the base material properties and the FSW process parameters. Different types of material were used for investigation. Microstructure analysis was carried out to check defects 698


699

Downward force z

x

Tool rotation y Friction stir welded region

Retreating side

Welding direction Shoulder

Pin Nugget Advancing side

22.1 Schematic drawing of friction stir welding (Mishra, 2005).

Laser beam Joining force Transparent joining partner

Diffusion zone

Absorbent joining partner

22.2 Principle of transmission laser welding of thermoplastics (Boglea, 2007).

occurring during the FSW process and also to predict defects in the weld connection point. Laser welding In this process (Boglea, 2007), the joining of the two parts uses an optical signal from a laser beam. The two parts are attached and force is applied to control and constrain the parts during the welding process. The laser beam passes through the transparent partner and is absorbed by the absorbent partner, from which heating spreads into the transparent partner. During heating, mixture of materials occurs. The process is depicted in Fig. 22.2.

700


The parameters of laser welding are optimal energy use and minimal thermal diffusion in the joining process. With a high quality laser beam and very small focal spots, the power densities are in the order of I = 106 W/cm2. Laser beams are used for cutting or welding. In the case of polymer welding they should be used ideally in the range of I = 100–300 W/cm2, otherwise they cause material overheating and degradation. Laser energy is diffused into the materials with optimized thermal laser welding. The irradiation strategy determines the movement of the laser beam on the materials and the focus diameter of the laser beam, as shown in Fig. 22.3. The frequency of the circular movement is chosen to be between 1 and 2 circles per second: v (mm/s) 22.1 r (mm) where f is the frequency of the circular movement (the number of circles per second), r is the radius of the circles and v is the feed rate. f = [1, …, 2]

Resistance welding Resistance welding of APC-2/AS4 PEEK/carbon fiber composite using a stainless steel mesh heating element by special specimen geometry gives the skin/stringer configuration used to represent a reinforced aerospace structural joint, as shown by Dube (2007). The welding parameters such as the input power level and clamping distance are the key factors in determining the weld quality and performance. Optimized parameters use short-beam shear tests, ultrasonic C-scan inspection and optical microscopy. Others types of welding process are magnetic pressure seam welding, pulsed current welding, explosive welding, brazing, gas welding, riveting and adhesive joining.

22.1.2 Joining of moldings

Weld seam width (mm)

Mold production is necessary in the joining process. Materials are made up of 316L/17-4PH and 316L/Fe powders, as done by Imgrund (2007). This 0.2 0.1 0

–0.2

0

0.2

0.4

0.6

–0.1 –0.2 Weld seam length (mm)

22.3 Irradiation strategies (Boglea, 2007).

0.8


701

experiment used a fine powder of micron size. The mean particle size and chemical composition of the materials are illustrated in Table 22.1, and all sintering cycles of the experiments are shown in Table 22.2. The powder– binder ratio of the feedstock, the master alloy addition, and the sintering conditions, were all considered in the experimental molding process for joining. The experiment was conducted using co-sintering of the 16L/174PH component. The magnetic properties of the co-injection mold are important to control sintering.

22.1.3 Joining applications Adaptive control for resistant welding process Techniques such as resistance welding are based on an electrical current signal. The design of control systems is a critical issue in the welding technique. Traditional control systems have been superseded by digital technology. In resistance spot welding the nugget size of the arc is difficult to measure accurately. Secondly, input welding energy may not be optimized, causing material distortion. Modern adaptive control systems overcome these problems. Adaptive control can be applied to nonlinear systems. The optimal welding energy model for soft-steel spot welding is determined by equation 22.2: Table 22.1 Mean particle size d50 (µm) and chemical composition (wt%) of powder used in this study (fraction missing to 100%: trace elements P, O, N) Powder

d50*

C

Cr

Ni

Mo

Cu

Mn

Si

Nb

Fe

Fe 316L 17-4PH CrNiMo

6.6 3.6 3.3 3.2

0.05 0.03 0.04 0.04

– 17.2 16.7 52.6

– 11.2 4.9 38.0

– 2.6 0.17 7.0

– – 4.6 –

– 1.8 0.63 0.38

– 0.76 0.59 0.84

– – 0.3 –

99.74 balance balance 0.1

* d50 means that the powders used are of ultra-fine grade (d50 < 7 µm) having almost spherical particles.

Table 22.2 All sintering cycles used for manufacturing magnetic and non-magnetic bimetals Sintering cycle

Bimetal

Heating rate (K/min)

Sintering temp. (°C)

Sintering time (min)

1 2 3 3 4 5 6

316L/17-4PH 316L/17-4PH 316L/17-4PH 316L/Fe 316L MA/Fe 316L MA/Fe 316L MA/Fe

5 5 5 5 5 0.5 0.5

1050 1150 1300 1300 900 900 900

60 60 60 60 60 60 300

702

Advances in polymer processing – α ( h ) 8 2k t   d W ( t ) = Wo ( h )  1 – e   

22.2

where: d is the nugget diameter, d = (5–6) h (mm) h is the thickness of welded plate (mm) k is the heat diffusivity (mm2/s) α(h) is the thickness weighting factor, α(h) = 2h ± 0.5 Wo(h) is the total input energy to the welded plates in the entire welding process (including the energy of making the nugget and loss energy). The model reference adaptive fuzzy control system (MRAFCS) for spot welding by Chen (1997) is shown in Fig. 22.4. This was derived by combining the welding energy reference model with a fuzzy controller. The input energy reference model, the fuzzy controller and some calculations are executed by the computer.

22.1.4 Bond strength of joining methods Combination of materials is very important in modern manufacture. Of critical importance is the performance of the joining interface between similar or dissimilar materials. Measurement of the bond strength is needed to verify the strength and durability at the joining point. The bond strengths of the combined materials are tested and revealed by experimental methods. Polymers such as poly(methyl methacrylate) (PMMA) and Homalite materials are used to test bond strength as shown in Fig. 22.5. The Iosipescu shear test was Ugd

e (t ) Fuzzy rule

Fuzzy

de (t )

s

Error

inference Output Control algorithm

IERM

u (t )

Controller

–

plant

+

Expect output

Input energy reference

22.4 Block diagram of MRAFCS architecture.


(a)

703

(b)

Adhesive

Polymer

Polymer Polymer

x2

135°

Adhesive bonding

x1

Interface

(c)

22.5 Iosipescu shear test for (a) pure and (b) bonded polymers; (c) butt-joint tensile test (Roy Xu, 2004). Table 22.3 Measured bonding strength data Polymer // adhesive // Polymer


Shear strength (MPa)

Homalite Homalite Homalite Homalite Homalite

28 7.74 6.99 6.75 1.53

>23.26 >21.65 12.58 7.47 0.81

// // // // //

Polyester // Homalite Wedon-10 // Homalite 330 // Homalite 384 // Homalite 583 // Homalite

used for a uniform shear stress distribution to find the strength of materials in which five different kinds of interfacial bonding adhesives – Weldon-10, polyester, Loctite 330, Loctite 384 and Loctite 5083 – were used to bond the interfaces. The strengths of the bonding adhesives and polymers are given in Table 22.3. As an example, the bond strengths of PMMA and Homalite were tested with a shear load applied to the specimen materials, the result being shown in Fig. 22.6. Loads applied to test the specimens varied from 300 N to 500 N. The severity of the cracking observed from the fringe pattern increased when the applied load was increased.

22.1.5 Joining of plastic There are many methods of joining plastic materials using advanced technology, such as thermal, microwave, ultrasonic, etc. These are limited by the energy

704

PMMA

Bonding Homalite

PMMA Field of view Field of view

Simulation

Experiments

(b)

(a) Applied load = 150 N

(d)

(c) (a)

Applied load = 400 N (b)

22.6 Result of bond strength test: (a) PMMA joining; (b) Homalite joining (Roy Xu, 2004).


Homalite

Bonding


705

source in the joining process. To cope with this limitation, a new plastic joining method for fixing bars with a hot forged plate is proposed by Matsumoto (2008). Figure 22.7 shows ‘indentation joining’, a newly developed plastic joining method for fixing a cold bar with a hot forged plate without lubrication on press. High strength and seizure surface of the joining material are achieved with good quality after the cooling process. The relationship between indentation pressure and plate temperature is shown in Figs 22.8, 22.9 and 22.10. The indentation pressure is calculated by equation 22.3:

Indentation pressure = maximum indentation of load bar cross-sectional area of bar

22.3

where the cross-sectional area is π(DB/2)2. To evaluate the bonding strength between the bar and the plate, the shear bonding stress (PD in Fig. 22.10) is calculated as equation 22.4:

Shear bonding stress =

maximum drawing load of indented bar interface surface area of bar-plate 22.4

Figure 22.10 shows two ways of joining a plate and bar. In the case of the plate without a hole, the plate sinks significantly and a long gap between the bar and the plate is observed around the front end of the bar. This does not occur when a plate with a hole is used in the joining process. Thus the higher bonding strength is achieved using the plate with a hole in joining. In the Bar (room temp.)

Indentation

Cooling Slug

Hot forged part (high temp.)

Seizure

Clamping Cross-sectional view Cross-sectional view

22.7 Details of the bar and the plate after indentation (Matsumoto, 2008).

706


Bar

Bar

Fattening of bar

Plate

Plate

10 mm 10 mm Slug (a) Failure (Plate temperature in indentation TP = 750°C)

(b) Success (TP = 950°C)

22.8 Details of the bar and the plate after indentation joining (clearance between bar and die. CL = 0.2 mm, plate thickness/bar diameter tP / DB = 1.0 ) (Matsumoto, 2008). 1200

1.4 1.2

Failure

Yield srength of bar material at room temp.

800

1.0 0.8 0.6

1000

Successful (Bar fattening)

Successful

600 400

0.4 0.2

200

Indentation pressure PI/MPa

Indentation pressure/ Yield strength of bar PI/σy

1.6

0 0 600 700 800 900 1000 1100 1200 Plate temperature in indentation jointing TP /°C

22.9 Relation between plate temperature and indentation pressure in indentation joining (clearance between bar and die CL = 0.2 mm, tP/DB = 1.0, σY = 780 MPa).

fatigue test of bond strength as shown in Fig. 22.11, the shear bonding stress after 10 000 cycles of loading is lower than the initial condition by approximately 15%. The rate of reduction of the bonding stress after the fatigue test is similar to that due to mechanical clamping.

22.2

Machining process

Advanced machining processes can achieve a high material removal rate and good quality of surface smoothness. They can remove material with a high

Joining, machining, finishing and decorating of polymers Distance from upper surface of plate

707

Sinking of plate

0mm

Bar

2mm

Plate 4mm

6mm

Observation area

8mm

Gap (no joining)

Plate Bar (a) Hole diameter of plate/ Bar diametr DH/DB = 0

1mm

Plate Bar (b)D H/D B = 0.88

22.10 Cross-sectional views of interface of indented bar–plate (TP = 950°C, CL = 0.2 mm, tP/DB = 1.0) (Matsumoto, 2008).

Shear bonding stress PD/MPa

160 140 120 100 80

Plate Bar

60 Repeating load 40 20 0 2

4 6 8 10 Cycles of loading in fatigue test (×103)

12

22.11 Relation between cycles of loading in fatigue test and shear bonding stress of indented bar (TP = 950°C, CL = 0.2 mm, tp/DB = 1.0, DH/DB = 0.88) (Matsumoto, 2008).

performance level in the range up to 150–1500 cm3/min at cutting speeds up to 8 m/min. Dry and semi-dry cutting is widely used in order to be more environmentally friendly. The super-hard cutting tool materials embody hardness levels in the range 3000–9000 HV with toughness levels exceeding 1000 MPa. Coated tool materials can achieve high quality surface smoothness. Ultra-precision cutting can operate in the range

Advances in Polymer Processing: From Macro- to Nano-Scales

Polymer Composites: From Nano- to Macro-Scale

Polymer Structure Characterization: From Nano to Macro Organization

Polymer Processing

Processing of Composites (Progress in Polymer Processing)

Polymer Libraries (Advances in Polymer Science)

Polymer Thermodynamics: Liquid Polymer-Containing Mixtures (Advances in Polymer Science)

Fundamentals of Polymer Processing

Advanced Polymer Processing Operations

Advanced Polymer Processing Operations

Principles of Polymer Processing

Polymer Thermodynamics: Liquid Polymer-Containing Mixtures (Advances in Polymer Science)

Interphases and Mesophases in Polymer Crystallization. Advances in Polymer Science

Advances in Cardiac Signal Processing

Recent Advances in Signal Processing

Advances in speech signal processing

New Polymer Materials (Advances in Polymer Science,Vol.94)

Rheology for Polymer Melt Processing

Polymer Membranes Biomembranes (Advances in Polymer Science, Volume 224)

Biomedical Applications Polymer Blends (Advances in Polymer Science)

MACRO TO NANO SPECTROSCOPY

Polymer Physics and Engineering, Volume 154 (Advances in Polymer Science)

Polymer Membranes Biomembranes (Advances in Polymer Science, Volume 224)

Fuel Cells II (Advances in Polymer Science)

Silicon Polymers (Advances in Polymer Science, 235)

Multicomponent Polymer Materials (Advances in Chemistry 211)

Emissive Materials - Nanomaterials (Advances in Polymer Science)

Nanocomposites (Advances in Polymer Science) (vol 153)

Advances in Magnetism: From Molecules to Materials

Advances In Polymer Science Vol 125: STATISTICAL MECHANICS, DEFORMATION, ULTRASONIC SPECTROSCOPY (Advances in Polymer Science)

Emissive Materials - Nanomaterials (Advances in Polymer Science)

Advances in Polymer Processing: From Macro- to Nano-Scales

Polymer Composites: From Nano- to Macro-Scale

Polymer Structure Characterization: From Nano to Macro Organization

Polymer Processing

Processing of Composites (Progress in Polymer Processing)

Polymer Libraries (Advances in Polymer Science)

Polymer Thermodynamics: Liquid Polymer-Containing Mixtures (Advances in Polymer Science)

Fundamentals of Polymer Processing

Advanced Polymer Processing Operations

Advanced Polymer Processing Operations

Principles of Polymer Processing

Polymer Thermodynamics: Liquid Polymer-Containing Mixtures (Advances in Polymer Science)

Interphases and Mesophases in Polymer Crystallization. Advances in Polymer Science

Advances in Cardiac Signal Processing

Recent Advances in Signal Processing

Advances in speech signal processing

New Polymer Materials (Advances in Polymer Science,Vol.94)

Rheology for Polymer Melt Processing

Polymer Membranes Biomembranes (Advances in Polymer Science, Volume 224)

Biomedical Applications Polymer Blends (Advances in Polymer Science)

MACRO TO NANO SPECTROSCOPY

Polymer Physics and Engineering, Volume 154 (Advances in Polymer Science)

Polymer Membranes Biomembranes (Advances in Polymer Science, Volume 224)

Fuel Cells II (Advances in Polymer Science)

Silicon Polymers (Advances in Polymer Science, 235)

Multicomponent Polymer Materials (Advances in Chemistry 211)

Emissive Materials - Nanomaterials (Advances in Polymer Science)

Nanocomposites (Advances in Polymer Science) (vol 153)

Advances in Magnetism: From Molecules to Materials

Advances In Polymer Science Vol 125: STATISTICAL MECHANICS, DEFORMATION, ULTRASONIC SPECTROSCOPY (Advances in Polymer Science)

Emissive Materials - Nanomaterials (Advances in Polymer Science)

Recommend Documents