Studies in Surface Science and Catalysis 93 CHARACTERIZATION AND CHEMICAL MODIFICATION OF THE SILICA SURFACE
This Pag...
75 downloads
1547 Views
20MB 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
Studies in Surface Science and Catalysis 93 CHARACTERIZATION AND CHEMICAL MODIFICATION OF THE SILICA SURFACE
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis Advisory Editors: B. Delmon and J.T. Yates Vol. 93
CHARACTERIZATION AND CHEMICAL MODIFICATION OF THE SILICA SURFACE E. E Vansant, R Van Der Voort and K. C. Vrancken
University of Antwerp (U.LA.), Laboratory of Inorganic Chemistry, Universiteitsplein 1, B-2610 Wilrijk, Belgium
1995 ELSEVIER
A m s t e r d a m m L a u s a n n e - - N e w York m O x f o r d - - S h a n n o n - - T o k y o
ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 P.O. Box 211, 1000 AE Amsterdam, The Netherlands
ISBN 0-444-81928-2 91995 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department, P.O. Box 521, 1000 AM Amsterdam, The Netherlands. Special regulations for readers in the U.S.A.- This publication has been registered with the Copyright Clearance Center Inc. (CCC), Salem, Massachusetts. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the U.S.A. All other copyright questions, including photocopying outside of the U.S.A., should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. This book is printed on acid-free paper. Printed in The Netherlands
Preface
Oxide surface materials are widely used in many applications, in particular where chemically modified oxide surfaces are involved. Indeed, in disciplines as separations, catalysis, bioengineering, electronics, ceramics, etc. modified oxide surfaces are very important. In all cases, the knowledge of their chemical and surface characteristics is of great importance for the understanding and eventual improvement of their performances. The purpose of this book is essentially to cover techniques and procedures characterizing and modifying the silica surface. We hope that this book will be useful for all those who, working at the graduate student or research worker level, are interested in the chemistry of silica and chemically modified oxide surfaces. The content of this book reviews the latest developments in the characterization and chemical modification of silica surfaces. No attempt has been made to survey exhaustively the literature of any topic. The material has been collected from recent publications and own research work in this field. Also, recent disclosures of research activities in the former USSR are documented in detail in the text. The book is divided into three major parts. The first part (Part I) reviews the characterization of the silica surface, discussing the preparation and properties of pure silica (Chapter 1), the physical characterization of the silica surface (Chapter 2), the chemistry of silica (Chapter 3), the quantification of the silanol number (Chapter 4), the distribution of the silanol types and their desorption energies (Chapter 5), the effect of surface morphology on the dehydroxilation behaviour (Chapter 6) and the related silicate materials (Chapter 7). Part II discusses the chemical modification of
vi the silica surface, coveting the procedures for a chemical modification of the silica surface and their applications (Chapter 8), the modification with silicon compounds (Chapter 9), the modification with boron compounds (Chapter 10) and the use of other modifiers (Chapter 11) including the ammoniation of modified silicas to introduce functional groups on the surface (Chapter 12). The third part (Part III) describes the chemical surface coating technique (Chapter 14) with respect to other surface coating methods (Chapter 13). Furthermore, the principles of the most frequently used surface analysis techniques are briefly described in the annexes A, B, C and D. We are grateful to the authors and publishers (The American Chemical Society, Plenum Press, The American Institute of Chemical Engineering, Elsevier Science Publishers, The Royal Society of Chemistry, Academic Press, VCH Verlaggeselschaft mbH, VSP International Science Publishers, Marcel Dekker Inc., John Wiley and Sons, Degussa AG, Gordon and Breach Science Publishers, Springer Verlag INC, and The National Research Council of Canada) who granted us permission to reproduce illustrations from their books, articles and journals. We wish to express our appreciation to Prof. J. Riga (University of Namur), Prof. P. Grobet (Catholic University of Leuven), Dr. B. Gilissen (Flemish Institute for Technological Research), Prof. P. Geladi (University of Umeh-Sweden) and Prof. J. Van Landuyt (University of Antwerp) for their research collaboration. Very special thanks to Mr. K. Possemiers for his help in the writing of Chapter 10 and Mrs. L.F. Quiroz and Mrs. I. De Ridder for typing and setting every letter and every figure of this work. Finally, we want to thank the NFWO (National Science Foundation of Belgium), the University of Antwerp (UIA), the Commission of the European Communities and the IUAP for their financial support.
E.F. Vansant
P. Van Der Voort K.C. Vrancken
vii Etienne F. Vansant is presently Professor in Inorganic Chemistry at the University of
Antwerp (UIA), Belgium. He has served as Visiting Professor, Research Associate and Invited Consultant in several universities and companies in the field of material science. Professor Vansant's research interests include the optimalization of gas separation and purification techniques for both industrial and ecological purposes, the conditioning and purification of waste waters and soils, and the development of new materials. Pascal Van Der Voort is presently senior research assistant at the Laboratory of
Inorganic Chemistry (University of Antwerp). He was granted the title of Doctor in Sciences for his pioneering work on Chemical Surface Coating. After that, Dr. Van Der Voort stayed strongly involved in the research on chemically modified oxide surfaces and their application as ceramics and catalysts. Karl C. Vrancken is presently a researcher at the Laboratory of Inorganic Chemistry
(University of Antwerp). In this position he substantiated the development of the Chemical Surface Coating method. In his young career as a Doctor in Sciences, he gained full expertise in organosilane chemistry. His current work is concerned with surface modifications and advanced materials preparation.
This Page Intentionally Left Blank
ix
Table of contents PART I CHAPTER 1
CHARACTERIZATION OF THE SILICA SURFACE SILICA: PREPARATION AND PROPERTIES
1. Introduction: natural and synthetic silica 2. Preparation of amorphous silica 2.1 Silica sols and gels 2.2 Pyrogenic silicas 2.3 Precipitated silicas 2.4 Comparative properties of silicas 3. The effect of particle packing on the surface area 4. The sol - gel process 4.1 Hydrolysis 4.2 Condensation 4.3 Gelation 4.4 Aging 4.5 Drying 4.6 Stabilization, consolidation 4.7 Advantages and applications 5. Applications of pure silicas and powders 5.1 Porosity 5.2 Active surface 5.3 Hardness 5.4 Particle size References CHAPTER 2
PHYSICAL CHARACTERIZATION OF THE SILICA SURFACE
1. The isotherm 2. Determination of the specific surface area 3. Pore size characterization 3.1 Average pore diameter 3.2 Pore size distributions 3.2.1 Mesoporous silicas 3.2.2 Microporous silicas 3.2.3 Macroporous silicas 3.3 Modelling theories References
3 4 4 7 8 9 13 15 16 19 22 24 24 25 27 27 27 28 28 28 28
31 32 34 37 37 38 38 41 53 55 55
CHAPTER 3
THE SURFACE CHEMISTRY OF SILICA
1. Surface species 2. The hydroxylated surface 3. Dehydroxylation and rehydroxylation 3.1 Dehydration and dehydroxylation 3.2 Rehydration and rehydroxylation 4. Infrared study of the silica surface 5. Deuteration 5.1 Deuteration of hydroxylated silica 5.2 Deuteration of partially dehydroxylated silica References CHAPTER 4
QUANTIFICATION OF THE SILANOL NUMBER
1. Introduction 2. The silanol number- a physicochemical constant 3. Hexamethyldisilazane as a silanol titrator 4. Conclusion References CHAPTER 5
THE DISTRIBUTION OF THE SILANOL TYPES AND THEIR DESORPTION ENERGIES
1. The infrared models 2. The NMR models 3. T P D - models 4. Zhuravlev's model 5. Comparison and conclusion References CHAPTER 6
THE EFFECT OF SURFACE MORPHOLOGY ON THE DEHYDROXYLATION BEHAVIOUR
References CHAPTER 7 1. 2. 3. 4. 5. 6.
59 59 62 62 62 64 65 68 68 71 74 79 79 81 83 88 89
93 94 104 109 118 122 125
127 131
RELATED MATERIALS: SILICATES
Simple orthosilicates Other discrete, noncyclic silicate anions Cyclic silicate anions Infinite chain anions Infinite sheet anions Minerals 6.1 Clays 6.2 Zeolites References
133 133 133 134 134 135 135 136 137 145
xi PART II
CHAPTER 8
CHEMICAL MODIFICATION OF THE SILICA SURFACE CHEMICAL MODIFICATION OF SILICA: APPLICATIONS AND PROCEDURES
1. Applications of modified silica gel 1.1 Introduction 1.2 Analytical 1.2.1 High Performance Liquid Chromatography (HPLC) 1.2.2 Ion Exchange Chromatography 1.2.3 Size Exclusion Chromatography 1.2.4 Gas Chromatography 1.2.5 Preconcentration of trace metals 1.3 Chemical 1.3.1 Heterogeneous catalysis 1.3.2 Phase transfer catalysis 1.4 Biochemical 1.4.1 Immobilization of enzymes 1.4.2 Affinity Chromatography (AC) 1.5 Industrial 1.5.1 Composites 1.5.2 S-emission monitoring 1.5.3 Modification of thin SiO/layers 2. Modification procedures 2.1 Introduction 2.2 Sol- gel 2.3 Aqueous solvent 2.4 Organic solvent 2.5 Self- assembled monolayers 2.6 Hydride intermediate 2.7 Vapour- phase reactions References CHAPTER 9
MODIFICATION WITH SILICON COMPOUNDS: MECHANISTIC STUDIES
1. Modification with aminosilanes 1.1 Modification of silica gel with aminosilanes from aqueous solution 1.1.1 The APTS structure in aqueous solution 1.1.2 Modification from aqueous solvent 1.2 The influence of water in the reaction of silica with APTS 1.2.1 Reaction on hydrated silica gel 1.2.2 Water in the solvent- silica system 1.2.3 Air humidity in the curing stage 1.2.4 Conclusion
149 149 149 151 154 158 158 159 159 160 160 162 163 163 165 167 168 171 171 172 172 173 175 177 180 183 185 187
193 194 196 196 198 200 200 205 206 209
o~
Xll
1.3 Modelling of the interactions in the loading step 1.3.1 Adsorption kinetics and isotherms 1.3.2 Limitations by substrate structure 1.3.3 Conclusion 1.4 Characterization of the aminosilane modified silica 1.4.1 Chemical condensation of aminosilanes with the silica surface 1.4.2 Limitations by substrate structure 1.4.3 Conclusion 1.5 Interactions at the amine side 1.5.1 Stability and surface interaction 1.5.2 Coordination of the amine group 1.5.3 Conclusion 1.6 The role of silanols in the modification of silica with aminosilanes 1.6.1 The role of surface silanols 1.6.2 The role of silane silanols 1.6.3 Conclusion 2. Modification with chlorosilanes 2.1 Vapour- phase reactions with (methyl) chlorosilanes 2.1.1 A review 2.1.2 Effectiveness, surface coverage and stoichiometry 2.2 Liquid - phase reactions with alkylchlorosilanes (Cg - C~g) 3. Modification with other silicon compounds 3.1 Halogenosilanes (R4.,SiBr., R4_.SiI.) 3.2 Alkoxysilanes (R4_,Si(OR).) 3.3 Alkylsiloxanes References CHAPTER 10
MODIFICATION WITH BORON COMPOUNDS
1. Introduction 2. Modification with boronhalides 2.1 Reaction mechanism 2.2 Influencing factors 2.2.1 The effect of the pretreatment temperature 2.2.2 The effect of the reaction temperature 2.2.3 The effect of the degassing temperature 2.2.4 The effect of the transition state 2.2.4 The effect of the reaction time 2.3 Stability of the modified surface 3. Modification of the silica surface with diborane 3.1 Introduction 3.2 Reaction mechanism 3.2.1 Reaction with hydroxyls 3.2.2 Reaction with siloxanes 3.3 Hydroxyl specificity 3.4 Kinetic study of the boronation reaction
209 210 219 225 225 226 233 239 240 240 243 244 255 256 262 265 266 266 266 270 282 288 288 289 290 292 299 299 300 300 304 304 311 312 313 315 315 319 319 320 320 331 335 339
xiii
CHAPTER 11
3.5 Influence of the boronation reaction on porosity of the silica References
347 353
MODIFICATION WITH OTHER COMPOUNDS
357
1. Modification with T i - compounds 1.1 Reaction mechanisms 1.2 Analysis techniques 1.3 Applications 2. Modification with A1- compounds 2.1 Reaction mechanisms 2.2 Analysis techniques 2.3 Applications 2.3.1 Model systems for cracking catalysts 2.3.2 Ziegler - Natta catalysts 3. Modification with P - compounds 3.1 Reaction mechanisms 3.2 Analysis techniques 3.3 Applications 4. Modification with transition metal compounds References CHAPTER 12
AMMONIATION OF MODIFIED SILICA: INTRODUCTION OF FUNCTIONAL GROUPS
1. Introduction 2. Ammoniation of the unmodified silica 2.1 At room temperature 2.2 At moderate temperatures 2.3 At high temperatures 2.4 At very high temperatures 2.5 Summary 3. Ammoniation of halogenated silica 4. Ammoniation of chlorosilylated silica 4.1 Room temperature ammoniation 4.2 Ammoniation at higher temperatures 4.2.1 Silazane functions 4.2.2 Nitride functions 4.3 Diagnostic value of the Si - H vibration 4.3.1 Quantitative determination of the surface species, using PLS 4.3.2 Mechanistic study of the perturbation of the silane band 4.3.3 Conclusion 5. Ammoniation of silica, activated with boron compounds 5.1 The boron - nitrogen chemistry 5.2 Ammoniation of boronated silica 5.3 Ammoniation of silica, activated with BC13
357 357 361 363 364 364 368 368 368 369 370 370 374 375 375 378
383 383 384 384 386 388 389 389 390 393 393 396 399 401 404 405 413 417 418 418 422 425
xiv PART IH
CHEMICAL SURFACE COATING
CHAPTER 13
COATING TECHNIQUES
437
1. Chemical Vapour Deposition (CVD) 1.1 Conventional Chemical Vapour Deposition 1.2 Metal - organic CVD (MOCVD) 1.3 Plasma- Enhanced CVD (PECVD) 1.4 Laser CVD (LCVD) 1.5 Fluidized CVD (FBCVD) 1.6 Chemical Vapour Infiltration (CVI) 1.7 Materials and applications 2. Physical Vapour Deposition (PVD) 2.1 Thermal Physical Vapour Deposition (PVD) 2.2 Sputtering 2.3 Ion plating 3. Atomic Layer Epitaxy 4. Molecular Layering 5. Chemical Surface Coating References
437 438 440 441 442 443 445 445 449 450 451 452 453 455 458 458
CHAPTER 14
CHEMICAL SURFACE COATING
1. Principles of Chemical Surface Coating 2. Gas-phase modification 2.1 Creation of the precursor 2.1.1 Cycle 1 2.1.2 Cycles 2 - 4 2.1.3 Further optimization of the CSC precursor 2.2 Thermal conversion of the precursor 2.2.1 Quantitative determination of the surface species 2.2.2 Conversion of silanes, some reflections 2.2.3 Conclusion 3. Liquid phase CSC 3.1 The synthesis of silicon carbides 3.2 Silicon (oxy)carbide coatings by Chemical Surface Coating 3.3 Conclusion References
461 461 463 464 464 465 469 471 471 474 475 476 476 478 485 485
XV
APPENDICES
SURFACE ANALYSIS TECHNIQUES
APPENDIX A
FTIR-PAS Fourier Transform Infrared Spectroscopy with Photoacoustic detection
1. Introduction 2. Fourier Transform Infrared Spectroscopy 2.1 Principles 2.2 The interferogram 2.3 Apodization and phase correction 3. Photoacoustic spectroscopy 3.1 Principles Bibliography APPENDIX B
XPS X-Ray Photoelectron Spectroscopy
1. Principles 2. Instrumentation Bibliography APPENDIX C
489 491 491 492 495 496 496 499
501 501 503 504
29Si CP MAS NMR
Cross Polarization Magic Angle Nuclear Magnetic Resonance 1. Principles 2. Magic Angle Spinning 3. Cross Polarization Bibliography APPENDIX D
489
SURFACE SCIENCE TECHNIQUES
505 505 507 509 510 511
AUTHOR INDEX
529
SUBJECT INDEX
543
This Page Intentionally Left Blank
9iiiiiiiiiiiiiiiiiiii!i!i!iiiiii#ili!iiiii!iiiiiiiiiiiiiiiiiiii
..........................
iiiiiiiiiii!iiiiiiii!iiiiiii!iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
~
i~!i~ii~i~iiii~iiiii~iiiiiiii~iiiiiii~ii~iii~i~~iiii!~iiii~liliii~!iii~!i!ii!iiii~ii~ii~ii!iiiili~i~i~i~ii~i~i~i~i~i~iiii!iiiii~~~iii~i~ i~iiii!i!iiiiii~iiiii!i~i!~i~ii~iiiiii~i~!iii~i~ii!i~iiiiii~ii~iiiiiiii~iii~li~iii~iiiiiiiiiii
~iii~iii~i~iiiiiiiiiiii~iiiiiiii!ii~iiiiiiiiiiiii~i~iiiiiii~i~ii~iiiiii~i~i~iiiiii~iii!iiii~iiiiiii~iiii~ii~
'.iiiiiii . . .i.ii.i.ii. .iiiiiiiiiiiii . . . . . .ii!iii!iiiiiiiiii!i . . . . . . . .iiiiiiii!iiiiiiiiii!iliiiiii!iiii~, . . . . .~. . . .~. U ~~,, ~ ~,~~,~,~,, ~,~,~i,~,~~,~,, ~,~,i~,~~,~,~,~,~,~,,,~,~, ,~, , , ,................... , ~,............. 9 ~i~i~i~i~i~i~i~ t~,~,~,~,l~oooo,,~,~t,oooo,l~,~~,~l,l,t,~,~,~i~~ ,~,~,t~,l ~iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii~iiiiil ~i~i~i~i~i~i~i~i~i~i~i~i~i~tI~1I~i~iii~Ii~...... I~oi~oo~o~io~II~ ~'~ '~""~"''"'iiiiiiliiiiiiiiiiiiiiiiiiiiii#iiiiiiiiiiiiiiill i ~ii~iiiiiiii~iii~i~iiii!iiiii~iiiiiiiiii~i~i~!iiiiiiiii~iG~iiii~iiiiiiii~i~i~iiiiii~iiii~i~iii~iiiiii~iiiiii~iiiiiiiiii~i!iiiiii~i~i~iiiiiiiiiiiiiii~i~i~iiiiiiiiiii~i~i~iiiiiiiiiiii~iiiiiii~iii~iiiiiiiiiiiiiiiiiiiii~iii
iiiiiiiiiiiili!ii!i!!iiiii!iii!i!i!iiiiiii!ii!i!i!i!iil!i!iiiiiiiiiiiiiiiiiiiiii~iii!ii!ii i!ii!ii!i!ii iii!ii!i!i!iiiiiiii~iii!i!iiiiiiiiiiiiiii~iiiiiiiiiii!~iiiiiiiii!iiiiii•iiii•~•iiiii•i•i••••iiii••iiiii•iii•!iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii i!iii•iii•i•i•i•i••iii•i•ii•iiii•i•iiiiii•iiii•iiiiiii ~i~i~!~i~i~i~i~i~i~i~i~i~i~i~!i~iii~i~i~i~i~i~i~i~,i~,,i~li~ii~ii~ii~iiiiiii~ii~ii~ii~iiiiii~ili~iii~ilii~ii~ii~i~ii~iii~iii~!i~iiii~ii~ii~ii~iii~iiiiiiiii!i~ii~!~~ii~l~i~i~i~176 iii
iiiii~iii!iii~iii!ii!ii~~i!iiii~iiiii~iiii!~iiiiiiiiiiiiiiiii!iiiiiiii~iii!iiii~iii!iiii
This Page Intentionally Left Blank
PART 1: THE SILICA SURFACE
Chapter 1
Silica:
preparation and properties
1 Introduction: natural and synthetic silica The name silica comprises a large class of products with the general formula SiO2 or SiO2.xH20. Silica is a naturally occurring material in minerals, such as quartz and flint, and in plants such as bamboo, rice and barley. Most of the silica used in chemical applications however, has a synthetic origin. In its natural form it mostly occurs as a crystalline phase. Various phases may be formed, depending on temperature, pressure and degree of hydration. At atmospheric pressure the anhydrous crystalline silica may be classified in the following phases, according to the temperature:
1143 K quartz
,
'~
1743 K tridymite
"
cristobalite
At 1973 K cristobalite is transformed to amorphous vitreous silica glass. The crystalline form involves a high degree of ordering in a dense structure. The active surface, which may participate in any chemical or physical interaction, is limited to the external surface of the crystalline particles. The specific surface area therefore is similar to the geometric surface.
Amorphous silica occurs in various forms. According to the application, fibres, sheets, sols, gels and powders may be fabricated. A clear view upon the properties of all amorphous forms can be obtained from the process used for their preparation. A main feature of interest in this work is the porosity of the amorphous silica forms. Porosity introduces a large surface area inside the silica particles. As interphase processes require a large surface/mass ratio, amorphous silicas are far more interesting for chemical and physical applications than their crystalline counterparts.
2 Preparation of amorphous silica By changing the method and specific parameters of the silica preparation, surface area, pore volume, pore size and particle size are, to some extent, independently controllable. These are the four variables, governing the chemical and physical behaviour of silica. Each of them will be discussed in detail in the subsequent paragraphs. Initial interest is directed to the methods of preparation and the various types obtained. A survey of these methods and their main characteristics is presented in table 1.1.
2.1 Silica sols and gels The most documented method for the preparation of silica is the sol-gel route. In this process Si(OH)4 molecules condense to form a siloxane network. As a starting compound a soluble silicate (generally sodium silicate) has mainly been used. Over the last decade, however, the use of alkoxysilanes (Si(OR)4 , with R = CH3, C2H5 or C3H7) has gained importance rapidly. Hydrolysis of the alkoxy-group precedes condensation with a neighbouring silanol. Hydrolysis and condensation occur simultaneously, in the aqueous alkoxide solution. Silica sols thus are formed by the mixing of the silicate salt with an acid or the liquid alkoxide with water. By condensation, stable particles of colloidal size are formed. As condensation proceeds, small three-dimensional siloxane networks are gradually formed. The condensation reaction may be influenced by the addition of an electrolyte
or a change of the pH. A control of both factors will either favour the growing of the particles or the linkage of particles to form chains. This is accompanied by an increase in the viscosity of the medium. At the point where elastic stress is supported, the sol has condensed to a gel. Because of the gradual sol-to-gel transition, the gelation point is difficult to measure analytically. The as-formed gel is termed a hydrogel (or alcogel if an alcohol has been used as solvent). The hydrogel structure is controlled by temperature, the pH of the medium, the nature of the solvent, the nature of the added electrolyte and the type of the starting salt or alkoxyde. Aging or drying of the hydrogel results in a loss of the pore-filling liquid. The pores are narrowed by capillary forces exerted by this liquid. A xerogel is formed, with porosity and surface area depending on the aging and drying conditions such as: application of hydrothermal aging (aging at high temperature in aqueous medium), pH and concentration of the hydrogel particles. If an alcogel is dried under supercritical conditions, the pore narrowing by capillary attraction is excluded. The thus formed gel has a very large pore volume (up to 98 % of the total volume) and is therefore named an aerogel. Following Barby, ~ a somewhat broader definition for this type of gel is needed, the main feature being that the pores of an aerogel do not collapse upon immersion in water and redrying, as can be accurately measured by direct porosimetry. A more profound survey of the sol-gel process will be given below.
Table 1.1 Classification of silicas, according to Barby I I.
NATURAL SILICAS
Have a specific surface area, similar to their geometric surface.
II.
SYNTHETIC SILICAS (mostly amorphous)
Surface area, pore volume, pore size and particle size are to some extent independently controllable (commercially interesting).
1.
Colloidal silica Silica sols
Stable dispersions or sols of discrete particles of amorphous silica.
2.
Silica gels
Coherent, rigid, 3D network of contiguous particles of colloidal silica.
2.1. Hydrogels
Silica gel, in which the pores are filled with the corresponding liquid (water).
2.2. Xerogels
A gel from which the liquid medium has been removed, resulting in a compressed structure and a reduced porosity.
2.3. Aerogels
A special form of xerogel, from which the liquid has been removed in such a way as to prevent any collapse or change in the structure as liquid is removed (Iler). A gel with water-collapsible macropores as judged by direct porosimetry (Barby).
Pyrogenic silicas
Silicas made at high temperatures.
3.1. Aerosils
| flame hydrolysis products of SIC14; very pure materials.
3.2. Arc silicas
Made by the reduction of high purity sand.
3.3. Plasma silicas
Ultra fine silica powders, made by the direct volatilization of sand in a plasma jet.
Precipitates
Made by the precipitation of the silicic acid solution.
3.
4.
2.2 Pyrogenic silicas Besides the preparation in the liquid phase, silica may also be formed with high temperature processes, using a flame, arc or plasma. One of the most widely used sources of pure silicas is the burning of SiCI 4 with hydrogen and oxygen. In the flame the following reactions take place" 2 H2 SiC14 2 H2
+ O2
-
+2 H20
+ 02
+ SiC14
2 H20 -
SiO 2 -
+ 4 HC1 SiO 2
+4 HC1
In this way 'fumed' silicas are formed. This process was developed by Degussa and the thus formed silicas were marketed as Aerosil. | A flow-chart of the production process is displayed in figure 1.1. In the reaction process HCI is formed, which is evacuated from the system.
The
characteristics of the produced silica may be controlled by a variation of the reagent concentrations, the flame temperature and the time of presence in the combustion chamber. Thus, the specific surface area, particle size and particle size distribution may be varied. In this type of silica, primary particles are linked into linear chains and a nonporous structure is produced (figure 1.2).
Arc silicas are formed by the reduction of high-purity sand in a furnace. This type of silica shows a greater variation in particle size. Primary particles do not form chains but form dense, non-microporous secondary particles. The volatilization of sand in a plasma jet produces plasma silicas. These are ultra fine silica powders.
Hydrogen
HCl-adsorption
Oxygen (air) Si-tetra chloride
b
~
~ I
I
l
]
I f
i,I I
...........
pyrogenic silica gel a: vaporiser b: mixing chamber c: combustion chamber
d: cooling e: separation
f: purification g: silo
Figure 1.1 Flow - chart of production process. Taken from ref (2), with permission.
2.3 Precipitated silicas
The precipitated silicas include a wide range of silicas with a variety of structural characteristics. Most of the preparation methods are patented. In general the formation involves a coagulation and precipitation from silica solutions. Properties may therefore be supposed to be similar to those of the gels. For these silicas however, preparation conditions are such as to avoid gel growth and stimulate precipitation. As an overall definition, Barby proposed: dry silicas with no long or short distance characteristic structure.
1
.... ~~::, ~
~-'iiiiiii~iiiiii~ii!::~i !~............ .i~i!~!i~ill ........
!!!iiii~i!ii:,i~iii~i...... ~iii.!i~:ii."~. i~i:. . . . . . . .
Figure 1.2 TEM- picture of AEROSIL 380. Takenfrom ref (2), with permission.
2.4 Comparative properties of silicas The physical properties of the different silicas, prepared by the above mentioned methods, are compared in table 1.2. The variation in physical properties between various silicas is caused by the different way in which particles are aggregated and agglomerated. Agglomeration is the sharing of a plane or side between two particles, while aggregation indicates one-point linking of particles or agglomerates. The size of the primary particles, together with the density and degree of agglomeration and aggregation, determines the porosity and specific surface area of the silicas. The chain-like agglomeration of primary particles in fumed silicasresults in a non-porous structure. Aerosils are finely divided and have no interior surface. For arc silicas, primary particles agglomerate in spheres. Because of the dense packing of the particles, those spheres do not show any porosity. The specific surface is the external surface of the spheres. Xerogels and aerogels have a porous structure. The processes governing this structural build-up in the silica gels will be discussed below.
Table 1.2 Physical properties of various silicas2 Characteristics
Pyrogenic silicas fumed silica
Specific BET area Size primary particles Size aggregation /agglomeration Density Volume Mean pore diameter Pore diameter distribution Shape of interior surface Aggregation and agglomeration structure
* not applicable
m2/g nm /zm
50 to 600 5 to 50
g/cm 3 ml/100g nm
2.2 1000 to 2000 non porous till ca. 300 m2/g 0 chain-like agglomeration (open surface)
Silicas made by wet methods arc silica
precipitated silica
25 to 300 5 to 500 2 t o 15
30 to 800 5 to 100
2.2 500 to 1000 non porous
0 dense spherical aggregates/particles non-agglomerated
xerogels
aerogels
250 to 1000 3 to 20 1 to 20
250 to 400 3 to 20 1 to 15
1.9 to 2.1 200 to 2000 > 30
2.0 100 to 200 2 to 20
2.0 800 to 2000 > 25
very broad poor
narrow very much
narrow much
slightly aggregated nearly spherical particles
highly porous agglomerated particles
macroporous agglomerated particles
1 to 40
11
The TEM microstructure of three porous silica xerogels is shown in figure 1.3.
Figure 1.3 TEM picture of porous silica xerogels, (a) microporous," (b) mesoporous and (c) macroporous (see next page).
12
Figure 1.3. TEM-picture of porous silica xerogels (continued).
It is apparent that the average pore size is related to the size of the agglomerated particles. While in the microporous (a) sample the particles may not be discerned individually, the macroporous (c) samples clearly show agglomerated particles with a 20-30 nm diameter. Macropores are also visible on this picture. The mesoporous sample (b) has intermediate characteristics. The agglomerated clusters have a particle size in the 1-20/zm range. The specific surface area is mainly due to the interior surface of these clusters. This property is responsible for the fact that silica gels are mainly applied for adsorption while fumed silicas are used in rheological applications. In these the presence of highly dispersed small particles is important. The applications of silica will be discussed in detail furtheron.
13 3 The effect of particle packing on the surface area
Whereas the surface area of a crystalline silica is in fact the external surface area, the surface area and the pore size distribution of an amorphous silica are actually determined by the dimensions of the silica spheres (primary particles) that build up the network. For non-aggregated spherical particles, this relationship is very straightforward. In this silica type, the primary particles are not clustered and Sheinfain's 3 globular theory can be applied. The globular theory predicts an inverse relationship between surface area and the primary particle size by the following equation: S -
3
(1)
a.p
with S the specific surface area (m2/g), a the particle radius (nm) and p the density (2.2 * 106/m3). Upon clustering of the primary particles towards secondary structures (aggregation), obviously the surface area will drop. This is illustrated in figure 1.4, showing the area lost at the point of contact between two spheres. The approximate area lost per contact and per particle is 7rb2. A slightly more accurate value is given by equation 2" fraction of area lost per contact -
aR
2(R+a)2
(2)
Based upon the theoretical maximum surface area (Sm) and the measured surface area (SBEr, cfr. chapter 2), the average coordination number of the primary particles in the aggregate can be calculated as: 1 jv~
--
-
SBET Sm aR
total loss of surface area loss of surface area per coordination
(3)
2 (R + a)2 This number should be interpreted as the average number of primary particles that surround the particle under study.
14
SiO 2
R
R R
~l,~ I
;"
2b
r
I
Figure 1.4 Area lost to nitrogen adsorption at the point of contact between spherical particles. Taken from ref . (6), with permission.
15
4 The sol-gel process From the above mentioned silica preparation paths, the sol-gel route is the most studied and consequently most documented. Sol-gel processing is used not only for the preparation of silica gels, but also for the synthesis of ceramic products, ranging from thin films and coatings over porous membranes to composite bodies. 4 This broad success of the method is due to its ability to form pure and homogeneous products at very low temperatures. 5 Thus, the sol-gel technology is replacing the millennia old ceramic fabrication processes in which powders are shaped into objects and subsequently densified at temperatures close to their liquidus. This allowed a transformation of ceramics and glasses from 'stone age materials' to 'space age materials'. The sol-gel process is a wet chemical method, involving hydrolysis and condensation of metal alkoxides and inorganic salts. The synthesis of silica gel using sodium silicate as a starting product has been well documented by Iler, 6 Unger 7 and Barby ~ at the end of the 1970's. Though St6ber et al. 8 had reported the controlled synthesis of spherical silica powder from tetraethoxysilane (TEOS), using ammonia as a catalyst in 1968, the use of alkoxide precursors got minor attention in their works. The activity in the research on these precursors was definitively initiated by the work of Yoldas 9'1~ and Yamane et al., II who demonstrated the preparation of silica monoliths by careful drying of gels. During the 1980's until now the sol-gel process, using alkoxysilane precursors has been a major research topic, aiming at the thorough understanding of the various process steps and at a broadening of the field of applications. The huge amount of literature in this context has been excellently reviewed and synthesized by Brinker and Scherer, ~2Hench and West 13and Scherer. 5'26 We will restrict ourselves to giving a survey of the sol-gel synthesis of silica gel, emphasizing the control of pore structure, specific surface area and particle size in the course of the process.
16
4.1 Hydrolysis Sol-gel silica synthesis is based on the controlled condensation of Si(OH)4 entities. These may be formed by hydrolysis of soluble alkali metal silicates or of alkoxysilanes. The commonly used compounds are sodium silicates and tetraethoxysilane (TEOS). Both reactions are governed by specific parameters, which will be discussed briefly:
(1)
hydrolysis of sodium silicate:
Silicate solution species are controlled by the pH of the medium and silicon concentration. ~4 Figure 1.5 displays the aqueous silicate species as a function of pH for concentrations of 0.01 and 10.5 M of silicon. Monomolecular Si(OH)4 is the predominant solution species below pH 7. At higher pH, anionic and polynuclear species are formed. Sodium silicate solutions are neutralized using sulphuric acid to obtain the Si(OH)4. The sodium silicate solution characteristics are dependent on the SiO2:Na20 ratio. For production of precipitated silicas, sols and gels, a sodium silicate ratio 10:3 is usually employed rather than lower ratios, since less acid is required for neutralisation. Furthermore this ratio is available at low cost, since large quantities are produced.
(2)
hydrolysis of alkoxysilanes:
Alkoxysilanes have the general formula Si(OR)4. In most cases the organic R group is methyl or ethyl. Nomenclature and abbreviations are as follows" Si(OCn3) 4 tetramethoxysilane TMOS Si(OCH2CH3)4 tetraethoxysilane TEOS
Because water and alkoxysilanes are immiscible, a mutual solvent such as alcohol is usually used as a homogenizing agent. ~2 However, gels can be prepared from
17
t00
(a) OJm Si(~r) [~.,j
i
80/,
60
....
\
I ....
I .... I ....
Si408(OH ~-
/
rlt3HL, --
I ....
"4il
,/
si,O(OH- :
(b) t0 -Sin Si(T~)
80
\
t
~
2O 6
O0
~ ....
60
20 8
I0 pH
t2
6
8
t0 pH
12
Figure 1.5 Distribution of aqueous silicate species at 298 K in (a) O. 01 M Si(IV) and (b) 10 s M Si(IV). Ionic strength, I = 3 M.
siliconalkoxyde-water mixtures without added solvent, since alcohol produced as a byproduct of the hydrolysis reaction is sufficient to homogenize the initially phase separated system. In order to get a rapid and complete hydrolysis, an acid or a base catalyst may be used. In both cases the reaction occurs by a nucleophilic attack of the oxygen contained in water, to the silicon atom. Hydrolysis mechanisms were studied by Osterholz and Pohl, ~5using alkyltrialkoxysilanes instead of tetra-alkoxysilanes. Both types of compounds are following the mechanism below. The base catalyzed hydrolysis (figure 1.6) is a two-step process, with formation of a pentacoordinate intermediate. Acid catalyzed hydrolysis proceeds by an SN2-type mechanism. The leaving alkoxy group is rapidly protonated and a water molecule performs a nucleophilic attack at the central silicon atom.
18
kl _~SiOR + B: + H20
['-~+ H 6 _ / --] z/:: ~:--H--O--~~RA
\
\
Stepl
T.S.1 [B:H++ HO-Si-OR/.
k2
x 6 - R 8+ :/: ~HO-Si--6--H--:B~~.
" "
/A
Pentacoordinate Intermediate
\ Step2
"
T.S.2
SiOH + B: + ROH Figure 1.6a Proposed mechanism for the base catalyzed hydrolysis of silane esters. +
+ "SiOR
kl
+ H
"
H +
~--SiOR +
~--SiOR/+ H20 .
+ H! --~ ~ SiOH
H Step 1 +
~
H+
1 "" ..
. ~SiOH
+ HOR Step 2
k3 -x SiOH + H +
Step 3
Figure 1.6b Proposed mechanism for the hydronium catalyzed hydrolysis of silane esters.
19 4.2 Condensation
Silicic acid molecules condense to form siloxane bonds, with release of water (A). 6 --Si-OH
+
HO-Si-
~
-Si-O-Si-
+
1t20
(A)
Condensation may also proceed by the reaction of the alkoxysilane with a silanol group, releasing an alcohol (B). - Si-OH
+
RO-Si--
--,
-- S i - O - S i -
+
ROH
(B)
The condensation may be acid or base catalyzed. Osterholz and Poh115 concluded that acid catalyzed condensation proceeds through an SN2-Si type mechanism, while base catalyzed condensation is less well understood. They also found that the reaction goes to completion in protic solvents, while in aprotic solvents equilibrium is reached. Hydrolysis and condensation occur simultaneously. The relative rate of both processes determines the sol structure. As discussed by Ying et a l . , 16 in acidic conditions, hydrolysis is faster than condensation. The rate of condensation slows down with increasing number of siloxane linkages around a central silicon atom. This leads to weakly branched polymeric networks. In the basic conditions, on the contrary, condensation is accelerated relative to hydrolysis. The rate of condensation increases with increasing number of siloxane linkages. Thus, highly branched networks with ring structures are formed. This generates large, bulkier, more ramified polymers. According to Ro et al. 17 gels made from low water content sols contain residual organic groups, caused by incomplete hydrolysis, which contributes to the formation of micropores during the thermal treatment. Acid-catalyzed gels show slit-shaped micropores and have a fibrous or plate-like structure. Base-catalyzed gels have cylindrical pores and spherical particles. An understanding of these trends may be gained from the polymerization and growth mechanism of silica sols. Iler 6 stated that the Si(OH)4 condensation and polymerization is not analogous to the linear branching and cross-linking in organic polymers. Three stages are recognized in the silicic acid polymerization"
20
1) 2) 3)
polymerization of monomers to form small primary particles; growth of primary particles; linking of particles into branched chains, then networks, finally extending throughout the liquid medium, thickening it to a gel.
By controlling the pH and with the addition of electrolytes that induce flocculation, the relative importance of steps 2 and 3 may be varied. This control is displayed in figure 1.7.
Monomer Dimer Cyclic pH < 7 or pH 7-10 with salts present
Particle A
~.
~/ ~ ~
".
1 nm 5 nm
pH 7-10 with salts B absent
00 nm
three-dimensional ~ gel networEs..........
Figure 1.7
~(-~ ~
~'/~ /
,,~ors
Polymerizationbehaviour of silica. Takenfrom ref (6), with permission.
21 Condensation takes place in such a way as to maximize the number of Si-O-Si bonds and minimize the number of terminal hydroxyl groups by internal condensation. Thus rings are quickly formed to which monomers add, creating three-dimensional particles. ~2 By a variation of pH and salt addition the particle aggregation into secondary particles (A) or further particle growth (B) is controlled. Thus the particle size and pore structure of the silica is determined. Iler discerned three pH domains in the polymerization process. At pH below 2 the gel time, which indicates the rate of polymerization, is proportional to the proton concentration [H§ Brinker and Scherer ~2 proposed a mechanism involving an associative m SiOHR(OH2) + intermediate. At pH 2 the particles reach the isoelectric point. Above this pH value the condensation rate is proportional to [OH], following: fast - S i - O H + O H - - , - S i O + HzO (C) slow -Si-O
+
HO-Si~
-~
-Si-O-Si-
+
Off
fa)
Reaction (D) occurs preferentially between more highly condensed species and less highly condensed neutral species. I.e. monomers add to small polymerized particles (trimers, tetramers). Further growth involves continued monomer deposition and particle aggregation. Particle growth is limited to about 2 nm by the solubility of the silica in this pH range. Above pH 7, particles are charged by ionisation. Aggregation is therefore reduced. Growth occurs by monomer deposition and Ostwald ripening. This involves the dissolution of small particles and redeposition on larger ones. Aggregation can only take place after the addition of salts, which reduce the stabilizing double layer thickness. Following Murakata et al., TM the pore size distribution is controlled by the addition of several kinds of inorganic salts and surfactants in the sol-gel reaction mixture. Inorganic salts depress the formation of mesopores. The influence of surfactants is dependent on the charge of the head group. The fractional pore volume of large pores is increased in using long-chain diols in the starting solution. ~9
22 Solid silica spheres are prepared by the St6ber s method. This involves the condensation of TEOS in alcoholic solution of water and ammonia. The mechanism of this synthesis was elucidated by Van Blaaderen et al) ~ The particle growth is rate-limited by the production of hydrolysed monomer molecules. It involves a surface condensation of monomers and oligomers, while aggregation of particles is only occurring in the early stages of the condensation.
4.3 Gelation As the polymeric network extends throughout the total volume, the sol thickens to a gel. 6 The sol-to-gel conversion is a gradual process, which is easily observed qualitatively, but difficult to measure analytically. The gelation point tg~ is defined as the point where elastic stress is supported. ~3 Thetg~ is not an intrinsic property of a sol. It is influenced by the size of the container, the solution pH, the nature of the salt concentration, the anion and solvent, the type of initial alkoxy group and the amount of water. A lot of work has been directed to the modelling of the gelling process. The percolation model offers the most widely accepted theory. A review of the model is beyond the scope of this book. The reader is referred to the reviews of Zallen, 22 Staufer et al. 23 and Brinker and Scherer. 12 The gelation rate influences the pore structure. Fast gelation gives an open structure because the particles are quickly connected and cannot undergo further rearrangements.U
4. 4 Aging When the gel is kept in contact with the pore-filling liquid, its structure and properties keep changing as a function of time. This process is called aging. During the aging period, four processes affect the porous structure and surface area of the silica gel. These are polycondensation, syneresis, coarsening and phase transformation. 6'~2'~3
23 Polycondensation is the further reaction of silanols and alkoxy groups in the gel structure to form siloxane bonds by reactions 3 and 4. These reactions result in a densification and stiffening of the siloxane network. Syneresis is the shrinkage of the gel network, resulting in the expulsion of the pore liquid. This shrinkage is caused by the condensation of surface groups inside the pores, resulting in a pore narrowing. In aqueous gel systems the pore size is controlled by an equilibrium between electrostatic repulsion and Van Der Waals attractions. In this case, shrinkage is produced by addition of an electrolyte. The syneresis contraction rate is affected by the sol concentration, temperature and type of solvent. ~3 Besides the condensation, shrinkage may also be caused by capillary forces exerted by the pore-filling liquid on the walls. ~2 Coarsening or Ostwald ripening is the solution and redeposition of small particles. The cause of this process is the higher solubility of convex surfaces, compared to concave surfaces. Therefore, the small particles are dissolved and redeposition on larger particles occurs. Also necks between particles will grow (figure 1.8) and small pores may be filled in. This results in an increase in the average pore size and a decrease in the specific surface area. Summarizing, during aging the pore structure, surface area and stiffness of the gel network are changed and controlled by the following parameters: time, temperature, pH, added electrolyte and pore fluid. 25
Figure 1.8 A chain of small particles is converted to a fiber or rod by the laws of solubility of convex and concave surfaces.
24
4.5 Drying In the drying step the hydrogel or alcogel obtained by gelling and aging is converted to a xerogel. Drying of silica gels is a three stage process. 13 In stage 1 the decrease in gel volume is equal to the volume of the liquid lost by evaporation. Shrinkage of the gel structure occurs due to the large capillary forces exerted by the pore liquid. The greatest changes in volume, weight, density and structure occur in this stage. According to Deshpande et al. 25 a linear decrease in xerogel surface area is observed with increasing solvent tension. Pore volume and pore size distribution follow a similar trend. Stage 1 stops when shrinkage ceases because of the increased stiffness of the dried gel network. This is noted as the 'critical point'. Up to this point the pores remain full of liquid. 26 In stage 2 the pores are emptied. Though the capillary pressure is at its highest level, at the critical point the network may not be compressed further, and the pore liquid evaporates. Liquid is transported through films that cover partially empty pores and evaporates at the surface. The capillary forces reduce consequently. Stage 3 starts when the pores have substantially emptied and surface films can not longer be sustained. The liquid escapes by evaporation within pores and diffusion of vapour to the surface. During this stage there is no further change in structure. The high shrinkage of gels in this drying process has prevented the economical production of large objects and thick films by sol-gel processing. For thin films shrinkage results in thinning of the film rather than cracking. However, processes have been developed to reduce these problems. TM
4. 6 Stabilization, consolidation In order to convert xerogels to dense, stable, solid materials, a further thermal treatment is required. In this stabilization step the gel is dehydrated, dehydroxylated and sintered. Thus all surface hydroxyls are removed and the material is fully densified. Since this work discusses the characterization and modification of porous silica xerogel we will not further discuss the processes in this step. The reader is
25 referred to the work of Brinker and Scherer n and Hench and West ~3 for further details.
4. 7 Advantages and applications A general survey of the advantages of the sol-gel route over the traditional ceramic synthesis processes, concerning processing temperature, is presented in figure 1.9. The huge advantage of sol-gel processing is evident from its step-wise temperature profile, with gradually increasing temperatures. Traditional processes require much higher temperatures and major temperature variations in the course of the process. Further advantages and disadvantages of the sol-gel process were summarized by MacKenzie27 and are presented in tables 1.3 and 1.4.
Melt
1773 TL
Quench
T
Melt-formed Glass b
1
~
!
Time
Figure 1.9 Comparison of the processing of melt-derived glasses and gel-derived 'glass-like' solids.
26
Table 1.3 Some advantages of the sol-gel method over conventional melting for glass .
2. 3.
.
5. 6. 7.
Better homogeneity from raw materials. Better purity from raw materials. Lower temperature of preparation: a. Save energy; b. Minimize evaporation losses; c. Minimize air pollution; d. No reaction with containers, thus purity; e. Bypass phase separation; f. Bypass crystallization. New crystalline solids outside the range of normal glass formation. New crystalline phases from new noncrystalline solids. Better glass products from special glass properties. Special products such as films.
Table 1.4 Some disadvantages of the sol-gel method .
2. 3. 4. 5. 6. 7.
High cost of raw materials. Large shrinkage during process. Residual fine pores. Residual hydroxyl. Residual carbon. Health hazard of organic solutions. Long processing times.
Sol-gel derived materials may be used in the form of films, fibres, monoliths, powders, composites and porous media. The most successful applications are those that utilize the potential advantages of the sol-gel processing such as purity, homogeneity and controlled porosity, combined with the ability to form shaped objects at low temperatures, while avoiding inherent disadvantages such as cost of raw materials, long processing times and high shrinkage. ~2
27
5 Applications of pure silicas and powders Due to the variety in porous structure, particle size and surface area, pure silica gels and powders find a very wide range of applications. Variation in preparation methods and parameters allows the tailoring of the substrate properties for specific application needs. The main features in the silica applications are its porosity, active surface, hardness, particle size and the viscous and thixotropic properties. Although most applications are based on a combination of those, a classification according to the main properties of interest may be set up. For references, the reader is referred to the works of Iler 6 and Unger 7 and to the references cited in chapter 8. 5.1 Porosity
Pure silica is used as a stationary phase in various types of liquid chromatography. 2g In size exclusion chromatography 29the separation of polymer compounds is effectuated on basis of diffusion rate of variably-sized molecules through a microporous silica packing with uniform pore size. 5. 2 Active surface
The silica surface allows variable types of interactions in which the large specific surface area as well as the surface site chemistry play an important role. Both topics will be discussed in detail below. Concerning the applications, three types of interactions may be discerned" * adsorption" in liquid-liquid chromatography, silica packings are used, which acts as a sponge to hold the static liquid phase as an active adsorbent. The pure silica may also be used as a stationary phase in column liquid chromatography. The adsorption characteristics are also applied in the use of silica as a catalyst base. Active catalyst species are adsorbed onto high-area silica. The adsorption of dye molecules on silica powders in coatings, paintings and painting inks results in less intense and flatted color. * absorption: silica is used in various forms (from finely divided powders to small granules) as a desiccant. This is probably the largest of all applications. Water is
28 physically and chemically bound to the silica surface. Silica is added to any package product, which may be subjected to corrosion or deterioration by moisture. * ion-exchange: the ion-exchange capacity of the silica surface is used for the bounding of metal complexes and surface metal cations. This property is applicable in the chromatographic separation of cationic species. These have been reviewed by Bannasch and Stam3~and Markova and Vydra. 3~ The nature of the silica surface may be changed by modification with a variety of molecules, causing a still broader range of applications. A survey of these will be given in chapter 2. 5.3 Hardness
The hardness and abrasive properties of silica gels and powders are of interest in their use as reinforcing agents in composite materials. Wear resistance and tensile strength of rubbers and polymers are increased by the introduction of silica in the organic network. In order to preserve these improved qualities in wet conditions, however, the silica surface has to be modified to assure interfacial bonding. 32 5. 4 Particle Size
Due to their disperse character and small particle size, silicas are used as flow aids, i.e. they are used to improve the flow behaviour of other materials. The adsorption of the fine silica particles on other type powdered compounds reduces interparticle interactions. Particle adhesion, electrostatic adhesion, Van Der Waals forces and liquid bridge formation is reduced or avoided. 33 This allows free-flowing behaviour of strongly interacting or irregularly shaped powdered materials. 5.5 Viscosity andthixotropy
When suspended colloidal particles form a network through the liquid, viscosity is increased. For silica gels the thickening and thixotropic effects are used in a large number of applications. The silica is mixed with paints, coatings, inks, pharmaceutical and cosmetics with this purpose. 34
29 References D. Barby, Silicas, in 'Characterization of Powder Surfaces', G.D. Parfitt and G.S.W. Sing eds., Academic Press, London, UK, 1976, 353. 2. .
4. 0
0
.
G. Michael and H. Ferch, Schriftenreihe Pigmente, Degussa, 1991, 11. R. Yu. Sheinfain, N.S. Kruglikova, O.P. Stas and I.E. Neimark, Kolloid Zhur., 1963, 25, 732. J.D. MacKenzie, J. Non-Cryst. Sol., 1988, 100, 162. G.W. Scherer, in Drying '92, A.S. Mujumdar ed., Elsevier, Amsterdam, The Netherlands, 1992, 92. R.K. Iler, The Chemistry of Silica, John Wiley & Sons, New York, 1979. K.K. Unger, 'Porous Silica, its properties and use as a support in column liquid chromatography', Elsevier, Amsterdam, The Netherlands, 1979.
8.
W. St6ber, A. Fink and E. Bohm, J. Coll. Interface Sci., 1968, 26, 62.
9.
B.E. Yoldas, J. Mater. Sci., 1975, 10, 1856.
10.
B.E. Yoldas, J. Mater. Sci., 1977, 12, 1203.
11.
M. Yamane, A. Shinji and T. Sakaino, J. Mater. Sci., 1978, 13, 865.
12.
C.J. Brinker and G.W. Scherer, 'Sol-gel Science', Academic Press, New York, 1989.
13.
L.L. Hench and J.K. West, Chem. Rev., 1990, 90, 33.
14.
C.F. Baes and R.E. Mesmen, in 'The Hydrolysis of Cations', John Wiley & Sons, New York, 1976, p 342.
15.
F.D. Osterholz and E.R. Pohl, in 'Silanes and other Coupling Agents', K.L. Mittaled VSP, Utrecht, The Netherlands, 1992, 119.
16.
J.Y. Ying, J.B. Benzingen and A. Navrotsky, J. Am. Ceram. Soc., 1993, 76(10), 2571.
17.
J.C. Ro and I.J. Chung, J. Non-Cryst. Solids, 1991, 130, 8.
18.
T. Murakata, S. Sato, T. Ohgawara, T. Watanabe and T. Suzuki, J. Mater. Sci., 1992, 27, 1567.
19.
K. Kwabata, H. Yamashita and T. Maekawa, J. Ceram. Soc. Jpn. Int. Ed., 1993, 101,490.
30 20.
A. Van Blaaderen, J. Van Geest and A. Vrij, J. Gel. Interface Sci., 1992, 154, 481.
21.
A. Van Blaaderen and A.P.M. Kentgens, J. Non-Cryst. Sol., 1992, 149, 161.
22.
R. ZaUen, 'The Physics of Amorphous Solids', John Wiley & Sons, New York, 1983.
23.
D. Stauffer, A. Coniglio and M. Adam, Advances in Polymer Science, 1982, 44, 103.
24.
G. Dessacles, I. Biay, F. Kolenda, J.F. Quinson and J.P. Reymond, J. Non-Cryst. Solids, 1992, 147&148, 141.
25.
R. Deshpande, D.V. Hua, D.M. Smith and C.J. Brinker, J. Non-Cryst. Solids, 1992, 144, 32.
26.
G.W. Scherer, J. Non-Cryst. Solids, 1992, 147, 363.
27.
J.D. MacKenzie, in 'Ultrastructure Processing of Glasses, Ceramics and Composites', L.L. Hench and D.R. Ulrich eds., John Wiley & Sons, New York, 1984, 15.
28.
R.M. Smith, Gas and Liquid Chromatography in Analytical Chemistry, J. Wiley & Sons, 1988.
29.
W.W. Yau, J.J. Kirkland and D.D. Bly, Modem size-exclusion liquid chromatography. Practice of gel permeation and gel filtration chromatography., J. Wiley, Chichester, 1979.
30.
W. Bannasch and H.H. Stamm, AEC Accession No. 15760, Rep. No. KFK-233, Avail AEC, 54 pp, 1964.
31.
V. Markova and F. Vydra, Chem. Listy, 1966, 60, 860.
32.
E.P. Pleuddemann, Silane Coupling Agents, Plenum Press, New York, 1991.
33.
H. Ferch, R. Oelm/iller and B. Grinschgl, Technical Bull. Pigments, Degussa, 1993, 31.
34.
H. Fratzsceh, Schriftenreihe Pigmente, Degussa, 1983, 23.
31
Chapter 2
Physical characterization of the silica surface
The three parameters specific surface area (m2/g), pore size distribution (f~(rp) and fA(rp)-distributions based on pore volume and pore area respectively) and particle size are sufficient to fully characterize the physical properties of the silica matrix. The exact knowledge of the specific surface area of a silica gel, treated at a given temperature, is essential to express the concentration of the reactive surface species in (#/nm2). This unit is in many cases superior to the more conventional (mmol/g), since it gives a microscopical image of the surface loading, and since it is directly related to effects such as sterical hindrance. The pore size distribution gives a deeper insight in the availability of the reactive surface silanols towards the reacting molecules. It also explains the sorption characteristics of the silica. In fact, pore size and surface area are established by an underlying parameter: the particle size of the (secondary) particles. Direct measurement of particle sizes yields information on surface area and porosity and vice versa.
32 1 The isotherm
In sorption, the amount of gas adsorbed, X,, on a given adsorbent is measured as a function of the equilibrium partial pressure, p, of the adsorbate at constant temperature. The equilibrium partial pressure is preferably related to Po, the saturation vapour pressure of the adsorbate. Measurements are made at temperatures at which the gas at atmospheric pressure is in the liquid state. For nitrogen, this is 77 K. The sorption isotherms can be grouped into five types, according to the classification of Brunauer, Emmet and Teller. ~a'3'4 However, we prefer a classification, based on the pore size of the adsorbent. 5 The IUPAC classification 6 of pores is given in table 2.1.
Table 2.1 IUPAC classification of pores Micropores Mesopores Macropores Megapores
Pore-diameter: Pore-diameter: Pore-diameter: Pore-diameter:
0- 2 2 - 50 50 - 7500 > 7500
nm nm nm nm
Figure 2.1 shows three nitrogen adsorption isotherms that were obtained on a purely microporous, a purely mesoporous and a purely macroporous silica sample. There are distinct differences in the shapes of these isotherms. Their different slopes can be interpreted by means of three mechanisms that may occur during the adsorption and desorption runs. On a purely microporous silica, a Langmuir 7,g type isotherm is observed. This is the type I isotherm, according to the BDDT classification. 2 Starting at relatively low pressure, a sharp increase in adsorption is observed, which is due to the gradual filling of the micropores with adsorbate. Adsorption proceeds until all pores are filled, which is indicated by a fiat region, reflecting a constant adsorption. The adsorption and desorption branches are identical, so no hysteresis is observed. On the purely mesoporous silica, a multilayer of adsorbate is formed, with increasing relative pressures. Depending on the mean pore diameter, but generally above P/P0 = 0.4, capillary condensation takes place, resulting in a further increase of X~. The
33
Xa (g/g)
1.2
_
9
0.80.60.40
0
0.2
0.4
P/P0
0.6
0.8
1
Figure 2.1 Nitrogen sorption isotherms at 77 K for a purely microporous (+), mesoporous (m) and macroporous (o) silica.
horizontal branch near the saturation pressure (P/P0 = 1) indicates that all mesopores are filled with liquid adsorbate. The desorption branch does not follow the adsorption branch, but gives a distinct hysteresis loop" the amount adsorbed is always greater along the desorption branch compared to the adsorption branch. This typical pattern for mesoporous solids is explained by capillary condensation of the adsorbate in the pores, causing an increase of adsorbate during the adsorption run, and a consequent retardation during the desorption. This is a type IV isotherm in the BDDT classification. On the purely macroporous substrate, multilayer formation takes place in the adsorption run, whereas capillary condensation only occurs at relative pressures near unity. The desorption follows the adsorption branch (type II isotherm in the BET classification).
34 Figure 2.2 shows the nitrogen isotherm for a mesoporous silicagel (Kieselgel 60, Merck) thermally treated at 973 K. In fact, in the remainder of this book, we will often refer to this specific silica, since it was used as the starting material in many of our studies. This isotherm is typically a mesoporous isotherm, according to the classification of Unger, or a type IV isotherm, according to the BET classification. The three main areas - formation of monolayer, multilayer, and capillary condensation- are indicated in the figure.
ml/g (at STP)
600
MULTILAYER
CAPILLARY CONDENSATION
500 400 300 200 100 0
0
0.2
0.4
1
1
0.6
0.8
1
P]Po Figure 2.2 Nitrogen sorption isotherm (77 K) on Kieselgel 60, treated at 973 K for 17 h.
2 Determination of the specific surface area
The most widespread method in determining the specific surface area of solid substrates is without doubt the Brunauer-Emmet-Teller (BET) method. 3 It is based on a kinetic model of the adsorption process by Langmuir, 7 in which the surface of the solid was regarded as an array of adsorption sites. A state of dynamic equilibrium
35 was postulated, in which the rate of molecules arriving from the gas phase and condensing onto the bare sites (00) is equal to the rate at which molecules evaporate from occupied sites (00. When extended to a two-layer adsorption, the Langmuir mechanism requires that the rate of condensation from the gas phase on molecules, already adsorbed in the first layer, equals the rate of evaporation from the second layer. For multilayer adsorption, the model implies that at any pressure, the fractions of the surface covered with 1,2,...,i molecules will be 0~, 02, ..., 0~ respectively, so that the thickness of the adsorbed layer will not be constant throughout. Since this model was far too complex to serve any practical purpose, Brunauer, Emmet and Teller made some simplifying assumptions (the main one being that in all layers the evaporation-condensation mechanisms are identical) to derive their famous BET equation, to be used in the multilayer-adsorption region of the adsorption isotherm:
P v,,(Po-P)
-
1 v,..c
+
(c-1)P Vm.C.Po
(1)
where v, is the number of moles adsorbed per gram adsorbent at gas pressure P, Vm is the monolayer capacity of the surface (the number of moles of gas per gram of adsorbent, required to form a monolayer), P0 is the saturation pressure and C is the BET constant, which is a function of the heat of adsorption. Mathematical treatment of the adsorption points in the P/Po region 0.05-0.35 with the BET equation, yields the two unknowns C and Vm. Using the monolayer capacity, the specific surface area (SB~v) can be easily calculated, according to
SBrr = vm.am.Na.lO-2O
(m2[g)
(2)
36 with NA the Avogadro number (6.02"1023 molecules/mol) and am the molecular crosssectional area of the gas molecule. This value is commonly assumed to be about 0.162 nm 2 for nitrogen on an oxide surface. 9'1~ In spite of some grounded criticisms 11'12'13'14'15'16(the BET model assumes energetically identical adsorption sites, neglects lateral interactions and assumes identical behaviour in all layers of the multilayer adsorption), and the proposal of alternative models, 17'1s'19'2~the BET equation has retained its utility. It is a relatively easy approach and the method is applicable to a great variety of adsorption isotherms. Figure 2.3 shows the surface area (SBEv) of Kieselgel 40, 60 and 100 as a function of the pretreatment temperature. These numbers (40, 60 and 100) refer to the mean pore diameter, as will be discussed further in the text. It can be seen from figure 2.3 that the surface area is constant up to temperatures of 873-973 K, after which the surface area collapses. The decrease in surface area is influenced by the pore size of the sample. 700 600
BET --
surface area (m2/g) "
500 400
+
300 200 100 0
273
373
473
573 673 773 Temperature (K)
873
973
1073
Figure 2.3 Specific surface area (SBer)of Kieselgel 40 (m), 60 (+) and 100 (*) as a function of the temperature.
37
In the temperature range 473-673 K, mainly vicinal hydroxyl groups condense at the surface, forming strained siloxane bridges. This process does not influence the pore structure parameters. Above 873 K, interparticle condensation of free hydroxyl groups occurs, which is combined with a rearrangement of silica globules to produce a more stable configuration. Obviously, this interparticle condensation is highly favoured by smaller pores, resulting in a faster collapse of the pore structure. At temperatures higher than 1473 K, non-porous products are obtained.
3 Pore size characterization
3.1 Average pore diameter
Assuming that the external surface of a porous substrate is negligible, and that the pores are cylindrically shaped, the average pore diameter of a substrate can be calculated by the Wheeler formula: 21
rp _ 2"103 lip
(3)
SBET
Pore diameter and pore radius are expressed in nm. Vp is the volume (ml/g) of liquid adsorbate, and can be calculated from the volume of adsorbed g_~ (V~) by means of the Gurvitsch rule 22 Vp = 1,54.10-3 . Va
(4)
The average pore radii of Kieselgel 40, 60 and 100, degassed for 17 h at room temperature were 2.14, 3.59 and 6.22 nm respectively.
38 3.2 Pore sizedistributions
3.2.1 Mesoporous silicas Most models to calculate the pore size distributions of mesoporous solids, are based on the Kelvin equation, based on Thomson's 23 (later Lord Kelvin) thermodynamical statement that 'the equilibrium vapour pressure (p), over a concave meniscus of liquid, must be less than the saturation vapour pressure (P0) at the same temperature'. This implies that a vapour will be able to condense to a liquid in the pore of a solid, even when the relative pressure is less than unity. This process is commonly called the capillary condensation. The mathematical formulation correlates the core radius of the liquid adsorbate (r0 to the relative pressure"
trip
_ -2~,V~cos(O)._l
Po
0
=
Vl
--
l'k
=
RT
(5)
rt
surface tension of liquid nitrogen at 77 K (8.85"10 .3 N/m) contact angle; assumed to be 0 molar volume of liquid nitrogen at 77 K (34.65"10 .3 l/mol) Kelvin radius (nm)
Thus, for liquid nitrogen at 77 K, this equation reduces to" ~
_=
0.415
log P---. Po
(6)
Unfortunately, the Kelvin radius (r0 does not equal the actual pore radius (rp), one would like to measure. This is due to the fact that multilayer adsorption occurs, prior to the capillary condensation, resulting in a pore narrowing. Therefore, if t is the thickness of the adsorbed layer, then rp equals
39
r p = r , +t
Numerous scientist have calculated t-values as a function of
(7)
p / p 0 .24'25'26'27
A
general
accepted formulation is the Halsey equation: 2g
(8)
X
m
=
n
-"
Halsey exponent (1/3 for N 2 at 77 K) monolayer thickness (0.354 nm for N 2 at 77 K) Halsey constant (-5.0 for N 2 at 77 K)
Based on the above general principles, quite a number of models have been developed to estimate pore size distributions. 29'3~ They are based on different pore models (cylindrical, ink bottle, packed sphere, ...). Even the so-called 'modelless' calculation methods do need a pore model in the end to convert the results into an actual pore size distribution. Very often, the exact pore shape is not known, or the pores are very irregular, which makes the choice of the model rather arbitrary. The model of Barett, Joyner and Halenda 34 (BJH model) is based on calculation methods for cylindrical pores. The method uses the desorption branch of the isotherm. The desorbed amount of gas is due either to the evaporation of the liquid core, or to the desorption of a multilayer. Both phenomena are related to the relative pressure, by means of the Kelvin and the Halsey equation. The exact computer algorithms 35 are not discussed here. The calculations are rather tedious, but straightforward. Figures 2.4 and 2.5 present the pore size distributions of the three silicas under study, based on the pore volume (fv(rp)) and on the pore area (fA(rp)) respectively. It is obvious from these figures that the pore size distributions of Kieselgel 40 and 60 are centered reasonably well around the average (Wheeler) pore diameter.
40
Pore volume (ml/g)
0.14
1
0.12
3
0.I0
0.08 0.06 0.040.02 0
0
5
10 15 Average pore diameter (rim)
20
Figure 2.4 Pore size distribution (RIH-method), based on pore volume of KG 60 (1), KG 40 (2) and KG 100 (3), degassed at 373 K.
140
Pore area (m2/g)
120 100 8O 604020 0
0
5
10 Average pore diameter (rim)
15
20
Figure 2.5 Pore size distribution (BJH-method), based on pore-area of KG 60 (1), KG 40 (2) and KG 100 (3), degassed at 373 K.
41
The pore size distribution of Kieselgel 100 is also centered around the Wheeler pore diameter, but the distribution is very broad, indicating that extreme caution is needed when interpreting average pore diameters. This latter value has only a statistical relevance, and is by no means a reflection of the actual pore size. Both the BET model and pore size distribution models have a rather low degree of accuracy. A divergence of at least 10% from the actual area is not exceptional. 36 A way of checking the validity of the different models consists in comparing the cumulative pore area, as calculated by the BJH model, with the BET surface. Some results are presented in table 2.2. Table 2.2 Comparison of BET-surface area with BJH cumulative pore area
Sample
KG KG KG
SBET (m2/g) 40 60 100
582 403 313
cum. pore area (m2/g) 564 466 433
% deviation
3 16 38
It is obvious that the deviation between the two models becomes larger with an increasing broadness of the pore size distribution. The results only agree very well for the fairly narrow pore size distribution of Kieselgel 40. However, for the broad KG100 distribution, the BJH model overestimates the surface area by almost 40%. Repeated measurement on independently prepared samples clearly demonstrated that this overestimation was an artifact of the model, and was not caused by (erroneous) sample preparation.
3.2.2 Microporous silicas Models to evaluate the microporous volume exist for several decades. They do not yield a micropore size distribution, but simply quantify the pore volume of all pores with a diameter < 2 nm.
42 Typical methods for testing microporosity consist in the comparison of a standard isotherm (mostly an isotherm of a non-(micro)porous sample of identical chemical composition), with the isotherm under study. t-plot A convenient method is provided by the t-plot of Lippens and De Boer. 37 It consists of plotting the volume of gas adsorbed vs t, the statistical thickness of the adsorbed film. t is a function of P/Po, as measured in the standard isotherm, according to the Halsey equation or one of its modifications. 3s'39'4~ In any normal case of multimolecular adsorption, the experimental points (t vs. P/P0) should fall on a straight line through the origin. If the adsorbent contains micropores, the uptake is enhanced in the low pressure region and the isotherm is correspondingly distorted. In the t-plot, extrapolation of the curve gives a positive intercept which is equivalent to the micropore volume by means of the Gurvitsch rule. When mesopores are present, capillary condensation will occur according to the Kelvin equation. A t-plot will therefore show an upward deviation, starting at the relative pressure at which the finest pores are just being filled. Figure 2.6 schematically illustrates the effects of micro- and mesoporosity on a t-plot.
~,-plot
The t-curve was originally developed to evaluate the thickness of the adsorbed layer on the walls of the pores of mesoporous solids. However, for the purpose of testing for conformity to the standard isotherm, a knowledge of the numerical thickness is irrelevant. Sing 42 has pointed out that it is sufficient to use as a normalizing factor, the amount adsorbed at some fixed relative pressure, in practice taken as (p/po)~= 0.4. The
43
ml/g 400 350 mesopores 300 250 200 150 100 ....... ~ I micro~pore volume
50 /,"
0
I
0.2
0.4
0.6
0.8
I
1.0
I
1.2
1.4
De Boer statistical thickness (rim) Figure 2.6 Effects of micro- and mesoporosity on a t-plot.
normalized adsorption n / n o . 4 - - Ors, obtained from the isotherm on a reference sample of the solid, is then plotted against P/Po, to obtain a standard c~ccurve, rather than a t-curve. The as-CUrve can then be used to construct an ctcplot from the isotherm of a test sample of the solid, just as the t-curve can be used to produce a t-plot. If a straight line through the origin results, one may infer that the isotherm under test is identical in shape with the standard. A particular advantage of the ors-method is that its applicability is not restricted to nitrogen. 42 It offers a simple but effective means of testing for the identity in shape of the isotherms of any suitable adsorptive on a given set of samples of the same substance. In view of the widespread use of nitrogen and argon in surface area and porosity studies, data for the construction of the standard cry-curves for these adsorbates on hydroxylated silica are given in tables 2.3 and 2.4.
44 Table 2.3 Standard data for the adsorption of argon at 77 K on nonporous hydroxylated silica 43 Relative pressure p/po
~, ( =n/no 4)
Relative pressure P/Po
(=n/no.a)
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.20 0.22 0.24 0.25 0.26 0.28
0.243 0.324 0.373 0.413 0.450 0.483 0.514 0.541 0.563 0.583 0.602 0.620 0.638 0.657 0.674 0.689 0.705 0.719 0.733 0.748 0.773 0.801 0.813 0.826 0.851
0.30 0.32 0.34 0.36 0.38 0.40 0.42 0.44 0.46 0.48 0.50 0.52 0.54 0.56 0.58 0.60 0.62 0.64 0.66 0.68 0.70 0.72 0.74 0.76 0.78
0.876 0.900 0.923 0.948 0.973 1.000 1.022 1.048 1.064 1.098 1.123 1.148 1.172 1.198 1.225 1.250 1.275 1.300 1.327 1.354 1.387 1.418 1.451 1.486 1.527
45
Table 2.4 Standard data for the adsorption of nitrogen at 77 K on nonporous hydroxylated silica *~'~s Relative P/Po 0.001 0.005 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30 0.32 0.34 0.36 0.38 0.40 0.42 0.44 0.46 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90
Adsorption per unit area mol m2 4.0 5.4 6.2 7.7 8.5 9.0 9.3 9.4 9.7 10.0 10.2 10.5 10.8 11.3 11.6 11.9 12.4 12.7 13.0 13.3 13.6 13.9 14.2 14.5 14.8 15.1 15.5 15.6 16.1 16.4 17.0 17.8 18.9 19.9 21.3 22.7 25.0 28.0 37.0
as (=n/no.4) 0.26 0.35 0.40 0.50 0.55 0.58 0.60 0.61 0.63 0.65 0.66 0.68 0.70 0.73 0.75 0.77 0.80 0.82 0.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00 1.01 1.04 1.06 1.10 1.14 1.22 1.29 1.38 1.47 1.62 1.81 2.40
46
Other -more complicated- models to evaluate the microporous volume exist.
The
Dubinin-Radushkevich model 46'47'48,49 is based on thermodynamical considerations
concerning the process of micropore filling. Full discussion of this model is beyond the scope of this book. The reader is referred to the standard work of Gregg and Sing5~ on adsorption for a detailed treatment.
Micropore size distributions Many technologically important solids (including silica) have a micropore structure that is well below the lower limit of pore size at which the Kelvin equation (equation 5) is applicable. The basis for the Kelvin equation is a thermomechanical equilibrium across the hemispherical meniscus of a capillary condensate within a cylindrical pore. Below a pore size of 2 nm diameter, the liquid cannot be considered as a fluid with bulk properties because of the forces exerted by the wall. Theoretical calculations suggest that the properties of fluids in microporous structures are highly dependent on the size of the pore. Useful information about micropore structures can be derived from nitrogen or argon isotherm data in terms of the C-constant (BET), t or a:plots and the DubininRadushkevich models. Yet these models do not provide a straightforward relation between the logarithm of relative pressure and the pore size, as does the Kelvin equation. Everett and Pow151 (1976) have developed a pore size distribution model for the slit shaped pores of ultramicroporous carbons. This model has been further elaborated by Horvath and Kawazoe. 52 Because of the conceptual and mathematical simplicity of these models, they have been used recently to describe adsorption in zeolite and clay structures, 53'54'55'56 based on either a cylindrical pore model or a packed sphere model. These models provide an easy analysis of the microporous structure in terms of the argon adsorption isotherm.
47
The slit model Adsorption in ultramicroporous carbon was treated in terms of a slit-potential model by Everett and PowP ~ and was later extended by Horvath and Kawazoe. 52 They assumed a slab geometry with the slit walls comprised of two infinite graphitic planes. Adsorption occurs on the two parallel planes, as shown in figure 2.7.
A
dE
A :: d
dA
v
O
do
(~
A: Adsorbatc
E: Adsorbent
Figure 2. 7 Slit-shape pore model (Horvath and Kawazoe).
The potential energy of interaction, e, for one adsorbate molecule at a distance z from an atom in the surface layer can be expressed as
c (z) -- K c*
-
+
(9)
48
where E* is the potential energy minimum m
r = (--n--n)c-n) c'-') ?l-Dl
(10)
/71
and
1
o = (m__)(,-m) do
(11)
n
The parameters n and m are the order of the dispersion and repulsion terms in equation 9, respectively, and do is taken as the arithmetic mean of the diameters of the adsorbent atoms in the wall,
de,
and the adsorbate atoms,
dA, as
shown in figure 2.7.
With n = 10 and m = 4, K = 3.07 and tr = 0.858 do. If the distance between the nuclei of the two parallel layers is L, then the potential function of one adsorbate molecule between the two layers is"
()lO (,o),
c(z) -Ke*
(12)
In the case of the pore filled with adsorbate molecules one must account for adsorbateadsorbent, and adsorbate-adsorbate-adsorbent interactions. Walker 57 expressed the potential energy minimum, : , corresponding to these two respective interactions in terms of two dispersion constants, AE.A, and AA-A, as well as the number density of adsorbate and adsorbent atoms per unit area. This potential minimum takes the form: -
(13)
10
do4
49 where the dispersion constants can be expressed according to the Kirkwood-Miiller formalism as"
AF_A --
AA-A= with
M
c--
6Mc 2t~eaa
(14)
3Mc20~aXa
(15)
mass of an electron (kg) speed of light (m/s)
polarizability (cm3)
X=
magnetic susceptibility
(cm 3)
Using equation 11 to reexpress do in terms of a and by replacing E" in equation 12 by the expression in equation 13, one obtains:
r
-
+
NA a~_A + Ne ae_a 204
+( 0 )IO
(Z)4 (Z) 10 (~-Z)' ~'~-Z)] -
(16) .
If the free energy of adsorption is taken as equal to the net energy of interaction between the layers, then the energy balance yields"
50
f R r In
P
(17)
-~ N . ~ (L - do) - do
With substitution of equation 16 into equation 17 and integration this yields the Horvarth and Kavazoe result for slitlike geometry: RT
•(To) P
04
3 (L-do) 3
NA aA_~ + N~ a~_A
= Nay __
o 4 (L-2do) 1210
9(L-do) 9
__
124
3d3o
(18) +
1210 ]
9dgoJ
This has Kelvin-like simplicity in that it relates the relative pressure, P/Po, to the relevant pore size, L, in terms of the physical properties of the adsorbate-adsorbent couple.
Cylindrical model
To test the geometric sensitivity of the calculated pore size for a cylindrical model, a cylindrical potential was utilized along with the following assumptions: ~
2.
@
Q
The pore is a perfect cylinder of infinite length but finite radius, rp. The inside wall of the cylinder is a single layer of atoms, which is taken as a continuum of potential interaction. The interaction with the pore wall is taken to be due only to the dispersion forces. Adsorption occurs only on the inside wall of a cylinder in the micropore region. The interaction of adsorption is taken to be due only to that between the adsorbate and the adsorbent, with the latter considered to be the oxide ion of the zeolite.
51
Based on these assumptions, the potential energy for the cylindrical pore displayed in figure 2.8 is a function of r, the distance from the central axis. A general expression for a cylindrical potential was provided by Everett and Powl. 5~
5 r
r (r) = ~ ~
6 ttk, [-21~ (trp) -d ~ 1 7k--o ~
-
tr,)
~
--
[3k
(19)
where the constants Ctk and ~k are given by:
o.,
tg k
=
13o.5 =
r (-4.5) r(-4.5 - k) r(k+ 1)
(20)
r (-1.5) r (-1.5-k) r (k+l)
(21)
The potential energy minimum, e*, can be expressed in term of equation 13, using the dispersion constants from equations 14 and 15. Once again, if one assumes that the free energy of adsorption is given by the average of the intermolecular potential in the cylindrical pore, then one can take either a line average or an area average. In the latter, one can envision the molecules as having two degrees of freedom to move across the pore, while in the former they may have only one degree of freedom, along the diameter of the pore. This could correspond to the physically real case in which a portion of the pore is blocked by foreign atoms such as carbon.
For the line-average case, the potential for the cylindrical model becomes: rp-d o
f e(r) dr 0
ELl
=
r,-do
(22)
52
Ido
P
O
E: Adsorbent
O
A: Adsorbatr
Figure 2.8 Cylindrical pore model.
and the model equation that results is:
In(
RT
)-4
d4o
1 (1_ do)2~ (-~ 21 IXk ( d~ x E** [2k+1 k--o
r.
r.
(23) - [3k (d~ r.
For the case of the area-average cylindrical potential"
ELA =
f
e (r) 2~ rdr
o
(24)
r,-d,
f o
2~ rdr
53 and the model equation that results is" In(
RT
)-4
d~ (25)
x ~
,=o
~
1
rp )
"t
--
~ rp )
Pt ~ rp ) j
The Horvath-Kawazoe pore size model (or one of its modifications) has become recently available as a fully implemented software package.
3.2.3 Macroporous silicas Mercury porosimetry is generally regarded as the best method available for the routine determination of pore sizes in the macropore and upper mesopore range. The idea of using mercury intrusion to measure pore size appears to have been first suggested by Washburn, 58 who put forward the basic equation (the Washburn equation): rp = _ 2ycos0 Ap
with 3,
=
0
=
(26)
the radius of a cylindrical pore surface tension of Hg (484 mN/m = dyne/cm) contact angle between mercury and the wall, assumed to be 130 ~ or 140" depending on the source. 42
The principles of mercury porosimetry are further illustrated in figure 2.9. The technique of mercury porosimetry consists essentially of measuring the extent of mercury penetration into an evaluated solid as a function of the applied hydrostatic pressure. The full scope of the method first became apparent in 1945 when Ritter and Drake 59 developed a technique for making measurements at high pressures. The method has enjoyed increasing popularity with the passing of years, and automatic porosimeters are now in use for the routine examination of the pore structure of
54
catalysts, cements and other porous materials. The range of such porosimeters extends from r p --- 3.5 nm (corresponding to the usual maximum pressure - 2000 bar) to r p - 7 . 5 / ~ m , the size of pores penetrated at atmospheric pressure. One problem is that if the silica gel is not very strong, the structure collapses by the external pressure of mercury before pores are penetrated. It is for this reason that the nitrogen adsorption isotherm method is preferred for research purposes. Nevertheless, for strong bodies like industrial catalyst gels, the mercury penetration method is far more rapid not only in execution, but also in converting results to pore size distribution curves.
ii i ii!iiiiiiiii Sig:::i~ii~i~:t':~!i::::% Gillis D 'Hamers (H20 desorption); 1H NMR model.
At higher temperatures (> 673 K) the free silanols start to condense. This process is relatively slow, since a certain proton mobility is required. At low concentrations of hydroxyl groups, such an activated diffusion of protons along the surface strongly limits the condensation process. The absolute concentration of the geminal silanols decreases over the entire temperature region, but also very slowly. Since geminal hydroxyls may be considered as a special form of free hydroxyls, their condensation is limited by the same factors. The bridged silanols condense very quickly and disappear completely after a treatment in vacuo at 673 K. The suggested distribution of the silanol types as a function of temperature is numerically presented in table 5.5.
124
Silanol number (OI-I/nn/)
!
473
573
Figure 5.27 In vacuo. vicinal silanols.
673
773 873 Temperature (K)
973
1073
9total hydroxyls; + geminal silanols; * isolated silanols;
Table 5.5 Concentration of the different silanol types, as a function of treatment temperature in vacuo
Temperature (K)
Total OH (OH/nm 2)
Isolated OH (OH/nm 2)
Vicinal OH (OH/nm 2)
Geminal OH (OH/nm2)
473 573 673 773 873 973 1073
4.60 3.55 2.35 1.80 1.50 1.15 0.70
1.15 1.65 2.05 1.55 1.30 0.90 0.60
2.85 1.40 0 0 0 0 0
0.60 0.50 0.30 0.25 0.20 0.25 0.10
125
References
P. Van Der Voort, I. Gillis-D'Hamers and E.F. Vansant, J. Chem. Soc. Faraday Trans., 1990, 86, 3751, 0
P. Van Der Voort, I. Gillis-D'Hamers, K.C. Vranken and E.F. Vansant, J. Chem. Soc. Faraday Trans., 1992, 88, 723. D.A. Gardella, D.Z. Jiang, and E.M. Eyring, Appl. Spectrosc., 1983, 37, 131.
.
D.E. Leyden, R.S. Shreedhara, J.P. Blitz, and J.B. Atwater, Microchem. Acta Wien, 1988, II, 53.
.
.
0
.
.
0
D. Gorski, E. Klemm, P. Fink, and H. H6rhold, J. Colloid Interface Sci., 1988, 126, 445. I. Gillis-D'Hamers, P. Van Der Voort, P. De Hulsters and E.F. Vansant, A high resolution FTIR-PAS study of the free hydroxyl group on the silica gel surface, in Proc. Int. Workshop on FTIR, ed. E.F. Vansant, U.I.A., Wilrijk, 1990. I. Gillis-D'Hamers, K.C. Vrancken, E.F. Vansant and G. De Roy, J. Chem. Soc. Faraday Trans., 1992, 88, 2047. P. Fink, H. Hobert and G. Rudakoff, Wiss. Ztschr. F.S.U., 1987, R36, 581. (in German) J.B. Peri and A.L. Hensley, J. Phys. Chem., 1968, 72, 2936.
10.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1973, 77, 1465.
11.
F.M. Van Cauwelaert, P. Jacobs and J.B. Uytterhoeven, J. Phys. Chem., 1970, 76, 1434.
12.
H.P. Boehm and H. Kn6zinger, Catalysis-Science and Technology, Vol 4 Eds. J.R. Andersen and M. Boudart (Springer, Berlin, 1984).
13.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1976, 80, 1998.
14.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1975, 79, 761.
15.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1976, 80, 1995.
16.
B.A. Morrow, I.A. Cody and L.S.M. Lee, J. Phys. Chem., 1976, 80, 2761.
17.
P. Hoffmann and E. Kn/Szinger, Surface Science, 1987, 188, 181.
126 18.
J. Sauer and K.P. Scr6der Z. Phys. (Leipzig), 1985, 266, 379.
19.
R. West and R.H. Baney, J. Am. Chem. Soc., 1959, 81, 6145.
20.
G.I. Harris, J. Chem. Soc., 1963, 5978.
21.
D.W. Sindorf and G.E. Maciel, J. Phys. Chem., 1983, 105, 1487.
22.
G.E. Maciel and D.W. Sindorf, J. Phys. Chem., 1983, 87, 5516.
23.
J.S. Waugh, L.M Huber and U. Haeberlen, Phys. Rev. Lett., 1968, 20, 180.
24.
B.C. Gerstein, C. Chou, R.G. Pembleton and R.C. Wilson, J. Phys. Chem., 1977, 81,565.
25.
B. Schnabel, U. Haubenreisser, G. Scheler and R. Muller in Proceedings 19~ Congress Ampere, H. Brunner, K.H. Hauser and D. Schweitzer ed., Heiselberg, 1976.
26.
C. Bronnimann, R.G. Zeigler and G.E. Maciel, J. Am. Soc. 1988, 110, 2023.
27.
S. Haukka, E. Lakomaa and A. Root, J. Phys. Chem., 1993, 97, 5058.
28.
S. Haukka and A. Root, J. Phys. Chem., 1994, 98, 1695.
29.
R.V. Cvetanovic and Y. Amenomiya, Adv. Catal., 1967, 17, 103.
30.
R.E. Richards and L.V.C. Rees, Zeolites, 1986, 6, 17.
31.
F. Duprat and G. Gau, Ind. Eng. Chem. Res., 1990, 29, 1424.
32.
I. Gillis-D'Hamers, PhD-thesis, University of Antwerp, Antwerp, 1993.
33.
R.C. Weast, Handbook of Chemistry and Physics, CRC-press, 1984, p. D-47.
34.
E. Dima and L.V.C. Rees, Zeolites, 1987, 7, 219.
35.
L.T. Zhuravlev, Colloids and Surfaces A: physicochemical and engineering aspects, 1993, 74, 71.
36.
A.A. Agzamkhodzhaev, G.A. Galkin and L.T. Zhuravlev, in M.M. Dubidin and V.V. Serpinsky (Ms.), Main Problems of Physical Adsorption Theory, Nauka, Moscow, 1970. Proc. All-Union Conf., Moscow, 1968.
37.
E.S. Freemann and B. Carrol, J. Phys. Chem., 1958, 62, 394.
38.
D.A. Anderson and E.S. Freemann, J. Polym. Sci., 1961, 54, 253.
127
Chapter 6 The effect of surface morphology on the d e h y d r o x y l a t i o n behaviour
In the previous chapter, it was concluded that the silanol number (i.e. the total concentration of silanol groups on the silica surface, expressed in OH/nm 2) can be considered as a physicochemical constant, within certain margins of error (+ 0.5 OH/nm2). Examining the dehydroxylation behaviour of various silica structures in more detail, small differences in the dehydroxylation rate can be observed. Gillis-D'Hamers ~ has calculated the dehydroxylation energies (E~(OH)) of silica gels with varying porosity (cfr. chapter 5). The Ea(OH) curves as a function of the silanol number are presented in figure 6.1. In this figure, KG 40 and KG 60 are the acronyms for Kieselgel 40 and 60, which are silica gels obtained from Merck with an average Wheeler pore diameter of 2.1 nm and 3.6 nm respectively. The abcis of the figure can be converted to treatment temperature, using the data in table 5.5 (Chapter 5). For instance, the silanol region 4 - 5 0 H / n m z corresponds to a treatment temperature of 373 -573 K. It is obvious from figure 6.1 that there is no difference in desorption energies in the silanol region from 0 to 4 0 H / n m 2, confirming Zhuravlev's statement that the silanol number can be considered as a physicochemical constant.
128
Desorption energies (kJ/mol)
350
1001
~176
150-
0[ 0
I
I
1
I
I
2 3 4 Silanol number (OI-I]nm2)
I
5
Figure 6.1. Water desorption energies as a function of the silanol number for Kieselgel 40 (o) and Kieselgel 60 (0).
In the high silanol region, the KG 40 seems to dehydroxylate a little easier. This silanol region corresponds to a thermal treatment at 373 - 573 K, suggesting that in the very early stages of dehydroxylation, the surface morphology has a small impact on the rate of water desorption. However, quite contradictory conclusions have been drawn in literature concerning the dehydroxylation as a function of porosity. Snyder 2 stated that the relative ratio of free to bridged silanols is closely related to the pore diameter. He claimed that the surface of a wide pore is comparable to crystalline silica, and will therefore contain more free silanols. Dzis'ko et al. 3 found that smaller pores undergo slower dehydroxylation, whereas Bakyrdzhiev 4 reported that dehydroxylation occurs at lower temperature in smaller pores. In order to comprehend and explain these different statements, it is essential to consider the actual physical reason for bridged silanol condensation. The dehydroxylation of bridged silanols is mainly determined by the strength of the hydrogen bonding. The closer the vicinal hydroxyls, the stronger the hydrogen bonding, resulting in a greater stability towards dehydroxylation.
129 It is obvious that pore size (especially pore diameter) plays an important role in the spacing of the bridged silanols. Smaller pores exhibit a more negative radius of curvature, as shown in figure 6.2. This negative radius causes a decrease in the intersilanol distance, and thus a strengthening of the hydrogen bonding. This is reflected in a slower dehydroxylation behaviour.
i::~:, i::~::.
Figure 6.2.
H
H
H
H
H
:~il ....~::i
Stabilizing effect of small pore radii on the hydroxyl spacing.
However, the experimental data of figure 6.1 show exactly the opposite: the bridged silanols of Kieselgel 40 (smaller pores) condense more readily than the ones of Kieselgel 60. In view of this apparent contradiction, supported by the ambiguity in literature, one must conclude that the pore size cannot be the only parameter, influencing the relative distance between the vicinal silanols. Another factor must be involved, causing the destabilization of the bridged silanols in Kieselgel 40. Microscopically, a silica (gel) can be considered as a cluster of primary particles. The pores can be regarded as open spaces between these particles. The particle size itself has a destabilizing effect on the bridged silanols (figure 6.3). Smaller particles exhibit a more positive radius of curvature, resulting in an increase of the intersilanol distance. Precise data on the primary particle size of the silica gels could not be obtained from the manufacturer. Fortunately, particle size can be estimated, based on theoretical considerations.
130
H
H
H
H
H ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::.
Figure 6.3 Destabilizing effect of small particle sizes on the hydroxyl spacing.
It was pointed out in chapter 1 that Sheinfain's globular theory 5 allows an estimation of the average primary particle size if the specific surface area is known, a according to
S -
3 a.p
(1)
with S the specific surface area (m2/g), a the primary particle radius (nm) and p the density (2.2"106 g/m3). Putting the specific surface areas of 400 m2/g and 610 m2/g (cfr. chapter 2) for Kieselgel 60 and 40 in equation 1, one obtains primary particle diameters of 6.8 and 4.5 nm respectively. This means that Kieselgel 40 consists of considerably smaller primary particles, explaining the destabilization of the bridged silanols.
The loss in surface area upon aggregation, according to equation 2 (chapter 1) can be neglected if the particle radius R > > adsorbate radius a.
131 References
I. Gillis-D'Hamers, PhD-thesis, University of Antwerp, Antwerp, 1993. ~
~
~
~
L.R. Snyder, Principles of adsorption chromatography, Dekker, New York, 1967. V.A. Dzis'ko, A.A. Vishnevskaya and V.S. Chesalova, Zh. Fiz. Khim., 1950, 24, 1416. I. Bakyrdzhiev, Dokl. Acad. Nauk. SSSR, 1966, 168, 618. R.Yu. Sheinfain, N.S. Kruglikova, O.P. Stas and I.E. Neimark, Kolloid Zhur., 1963, 25, 732.
This Page Intentionally Left Blank
133
Chapter 7 Related Materials : Silicates
Most of our understanding of silicate structures comes from studies of the many naturally occurring or synthetic silicates. As for silica, the basic unit of the structure is the SiO 4 tetrahedron. These tetrahedra occur singly or by sharing oxygen atoms, in small groups, in small cyclic groups, in infinite chains, in infinite sheets or in structures enclosing cavities or channels.
1 Simple orthosilicates A few silicates are known in which there are simple, discrete orthosilicate, SiO44In such components, the associated cations are coordinated by the oxygen atoms, and various structures are found depending on the coordination number of the cations. There are a number of compounds of the type M2SiO4, where M 2§ is Mg 2§ Fe 2§ Mn 2§ or some other cation with a preferred coordination number of 6, in which the SiO44- anions are arranged to provide an interaction with six oxygen atoms in an anions.
octahedron symmetry. 2 Other discrete, noncyclic silicate anions The simplest of the condensed silicate anions are formed by combining two or more SiO4 tetrahedra by sharing oxygen atoms. The pyrosilicate anion, in thortveitite
(Sc2Si207) and hemimorphite (Zna(OH)2Si207).
Si2076-, is present
134 3 Cyclic silicate anions
The structures of two such cyclic ions, 5i3096 and Si601812-, are shown schematically in figure 7.1. The general formula for any such ion is SinO3~n-. The ion 5i3096" forms the structure of benitoite (BaTiSi309).
0-o "
=Si
Figure 7.1 Examples of cyclic silicate anions. 4 Inf'mite chain anions
These are of two main types" the pyroxenes, which contain single-strand chains of composition (SiO32-)n (figure 7.2) and the amphiboles, which contain double-strand, cross-linked chains or bands of composition (Si4O116-)n.
Figure 7. 2 A linear chain silicate anion.
Examples of pyroxenes are enstatite (MgSiO3), diopside [CaMg(SiO3)2], and spodumene [LiAI(SiO3)2]. A typical amphibole is tremolite Ca2Mgs(Si40~)2(OH)2.
135 5 Infinite sheet anions
When SiO 4 tetrahedra are linked into infinite two-dimensional networks as shown in figure 7.3, the empirical formula for the anions is (SiOsZ-)n. Many silicates have sheet structures with sheets bound together by cations that lie between them.
Figure 7.3 Sheet silicate anion structure idealized. 6 Minerals
The next logical extension in the progression above from simple SiO4 4- ions to larger and more complex structures would be to three-dimensional structures in which each oxygen is shared between two tetrahedra. The empirical formula for such a substance is simply (SiO2)n. However, if some silicon atoms in such a three-dimensional framework structure are replaced by aluminium, the frame must be negatively charged and neutralized by cations. Aluminosilicates of these type are clays and many zeolites.
136 6.1 Clays
Clays are aluminosilicates with a two-dimensional or layered structure including the common sheet 2:1 alumino- and magnesium- silicates (montmorillonite, hectorite, micas, vermiculites) (figure 7.4) and 1:1 minerals (kaolinites, chlorites). These materials swell in water and polar solvents, up to the point where there remains no mutual interaction between the clay sheets. After dehydration below 393 K, the clay can be restored in its original state, however dehydration at higher temperatures causes irreversible collapse of the structure in the sense that the clay platelets are electrostatically bonded by dehydrated cations and exhibit no adsorption.
Mr~. xH20
Figure 7. 4 The cross-linked structure of a 2:1 layer lattice silicate. Oxygen atoms in the central two planes link the octahedral sheet with the tetrahedral sheets. Metal ions in the octahedral and tetrahedral positions are not shown.
Pillared clays Pillared clays or in general pillared interlayered compounds have a virtually constant distance between the layers and therefore interlayer space. It is probably the most interesting group of new stable micro- and meso-porous materials available for adsorption and separation applications (figure 7.5). The great diversity of layered compounds and pillar combinations so far used in PILC synthesis were recently reviewed by Vaughan. ~ Pillared interlayered compounds are prepared by swelling the layered compounds in water or by amine intercalation and the exchanging of the
137 interlayer cations or anions with a large polymeric inorganic cation or anion. After filtration, washing and drying, a calcination at about 773 K is required to cross-link the pillars with the layers (figure 7.6). The overall process is quite complicated and can be tuned according to the utilization of the material.
i~o~O day" sweOOing
dehydraltion
b~\\\\\\\\\\\\\\\NNNN\\NN\\\"x\X~
~ ~~~
o~
~ - ~ ~ _
~KK';KK';~KK~-;';K'~';KK';'~i
~\\\\\\\\\\\\\\\\\\\\\\\\\\\\'q
~N\\\\NN\\NN\\\N\\~ ~\\\\\\\\\\\\\\\\\\~
,:.:.:.:.-.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:
~\~,,NN\\\\\\\\\\\\~
~NN\\\\\\\\\\\\\\\~
D iiiiiiii ii',iiiii /ii~,i',ii~Hiii,',i', ~ ~\\\\\\NN\\\\\\\\\~
PiO0~~l cll~y dimensien~J 9 s~s~biOi~ 9
9
Figure 7. 5 Swelling effects in normal clay and dimensional stability in pillared clays. 6. 2 Zeolites Since their discovery more than a century ago, many studies on zeolite minerals have been carried out. Moreover, in the late 1940's the preparation of synthetic zeolites was the start of a large range of zeolitic materials reflecting the complete range of framework substituted tectosilicate including phosphates. From their structural formula nx M n+[(At0~)~ (Si02)y ] wH20
138 I
12_-
I
micropore !_ m
"
m
1
mesopore
Figure 7. 6 Schematic representation of pores in pillared clays. adsorption and ion exchange properties were recognized as their specific characteristics. The possibility of zeolites to adsorb molecules of relative small size over molecules of larger size formed the basis for the introduction of the term molecular sieves. The first generation of zeolites (1940-1950) reflects the zeolites with low silica to alumina ratios (SIO2/A1203 -< 10), prepared from a crystallization of reactive aluminosilicate gels with alkali and alkaline earth metal hydroxides. Such zeolites are characterized by a high ion-exchange capacity, an extremely hydrophilic surface and many acid sites with a wide variety of acid strength. Such frameworks show different thermal and hydrothermal stability, depending on structural and compositional factors. Typical zeolites are zeolite A, X, Y, chabazite, erionite, mordenite (figures 7.7 and
7.8).
139
a
b
l
OH
0
[H]
0
It
S-NHCQN O It
2 §
CI
165 The stability of covalent bonds formed in method A have been tested by washing off the immobilized enzyme with detergents. 4~ This study indicated that glutaraldehyde reacts quickly with the APTS surface with the formation of stable bonds. The enzymes adsorb to the activated surface, with the majority being adsorbed within the first minute. Stable covalent bonds are formed between the glutaraldehyde and the enzyme. Besides the stability of the bonding of the enzyme to the support, the activity of the immobilized enzyme is an important parameter. Kallury 4~'42'43 studied the influence of immobilization on phospholipid-bound silica. He found a considerable stabilizing influence of the phospholipid environment on the enzyme activity. The results with urease demonstrated that a very high thermal stability could be achieved by combining the rigidity generated through immobilization with the stabilizing effect of the phospholipid molecules. The current large body of immobilization procedures has been reviewed by Cabral and Kennedy. 44 Besides facilitating and optimizing the practical application of enzymes, the developing of the immobilized enzymes also allowed a better study of enzyme functioning. 45 Many other new applications in the field of biotechnological products and processes have emerged. Enzymes immobilized on various supports are used e.g. for the purification of molecules that bind to immoglobulines, lecithins or other binding proteins, for improved biocompatibility of materials, to provide selective biosensors and in solid-phase analytical assays. Of these, silica is used as a porous support in the fabrication of analytical detection s y s t e m s . 46 1.4.2 Affinity Chromatography (AC) Affinity Chromatography was initially defined as a method based on specific and reversible molecular interactions between biologically active substances. However, the method has extended to non-biological stationary phases, such as metal-chelate complexes. This technique is used for separation and purification of proteins and other biologic materials, such as viruses and cells. 47'48 A survey of various stationary phases and affinity interactions is given in figure 8.2.
166
Motrix ~nmobiUzedIAffine
componentIcomponen{
E
(a)
(b)
Motrix Adsorption (or binding)
Protein
0 II
J
U
componen{
of {he of'fine
0
nu.,
L1
,:.me H20~" 1~"O4-1.L 2+3 ~ LJ = -
?A
~ ~
I
O L~
3HzO
(c)
(d)
V IOn rO'e'n=
(e)
~Ads~176
~ t
+
-I- R'-S-S-R' Figure 8.2 Variants of affinity chromatography, (a) biospecific A C," (b) metal chelate chrom. ;(c) charge transfer adsorption chrom.; (d) hydrophobic interaction chrom, and (e) covalent chrom. (chemisorption). Abbreviations: E = enzyme, L = amino acid group, me = metal ion, R~ = electron - withdrawing substituent, R~ = electron donating substituent; taken from ref (47) with permission.
167 Specific inhibitors for proteins are immobilized on a (inorganic or organic) porous substrate. In varying the inhibitor, enzymes can be selectively adsorbed and enzyme mixtures separated. The adsorbed enzyme may be eluted by a solution of free ligand or by a change in pH and/or ionic strength. 49 Affinity chromatography combines the analytical and chemical capacities of chemically bonded stationary phases and immobilized enzymes. Technology and methodology of both techniques are joined in the development of affinity stationary phases. Since steric requirements are even more determining than in simple immobilized enzyme systems, spacer molecules have great importance in these modifications. Commonly used spacer arms are summarized in figure 8.3.
The ultimate combination of HPLC and AC is effectuated in High Performance Affinity Chromatography (HPAC). 47 The development of this hybrid technique was highly assisted by the use of modified silica. Traditional polysaccharide supports may not be used for HPAC, because they lack mechanical stability to withstand the high pressure drops, inherent to this method. Modified silica beads are well suited. These may be coated with active groups as in normal AC applications. Additionally, if the separation requires the use of an organic stationary phase, the silica beads are modified with a silane or polymer with subsequent deposition of polysaccharides such as dextrans, agarose or cellulose. 5~
1.5 Industrial
In the industrial field, modified silica assists the development of advanced materials. Modified silica powders are used as a filler in rubbers and in new-type gas sensors. Most industrial applications involve the use of strongly related materials such as glass fibres or thin SiO2 layers. We wish to restrict ourselves to a discussion of the application of powdered silica and a short indication of related compounds. The chemistry behind the modification of powders, fibres or thin layers is similar. Therefore fibres as well as thin oxide layers have repeatedly been used to model silica modification mechanisms.
168
0 - CH 2" CH2"OH
(CH2) 3" N H 2 - (CH2) n- NH 2 [- (CH2) n- COOH -
CH 2- CH
-
CH
n = 2-12 n=2-12
\o / [" (CH2) 3" N H - (CH2) 3- N H 2 OH O OH II II I [- CH 2- CH - CH 2" NH - C - CH_2," NH - CH 2- CH - C H2-NH 2
-cH-c2 6 ~-N-N
-c5 ~
-cH-cooI 2 N-N+CI"
Figure 8.3 Some spacers used in affinity chromatography; takenfrom ref (47) with permission.
1.5.1 Composites Silica powder, glass beads and fibres are commonly used for the reinforcement of plastics. The produced composite materials have an increased thermal and mechanical stability, compared to the pure polymeric material. In order to bind the inorganic filler to the organic matrix, silane molecules, with both an inorganic and organic side, are used. The silane may be mixed with the matrix and filler material in the composite preparation, or be coated onto the filler prior to mixing. The application
169 of silane coupling agents perseveres the enhanced properties of composite materials in wet conditions. Glass fibres are used in the fabrication of tough materials, while silica powders are added to elastomers. Commonly used materials for glass fiber reinforcement include polyester, epoxy and phenolic resins. These are used in building and construction industry. Glass fibres are mostly of the E-glass type. The E-glass composition is displayed in table 8.6. 5~ The mechanism governing the reinforcing ability of coupling agents has been repeatedly studied, and falls beyond the scope of this book. The reader is referred to the work of Plueddemann 2'52 and Fowkes. 53'54 Table 8.6 Relative percentages of various oxides in E-glass composition51
E-glass composition (%) SiO2 A1203 CaO MgO Na20 K20 B203
55.2 8.0 18.7 4.6 0.3 0.2 7.3
In elastomers, such as natural and synthetic rubbers, modified silica as well as active carbon is used as a reinforcing material. It imparts utmost tear, tensile and abrasive resistance. The advantages of silica over carbon are its colourlessness and its constant isolating properties, even at high temperatures. Disadvantages of pure silica as a filler, such as the low network density, have been overcome by the application of the modified substrate. The difunctional tertrasulfane (bis(3-trimethoxysilylpropyl)tetrasulfane) is commonly applied. In composite formation, the sulfur bridges are split and react with elastomer carbon-carbon double bonds (figure 8.4). Other silanes, containing amino, sulfur, halogen or phenyl groups are also applied with various rubber types, as summarized in table 8.7. 55
170
--Si-'-Si-Si-
o
\Si" -O/ \ (CH2)3.S .~ S I -o\ (CH2)3- S ~S
+
o/Si /
--Si-
i/
\ OC2H5 $81 A
=Si-Si=Si-Si-
- oO\Si/OC2H5/\ / ~ (CH2)3- Sa- "Sxl Sx3 -O\o/ /OC(CH2) ~H3- Sb'I 2 ~ Si\ 5 -Sx2 Sx4
Figure 8. 4 The use o f bis(3-trimethoxysilyl)tetrasulfane) modified silica in the vulcanization of elastomers (idealized a +b = 4, 1 90 90 8); taken from ref (55).
Table 8.7 Use of organosilanes in reinforced rubbers55 (NR: natural rubber, SBR: slyrene-butadiene rubber, EPDM: ethylene-propylene-diene rubber)
ORGANOSILANE
RUBBER TYPE
FILLER MATERIAL
(RO)3Si-(CH2)3-C1
halogenrubber NR, SBR
SiO2 SiO2, clay
(RO)3Si-(CH2)3-NH2
NR, SBR
SiO2, clay
(RO)3Si-(CH2)3-SH
NR, SBR
SiO2, clay
(RO)3Si-(CH2)3-S-C(NH2)2+C1-
halogenrubber NR, SBR
SiO2 SiO2, clay
(RO)aSi-(CH2)3-Sx-(CH2)a-Si-SCN
NR, SBR
SiO2
(RO)3Si-(CH2)3C5H5
EPDM
SiO2, clay
171 1.5.2 S-emission monitoring The modification of silica gel with diethanolamine (DEA) resulted in a stable alkaline filter, capable to collect H2S, COS, CS2, SO2, CO2 and H20 from contaminated air. 56 The presence of odorous mercaptanes and organic sulphides may be selectively determined, as these compounds do not react with the DEA and therefore are not collected on the filter. By means of this scrubber an automatic survey system has been developed allowing the detection and determination of the nature of odour nuisance, caused by industrial emissions of volatile organic sulphur compounds, up to the ppb-level. The system consists of two continuous, highly sensitive and fast responding sulphur analyzers. One of the detectors is equipped with an SOx scrubber and measures the total amount of non-SO2 sulphur compounds. The other detects the organic S-compounds, using the amine-silica filter. 1.5.3 Modification of thin SiO2 layers In various industrial applications, silica modification is used on thin layers of SiO2, rather than on the powdered form. Monomolecular or thin layers of silicon oxide are thermally grown on Si-wafers for the production of high-tech materials. The surface chemistry of these layers is comparable to the powdered form, with the absence of a porous structure. This type of material is commonly used in the production of semiconductor devices. 57 The silica layer is used as a starting layer for integrated circuit (IC) build-up. IC layer materials range from single crystals and doped polycrystalline silicon, silicon nitride, thermally-grown oxide to vapour deposited or sputtered metal or metal silicide layers. Structural adhesion of the various layers is obtained by the application of organosilanes, such as AEAPTS, APTS and GPTS. The thin layer technology is also used in the fabrication of chemical sensors. Cleche: 8 defines a chemical sensor as a physical device, called a transducer which delivers an electric signal, which is controlled by its sensitized part, called the detector. This detector must be selective i.e. must be prepared, or functionalized in order to respond only to the substance or group of substances to be detected. Since sensor technology goes beyond the scope of this work, we will limit ourselves here to this definition.
172 In research on the mechanisms gouverning the modification reactions, the thin silica layers allow the application of various surface analytical techniques, which are of no use for analysis inside porous systems. Reaction mechanisms are simplified by the elimination of porosity and may be studied by direct surface techniques such as ellipsometry, as well as microscopic techniques such as Scanning Electron Microscopy (SEM) and Atomic Force Microscopy (AFM). 59
2 Modification procedures 2.1 Introduction
In the selection of a chemical modification procedure, two criteria have to be considered. (i) (ii)
the aimed coating morphology; the scale on which the modification has to be performed.
The coating morphology includes layer thickness (mono or multilayer), the modification density (molecules/nm2), the orientation of the surface molecules, and the type of interaction of the coating layer with the surface (relative amount physisorption/chemisorption). The second criterium is of equal importance. Laboratory-scale modifications require different procedures than industrial applications. Modifications for mechanistic studies may be more labour-intensive than large-scale fabrication of e.g. filler materials. In this perspective, various applications also require a variable control of the surface morphology. The production of stationary phases for HPLC needs careful control of the obtained surface structure, although they are produced on a relatively large scale. In industrial applications, the degree of precision which needs to be achieved, is of main interest. In this respect, the time/temperature profile of the procedure, the material movement inside the reaction vessel, and the means and form of introducing the reagent need to be considered. The cost of the surface modifying reagent also dictates the precision of the method and the monitoring which should be applied. 6~
173
The procedures used for the chemical modification of silica, will be discussed using these criteria. This survey is restricted to the preparation of chemically modified silicas, used as a base material in the above-cited applications. Procedures are ordered according to the possibility to control the ultimately formed layer.
2.2 Sol-gel The incorporation of organofunctional groups on the silica surface may be effectuated during the synthesis of the silica material. The addition of organofunctional alkoxysilanes to the TEOS solution in the sol-gel process, produces functionalized silica gels. This procedure does not allow a careful control of the obtained surface morphology. Since the relative amounts of silane and TEOS is the only variable parameter, neither layer thickness, nor modification density can be precisely tuned. This results in an irreproducible functionalization of the surface.
Base-catalyze~ RSi(OR' )3 1
H20
Acid-base RSi(OR' )3 I
1. H O 2. ~_i~catalyst
Hydrolysis
Hydrolysis
I
I Na4~
Na 4oH
Gelation Vacuum 353 - 373 K Dried gel
Gelation
Chlorosilane RSiC1
3
1. CHCI 2, H20 3
Hydrolysis and gelation I Vaemtm 353 - 373 K Dried gel
| Vaeanma 4 353 373 K -
Dried gel
Figure 8.5 Synthesis of organosilicon gels, takenfrom ref (61) with permission.
174 The organic functionality may also be incorporated by the sol-gel synthesis of the pure organosilane. The thus formed gels are referred to as organosilicon gels. These type of gels have been produced to serve as precursor gels in the preparation of silicon carbides. 6~ Three routes have been set up, as displayed in figure 8.5. Either alkoxy- or chlorosilanes are used as a starting material. Alkoxysilane gel synthesis may proceed via a base catalysed or an acid-base route. The effectiveness of each route is dictated by the functionality of the starting silane. The acid-base route is most generally preferred. 61'62 The obtained gels show very low porosity and surface area values (table 8.8). Therefore this procedure is only used for dedicated applications, such as the production of ceramics.
Table 8.8 Pore volume and surface area data for chlorosilane-derived gels 61
R-group
Pore vol. (cm3/g)
Surface area (m2/g)
Methyl Ethyl Vinyl Propyl Allyl Phenyl
0.45 0.26 0.21 0.09 0.22 0.05
2.20 0.33 0.27 0.24 0.08 0.15
Surface coating using sol-gel technology is mainly used for the preparation of oxide or organofunctional siloxane layers on solid materials. In this case, the material is dipped or spun in the TEOS sol. This procedure leads to the formation of multilayered coatings of irreproducible thickness. Since this type of coating may only be used for axially or radially symmetric materials, alternatives have been developed. Those include spraying, electrophoresis, thermophoresis and ultrasonic pulverization. 63 However, these are not of interest in powder modification.
175
2.3 Aqueous solvent The preparation of organofunctional silica gels on industrial scale is performed by liquid phase reaction. As a solvent, water, a water/ethanol or water/acetone mixture is used.
Chlorosilanes or alkoxysilanes are used for this type of modification.
In the aqueous solvent the silanes undergo hydrolysis and condensation, before deposition on the surface 64 (figure 8.6).
RSi(OMe) 3 3H20 -~,/[ k.~
Hydrogenbonding R I HO~Si
Hydrolysis
O
Condensation
H
~Si
I
H
~OH
O H
H
H
"O
O
"O "
I
I
I
2H 20
Substrate
2H20
R I
R I
R I
I
I
I
Bond formation R
HO - - S i - 0 -- Si - - 0 - - Si - - OH
OH
R I
O
H
!
OH + OH
O ~ S i ~ O I
I
RSi(OH)3 3MeOH
OH
R I
HO--
OH
R
I
R
I
Si--
0 -- S i -
O
t
0 --
O
OH
i
Si' m
I I
O I
H ~
Substrate
OH
H i
0 I Substrate
Figure 8. 6 Mechanism of silane deposition in aqueous solvent, taken from ref (64), with permission.
In contact with water, halogen or alkoxy groups are hydrolyzed.
The as-formed
silanol groups go into hydrogen-bonding interactions with neighbouring hydrolyzed silane molecules and with surface silanol groups. release of water.
Siloxane bonds are formed, with
The coating molecules are polymerized horizontally as well as
vertically. Thus, a three-dimensional polymeric silane network is formed on the silica surface.
The polymerization
reactions
are hard
to control
and a layer of
176 unreproducible thickness results. The control is ameliorated with the addition of a variable amount of a polar organic solvent, such as ethanol or acetone. Ethanol/water mixtures in a 70/30% ratio are commonly used. Covalent bond formation is not an immediate process. Silane coating layers consist of physisorbed as well as chemisorbed molecules. Physisorbed molecules go into condensation only slowly and chemical stabilization of the coating layer requires a post-reaction curing step. In this step the modified substrate is thermally treated at temperatures generally in the 353 - 473 K range. The rate of hydrolysis of silanes is influenced by the functionality of the organic group and the type of the leaving group. The stability of the silanols increases with increasing size of the organic group. Acyloxy- and aminofunctional silanes are far more susceptible to hydrolysis than other alkoxysilanes. They hydrolyse in minutes in water, whereas others are stable for several hours. The increased rate of hydrolysis of the former is due to the acid or basic character of the functional group. Hydrolysis of the other alkoxysilane molecules may be catalyzed by the addition of a base or an acid. In solution, acid addition accelerates hydrolysis but decelerates polymerization. In the reaction mixture, however, the acid evaporates from the surface after hydrolysis, thus allowing polymerization. The molecular weight of adsorbed species increases, with increasing concentration of HCI. 65 The amount of silane deposited from water on a mineral surface, increases rapidly with increasing concentration, up to certain transition concentrations. Above this concentration deposition increases more slowly. These type of transitions are attributed to the onset of micelle formation between the aqueous silanetriols, causing enhanced stability of the hydrolyzed molecules. 2 The detailed hydrolysis and condensation behaviour of silane molecules in solution has been studied repeatedly, those were brought together in the festschrift in honour of Prof. Plueddemann. 66 Morral167 reported the formation of APTS layers with a thickness on the order of
10nm, by modification from aqueous solution. Tutas 68 found that aminosilane layers reach maximal thickness within the first minute of reaction, while vinylsilane layers keep growing with time. Films, deposited on glass from 0.1 - 5 % aqueous solutions were 5 - 20 nm thick. The thick, polymerized layers are ideal for industrial applications, such as the coupling of filler materials. In the chemical or analytical
177 field, they are of very limited value. For catalyst development and chromatographic separations, well-defined, ordered layers are required. In order to combine the stability gained by cross-linking with the reproducibility and characteristics of a monolayer coating, modification procedures in organic solvents have been set up.
2.4 Organic solvent If modification with chlorosilanes or alkoxysilanes is performed in fully dry conditions (dry organic solvent, dehydrated surface), hydrolysis is prevented. Chemical bonding with the substrate should result from the direct condensation of the chloro- or alkoxygroups with the surface silanols. From experiments using methoxymethylsilanes, B l i t z 69 concluded that this direct condensation does not take place. Post-reaction curing only results in evaporation of the adsorbed molecules. Alkoxysilanes may only bond chemically to the silica surface if water is present at the interface. Thus adsorbed silane molecules are hydrolyzed before reaction with the surface. Hydrolysis, however also causes polymerization and therefore non-monolayer coverages are obtained. Another way to realize the direct condensation is using ammonia as a catalyst. 7~ Surface loadings of up to twelve times the uncatalyzed reactions have been found. In figures 8.7 and 8.8 two possible mechanisms of ammonia catalysis are proposed. 7~
OMe Me~, I ..,,, ~H MeOV Sir'~u " ~ ~ N 7--- H O J..L.. H ~ " " H ............
J ................................................................................................
Me
+ MeOH
MeO b'gSi~OMe :-
t
O I .....
Figure 8. 7 Amine catalysis of silylaytion reaction, mechanism #1.
+NH
3
178
80/-
H, ~H N- H H";" 8+
OMe MeO, ,[. , ~l-Me
MeO'~_.
A
H,,~ N- H /n .
>'.8 +
. .. Me H H Meu O \ 9 Me .,.~,,./ N.~H ~bl" H ")' 8 + MeO N O / +
B
MeO
Me-~ H
M e /~1 ----'MeO ~,-~H
o" D
--,~.m
C
H
OMe
Si I
H
MeO /
+ MeOH 3
\
o E
Figure 8.8 Amine catalysis of silylation reaction, mechanism #2. In the first mechanism (figure 8.7) the transition state involves the formation of a lowenergetic six-membered ring. The second mechanism (figure 8.8) proceeds by the formation of a pentacoordinate intermediate, after nucleophilic attack of the ammonia at the silicon atom. Aminosilanes contain the catalyzing amine function in the organic chain. The reaction of aminosilanes with silica gel in dry conditions is therefore self-catalyzed. They show direct condensation, even in completely dry conditions. Upon addition of the aminosilane to the silica substrate, the amine group may form hydrogen bonds or proton transfer complexes with the surface silanols. This results in a very fast adsorption, followed by direct condensation. This reaction mechanism of APTS with silica gel in dry conditions, is displayed in figure 8.9. After liquid phase reaction, the filtered substrate is cured, in order to consolidate the modification layer. concluded that the initial adsorption step for -y-aminopropyltriethoxysilane (APTS) takes less than 18 s, which was the shortest measurable time. The ultimate orientation of the aminopropylchain (amine near the surface or pointing away from it) is determined by the presence of water in the reaction and curing phase. 72 Kallury 73 demonstrated that the addition of a base in the reaction also reduces the amount of
Morral167
179 ) J
SI
APTS
H
0
1
NIt
2
Silica
Physisorptioa t-O, ~,O Si
& O~ S~,~
or
+
NIt 3
d
I
d-
Condensation
O"
rt, ,,H
o I
....N .~" ...... I H
,,H ........Nr'r'l H Curing
0
0
Figure 8.9 Modification o f silica gel with APTS.
amine-surface interaction. Further details on the reaction mechanism of APTS will be given extensively in the subsequent chapter. In order to exclude any hydrolyzed molecules in the reaction mixture, silanes are commonly distilled under reduced pressure before use. The toluene solvent may also be distilled prior to use or be stored on molecular sieve (5A zeolite) to assure complete drying. For extreme precaution, the silanation reactions may be performed under dry nitrogen atmosphere. A CaCI2 guard tube may also be employed. The use of hazardous organic solvents makes this procedure non-suited for large-scale industrial application. On the other hand, the possibility to control the modification structure, makes it very well suited for lab-scale mechanistic studies. The density of the alkyl-loading is strongly determined by the size of the alkylgroup. If bulky alkylgroups are used, the number of siloxane linkages with the surface is reduced, leaving unreacted alkoxy groups. Those are easily hydrolyzed, producing
180 residual silanols in the bonded phase. Since these have deleterious effects on chromatographic performance of n-alkyl stationary phases, a new procedure was developed to increase silane loading density for this type of application.
2.5 Self-assembled monolayers Among the liquid phase adsorption procedures, Self-Assembled Monolayers (SAMs) take a special place. This type of monolayer originated as an extension of LangmuirBlodgett film technology. TM In this latter technique, highly ordered films of large polar molecules are deposited on fiat surfaces. Being obtained as insoluble ordered floating films on the surface of a liquid, Langmuir-Blodgett monolayer films are transferred on solid supports from the water-gas interface, by dipping or immersion of the substrate (figure 8.10).
i (a)
Co)
I
I
I
|
Figure 8.10 Langmuir-Blodgett film formation on a solid support by transfer of a film at liquid-gas interface (a) to solid-gas (b) interface.
181
In the formation of SAMs, the film-forming molecules order themselves by chemical interaction with neighbouring molecules and with the substrate surface. This technique has been applied for a large variety of modifier/substrate combinations. Various sulphur compounds, like alkanethiols and (di)sulfides have been deposited on metals such as silver, copper and gold; isocyanides on platinum and carboxylic acids on aluminum oxide and silver oxide. 75 Alkyltrichlorosilanes have been deposited on gold, mica, aluminum, tin oxide and silicon oxide. The latter combination is of interest here. The main feature of the technique is the deposition of the coating molecules from organic solvent, on a cleaned hydrated surface. As a solvent dicyclohexyl, ethanol, n-hexadecane and n-heptane are used. In a typical synthesis procedure, the silica is first cleaned to remove all trace organic compounds. This cleaning step appears to be crucial in the formation of smooth, complete monolayers, especially on metal surfaces. Cleaning is performed by boiling in nitric acid or hydrogenperoxide/sulphuric acid ('piranha') solution. The cleaning step is followed by a careful rinse with distilled water and drying in a stream of dry nitrogen. The silica may be rehydrated by exposure to air with controlled humidity, shortly before use. Whereas the modification of flat solid substrates involves an immersion/retraction procedure, this is not possible for powdered substrates such as silica gel. Therefore, these are stirred in the reagent solution, filtered or decanted and rinsed with the pure solvent. Detailed descriptions of the reaction conditions may be found in 7 74 76 77 literature. ' ' ' The SAM is formed by using only the monolayer of surface water on the silica gel. Since no excess water is present, the polymerization reaction is strictly under control. Additionally, the water is present at the place where the surface bond has to be formed. Thus-formed SAM's are densely packed, ordered films, attached to the surface with chemical bonds. Alkyl chains are aligned paralel in a densely packed fashion. The surface is fully covered, irrespective of the number of hydroxyl groups. Not all silane molecules are covalently linked to the surface. Le Grange et al. 76 evidenced that on a dehydrated surface that was exposed to moisture, 1 in 5 octadecylchlorosilane molecules is bonded to the surface. Due to this dense structure and full surface
182 coverage, this type of layer is clearly different from other wet modification methods. In figure 8.11 the structure of a self-assembled monolayer is compared to that of a conventional polymeric layer.
0 " Si" ~ Si~ ~ Si~ ~Si
b ~ 1 7 6 1 7~6 1 7 o6
~Si.
..S~o.
o
ZZ~7
HO " i"o" b x OH
HO (~ "0
"O'~li"oOH
Figure 8.11 Schematic drawings of self-assembled monolayer (upper) and conventional monolayer. Takenfrom ref(8), with permission.
Wirth and Fatunmbi s defined the bonding of trifunctional silanes as self-assembly, when the bonding density of functional groups is made to be close-packed, approximately 8/~mol/m 2 (2.2 nmZ/chain). Conventional polymeric phases are no more than 5 /~mol/m2. Because of their close packing, these phases are very well suited for chromatographic separations, since all interference of surface hydroxyls is excluded. In order to avoid over-crowding of the alkyl chains, but retain the advantages of molecular self-assembly, mixed phases of long and short chain alkyls have been prepared. As one of such possible combinations, C3 chains have been mixed with
C18.7
183
2. 6 Hydride intermediate Alkyl chains are generally deposited on the silica surface using the corresponding alkylsilane. The hydride intermediate route, developed by Sandoval and Pesek 7g'79 gives an alternative for this method. The one-step alkylsilane route, is replaced by a two-step silanization-alkylation procedure. The loss of process simplicity is compensated by the advantages of an increased alkyl density and enhanced coating stability. In modification reactions using alkylsilanes, the silane density is influenced by the size of the alkyl group. The preparation of chromatographic stationary phases involves long chain alkyls (C8 or C~8) but requires a high alkyl density with a minimal amount of residual silanols. To overcome this apparent contradiction, alternatives to the alkylsilane methods have been developed. In the hydride intermediate route, the sterical limitations of the silanation reaction are overcome by applying the smallest possible silane. Triethoxysilane (HSi(OCH2CH3)3, TES) is used in a first step, producing a hydride surface. In order to make a clear distinction with normal alkylsilane reactions, this step is referred to as silanization. The silanization reaction is performed in dioxane/water solvent at reflux temperature, with acid catalysis. 8~ The reaction of the TES is analogous to other trialkoxysilanes. HSi(OCH2CH3) 3 -k 3 H20 ~ HSi(OH) 3 -I- 3 CH3CHzOH
(A)
HSi(OH) 3 -t- HO-Si --- --, HSi(OH)2-O-Si ---
(B)
After reaction (B) bidentate and intermolecular linkages are formed. This way to produce a hydride surface was proven to be preferential to the older procedure, involving chlorination and reduction of the surface. 8~ The deposition of alkylchains on the hydride surface is performed using heterogeneous phase hydrosilation of terminal unsaturated hydrocarbons (reaction (C)). cat. - Si-H + CH 2= CH-R
~
- Si-CH2-CHz-R
(C)
184 The reaction is influenced by various parameters, such as solvent, reaction time, reaction temperature and applied catalyst. 79 The reaction proceeds well in toluene solvent. However, if possible it is best to use the neat liquid olefin, depending on the physical state of the olefin at the reaction temperature. If a solvent is used, higher concentrations of the olefin lead to higher surface loadings. Various types of heterogeneous as well as homogeneous catalysts may be used. However, it appears that no catalyst of general use will be found. Every olefin has its own catalyst of choice. In this reaction a direct surface Si-C bond is formed. This type of bond appears to be more stable than the normal ---Si-O-Si-C bond. The stability of a Cs stationary phase, prepared both by hydrosilation and organosilanization are compared in figure 8.12. Both curves represent fraction of the initial surface coverage as a function of hydrolysis time. It is clear that the hydrosilated phase has a much better performance over the displayed time interval.
Fraction of initial surface coverage 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0
' 2'0'
4b ' 6'0'' Elapsed time (h)
80
'
100
Figure 8.12 Relative surface coverage of C-8 phase on Vydac TP as a function of hydrolysis time. A: C-8 prepared by hydrosilation, B: C-8 prepared by organosilanization. Takenfrom ref. (79), with permission.
185
2. 7 Vapour-phase reactions Modification of silica gel with volatile or gaseous compounds is performed in the vapour phase. Industrial-scale reactors and laboratory scale gas adsorption apparatus have been used. In the industrial field, fluidized bed and fluid mill reactors are of main importance. In fluidized bed reactors, g2 the particles undergo constant agitation due to a turbulent gas stream. Therefore, temperatures are uniform and easy to control. Reagents are introduced in the system as gases. Mass transport in the gas phase is much faster than in solution. Furthermore, gaseous phase separations require fewer procedural steps than solution phase procedures, and may also be more costeffective, due to independence from the use and disposal of non-aqueous solvents. All these advantages make the fluidized bed reactors preferential for controlled-process industrial modifications. Fluid mill reactors have comparable characteristics. In this type of reactor, the reagent/solvent mixture is introduced as a spray. The solvent is evaporated quickly upon introduction, leaving the reagent to interact with the substrate. 6~ For laboratory-scale modification, distinction has to be made between static and dynamic adsorption procedures. In a static procedure, the substrate is contacted with a known volume of gas at a well-defined pressure. The modifying gas may be stationary or circulating in a closed loop. Modification in a static gas adsorption apparatus allows the careful control of all reaction parameters. Temperature and pressure can be controlled and easily measured. Adsorption kinetics may be determined by following the pressure as a function of the reaction time. Figure 8.13 displays a volumetric adsorption apparatus, in which mercury is used, as a means to change the internal volume and for pressure measurement. In the dynamic gas modification procedure, the reacting gas is passed through the substrate and dissipated. Temperature and gas flow may be controlled, with the limitation of pressure build-up at the substrate site. Relatively large amounts of gas may be passed through the sample. The main difference between both procedures is the type of control of the reaction mechanism. For static reactions, the reaction velocity is controlled
186 thermodynamically. In the dynamic regime, on the other hand, the processes are under kinetic control.
F
[ A
C
Figure 8.13 Dynamic volumetric gas adsorption apparatus, (A) sample compartment, (B) calibrated volume bowls, (C) cryogenic trap, 09) manometer, (E) evacuation line, 07) circulation pump.
Small amounts of substrate (form 0.2 to several grams) may be modified. Therefore, this type of procedure is preferential in the fundamental research on modification mechanisms. For mechanistic study of reaction mechanisms various 'home-made' reactors have been build. Variable reagent dosing and spectroscopic detection systems have been
187 reported. 83'84'85 In these laboratory-scale procedures, either gaseous or volatile compounds may be used. Concerning volatiles, mainly volatile liquids, such as shortchain chloro- and alkoxysilanes have been reported. However some crystalline organic solids, which are of interest for silica modification, may also be deposited from the vapour phase. Solid amino acids, nucleic bases and carboxylic acids may be sublimed without decomposition at normal or reduced pressure and elevated temperature. The adsorption of carboxylic acids from the vapour phase, with formation of Si-O-CO-R bonds has been reported by Basyuk. 86 Studies on the modification with diborane, and various organosilanes have been reported and will be discussed in subsequent chapters. Hydrolysis and condensation behaviour are analogous to modification procedures in dry organic solvent. However, controlled gas phase procedures allow a better control of reaction conditions, leading to a more reproducible coating layer. While the possibility to form true monolayered coatings from solution has been doubted, 73'85gas phase modification is generally agreed to give monolayer modification.
References
H.A. Mottola, in Chemically Modified Surfaces, H.A. Mottola and J.R. Steinmetz eds., Elsevier, Amsterdam, The Netherlands, 1992, 1. 0
Q
.
0
0
E.P. Plueddemann, Silane Coupling Agents, Plenum Press, New York, 1991. R.M. Smith, Gas and Liquid Chromatography in Analytical Chemistry, John Wiley & Sons, 1988. K.B. Sentell, D.M. Bliesner and S.T. Shearer, in Chemically Modified Surfaces, J.J. Pesek and I.E. Leigh eds., Royal Society of Chemistry, Cambridge, UK, 1994, 190. T.S. Den, H.G. Kicinski and A.A. Kattrup, in Chemically Modified Surfaces in Science and Industry, D.E. Leyden and W.T. Collins eds., Gordon and Breach Science Publ., New York, 1988, 621. K. Albert, J. Schmid, B. Pfleiderer and E. Bayer, in Chemically Modified Surfaces, H.A. Mottola and J.R. Steinmetz eds., Elsevier, Amsterdam, The Netherlands, 1992, 105.
188 7. 0
M.J. Wirth and H.O. Fatunmbi, Anal. Chem., 1993, 65, 822. M.J. Wirth and H.O. Fatunmbi, in Chemically Modified Surfaces, J.J. Pesek and I.E. Leigh eds., Royal Society of Chemistry, Cambridge, UK, 1994, 203.
9.
B. Porsch and J. Kratka, J. Chromatogr., 1991, 543, 1.
10.
R.S. Shreedhara Murthy and J.P. Berry, in Chemically Modified Oxide Surfaces, D.E. Leyden and W.T. Collins eds., Gordon and Breach Science Publ., New York, 1990, 151.
11.
M.R. Schure, in Chemically Modified Surfaces, J.J. Pesek and I.E. Leigh eds., Royal Society of Chemistry, Cambridge, UK, 1994, 181.
12.
R.K. Gilpin and M.E. Gangoda, J. Chromatogr. Sci., 1990, 28, 277.
13.
H. Engelhardt, H. Low, W. Beck and W. Gotzinger, in Chemically Modified Surfaces, H.A. Mottola and J.R. Steinmetz eds., Elsevier, Amsterdam, the Netherlands, 1992, 225.
14.
M. Petro and D. Berek, Chromatographia, 1993, 37, 549.
15.
K.K. Unger, Porous Silica its properties and use as support in column liquid chromatography, Elsevier, Amsterdam, The Netherlands, 1979.
16.
W.W. Yau, J.J. Kirkland and D.D. Bly, Modem size-exclusion liquid chromatography. Practice of gel permeation and gel filtration chromatography., John Wiley, Chichester, 1979.
17.
H. Engelhardt and D. Mathes, J. Chromatogr., 1975, 108, 85.
18.
B. Porsch, J. Chromatogr. A, 1993, 653, 1.
19.
M.J.J. Hetem, J.W. De Haan, L.J.M. Van De Ven, J.A. Rijks and C.A. Cramers, in Chemically Modified Oxide Surfaces, D.E. Leyden and W.T. Collins eds., Gordon and Breach Science Publ., New York, 1990, 387.
20.
D.E. Leyden and G.H. Luttrell, Anal. Chem., 1975, 47, 1612.
21.
D.E. Leyden, G.H. Luttrell and G.E. Sloan, N.J. DeAngelis, Anal. Chim. Acta, 1976, 84, 97.
22.
I. Taylor and A.G. Howard, Anal. Chim. Acta, 1993, 271, 77.
23.
R.J. Kvitek, J.F. Evans and P.W. Carr, Anal. Chim. Acta, 1982, 144, 93.
24.
G.V. Kudryavtsev, D.V. Miltchenko, V.V. Yagov and A.A. Lopatkin, J. Colloid Interface Sci., 1990, 140, 114.
189
25.
A.M. Klonkowski and C.W. Schlaepfer, J. Non-Cryst. Sol., 1991, 129, 101.
26.
T. Seshadri and H.-J. Haupt, in Chemically Modified Surfaces in Science and Industry, D.E. Leyden and W.T. Collins eds., Gordon and Breach Science Publ., New York, 1988, 531.
27.
C. Kantipuly and S. Katragadda, A. Chow, H.D. Gesser, Talanta, 1990, 37, 491.
28.
J.R. Jezorek and H. Freiser, Anal. Chem., 1979, 51,366.
29.
L.L. Murrel, in Advanced Materials in Catalysis, J.J. Burton and F.L. Garten eds., Academic Press, New York, 1977, 235.
30.
M.T. Hoffmann, S.M.C. Neiva, M.R. Martins and D.W. Franco, in Chemically Modified Surfaces, H.A. Mottola and J.R. Steinmetz eds., Elsevier, Amsterdam, the Netherlands, 1992, 257.
31.
M. Kamada, H. Nishijama and Y. Kera, Bull. Chem. Soc. Japan, 1993, 66, 3565.
32.
E. Carlier, A. Guyot, A. Revillon, M.-F. Llauro-Darricades and R. Petiaud, React. Polym., 1991/1992, 16, 41.
33.
E. Carlier, A. Guyot and A. Revillon, React. Polym., 1991/1992, 16, 115.
34.
E. Carlier, A. Revillon, A. Guyot and P. Baumgartner, React. Polym., 1993, 21, 15.
35.
T.J. Pinnavaia, J. G-S. Lee and M. Abedini, in Silylated Surfaces, D.E. Leyden and W.T. Collins eds., Gordon and Breach, London, 1980, 333.
36.
P. Tundo and P. Vanturello, J. Am. Ceram. Soc., 1979, 101, 6606.
37.
L. Gianfreda and M.R. Scarfi, Mol. Cell. Biol., 1991, 100, 97.
38.
W.H. Scouten, R. Rines, C. Elhardt, J. Putker, E. de Vries, P. Jones, E. Edsmalm and B. Lindner, in Chemically Modified Surfaces in Science and Industry, D.E. Leyden and W.T. Collins eds., Gordon and Breach Science Publ., New York, 1988, 35.
39.
H.H. Weetall, in Immobilized Enzymes for Industrial Reactors, R.A. Messing ed., Academic Press, New York, 1975, 99.
40.
E. Vandenberg, H. Elwing, A. Askendal and I. Lundstr6m, J. Colloid Interface Sci., 1991, 143, 327.
41.
K.M.R. Kallury, W.E. Lee and M. Thompson, Anal. Chem., 1992, 64, 1062.
42.
K.M.R. Kallury, W.E. Lee and M. Thompson, Anal. Chem., 1993, 65, 2459.
190 43.
K.M.R. Kallury and M. Thompson, in Chemically Modified Surfaces, J.J. Pesek and I.E. Leigh eds., the Royal Society of Chemistry, Cambridge, UK, 1994, 58.
44.
J.M.S. Cabral and J.F. Kennedy, in Protein Immobilization: Fundamentals and Applications, R.F. Taylor ed., Marcel Dekker, New York, 1991, 73.
45.
A.M. Klibanov, Science, 1983, 219, 722.
46.
G. Gfibitz, E. Kfinssberg, P. Van Zoonen, H. Janssen, C. Gooijer, N.H. Velthorst and R.W. Frei, in Chemically Modified Surfaces in Science and Industry, D.E. Leyden and W.T. Collins eds., Gordon and Breach Science Publ., New York, 1988, 129.
47.
P. Mohr and K. Pommering, Affinity Chromatography: practical and theoretical aspects, Marcel Dekker, New York, 1985.
48.
W.H. Scouten, Affinity Chromatography, John Wiley & Sons, New York, 1981.
49.
P. Creatrecasas and C.B. Anfinsen, Ann. Rev. Biochem., 1971, 40, 259.
50.
D. Mislovicova, I. Novak and M. Pasteka, J. Chromatogr., 1991, 543, 9.
51.
K.K. Chawla, Composite Materials, Springer-Verlag, New York, 1987.
52.
E.P. Plueddemann, in Silanes and other Coupling Agents, K.L. Mittal ed., VSP, Utrecht, The Netherlands, 1992, p.3.
53.
F.M. Fowkes, J. Adhesion Sci. Technol., 1987, 1, 7.
54.
D.W. Dwight, F.M. Fowkes, D.A. Cole, M.J. Kulp, P.J. Sabat, L. Salvati Jr. and T.C. Huang, J. Adhesion Sci. Technol., 1990, 4, 619.
55.
U. Deschler, P. Kleinschmitt and P. Panster, Angew. Chemie, 1986, 98, 237.
56.
E. Vanderheyden, PhD thesis, University of Antwerp, Antwerp, 1989.
57.
J.H. Helbert and N. Saha, in Silanes and Other Coupling Agents, K.L. Mittal ed., VSP, Utrecht, The Netherlands, 1992, p.439.
58.
P. Clechet and N. Jaffrezic-Renault, Adv. Mater., 1990, 2, 293.
59.
E.T. Vandenberg, L. Bertilsson, B. Liedberg, K. Uvdal, R. Erlandsson, H. Elwing and I. Lundstr6m, J. Colloid Interface Sci., 1991, 147, 103.
60.
J. Griffiths, Ind. Min., 1987, 11, 23.
61.
D.A. White, S.M. Oleff, R.D. Boyer, P.A. Budinger and J.R. Fox, Adv. Ceram. Mater., 1987, 2, 45.
191 62.
H. Zhang and C.G. Pantano, J. Am. Ceram. Soc., 1990, 73, 958.
63.
P. Marage, M. Langlet and J.C. Joubert, Thin Solid Films, 1994, 238, 218.
64.
B. Arkles, Chemtech, 1977, 7, 766.
65.
N. Nishijama, R. Shick and H. Ishida, J. Colloid Interface Sci., 1991, 143, 146.
66.
K.L. Mittal ed., Silanes and Other Coupling Agents, VSP, Utrecht, The Netherlands, 1992, 119-198.
67.
S.W. Morrall, PhD dissertation, Fort Collins, Colorado, 1984.
68.
D.J. Tutas, R. Stromberg and E. Passaglia, SPE Trans., 1964, 4, 256.
69.
J.P. Blitz, R.S.S. Murthy and D.E. Leyden, J. Colloid Interface Sci., 1988, 121, 63.
70.
J.P. Blitz, R.S.S. Murthy and D.E. Leyden, J. Am. Chem. Soc., 1987, 109, 7141.
71.
J.P. Blitz, R.S.S. Murthy and D.E. Leyden, J. Colloid Interface Sci., 1988, 126, 387.
72.
K.C. Vrancken, P. Van Der Voort, K. Possemiers, P. Grobet and E.F. Vansant, in Chemically Modified Surfaces, J.J. Pesek and I.E. Leigh eds., The Royal Society of Chemistry, Cambridge, UK, 1994, 46.
73.
K.M.R. Kallury, P.M. MacDonald and M. Thompson, Langmuir, 1994, 10, 492.
74.
R. Maoz and J. Sagiv, J. Colloid Interface Sci., 1984, 100, 465.
75.
D.A. Offord and J.H. Griffin, Langmuir, 1993, 9, 3015.
76.
J.D. Le Grange, J.L. Markham and C.R. Kurkjian, I.angmuir, 1993, 9, 1746.
77.
S.R. Wasserman, G.M. Whitesides, I.M. Tidswell, B.M. Ocko, P.S. Pershan and J.D. Axe, J. Am. Chem. Soc., 1989, 111, 5852.
78.
J.J. Pesek, J.E. Sandoval, C.-H. Chu and E. Jonsson, in Chemically Modified Surfaces, H.A. Mottola and J.R. Steinmetz eds., Elsevier, Amsterdam, The Netherlands, 1992, 57.
79.
J.J. Pesek, in Chemically Modified Surfaces, J.J. Pesek and I.E. Leigh eds., Royal Society of Chemistry, Cambridge, UK, 1994, 1.
80.
C.-H. Chu, E. Jonsson, M. Auvinen, J.J. Pesek and J.E. Sandoval, Anal. Chem., 1993, 65, 808.
81.
J.E. Sandoval and J.J. Pesek, Anal. Chem., 1989, 61, 2067.
192 82.
K.G. Proctor and J.P. Blitz, in Chemically Modified Surfaces, H.A. Mottola and J.R. Steinmetz eds., Elsevier Science Publ., Amsterdam, The Netherlands, 1992, 209.
83.
J.B. Danner and J.M. Vohs, Appl. Surf. Sci., 1992, 62, 255.
84.
D.W. Sindorf and G.E. Maciel, J. Phys. Chem., 1982, 86, 5208.
85.
D.G. Kurth and T. Bein, J. Phys. Chem., 1992, 96, 6707.
86.
V.A. Basyuk, React. Kinet. Catal. Lett., 1993, 50, 83.
193
Chapter 9 Modification with silicon compounds: mechanistic studies
The organosilane modification of silica is of great importance in science and industry. Because of its widespread applications, there is a lot of interest in the underlying reaction mechanisms in the modification process. A fundamental understanding of these mechanisms can lead to an improvement of known applications and the development of new ones. Thus, fundamental knowledge supports industrial development. Mechanistic studies are performed on model systems with low complexity. Therefore organic solvent and vapour phase modification procedures are mostly used. Thus, the concentration of the reactants and the presence of water in the system may be easily controlled. From a careful description of the simple model systems, extrapolation towards large-scale industrial processes may be performed. In this part, we wish to focus on the study of two types of silanes. Aminoorganosilanes are special members of the alkoxysilanes group. They carry the catalyzing amine function, required for chemical bonding with the silica surface, inside the molecule. This makes them more reactive than other organosilanes and reduces the complexity of the liquid phase reaction system to be studied. Only three components, silica, silane and solvent, are present. Furthermore there is a large interest in the reaction mechanism of silica gel with APTS, since this aminosilane is the most widely used compound of the organosilane family. Chlorosilanes are discussed, considering both those with short-chain and long-chain organic groups. Short-chain chlorosilanes are important as model compounds in the general study of silica surface reactivity in vapour phase modification. Furthermore,
194 these compounds are very valid in the Chemical Surface Coating method, which will be discussed in part 3. The long-chain (Cs, C~s) organosilanes are very important in the fabrication of stationary phases. For these, special liquid phase modification procedures have been developed. The underlying mechanism deserves special attention. Finally, the modification with alkoxysilanes and alkylsiloxanes is briefly discussed.
1 Modification with aminosilanes Aminoorganofunctional alkoxysilanes (named as aminosilanes for clarity) carry one or more amine groups in the organic chain. This amine group is responsible for the specific chemical reaction behaviour and high reactivity of the aminosilane molecules. The electron rich amine centre can enter into hydrogen bonding interactions with hydrogen donating groups, such as hydroxyls or other amines. Because of its basic character (pI~ = 10.8 for a terminal primary amine), the amine is easily protonated. The three most widely used and studied aminosilanes are: ~ ~-aminopropyltriethoxysilane (APTS, (CH3CH20)3SiCH2CH2CH2NH2) ,
~,-aminopropyldiethoxymethylsilane (APDMS, (CH3CH20)2CH3SiCH2CH2CH2NH2) and N-~-aminoethyl-q~-aminopropyltrimethoxysilane, (AEAPTS ((CH30)3SiCH2CH2CH2NHCH2CH2NH2). Mixing of an aminosilane with silica gel results in a fast adsorption, by hydrogen bonding of the amine to a surface hydroxyl group. 2 After adsorption, the amine group can catalyze the condensation of the silicon side of the molecule with a surface silanol. Thus siloxane bonds with the surface may be formed in the absence of water. 3'4 For other silanes the siloxane bond formation requires an initial hydrolysis of the ethoxy groups or the addition of an amine in the reaction mixture. ~ This general reaction scheme has been presented in chapter 8. Here we will go into further detail on the types of interaction of the aminosilane with the silica surface and the characterization of the bonded silane species.
195 For the modification of silica with aminosilanes, the liquid phase procedure is usually applied. Only few studies have described the vapour phase APTS modification. 6'7 The modification proceeds in three steps. (i) A thermal pretreatment of the silica determines the degree of hydration and hydroxylation of the surface. (ii) In the loading step, the pretreated substrate is stirred with the silane in the appropriate solvent. (iii) Curing of the coating is accomplished in a thermal treatment. On industrial scale ethanol/water is used as a solvent, on lab-scale an organic solvent is used. The reasons for this discrepancy is the increased control on the reaction processes, possible in an organic solvent. This will be clarified by the discussion of the modification mechanism in aqueous solvent and the effect of water in the different modification steps. A correlation of data from literature is often hampered by the use of different reaction parameters or silica types. Therefore, a clear survey of the effect of parameters related to reaction conditions and substrate structure is given. Furthermore, a full description of the modification mechanism is only possible if the processes occurring in the loading step and the curing step are discussed separately. The study of each of these steps requires dedicated analytical procedures. For the study of the loading step, most analyses are performed on the silane/solvent mixture, while spectroscopic analyses are performed on the modified substrate after drying and curing. The amine side of the molecule requires special attention. The orientation of the aminosilane molecule on the silica surface is determined by the amine-surface interactions. This orientation is of great interest for any further application of the aminosilane modified silica substrate. In order to complete the full characterization of the aminosilane modification of silica, the specific role of silanols, both on the surface and on the silane molecule, will be regarded in detail. Silanol groups are the main reactive centers for chemical bonding, whatever the reaction conditions may be.
196
1.1 Modification of silica gel with aminosilanesfrom aqueous solution On industrial scale, water or water/ethanol mixtures are mostly used as a solvent for aminosilane modification of silica or glass fibres. Water has a profound effect on the silane structure in solution. Alkoxy groups are hydrolyzed to silanols. Those can combine to form a siloxane linkage between two silane molecules, with production of a new water molecule (reaction (A)). 2 -Si-OEt
+ 2 HzO--, 2 --Si-OH + 2 EtOH
- Si-OH + HO-Si-
~
- Si-O-Si-
+ H20
(A)
This reaction is generally applicable for organosilanes. However, APTS solutions are more stable towards polymerization than non-aminated silanes. This effect and the consequences of the APTS solute structure on the eventual modified layer structure will be discussed in this paragraph. 1.1.1 The APTS structure in aqueous solution Aqueous solutions of APTS have a clearly enhanced stability towards polymerization, compared to other, non-aminated, organosilanes. Based on his own work and on the FTIR study of Chiang et al., g Plueddemann 9 proposed the formation of an intramolecular seven membered ring in order to explain this special behaviour. This hypothesis was further supported by results of thermometric enthalpy titrations, performed by Kelly and Leyden. ~~ The ring conformation is given in figure 9.1. The bonding of a silanol hydrogen explains the stability against cross-linking. With a change of pH of the solution, acid-base equilibria govern the aminosilane conformation. This was studied by Chu et al., ~ using ~SN NMR on aqueous solutions of ~SN enriched APTS. NMR spectra at different pH are given in figure 9.2.
The nitrogen chemical shifts are -1.2, 2.8 and 6.8 ppm respectively. In 'neat' APTS, a resonance was found at-4.8 ppm. Both this peak and the -1.2 peak found in strong base (figure 9.2 a), are assigned to the free amine form. In acid (figure 9.2 c), the amine is in the protonated or 'open' amine cation form. The closed 7-membered ring,
197
..
"kc H
I-I
H
o
si /
n N
2.02 A
H H
O~
H/"2.45 ~
Figure 9.1 7-membered ring conformation of aminopropylsilanetriol, taken from ref (9) with permission. "~- NH 2
b C pH
13), (b) ambient (pH 10- 11), (c) acid (pH < 1), taken from ref (11) with permission. as proposed by Plueddemann, is expected to resonate between both of these forms. The position of the resonance is expected to reflect the state of hydrogen bonding. The closer the proton is to the amine, the more downfield the resonance. The resonance found at 2.8 ppm (figure 9.2 b) is assigned to an APTS having considerable cation character. Therefore both the closed hydrogen bonded form and an internal zwitterion structure may be supported. In the internal zwitterion a proton is transferred from a silane silanol to the amine group.
198 -O(OH)2Si-CH2CH2CH2NH3 +
Concluding, we may state that in aqueous solutions of APTS the amine is involved in acid-base interactions. The conformation of the molecule changes with changing pH of the medium. At natural pH (10.8) the hydrolyzed aminosilane forms a 6- or 7 - m e m b e r e d ring structure, stabilized by internal hydrogen bonding or protonation.
1.1.2 Modification from aqueous solvent For silica modification with APTS, the pH of the aqueous solution also has an effect on the eventual loading of the surface. ~2 The APTS surface loading on dehydrated silica, is given as a function of pH in figure 9.3. The loading was measured by integration of the CH2 vibration bands, after internal normalization.
40
xlO 17 Mole~afles/ nag silica
30-
20-
10,,
0
I
2
4
I
I
6
8
I
10
12
pH
Figure 9.3 APTS loading on dehydrated silica as a function of pH of the treating solution. taken from ref (12) with permission.
199 The surface loading appears to be highly dependent on the structure of the amine group. Maximal adsorption is found at a pH 10.6, which is the natural pH of the APTS solution. The closed APTS forms appear to show better adsorption to the silica surface than both, free and cation, open forms. The silica iso-electric point is about 2. The SiO- concentration increases with increasing pH. At low pH, the tendency of the -NH3+C1- to interact with the SiOH is low. At higher pH the concentration of surface SiO- increases, allowing a better interaction with the aminosilane. At high pH, some of the APTS silanol groups are in the form of SiO- and have the tendency to repel the negatively charged surface. Immobilization of the aminosilane molecule changes its interaction characteristics. Because the surface silanols are more acidic than silane silanols, the interaction with the surface silanols is thermodynamically favoured over intramolecular interaction. Kelly and Leyden ~~measured the enthalpy of adsorption of the aminosilane molecules. Their results indicate that interaction with the surface involves more proton transfer than in the closed form dissolved molecules. The ultimate structure of the aminosilane layer has been characterized using spectroscopic methods. Initial monolayer formation was indicated by Pantano and Wittberg, using X P S . 13 Upon modification with dilute aqueous solution of various silanes, the thickness of the silane layer was shown to be dependent on the length of the organic functional group. Insight in the coating structure was obtained from a solid-state NMR study on silica modified with deuterated APTS, by Kang and Blum. ~4 The aminosilane molecules are adsorbed and reacted directly on the silica surface, by siloxane bond formation, until a monolayer coverage is obtained. Then a polysilsesquioxane layer is formed on top of the monolayer. The absorbed amount increases with increasing initial concentration of the silane solution. The polymer layer is intermolecularly condensed and indirectly bonded to the surface. The following may be concluded. The reaction process for modification in aqueous solvent is affected by the pH of the solution. Maximal loading is obtained at natural pH. The modification layer is composed of a chemically bonded monolayer and an overlying physically adsorbed polymer network.
200
1.2
The influence of water in the reaction of silica with APTS
The modification in aqueous solvent causes uncontrollable hydrolysis and polymerization of the aminosilane molecules. Irreproducible coatings result. In order to control this parameter, modification procedures in dry organic solvents have been set up. The use of organic solvent allows a better control of reaction parameters and is therefore preferred in the study of the modification mechanism. Additionally, it allows the controlled adding of water in the reaction mixture. Thus, the exact role of water in the various modification steps may be assessed. The knowledge of the influence of water in the reaction mechanism forms the bridge from aqueous to organic phase modification. Many surface analysis techniques have been applied in studies on the reaction mechanism of APTS with silica gel. ~5 In order to study the chemical structure and interactions of the surface compound, FTIR has been recognized as a very powerful tooI.2,s,16,17, is
The infrared spectrum provides information on the presence and physical interactions of functional groups. Complementary information on the chemical environment and coordination of the Si and C atoms in the coating can be obtained from CP MAS N M R . 4'19'20'21'22'23'24 Thus a combination of both techniques allows a clear description of the physical and chemical interactions and coordination of silanes on silica. Both these techniques have been applied to study the influence of water in the three modification steps. 3 Every step will be discussed separately.
1.2.1 Reaction on hydrated silica gel Mesoporous silica gel was reacted with APTS (1%v/v APTS/solvent mixture), using various reaction and curing conditions. Sample-specific reaction conditions are indicated in table 9.1.
201 Table 9.1 Sample preparing conditions and integrated values of the siloxane band (1350-870 cm-1), ratioed to the reference band (1938-1766 cm-~)
sample
pretreatment
solvent
1 2 3 4
none 673 K air 673 K air 673 K air
C7H8 H20 1%H20/CTH 8 C7H8
curing conditions (20 h) 383 383 383 383
K air K air K air K air
SiOSi peak area 102.9 55.30 46.94 33.78
Hydrated silica gel was modified with APTS (sample 1) and studied by DRIFT (Diffuse Reflectance Infrared Fourier Transform) and CP MAS NMR. The IR spectrum of the modified silica (figure 9.4) shows silane NH, CH and Si-O-Si bands along with silica lattice and surface vibrations. Assignments of IR bands of APTS modified silica are given in table 9.2. Table 9.2 IR vibration bands: frequencies and band assignments 3'16'19
Wavenumber (cm1)
Assignment
3740 3740-3500 3373 3310 2978 2936 2869 1940-1770 1597 1471 1448 1413 1393 1250-1020 800
free Si-OH vibration (stretching) bridged Si-OH vibration (stretching) NH vibration (stretching) NH vibration (stretching) CH3 vibration (stretching) CH2 vibration (stretching) CH2 vibration (stretching) Si-O-Si vibration NH2 bending CH2 bending CH3bending Si-CH2 bending CH3 bending vibration Si-O-Si asymmetric stretching vibration OH bending
202
Kubelka - Munk units
f
0.8 -
0.6-
0.4
0.2
4000
'
3(~0
'
2000
'
10~
500
Wavenumber/cm"1 Figure 9. 4. DRIFT spectrum of hydrated silica, reacted with APTS in toluene solution, dried in air.
The influence of water can be investigated in studying the CH3 and Si-O-Si band areas, because hydrolysis and subsequent polymerization will decrease the former and increase the latter. In practice the asymmetrical CH 3 stretching (p = 2978 cm -~ ) and the Si-O-Si stretching (1350 - 870 cm ~) bands are most suited. The position of the NH 2 vibrational modes indicates a perturbation of the NH 2 group. Hydrogen bonding shifts the maxima of the stretching vibration bands to below 3370 and 3310 cm ~, respectively. The absence of the CH3 band and the magnitude of Si-O-Si band together with the position of the NH vibration modes (3367, 3300 cm ~) indicate the formation of a polymerized aminopropylsiloxane coating with NH2 groups interacting with Si-OH groups by means of hydrogen bonding (figure 9.5, structure I). In analogy with this, Kallury et al. 4 reported that surface water promotes amine-surface interactions and orients the amino moieties towards the surface. These conclusions were based on XPS results. Further details of the amine-surface interaction in loading and curing step will be given in the subsequent chapters.
203
* category
(i)
surface water, water solvent
~////////////////////////////~ = silica
R' O\Si/,R' /\ S i \/ / O
O O
surface
R,. H
* category (ii) Modificationof dry silica gel a) air curing H/ H\ CH CH O
o
\ Si/
/
\
(CI'I2) 3 - N ~ H
o
CH3 CH 2%
O o"
It' \ Si
R' Si/
/R'
O
c. c.:o
H
\ St'"
R'
b) vacuumcuring
Si O /
H
/
"O-Si-
\O
IV
Figure 9.5 Bonding models of APTS on silica gel.
These observations are further confirmed by the 29Si and ~3C CP MAS NMR spectra of the modified compound (figure 9.6 a, b). Band assignments of the 29Si spectra of APTS modified silica have been fully elaborated by De Haan et al. 2~ Comparing the shielding effects of ethoxy and hydroxy substituents, they proposed that hydrolyzed and non-hydrolyzed monodentate species have a different peak position. Thus, the following assignment was effectuated: R O - S i/\ Si / \
6
o (a) -67ppm
R
\Si / \ o
OR'
(b) -59ppm
R
R
EtO - S iI - OEt
HO - _~;i - OH
6
(c) -52.gppm
6
(d) -49.2ppm
R'-H or Et
In the spectrum of the modified hydrated silica (figure 9.6 a), the peaks corresponding to tridentate (a) and bidentate (b) species are clearly present. Peaks in the -100 ppm region are due to surface silicon atoms, with peaks at -100 ppm and -109 ppm for Q3 and Q4 sites, respectively (cfr. table 5.3). The peak corresponding to the monodentate (c) form is negligibly small, while the hydrolyzed monodentate (d) form is absent.
204
I
!
.
50
1
40
.
!
! . . . . . .
i
30
20
. . . .
1 .
.
10
.
.
L
=
0
6
I I
(a)
--4'0
'-
6o-i
,o"i,i0
....
Figure 9.6 Hydrated silica, modified in toluene solvent, air cured. (a) 29Si CP MAS NMR spectrum; (b) I~C CP MAS NMR spectrum.
The 29Si spectrum (figure 9.6 a) indicates the coating to be hydrolyzed and crosslinked. In order to conceive the contributions of the various species to every band, the spectra are deconvoluted. After deconvolution of the recorded spectrum into a sum of Gaussian curves, the different contributing signals can be integrated separately. For quantitative assessment some contact time and cross-polarization criteria have to be met. Optimal conditions are obtained from a variable contact time experiment. Full details have been worked out by Pfleiderer 2s and Caravajal et al. 2~a2 In general, quantitative correlations should only be made for species within one spectrum and not
205 between various spectra. Thus, the relative population of the different surface structures present can be determined. From the integration data a 40/60 ratio of the APTS molecules in the bidentate/tridentate form can be calculated. The presence of the monodentate form can be neglected in comparison to the amount of the other two forms. Because three covalent bonds with the surface can be excluded on steric grounds, the tridentate form implies polymerization of the APTS molecules. The ~3C spectrum (fig. 9.6 b) indicates a total loss of ethoxy groups as no peaks are observed at 57.6 ppm (-OCH2-) and 15.9 ppm (-OCHz_C_.CH3). The three observed peaks have been assigned ~9to the propyl carbons ( ctCH=9.5 ppm,/3CH=26.3 ppm, qtCH =43.3 ppm). The ~3C shift value of the ctCH is an indication of the stability of the aminosilane coating on the silica surface. Caravaja122 has shown that an increase in the APTS coating stability shifts the aCH peak to higher shift value (APTS solution ctCH=7.9 ppm). The observed peak position of the ctCH (9.5 ppm) indicates the covalent attachment of the silane molecules to the silica surface. Surface water can cause hydrolysis and polymerisation of the APTS coating, leading to a coating with 60% of the molecules having a structure of type I (figure 9.5). The residual 40% occur as the type II structure in the hydrolyzed form.
1.2.2
Water in the solvent-silica system
Various experiments were performed on dehydrated silica gel, using different amounts of water in the reaction phase. Polymer formation is evaluated by integrating the SiO-Si stretching band. The results are summarized in table 9.1. By rationing the integrated value to the internal reference band (1938 - 1766 cml) 16 variations in the silica surface (Si-O-Si) contribution to the different samples are ruled out. Integrated values show a clear increase with increasing water content of the solvent. It is therefore obvious that the amount of water is the determining factor in the polymerization process. Additionally, the DRIFT spectrum of the sample reacted in toluene solvent (sample 4) shows residual ethoxy groups, whereas the aqueous samples (samples 2, 3) do not have the CH 3 band. Comparing the Si-O-Si peak area of the dehydrated, aqueous modified sample (2) to the hydrated sample reacted in toluene
206 (sample 1) shows an increase in the amount of polymerized silane molecules on the silica surface for the latter sample. Comparable results have been obtained by MorraU 2 and Caravajal et al. 22 A higher APTS loading was found on a hydrated silica reacted in toluene solvent, compared to modification in aqueous solvent. Therefore, it can be concluded that the polymerization takes place at the silica surface, i.e. after adsorption of the aminosilane molecules. The surface effect can be explained by the interaction of the silane NH2 group with the substrate surface. As shown above, in water solvent the hydrolyzed aminosilane molecules are stabilized by internal hydrogen bonding of the amino group to the silane hydroxyls. When the amino group is H-bonded to a surface hydroxyl group this stabilization disappears and the silane silanols can condense to form a siloxane linkage. When the reaction is performed with hydrated silica in a dry solvent (sample 1), the hydrolysis only takes place at the silica surface and can immediately be followed by the condensation reaction. In both cases, structures of type I are formed. Reaction in a dry solvent with dehydrated silica (sample 4) minimizes the hydrolysis and polymerization in the reaction stage. Bonding takes place by direct condensation of the silane silicon with a silica surface hydroxyl, with loss of ethanol.
1.2.3 Air humidity in the curing stage In order to study the influence of air humidity in the curing stage, a post reaction curing was performed in air at 383 K (sample 4) and in vacuum at 423 K (sample 5). In both samples dehydrated silica was reacted in a dry solvent, minimizing the influence of hydrolysis in the first two modification stages. DRIFT spectra of both samples (figure 9.7) show a higher ethoxy content for the vacuum cured sample. Increasing the curing temperature has been reported to cause a decrease in the ethoxy content. 22 Therefore, a higher degree of covalent bonding to the silica surface, i.e. a lower ethoxy content, is expected for this sample. The observed loss of ethoxy groups can be attributed to the influence of air humidity in the curing stage, inducing hydrolysis, possibly followed by oligomerization, even for samples reacted in dry conditions. The polymerization, however, is reduced, comparing to the reaction in the presence of water. The existence of silane silanols,
207
3050
2920
2 7 9J 0
waven u mber/cm - 1
Figure 9. 7 DRIFT spectra of APTS modified silica gel: (a) air cured," (b) vacuum cured. oligomers and residual ethoxy groups can be concluded (figure 9.5, structure II, III). Only when the reaction and the curing stage are performed without any water present (sample 5) a hydrolysis can be prevented, leading to a type IV surface structure. These results are again confirmed by CP MAS NMR measurements. Vacuum pretreated vacuum cured samples were prepared, with minimal water contact. The influence of air humidity in the curing stage was tested by precuring one sample in air for 2 h (sample 7). ~3C NMR spectra are presented in figure 9.8. Ethoxy peaks are visible at 57.6 ppm (-OCH2-) and at 15.8 ppm (-OCH2..C_.CH3). The ethoxy content of the air precured sample (figure 5A) is clearly reduced, compared to the non-precured sample (figure 9.8 b). The z9si spectra, however, show no clear difference in relative species population. Both samples have 60% of the surface species in the bidentate form, with 20% both in mono- and tridentate form. The bidentate species however covers molecules with R' =H as well as R' = C H 2 C H 3. A decrease in the ethoxycontent with constant population of the bidentate species indicates the conversion of the-OCH2CH 3 into a - O H group, without further condensation. This conversion should be attributed to the influence of the air
208
I
I
I
I
) I
--
|
. . . . . . .
70
,
1
.
.
60
.
.
.
.
I
1 . . . . . . . .
50
!
. . . . . . . .
40
i
30
. . . . .
|
20
._
i
. . . . . . . . .
10
i
,_
0
6 Figure 9,8 13CCP MAS NMR spectra of dry treated APTS-modified silica: (a) 2h, 383K air precured; (b) 423K vacuum cured.
humidity. The coating in sample 7 mainly has the type II structure. The stability of the OH group is caused by a H-bond with the NH2 group of a neighbouring silane molecule or by the isolated character of these groups at the surface. Air humidity therefore, appears to have a clear influence on the hydrolysis of surface bonded APTS molecules.
209 1.2.4 Conclusion The effect of surface water and air humidity on the hydrolysis of APTS molecules adsorbed on the silica surface may be characterized as follows. Short time exposures to humid air cause partial hydrolysis of the modified layer. Extensive hydrolysis is only caused by surface adsorbed water. Hydrolyzed aminosilane molecules at the surface condense to form an aminopropylpolysiloxane layer. Only when all three modification stages (pretreating, loading, curing) are performed in completely dry conditions, hydrolysis of ethoxy groups can be prevented. The structures formed under the various modification conditions are summarized in figure 9.5.
1.3 Modelling of the interactions in the loading step In the broad range of literature on the organosilane modification of silica gel, most studies are based on spectroscopic analysis of the modified substrate. This requires a drying or curing step before analysis. This curing and drying may cause desorption and/or condensation of adsorbed molecules, thus changing the modification structure. Therefore, no data on the adsorption mechanism or the processes occurring in the loading step can be obtained from this type of studies. Still these processes are of interest because they control the ultimate structure of the modified layer. In literature distinction between loading and curing processes is not always made clearly. In order to clarify this interchange of information, we will discuss the mechanisms of loading and curing step in separate paragraphs. The bifunctionality of the aminosilane molecules is responsible for their specific chemical behaviour. The interaction of the amine group, and its role in the adsorption mechanism was already indicated in the previous paragraph. The different types of interaction of the amino group and its role in the modification mechanism will be evaluated in a separate paragraph. Here we consider the general physical and chemical adsorption of the silane molecule. Discussions will mainly be concerned with the silicon side of the molecule. In the few studies reporting on loading processes, specific analysis procedures have been developed in order to evaluate the course of the aminosilane deposition. Those include various in situ FTIR techniques as well as solvent analysis by various
210
analytical or spectrometric methods. These allow the measuring of adsorption kinetics and the building of adsorption isotherms. The correlation of data from literature is often hampered by the different reaction conditions and types of silica used in various studies. In order to overcome this problem, the effect of the porosity of the substrate on the reaction mechanism is studied. Also here a difference has to be made between the effect in the loading and curing step.
1.3.1 Adsorption kinetics and isotherms
Adsorption kinetics For the measuring of adsorption kinetics two criteria have to be met. The amount of compound adsorbed has to be quantified, and analyses have to be performed with short time intervals. Both these criteria are difficult to meet for reactions in solution. Various researchers have set up analysis procedures for this type of measurement. Morrall 2 used a HPLC system with two columns. The first column was loaded with the controlled pore glass (CPG) to be modified. The second column was used for separation of the reaction effluents. This column was coupled to a refractive index detector, allowing for quantitative detection of the effluents. The reaction was initiated by injecting an APTS/toluene mixture and stopped by injection of pure toluene. With this so-called stop-flow mechanism reaction times down to 18 seconds could be used. From these analyses it became evident that upon mixing of the aminosilane with the silica, a very rapid physisorption occurs. The initial adsorption of the APTS (from toluene solution on dried CPG) occurred before the 18 second minimum time delay of the stop-flow apparatus. For non-aminated silanes the adsorption proved to be much slower. This study also revealed the pivotal role of surface water in the modification of siliceous surfaces with alkoxysilanes, as discussed in the previous chapter. Vandenberg et al. 15used liquid/solid ellipsometry to obtain kinetic data. The thickness of the APTS coating on oxidized silicon wafers was measured, as a function of immersion time in APTS/toluene solution. The deposition profile showed a fast initial
211 adsorption, followed by a slower process. This was interpreted as an indication that the APTS first adsorbs non-covalently to the surface. After 30 min. reaction time, the deposition rate increases again. This additional adsorption was not further discussed. An analogous reaction profile is found using freeze sampling analysis. 26 In this technique, small samples are taken from the silane/toluene/silica gel reaction mixture, which are immediately frozen in liquid nitrogen to stop the reaction. The reaction mixture is stirred in order to secure a homogeneous mixing of the silica powder through the entire reaction volume. The samples are allowed to melt and are immediately centrifuged to separate substrate and solvent. The supernatant is than analyzed by classical Kjeldahl nitrogen analysis. From this analysis, surface loading is obtained as a function of reaction time. The surface loading (1) is generally expressed as amount adsorbed (mmol) per unit of mass (g) (or area (m2)) of the pure silica. The amount adsorbed is usually obtained from analysis of the residual amount of silane in the solvent at equilibrium (Cf). Various techniques have been applied to measure this quantity. Thus, surface loading 1 is given by" l=
(C i - C~
ms
(1)
With Ci the initially added amount of silane and ms the initial mass of the substrate. The reaction profiles of APTS and AEAPTS are displayed in figure 9.9. The reaction proceeds identically for both silanes in the first few minutes after mixing. At a reaction time of 6 min. a clear divergence occurs. Both compounds reach an equilibrium adsorption within one minute of reaction. This reflects the quick adsorption of the amine group, forming hydrogen bonds with the surface silanols. The lack of further deposition indicates that an equilibrium is reached. Because the reaction is performed in a dry solvent, on a dehydrated substrate, no oligomerization of the silane molecules can occur. It may be concluded, accordingly, that the equilibrium situation reflects the formation of a monolayer coating on the silica surface.
212
Loading (mmo1~g)
1.3
1.2
D 0
rl
[]
D 9
I.I []
D 9
1.0 0.9
1
10 100 Reaction time (m~n,)
1000
Figure 9.9 Silane loading on dried mesoporous silica gel as a function of reaction time," ( i ) APTS, (F-l) AEAPTS.
For the AEAPTS, carrying two amine groups, a two step adsorption is found. This indicates that an amine of the organic chain remains free in the first step (t < 6 min.) and is able to adsorb an additional layer of silane molecules. Secondary amines are better acceptors for hydrogen bonds than primary ones. The secondary amine will therefore enter in the initial hydrogen bonding with the surface silanols much more readily than the primary amine at the end of the organic chain. It will be the primary amine function, that remains free and is able to adsorb silane molecules from solution. Thus a secondary aminosilane layer is build up on the silica surface. Upon adsorption of the secondary layer, again an equilibrium situation is reached. For the monofunctional APTS, the observed deposition profile is identical to that reported by Vandenberg et al., using liquid/solid ellipsometry. ~5 The rapid, first equilibrium is followed by an additional adsorption. The equilibrium situation is related to the localized adsorption of silane molecules on the surface hydroxyls. Thus forming a monolayered coating on the surface. The extent of the additional adsorption was found to be not very reproducible and not strictly correlated to the initial silane concentration. At any time it showed not to have the step-wise profile, as found for the bifunctional aminosilane. Therefore, this last adsorption appears to be of a nonspecific type. We will discuss this further below.
213
Adsorption isotherms Further information on the adsorption process may be obtained from adsorption isotherm data. The construction of these isotherms for liquid phase processes differs from the well-known gas phase adsorption isotherms. For the liquid phase adsorption, the surface loading is plotted as a function of the equilibrium concentration of adsorbate (Cf). Adsorption isotherms for APTS and AEAPTS from acetone/water solvent on high surface area fumed silica were reported by Blum et al. 27 The aminosilane molecules show Langmuir type adsorption to the silica surface. Analysis data were recorded after shaking the silica with the treating solution, centrifugation and FTIR analysis of the supernatant. The absorbance values were determined by measuring the absorbance at 2877 cm -~ (methylene C-H stretch) in the difference spectra, after subtracting solvent contributions. With this method, the precision of a single adsorption experiment appeared to be very good, while reproducibility of the entire isotherm was poor. This was attributed to the wide variety of reactions undergone by the silanes, in this type of solvent. FTIR was also used by Trens. 7 Equilibrium concentrations were determined on the supernatant after centrifugation, by measuring the transmittance signal at 950 or 1104 cm -~. Isotherm data were recorded for modification of a precipitated silica gel (SBET = 25 m2/g) with APTS in aqueous as well as organic solvent. For the modification in aqueous solvent, a Langmuir type isotherm with a sharp rise at the origin, indicating the high affinity of the APTS for the silica surface, and a clear plateau was obtained. An analogous isotherm type was found for adsorption from organic solvent. A clear difference was observed for dehydrated and non-dehydrated silica. The plateau value decreases from 0.175 mmol/g to 0.057 mmol/g upon dehydration of the substrate. In the same work, isotherms for gas-phase adsorption of APTS are reported also. APTS loadings are lower than for the liquid phase adsorption. Relating data from different silicas and glass fibres, a clear correlation between water adsorption capacity of the substrate and silane affinity was found. Isotherm data at high equilibrium concentration for modification from organic solvent, have also been obtained by the freeze sampling method. The adsorption isotherm of APTS on dehydrated mesoporous silica from toluene solution is given in figure 9.10.
214
Loading (mmol]g) 2.0 1.6 1.2 9
9
9
m
9
9
0.8 0.4
0
2
,~
(i 8 1'0 Ceq (mmol)
1'2
1'4
16
Figure 9.10 Adsorption isotherm of APTS on dehydrated (673K) mesoporous silica gel from toluene solution, by freeze sampling method. The data points are obtained from adsorption values at equilibrium in the first 50 min. of reaction. At longer reaction times an additional adsorption was found, as reported above. The isotherm has two regions. For equilibrium concentrations up to 6 mmol, a plateau is observed, with a corresponding loading value of 1.02 mmol/g. This indicates the equilibrium formation of a monolayer coating of APTS on the silica surface, in the initial phase of the reaction. At higher concentration, the loading increases with increasing equilibrium concentration. At this excess of APTS over surface active sites, the equilibrium formation of a monolayer can not be pertained and the loading is proportional to the concentration of the aminosilane. Further insight in the structure of the coating, formed in both concentration regions, can be gained using two relative quantities, input ratio I, and degree of conversion X~. Both are defined as follows.
I-
NOn
(2)
Xa -
I
(3)
215 where NoH is the amount of OH groups (mmol) on the dehydrated silica surface (i.e. 1.44 mmol/g) and NAvrs the amount of APTS initially added to the reaction mixture. Modelling of the bonding type of the APTS molecules can be based on the relation between the values of I vs. X, for various concentrations of the silane used. In the case where all APTS forms a monolayer with one molecule of APTS occupying one surface silanol group over the whole range of concentrations, all experimental data would be situated on curve A (figure 9.11). A hypothetical monolayer having all APTS molecules coveting two hydroxyls is presented by curve B. It should be noted that curves A and B pass through the origin, because no intermolecular APTS reaction is assumed.
Xa
1.0 0.8 0.6 0.4 0.2 0.0
0.0
6.2
0'.4
I
6.6
6.8
i.0
i.2
1.4
Figure 9.11 Degree of conversion as a function of input ratio: ( ~ t ) monolayer modelling curves representing one (a) and two (b) hydroxy groups occupied by one APTS molecule; (- - - ) linear regression through experimental values for low and high concentrations.
216
Plotting of the experimental data for the low equilibrium concentration range (C~q= 0 - 6 mmol, i.e. I = 1.2 - 0.24) reveals that the data are linearly related. The linear regression curve passing through the data (X, = aI + b) has a slope a = 0.69 _+ 0.01, with a correlation r = 0.998. The good correlation between the data indicates that in this concentration range a similar type of coating is formed in the initial phase of the loading step, at all equilibrium concentrations. The data for the high concentration range (C~= 8 - 16 mmol, i.e. I = 0.18 - 0.09 ), can also be fitted with a linear curve. This curve, however, does not pass through the origin (figure 9.11), in contrast with both the modelling curves and the experimental curve at low concentration. Linear regression results in equation (4) : X,, = 0.73.1 + 0.04
(4)
with a correlation r = 0.991. The slopes of both curves through the experimental values (a~= 0.69 _ 0.01; a2 = 0.73 + 0.07) are equal within the margins of the respective standard errors. This indicates that for the high concentration range the first deposited layer is a monolayer, identical to the one formed at low concentrations. The positive intercept with the X~-axis (b) shows that at NoH = 0 (I = 0) still a conversion of 0.04 is reached, reflecting an intermolecular reaction of the APTS. Due to this intermolecular interaction, 4% of the initially added amount of APTS, is deposited on the first monolayer, leading to an equilibrium within the first minute of reaction. From these data the coordination of the APTS molecules with the surface can be evaluated. The inverse of the slope value gives the number of surface silanols bonded per APTS molecule, at initial monolayer coverage. A value of 1 . 4 0 H / A P T S is found. In the APTS monolayer, every APTS molecule is linked to an average of 1.4 surface silanols. This number gives no further information on the type or stability of interaction. Both the amine and the silicon side of the molecule may be involved. Generally, 60% of the monolayer molecules have a double interaction, and 40% interact with only one site. Further details have to be gained from other data. The adsorption isotherm for AEAPTS modification of mesoporous silica gel is given in figure 9.12.
217 3.5 loading (mmol/g)
3.0
x
(b)
2.5 2.0
[]
(a)
1.5 1.0 0.5 0.0
0
015
]
1'.5 Ceq (mmol)
2
2'.5
3
Figure 9.12 Adsorption isotherms of AEAPTS on dehydrated (673K) mesoporous silica gel from toluene solution, by freeze sampling method, (a) first equilibrium, (b) second equilibrium.
Since the reaction kinetic investigation showed a two-step adsorption, isotherms for both steps can be produced. Secondary adsorption is observed for equilibrium concentrations higher than 0.17 mmol. Both steps show an adsorption isotherm, analogous to the APTS adsorption isotherm, showing initial Langmuir behaviour, followed by additional adsorption at high equilibrium concentration. The equilibrium level reached in the first step is nearly identical to that of the mono-aminofunctional APTS, having a loading of 1.1 mmol/g. However, additional deposition occurs at far lower equilibrium concentration. The surface loadings above monolayer level at long reaction times or high concentrations have also been pointed out by Vandenberg et al. ~5 Using Atomic Force Microscopy (AFM) and Scanning Electron Microscopy (SEM), they found small aggregates of aminosilane on top of the coating layer. On these basis surface structures, as displayed in figure 9.13, were proposed.
218
//////////////////
I
////////////////// \\\\\ \\ \\\ \\ \\ \\\\ //////////////////
I
////////////////// \\\\\\\\\\\\\\\\\\ //////////////////
I
I
I
I
Figure 9.13 Schematic diagram of the structure of APTS on oxidized silicon wafers, after reaction in toluene under various conditions. (i) reaction time of minutes at RT, (ii) reaction time of hours at RT, (iii) reaction time of hours under reflux; taken from ref. a N with permission.
Some precaution has to be taken in extrapolating these data towards the porous silica gels discussed here. The reaction was performed on fiat thin SiO2 layers on silicon wafers. Therefore, no limitations by porosity are applicable. Secondly, the samples studied by AFM and SEM were dried before analysis. The effect of both these parameters will be evaluated further on. Summarizing, aminosilanes show a fast adsorption on the silica surface. An equilibrium monolayer coating is formed. Modification in aqueous solvent causes polymerization on top of the initial monolayer. For modification from organic solvent, the reactions can be better controlled. With the bifunctional AEAPTS, a secondary silane layer adsorbs on the free primary amine groups of the first monolayer. At high concentration and after long reaction times, for both aminosilane types, a further non-specific deposition occurs.
219 1.3.2 Limitations by substrate structure The effect of solvent type and aminosilane concentration has been evaluated. The third component in the reaction system is the silica substrate. The surface of the silica gel carries the active sites for adsorption. The concentration of these sites varies with varying silica type, its specific surface area and pretreatment temperature. Additionally, surface adsorbed water has a clear effect on the reaction mechanism. Isotherm data, reported in the previous paragraph, only accounted for fully hydrated or fully dehydrated silica. The effect of the available surface area and silanol number remains to be assessed. Information on these parameters allows the correlation of data from studies in which different silica types have been used. In this part the effect of these parameters in the loading step is discussed. Silica structural effects on the ultimate coating, after curing, are evaluated in the next paragraph. The effect of substrate structure in the loading step can be studied by measuring the total aminosilane loading on silicas with different topology. Analyses are performed on the solvent after the reaction, since any study of the modified substrate involves drying or curing. The quantitative analysis of the aminosilane in the toluene solvent may be performed using a spectrophotometrical method. 28 Thus an easy and quick alternative for the time-consuming and indirect Kjeldahl-N-analysis is applied. The determination of APTS by means of a colour reaction with salicylic aldehyde in protic solvent has been reported by Waddell and Leyden. 29 Upon reaction of an amine with salicylic aldehyde (reaction (B)) a yellow Schiff's base is formed. O II CH R-NH
2
+
CH 3 "-
R-NH=C
+ H20
03)
This reaction involves the formation of an imine, requiring proton catalysis and liberation of a molecule of water. Therefore, this procedure may not be applied directly in an aprotic solvent, such as toluene. With the addition of a small amount of diethylether, however, the reaction was found to reach equilibrium after one hour
220 of diethylether, however, the reaction was found to reach equilibrium after one hour and a linear correlation between the APTS concentration and the absorbance at ),=404 nm was found. After setting up a calibration curve (r = 0.996), unknown aminosilane concentrations in toluene solvent could be quantified. The total deposited amount of APTS was calculated from analysis of the residual amount of aminosilane in the solvent. Analysis was performed after two hours of reaction and consecutive filtration under ambient atmosphere. 150 td aliquots of the salicylic aldehyde and the diethylether were added to 10 ml samples of the filtrate. Absorbance was measured one hour after the ether addition. The calculated loading value yields the total surface loading, including chemical and physical deposition, in the loading step. Concerning substrate topology, three factors are of interest in the study of the aminosilane modification" surface hydration/hydroxylation, specific surface area and the mean pore size of the substrate. Comparative data on the effect of these parameters in the loading step have not been published, until recently. 2g Three mesoporous silica gels, with variable mean pore radius and specific surface area, have been studied. The substrates are named according to their approximated mean pore diameter. Measured values appeared to differ somewhat from the product names. 3~ The Kieselgels 40, 60 and 100 have a mean pore diameter of 4.2, 7.0 and 12.0 nm, respectively. Specific surface area increases with decreasing pore radius. Measured values, using the BET method, are given in table 9.3. Table 9.3 BET surface area of various silica gels used
silica type- pretreatment T
KG KG KG KG KG
60406010060-
473K 673K 673K 673 K 1073 K
SBET(before reaction, m2/g)
370 603 394 325 235
221 The influence of surface hydration and hydroxylation may be studied from the surface loading data, displayed in figure 9.14. Total surface loading was measured for the three silicas with variable pore size, at various pretreatment temperatures. 1 (retool/g) 1.9 1.7 1.5 []
1.3
123
1.1 06 200
'
4(10
'
6()0
'
8(10
'
IlXIO
pretreatment temperature (K) Figure 9.14 Total surface loading of APTS modified silica gel as a function of pretreatment temperature; *:Kieselgel 40, o: Kieselgel 60, ." Kieselgel 100.
Studying the total loading curve of the silica with intermediate pore size (KG60), three regions may be observed. At pretreatment temperatures below 473 K the total surface loading decreases with increasing temperature. A constant loading level (ltot -- 1.33 mmol/g) is obtained for pretreatment between 473 and 873 K. At 1073 K the total loading shows a reduced value. Samples with different pore size show a similar trend. At high pretreatment temperature (673 - 1073 K), the KG40 sample shows an identical loading as the KG60, while the KG100 has a reduced value. The influence of surface hydration is apparent in the 298 - 473 K region. Adsorbed aminosilane molecules are hydrolyzed and produce reactive sites for further condensation. Thus a vertically polymerised aminosilane layer is formed on the silica surface. Accordingly a higher loading value is found at low pretreatment temperature.
222 The total loading decreases with decreasing surface hydration, i.e. increasing pretreatment temperature. At 473 K the surface is totally dehydrated. As the pretreatment temperature is increased up to 873 K, the surface is gradually dehydroxylated. No change in total loading is obtained for either sample. Apparently, the number of hydroxyl groups has no effect on the total surface loading. The APTS deposition on the surface is non-hydroxyl specific. This agrees with the aspecific adsorption at long reaction times, as reported in the previous paragraph. The number of hydroxyls on the substrate surface plays no role in the total loading before drying or curing. As pretreatment temperature is raised to 1073 K, the total loading is reduced. After pretreatment at this temperature the silica shows a decreased specific surface area. In order to test whether this decrease can account for the reduction in total loading, surface coverage (C) was calculated, by rationing the loading (1) to the specific surface area (SB~x) of the sample. Multiplication by the Avogadro constant yields units of molecules/nm 2. C -
l
SBer
,NA
15)
Total surface coverage is plotted as a function of pretreatment temperature in figure 9.15. Within experimental error on surface area analyses, identical coverages are found for the KG60 and KG100 samples, over the entire pretreatment temperature range. At low temperature, the influence of polymerization is apparent again. Total coverage is constant in the 473 - 873 K range. In this temperature range loading as well as BET-surface area do not vary. A correlation between those two parameters is apparent. In this temperature region, the specific surface area appears to be of more importance than the silanol number. The total coverage at 1073 K, however, shows an increased value. The decrease in surface area can not account entirely for the reduced loading value. This indicates that specific surface area is not the only factor governing the APTS deposition at this high pretreatment temperature. A special type of control of the silane-to-surface reactivity
223
C (molec/nm2)
1
200
400
600
800
1000
pretreatment temperature (K)
Figure 9.15 Total surface coverage of APTS modified silica gel, with variable pretreatment temperature; *:Kieselgel 40, m: Kieselgel 60, A: Kieselgel 100.
may be inferred. It appears that this special type of interaction on dehydroxylated silica has a profound effect on the surface coverage, causing a higher coverage value than that found for the hydrated sample. Reasons for this typical behaviour of the dehydroxylated silica may be sought in a changed structure of the porous network or a higher reactivity of the surface. None of both hypotheses can be excluded on basis of these data. Concerning the coating structure, this effect may involve multilayer formation or a change in the molecular orientation at the surface.
1073 K pretreated silica has previously been reported to have an increased reactivity, compared to (partially) hydroxylated silica. This has been attributed to the formation of strained siloxane bridges upon dehydroxylation. 31 Strained siloxanes act as Lewis acid sites in reactions with compounds such as silanes 3~ and ammonia. 33'34 A possible explanation for the increased coverage may be that the amine group of the APTS may also enter into interaction with this type of surface species.
More
information on this effect will be obtained from the chemical loading values, after curing of the samples. This will be discussed in the following paragraph.
224
Coverages on the KG40 sample show a reduced value, compared to both other samples, at temperatures below 1073 K. This may be explained as a sterical hindrance of the modification process, due to the reduced average pore size. APTS molecules can not enter the porous structure of the KG40 as easily as they can in the larger pores of the KG60 and KG100 sample. Further evidence has to be obtained from the modified samples after curing, and will be discussed below. The above-cited behaviour of the dehydrated silica (pretreatment temp. = 473 - 873 K) is further exemplified by plotting the total coverage as a function of the silanol number (figure 9.16).
4.0
C (molec/nm2)
3.0.
A
2.0
m
-
1.0_
0.0
1
~2
3
4
5
Ix(OH) (#/nm2)
Figure 9.16 Total coverage of APTS modified dehydrated silica gel (pretreatment temperature = 473-873K) as a function of silanol number; m: Kieselge160, A "Kieselge1100.
The total coverage of APTS on dehydrated KG60 and KG100, clearly is independent of the silanol number. An average coverage of 2.05 APTS molecules per square nanometre is obtained. If we suppose that the silane molecules form one single layer on the surface, this implies that every APTS molecule covers an area of about 5.0 nm2, irrespective of the OH coverage of the surface. This value is reasonable, concerning the molecular dimensions of the APTS molecule. Chiang et al. s have previously used a hypothetical value of 4.0 nm2.
225
Correlating these data to the kinetic profile found previously, this would imply the following. In the initial phase of reaction, a monolayer coating is formed by instant adsorption of the amine groups to surface silanols. On a dehydrated surface, the silanes cover an average of 1.4 silanols per APTS molecule. After one hour of reaction the surface is further covered with silane molecules, to an extent, controlled by the specific surface area. This implies a surface filling, probably due to a rearrangement of the initially adsorbed molecules, which may reorient upon chemical bonding with the surface. In the ultimate layer each molecule covers an average area of 5.0 nm2. Although this descriptive model seems reasonable, it should be regarded as hypothetical, since it encompasses that deposition on top of the initial layer does not occur.
1.3.3 Conclusion Summarizing we may state that, as adsorption isotherms revealed the formation of an equilibrium in the initial stage of the reaction, total loading data after two hours of reaction and filtration reveal that specific surface area and mean pore size are the controlling parameters in the loading step. Surface water causes hydrolysis and polymerization. On a dehydrated surface, a surface coverage irrespective of the number of hydroxyls is formed. For silica dehydroxylated at elevated temperature (1073 K) a different behaviour is observed, suggesting the participation of strained siloxanes.
1.4 Characterization of the aminosilane modified silica For the stabilization of the silane layer on the silica surface, a post reaction curing step is required. After filtering and possibly washing, the reacted substrate is cured at temperatures in the 353-473 K range. The parameters used in different studies vary reasonably. The solvent used for washing, the curing temperature and surrounding atmosphere, predrying of the substrate before curing, all are used in a variable way. Together with the variability used in the loading parameters, this hinders a quantitative correlation of data from different studies. In order to allow a qualitative comparison, we are discussing the effect of various parameters on the modified structure.
226 In analogy to the discussion of the loading step processes, both the processes occurring in the curing step and the effect of substrate structure on these processes will be discussed. Whereas the study of the loading step processes involved analysis of the solvent, here the modified substrate is studied. For this, various spectroscopic techniques are applied. The discussion is focused on the silicon side of the aminosilane molecule. The interactions at the amine side will be discussed in the next paragraph. 1.4.1 Chemical condensation of aminosilanes with the silica surface
UV stability studies In the curing phase, the silane coating is stabilized by the formation of siloxane bonds between the silane and surface silicon atoms. Thus, physically adsorbed molecules are chemically bonded. The extent and rate of this process may be studied from the stability of the coating towards ethanol leaching under varying curing conditions. Detection of the aminosilane lost from the surface is performed by means of a spectrophotometric method, analogous to the one used in the measuring of the adsorption isotherms. Amines react with salicylic aldehyde with the formation of a yellow Schiff's base, having ~,max= 404 nm (reaction B). In toluene solvent, ether was needed to initiate the reaction. In a protic solvent, such as ethanol, the reaction proceeds readily. After calibration, the amine function thus allows the quantitative determination of the amount of an alkylamine or aminosilane liberated from a modified substrate upon ethanol leaching, by UV detection. Using this method, Waddell et al. 29 compared the stabilities of mono-, di- and trialkoxyaminosilanes, upon curing at 353 K in air. Reaction was performed on ovendried mesoporous silica gel in dry toluene. Their results indicated the progressive stabilization of the aminosilane layer upon heating. Under these circumstances, they found that 3 hours of curing were sufficient to reach optimal coating stability. The number of alkoxy groups in the silane molecule was shown to determine the overall stability of the layer, even before heating of the reacted substrate. After optimal stabilization, the trialkoxysilane (APTS) showed a comparable stability, to the
227
dialkoxysilane (APDMS). This indicated that the APTS does not form long polymers on the surface, since the APDMS can only form dimers. These results were further elaborated by measuring the silane loss over short leaching times for samples cured in air and under vacuum. 35 Dehydrated mesoporous silica gel was modified with APTS or APDMS in dry toluene (1% v/v silane/toluene). Curing was performed for variable times in air at 383 K or under vacuum at 423 K. For UV tests the modified silica was stirred in a salicylic aldehyde/ethanol solution. At indicated times a 5ml sample was taken, centrifuged and the supernatant was measured at 404 nm. The loss curves of APTS for variably cured samples are displayed in figure 9.17. The position and profile of the absorbance curves are indicative for the stability of the coating under study.
abs. A
0.8 0.6 !
0.4
[]
I._lr"n
.
0.2
I
0
50
I
I
I
100 150 200 reaction time (rain,)
I
250
300
Figure 9.17 Loss curves for short term ethanol leaching on APTS modified silica: (x) uncured," (+) 3h air cured," (*) 20h air cured," (m ) 20h vacuum cured.
The amount of product lost as well as the course of the loss, upon stirring of the modified substrate in ethanol, are indicative of the stability of the coating. All samples show a rapid initial loss, followed by a slower process.
228 The curve of the air dried, uncured sample (x) shows a clear sloping over this relatively short reaction period, indicating the occurrence of an easy removable and a more firmly bound component. However both are not stable towards short term ethanol leaching. For the sample which has been cured for 20 h under vacuum (o), thus securing the absence of air humidity, the coating is stabilized. An invariable amount of silane is detected over the entire leaching period, indicating only one type of physical interaction. The increased coating stability after 20 h of curing is due to chemisorption to the silica surface. Still a small amount of the coating molecules is easily removed. The absorbance value found, corresponds to a concentration of 0.02 mmol/g. From correlation with the loading values, cited earlier, this would correspond to about 2 % of the coating, which is not irreversibly bound to the surface. These molecules are physisorbed to the surface by hydrogen bonding of the amine group to a surface hydroxyl. For the sample cured in air for 20 h (*), an additional stabilization is obtained. This stabilization is attributed to oligomerization of the silane molecules on the silica surface. Oligomerization occurs under the influence of air humidity. From the difference in position of the curves for 3 h (+) and 20 h (*) cured sample, it can be concluded that there is a progressive oligomerization during the air curing period. The air cured APTS curves have a more clear sloping than those of the vacuum cured samples. The rapid release of loosely physisorbed molecules is followed by a more slow loss. Because the slow process is measurable within this short ethanol leaching time, it can not be attributed to the breaking of chemical bonds. The occurrence in humid conditions lead us to conclude that this slow loss is caused by hydrogen bonding interactions of the silane molecules via the silane silanols. These silanols are formed under the influence of air humidity. Hydrolysis of silane alkoxy groups by air humidity, therefore, causes a double stabilization. Silane silanols can form hydrogen bonds to the surface and can cause oligomerization of the silane coating. Figure 9.18 gives a clear view on the stabilization of the coating by condensation with the surface hydroxyls during curing under vacuum conditions. The amount of hydrogen bonded aminosilane, as calculated from the absorbance data, is plotted as a function of curing time in vacuum.
229
0.028 mmol/g
I
0.026 0.024 0.022
a []
0.020
[]
I
0
10
I
I
20 30 curing time (h)
I
40
50
Figure 9.18 Amount of physisorbed silane molecules per g of modified silica as a function of curing time in vacuum, (a) APTS; (b) APDMS.
In evaluating the data for the vacuum curing only, contributions from oligomerization and hydrogen bonding of silane silanols are ruled out. Additional information is obtained in comparing the data for APTS modified silica to those measured on silica modified with APDMS. The APDMS molecule has only two ethoxy groups and therefore can only form two chemical bonds to the silica surface. For APTS a maximal stability (minimal amount of physisorbed species) is obtained within 3 h of curing at 423 K under vacuum. In the APDMS case the physisorbed amount decreases with curing time increasing up to 20 h. Because of the presence of only 2 ethoxy groups in the APDMS molecule, compared to 3 for the APTS, stabilization by formation of a siloxane bond is at maximum only after prolonged curing. For vacuum curing times above 20 h, the same level as obtained for APTS is maintained. Both silanes have an identical stability towards short-term leaching. About 2 % of the coating layer remains physically adsorbed. Concluding, it may be stated that the amount of non-chemically bound silanes decreases upon curing of the samples, due to condensation with the surface. Extra stabilization is obtained by oligomerization of the silane molecules, for air cured samples. Also for these samples, silane hydroxyls, formed upon hydrolysis of ethoxy
230 groups, are in hydrogen bonding interaction with the silica surface. For vacuum curing, the time needed for full stabilization is related to the number of alkoxy groups in the silane molecules. After curing for 20 h both APTS and APDMS layers on silica have an identical stability.
29Si CP MAS NMR study of the chemically bonded species After characterization of the physisorption, we wish to get information on the ultimate chemical structure of the coating. Spectroscopic analysis of the modified substrate is used to complement the indirect analytical data, reported above. Further elucidation of the chemical bonding of the silanes to the surface is obtained from 295i CP MAS NMR. With this technique, the formation of siloxane bonds with the surface can be modelled.
29Si CP MAS NMR spectra of 20 h air and vacuum cured APTS modified silica and 20 h air cured APDMS modified silica were recorded. The 29Si NMR spectra (figure 9.19) give information on the coordination at the silicon atoms of the silane molecules and in the silica gel. Band assignments for the APTS modified silica haven been given earlier in this chapter. For the APDMS modified sample, the band at -19.7 ppm is assigned to the bidentate bonded form, the monodentate form is found at -12.4 ppm:
R CH3 ~
I
S l - - OH
-12.4 ppm
.\
/
~
-19.7 ppm
R
R
I
I
CH3 ~ Si--- O ~ S I ~ CH3
-19.7
ppm
From peak deconvolution, the relative percentages of the different bonding forms have been calculated. Results are given in table 9.4 In the vacuum cured sample (figure 9.19 a), the APTS is mainly in the bidentate form (-59 ppm). From curve fitting and integration a relative amount of 60% was calculated. Monodentate and tridentate forms show only minor contributions (20% each).
9
~marl
~
o~
c~
co
olli~
o~
~
232 The previously discussed hydrolysis and oligomerization are evident from the spectrum of the air cured sample (figure 9.19 b). The silane region shows clear contributions from the hydrolyzed monodentate (-46 ppm) and oligomerized (tridentate, -66 ppm) forms. The APDMS is found to be almost entirely (90%) in the bidentate form. Because of the bifunctionality of the APDMS this is the most plausible form. Any hydrolysis and condensation between silane molecules can at most lead to surface dimers. A minor contribution of the monodentate form (-12.4 ppm) is visible.
Table 9.4 ~gsi CP MAS NMR peak positions, with relative contributions of bonding form, for APTS and APDMS modified silica gel silane used + curing conditions APTS, air 20 h APTS, vacuum 20 h APDMS, air 20 h
chemical shift (ppm) / relative contribution of bonding form (%) -49 ppm 11 -12 ppm 10
-53 ppm 20 -20 ppm 90
-59 ppm 52 60
-67 ppm 37 20
Caravajal et al. 22 discussed the effect of curing temperature on the silane conformation. APTS modified silica was cured at variable temperature and 29Si CP MAS NMR spectra were subsequently recorded. Results for the modification of silica gel pretreated at 473 K are given in figure 9.20. Other reaction parameters were comparable to the study reported above. A clear shift of intensity is visible from the -49 ppm peak to the -58 ppm peak to the -66 ppm peak as the curing temperature is increased. This clearly shows that the number of siloxane bonds between the silane silicon atoms and the surface (or other silanes) increases as the curing temperature increases. Concluding, the processes occurring under various curing conditions have been modelled. In the curing phase, the aminosilane layer is stabilized by covalent bonding to the surface. Siloxane bonds are formed between silane and silica surface atoms. The velocity of the siloxane formation is dependent on the initial number of ethoxy
233 200 Series POST RE ACT l ON CUR 1 NO TEMPERATURE ! "Cl
200
-s8
150 110 65
......
-50
-I00
-150
Figure 9.20 29Si CP MAS NMR spectra of mesoporous silica gel, pretreated at 473 K and modified with APTS in toluene solvent, after curing under vacuum at indicated temperature; taken from ref (22) with permission.
groups in the silane molecule. Upon curing in air the physisorbed silane molecules hydrolyze and form short oligomers. This process enhances the stability of the layer. Curing under vacuum results in a majority of silane molecules bonded in a bidentate form. The average number of siloxane bonds per molecule increases with increasing curing temperature.
1.4.2 Limitations by substrate structure As the effect of curing parameters has been assessed, the role of the silica structure remains to be studied. In the previous chapter, we discussed the effect of substrate structure in the loading step. Data were obtained from solvent analysis after reaction. Those revealed information on the total loading of the surface. Concerning the curing phase, it has been shown that physically adsorbed molecules are progressively chemically bonded. No distinction has been made hereto between the total deposition, i.e. the amount of silane adsorbed from the solution, and the ultimate number of chemically bonded molecules. Comparing these figures should also yield information on the loss of physically adsorbed molecules by desorption during curing.
234
Chemisorption and physisorption The silane loading on the surface after curing may be analyzed using various techniques. Elemental analysis has been used repeatedly, zs The Leyden group introduced quantitative analysis from Diffuse Reflectance FTIR spectroscopy. ~6'17 Recently, Kallury calculated quantitative loading data from XPS spectral data. 4 Elemental analysis, however, gives the most direct results. This technique involves the catalytic combustion of the modified sample, with selective adsorption of the evolved gasses. The relative percentages of C, O and N may thus be obtained. From this percentage, the amount of APTS per gram of modified silica may be derived directly. Total loading data, however, have been given in mmol APTS per gram of initial pure silica. In order to allow comparison, this value should also be used here. To calculate the APTS loading per gram of pure silica, the molar mass of the deposited molecule is needed. Since APTS may be deposited in three configurations (mono-, bi- or tridentate form) this molar mass is not defined unequivocally. For a dehydrated silica it was shown previously that the relative mono/bi/tri ratio equals 0.2:0.6:0.2. The weighted average molar mass of the molecule fragment, MM = 131, is used. For the dehydroxylated substrate, this approach will cause a slight underestimation of the loading, since the relative contribution of the monodentate form increases with increasing pretreatment temperature. The elemental analysis gives the surface loading after curing of the modified sample. Therefore only chemically bonded species are measured. In analogy to the loading step study, the effect of substrate structure has been studied by modifying mesoporous silica gels with a variable mean pore diameter, zg Sample pretreatment and curing (20 h, 423 K) were performed under vacuum. Variation of the pretreatment temperature causes a change in specific surface area and silanol number. While the hydroxyl content did not influence the total deposition, it is expected to affect the chemical load of the surface. The effect of the hydroxyl content on the chemical load may be obtained, after plotting the load as a function of the input ratio (I). As defined earlier, I is the initial relative ratio of the number of surface hydroxyl groups (NoH) to the total amount of silane in the reaction mixture (NAvrs)(eq.2). Results are displayed in figure 9.21.
235
1 (retool/g) 1.3 1.1 0.~ 0.'~ 0.5 0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Figure 9.21 Chemical loading of APTS modified mesoporous silica gel as a function of input ratio OH:APTS.
The chemical loading increases with increasing relative number of hydroxyls. The curve levels off at a maximum of 1.16 mmol/g, for I >__ 1. This shows that the chemical bonding of the silane molecules is controlled by the hydroxyl content, as long as the aminosilane is in excess of the number of hydroxyls. As the hydroxyl number is further increased, with constant concentration of the reagent solution, the chemical loading remains constant. Reaction will result in a higher coordination of the silane molecules. The numerical values of total and chemical loading are displayed in table 9.5. The total loading reflects the sum of physically and chemically bonded species. During curing at high temperature and reduced pressure, physically adsorbed species are desorbed from the surface, or form chemical bonds. By subtracting the chemical loading after curing from the total loading before curing, the amount of APTS lost by desorption may be calculated. This amount is given as a relative percentage of the total deposition in table 9.5.
236
Table 9.5 Pretreated silicas, with their specific surface area and the total, chemically bonded and (relative) physically bonded loading of APTS a~er modification silica type pretreatment T
silica silica silica silica silica
60 - 473 K 40 - 673 K 60 - 673 K 100- 673 K 60 -1073 K
SBEr
Itotn1
(before reaction) (m2/g)
(mmol/g)
370 603 394 325 235
1.27 1.27 1.31 1.06 1.11
Ich~ (mmol/g)
1.14 1.18 1.09 0.96 0.72
Iphy,/Itot~ (%)
10 7 17 9 35
From the tabulated data, it may be observed that the amount of desorbed molecules varies a little with pretreatment temperature. For the 473 - 673 K pretreated samples about 10% of the total loading is lost during curing. For the dehydroxylated silica (1073 K), a relative loss of 35 % is observed. Similarly to the processes reported in the loading step, the behaviour of the dehydroxylated surface clearly differs from the other samples. This will be discussed below. Concerning first the dehydrated substrates (473 - 673 K), we reported above that in the loading step, a state of equilibrium is reached during the first hour of reaction. This was attributed to the formation of a monolayered coating. At longer reaction times, an additional deposition was observed. For the dehydrated substrate with intermediate porosity, (silica gel 6 0 - 673 K) at equilibrium, a silane loading of 1.02 mmol/g was obtained. Since the chemical loading value (1.09 mmol/g) corresponds to this monolayer equilibrium value, the following conclusions may be drawn. Upon mixing of the silane with the silica, the APTS adsorbs very rapidly on the silica surface by hydrogen bonding or ionic interaction of the amine group with a surface hydroxyl. The adsorption is hydroxyl specific and a monolayer is formed. Adsorbed silane molecules may self-catalyze the direct condensation of the silane with the surface, with the formation of siloxane bonds. This causes some reorientation of the adsorbed layer. At long reaction time additional silane adsorbs on the surface, filling the 'holes' between the hydroxyl-adsorbed species. Thus, a total loading controlled by the specific surface area of the substrate is obtained. Non-specific adsorbed molecules desorb quickly upon drying or curing, while hydroxyl-adsorbed species are further chemically bonded by siloxane formation.
237 As the additional physisorption has built up additional layers on top of the first monolayer, as for modification at higher concentration, the additional layers will analogously be desorbed. This mechanism is pointed out by thickness measurements, performed on non-porous modified silica layers. ~5 Upon curing at 473 K on air the thickness of the APTS layer appears to be reduced from 1.8 to 0.7 nm, indicating that after curing only one layer of APTS is left. The modification of the dehydroxylated substrate (pretreatment temperature = 1073 K) appears to follow a different mechanism. As stated above, the total coverage of this sample has a very high value. However, upon curing, as much as 35 % of the total loading is desorbed, leaving a chemical loading of 0.72 mmol/g. The mechanism of enhanced loading causes mainly a physical interaction with the surface. If we consider the ratio of the initial hydroxyl number to the number of chemically bonded molecules, the average number of reacted silanols per silane molecule is obtained. For this sample, a value of 0.6 is obtained. This indicates that at least 40% of the silane molecules have not coordinated with a surface hydroxyl group. The coordination with strained siloxanes may be an explanation for this phenomenon. However, based on these data, no absolute proof can be given. The physical and chemical bonding would than proceed by reactions (C) and (D), respectively.
- - SI
- - SI \ . . - - N H 2 R S i ( O R ' ) 3 0
+
H2NRSI(OR')3
- - Si /
0
=
(c)
- - Si /
= SI ~ N H 2 R S i ( O R '
)
= SI-- NHRSi(OR') 3 (D)
0 SI /
---- 81-- OH
Pore size distribution
As the effect of surface hydration and hydroxylation has been characterized, the influence of the porous structure remains to be studied. Sterical hindrance of the modification process was suspected from the total coverage values. This is further evidenced by measuring the pore size distribution of the three silicas before and after modification, as displayed in figure 9.22.
238
lXl~volut~ dltltlllit~ 0,8.
0,6 f
0,4
17h-FOH).
Similar observations can be made for the bridged silanols. One notices a significant enhancement in the conversion degree when the reaction is extended to 3 h, after which the conversion degree approaches a constant value. It is noticeable that-even after 17 h of reaction- only 50% of the bridged silanols have reacted (pretreatment temperature = 673 K), whereas sterically, total conversion should be possible. This could explain the statement of some authors that chlorosilanes react exclusively with free silanols. Most of cited studies described the chlorosilylation of a silica gel, pretreated at 673 K. After the reaction, the infrared spectrum indeed shows only a
282 large remaining bridged silanol band and no free silanol peak. Since most of the cited studies have been performed some decades ago, one could reasonably state that the small diminution of the BOH band could not be detected by older type spectrometers.
2.2 Liquid-phase reaction with alkylchlorosilanes (Cs- Cls) The formation of monolayers by self-assembly of organochlorosilanes on various surfaces 62'63'64'65 and organosulfur compounds on gold 66'67 is well established. The durability of the self-assembled monolayer is highly dependent on the effectiveness of the anchoring to the surface. On gold, the attachment to the surface is due to an interaction of sulphur end groups with the gold surface. However, the nature of the attachment of the organochlorosilane with the surface is ill-defined. 68'69'7~ Octadecyltrichlorosilane (C18H37SIC13, henceforth denoted OTS) is the most common
organosilane used for the formation of self-assembled monolayers and, when reacted with silica surface, finds extensive use as a bonded phase in liquid chromatography applications. 7~ A common mechanism proposed for attachment of the chlorosilane to the surface 69'72 involves the hydrolysis of the chlorosilane groups with water which is already on the surface of the substrate. The silanols which are formed then condense with the surface hydroxyls groups to form stable linkages to the substrate. In practice, a curing process is usually required to condense adjacent silanols attached to the organosilane to form a cross-linked 'mat' on the surface. Part of the difficulty in determining the nature of the attachment to the surface arises from the lack of direct spectral evidence. For high surface area metal oxides such as fumed silicas, infrared spectroscopy 49'73'74'75 and NMR 76'77 have been very successful in determining the nature of the attachment of chlorosilane molecules to the surface. The main problem in using infrared spectroscopy in the study of self-assembled monolayers is that the substrates have low surface area and therefore the bands due to the adsorbed species are low in intensity. Although it is generally believed that the water content on the surface is vital for the formation of a monolayer, and that OTS reacts with the water layer and forms only few bonds with the surface, direct spectroscopic evidence was lacking.
283
In 1992 Tripp and Hair TM unified the two chlorosilane approaches by reacting OTS with a high surface area amorphous silica gel, in order to probe spectroscopically the different surface species. Using a home-made in situ liquid infrared cell, they derived following conclusions: 1. OTS does not react with degassed silica at room temperature. The infrared band of the free hydroxyls shifts to 3690 cm ~ but does not change in intensity. This indicates that the chlorosilane is physisorbed (H-bridged) on the silica surface. Subsequent degassing removes all adsorbed species. This conclusion is not surprising. Also (methyl)chlorosilanes do not react with the silica surface at room temperature. Reaction temperatures > 473 K are required to achieve noticeable reaction. The boiling point of octadecyltrichlorosilane is 433 K. It would be very interesting to see what happens at reflux temperature.
0
OTS does react slightly with 'wet' silica at room temperature, containing multilayers of water on the surface. The broad band at 3650 cm ~ decreases slightly, and a band at 3350 cm ~ arises, attributed to trisilanols. 79'8~Tripp and Hair TM state that the first layer of water is strongly bonded to the surface and does not participate in the hydrolysis of the chlorosilane headgroup of the OTS molecule. Subsequent layers would be less strongly bonded to the surface and would be able to participate in direct hydrolysis of the OTS.
Since the hydrolysis and adsorption of the OTS occurs with the subsequent layers of water, an optimum level is necessary to form robust films" too little water results in the formation of an incomplete monolayer, whereas a thick water layer causes a polymerization of the OTS with the water, resulting in a very poor adherence to the silica surface. Not only octadecyltrichlorosilane is unreactive towards dry silica at room temperature. This is also the case for the chlorosilanes and the methylchlorosilanes. It was stated earlier that the vapour phase reaction occurs at elevated temperatures (> 473 K). This high-temperature constraint limits potential gas phase silanizing agents to those which have a high thermal stability and sufficient vapour pressure.
284
In practice, the common method for silanization of silica is to mix a chlorosilane with a hydrated silica in a suitable organic solvent. A common mechanism reported for this reaction is that the chlorosilane is first hydrolyzed by the water and this is followed by the condensation with the surface hydroxyl groups to form a strong S i - O - S i surface bond. 72 If the starting silane contains a trichlorosilyl headgroup, then further condensation between adjacent silanes can occur, yielding a two-dimensional polysiloxane network. The occurrence of the first step, the hydrolysis of the chlorosilane to a silanol by the surface water is amply supported by the literature. 6s'69,Ts'gl At either the solid/gas or solid/liquid interface the chlorosilane does not adsorb on a completely dehydrated silica and is hydrolyzed to the silanol with the surface water of a hydrated silica. However, Tripp and Hair s2 have shown that the second critical step (i.e. condensation of the silanol with the surface hydroxyls groups) does not occur. At the solid/gas interface, the silanol adsorbs on the surface but does not undergo condensation or polymerization, whereas at the solid/liquid interface, the silanol polymerizes in solution and adsorbs on the surface. In neither case there is a strong Si~-O-Si bond formed with the substrate. It is the absence of Si~-O-Si surface linkages that is responsible for the general lack of robustness of silanized surfaces prepared from solution. In a subsequent publication, Tripp and Hair 83 describe a new method to chemically bind chlorosilanes to the surface under mild reaction conditions. In fact, there are two possible ways for a base-catalyzed chemisorption of chlorosilanes on silica. One possible strategy is to use a base to promote the reaction of chlorosilanes with the surface silanols. In essence, this reaction proceeds by a one-stage nucleophilic mechanism through the formation of a pentacoordinate silicon intermediate. The Si-CI bond is lengthened in the intermediate and is susceptible to attack by a second nucleophile: s4
285
R -
R3SiC1 + Nu
Cl I
\Si-
R
(Q)
RW'Nu ~
R
C1 I
\Si- R
RW'Nu ~
+ -Si-OH
~
-Si - OSiR3
+ HC1 + N u
(R)
An alternative mechanism for the base-promoted reaction of silanes with silica has been described by Blitz et al. 85 In this mechanism, the base attacks directly to the surface silanols. The bonded amine renders the silanol more nucleophilic which then attacks the silicon atom of an approaching silane, giving rise to a pentacoordinate intermediate.
/ O"
.
N(Et)3 / 8 H + ~ C 1 ~ 3
+ N(EO 3_.
/ O 8-
CI "I N(Et)3 H 3C uSi,. ~ H _ /6 +
Cl1"~)/6_
CI CH 3 HCI NSi' N(Et)3 / O "el
In both mechanisms a pentacoordinate intermediate is postulated. The main difference is that the pentacoordinate intermediate is formed by attachment of the amine to the
286
chlorosilane in one case and by attachment to a surface Si-O group in the other. The main problem associated with base-catalized silanization is that it is very difficult to prevent polymerization of the silane in solution. Rapid polymerization of the chlorosilane occurs in solution unless extreme precautions are taken to exclude contact with residual water. Thus, in a typical base-promoted silanization on silica, it is more likely that both polymerization and surface reaction occur to some extent. Both mechanisms can account for polymerization. The intermediate formed by attachment of the amine to the chlorosilane could react with nucleophiles (i.e., molecular water) other than the surface silanols. In the mechanism described by Blitz et al. the chlorosilane (either attached or in solution) could be hydrolyzed to the trisilanol by molecular water and the trisilanol offers an additional source of silanols for base attachment and subsequent polymerization. Polymerization often results in a thick silane layer on the surface that in many cases is undesirable. Polymerization is not possible in the complete absence of water or when reactions are carried out using monochlorosilanes. However, trichlorosilanes are attractive because it is possible to increase the strength of the adsorbed silane layer through cross-linking between adjacent molecules. The other approach, the exclusion of trace quantities of water, especially in solution, is extremely difficult and costly. In 1993, Tripp and Hair g3 described a method to promote the direct reaction of the chlorosilyl headgroup with the surface hydroxyls groups, using a nitrogen-containing base (triethylamine). In this method, polymerization is avoided because the base and chlorosilane are not added simultaneously but subsequentially in a two-step process. The silica is evacuated and exposed to the vapour of a nitrogen-containing base which forms a strong hydrogen bond with the surface silanols. Excess base is removed by evacuation and the silica is then placed in contact with the chlorosilane. It is the evacuation of the excess amine which suppresses polymerization. The base catalyzes the reaction immediately and leaves the silane chemically attached to the surface of the silica. Although the coaddition of the base and chlorosilane under anhydrous conditions gives similar adsorbed species, the two-step process has a number of advantages:
287 in the presence of surface water, polymerization is avoided; in the absence of surface water, the two-step process can be used to build a multilayered silane in a well-defined manner.
C1 / C H 3 \Si ~ I
(Et)3N H 2~ =
C1
O
CH3 H\ O [
~ Si f O
N(Et)3
,I
H
I
o
CH 3 C1--si~C1 O~ CH 3SiCI_3
~ H3
C1
Si OI
0 ~~ S~~'m CH 3 / C1
The creation of such a network of attached silanes on a substrate by consecutive single-step adsorptions is in fact a Chemical Surface Coating (CSC). The principles of Chemical Surface Coating 86'87are explained and exemplified in detail in part 3 of this book.
288
3 Modification with other silicon compounds
3.1 Halogenosilanes (e4_nSinrn, e4.nSiIn) The interaction of the silica surface with trimethyliodosilane and trimethylbromosilane has been described by Tertykh and Belyakova. s8 Both reactions proceed much faster than the previously described chlorosilane reaction. The reaction of silica with trimethylbromosilane is completed after 30 minutes at 323 K and the reaction with trimethyliodosilane even after 10 minutes at room temperature. The authors evidenced, using a methoxylated silica as a substrate, that at these temperatures no reaction occurs of the halogenosilanes with the siloxane bridges of the silica. No pretreatment temperature was mentioned, however. Therefore we cannot exclude a small reaction of the bromo- or iodosilanes with the siloxane bridges of the silica, thermally pretreated at high temperatures. The reactivity sequence of the halogenosilanes can be ordered as (CH3)3SiI > (CH3)3SiBr > (CH3)3SiCI It can be inferred from the data in table 9.10 that the decrease in reactivity can be explained in terms of decreasing Si-X bond length or increasing Si-X bond energy. Also the higher polarizability and proton accepting properties of the larger halogens have to be taken into account. Table 9.10 Some physical and chemical characteristics of silicon-halogen bonds Bond
Length of the bond, nm
Si-C1 0.2020+0.0005 Si-Br 0.216+0.001 Si-I 0.246___0.002
Energy of the bond, kJ/mol
396 289 213
Polarizability, cm3
7.11 10.03 -
Dipole moment of the (CH3)3SiHal molecule, D 2.09 2.36 2.46
289
3.2 Alkoxysilanes (R4..Si(OR).) In a number of cases, (e.g. modification of a silica surface, containing N-functions), alkoxysilanes are more useful than the corresponding chlorosilanes for reacting with the silica surface. The reaction of dimethyldiethoxysilane ((CH3)2Si(OC2Hs)2) with the isolated silanols proceeds under milder conditions than the chlorosilane analogue: dimethyldichlorosilane ((CH3)2SIC12). In the former case, complete conversion is obtained at 373 K, whereas the chlorosilane needs temperatures up to 573 K. 88 However, in the case of a silica containing vicinal silanols, the reactivity is reversed and the chlorosilanes become more reactive than the corresponding alkoxysilanes. The increased reactivity of alkoxysilanes in the reaction with isolated silanol groups can be explained by the higher proton affinity of the oxygen atom of the Si-O-Si group, compared to the Cl-atom in the Si-CI group. This is evidenced by the observation that the physisorption of an alkoxysilane on a silica surface results in a free hydroxyl band shift of approximately 300 cm ~, whereas the physisorption of a chlorosilane induces a shift of'only' 100 c m l . 89 The presence of significant quantities of physisorbed water promotes the chemisorption of alkoxysilanes, especially of the ones with two or three alkoxy groups. Engelhardt and Orth 9~ reported that chemisorption of n-octyltriethoxysilane from a dry toluene solution on a dehydrated silica amounts 0.30/~mol/m 2. However, if a toluene solution containing traces of water is used, the amount of chemisorbed silane increases up to 1.25 #mol/m 2. The process of oligo- and polymerisation can occur either on the surface of the wet silica or in the toluene solution. The formation of polylayered coatings was investigated in detail by Gorski et al., 91 reporting on the chemisorption of 3-methylacryloxypropyltrimethoxysilane from acetone solutions, containing various amounts of water.
290 The rather difficult question how many alkoxygroups of one trialkoxysilane molecule participate in the chemical bonding has been studied by high resolution NMR on 29Si and ~3C nuclei. 92'93'94 Sindorf and Macie192 concluded that in the case of dehydrated silicas, mainly monodentate and some bidentate species are formed. Reactions involving all three alkoxy groups do no occur on the surface of dry silica. However, using a hydrated silica as a substrate results in the complete disappearance of the ~3C resonance of ethoxy groups, indicating that all alkoxygroups have participated in the reaction. In the last decades, the interest in the chemisorption of alkoxysilanes on the silica surface has increased significantly. Many organophylic silicas have been synthesized using alkoxysilanes as modifying agents. They have found applications as fillers of electric isolating resins or as thickeners for organosilicon vaselines. 88
3.3 Alkylsiloxanes The electron-donor ability of oxygen decreases according to the series C - O - C > C-O-Si > > Si-O-Si
This statement was evidenced by a study of West et al. 95 They noticed that the O H - b a n d shift of phenol amounted 329, 261 and 169 cm ~ upon interaction with (CH3)3COC(CH3)3, (CH3)3COSi(Cn3) 3 and (CH3)3SiOSi(CH3) 3 respectively.
It is assumed that this behaviour can be explained by an additional interaction between the undivided p-electrons of the oxygen atom and the vacant 3d-orbitals of the silicon atom (the so called (p:d,)-interaction). 96 The ideas about the nature of the siloxane bond were developed by Voronkov. 97 He stated that (p:d~)-bonds of a donor-acceptor type are formed in a Si-O-Si system owing to the undivided electron pairs of oxygen and the vacant d-orbitals of the silicon atoms. In this case, the oxygen atom is in a state of hybridization, intermediate between sp 2 and sp.
291
This leads to a significant weakening of the electron donor properties of oxygen and to a significant weakening of the electron accepting properties of the silicon. This means that the heterolysis of a siloxane bond proceeds under much more severe conditions than their carbon analogues. However, the heterolytic splitting of Si-O-Si bonds is not impossible. Strong electrophilic reagents, such as BF3 or A1C13, or strong nucleophilic reagents with a possibility of simultaneous electrophilic attack, such as HF, are able to break the Si-O-Si bond quite easily. Moreover, the properties of the Si-O-Si bond are strongly influenced by the nature of the substitutors near the silicon atom. Electron accepting substitutors lead to an increase of the siloxane bond strength, whereas electron donating substitutors lead to a decrease of the siloxane bond strength. Kriegsmann 9s noticed that the siloxane bonds of hexamethyldisiloxane are easily broken in the reaction with BF3, whereas hexachlorodisiloxane does not react at all. Room temperature adsorption of hexamethyldisiloxane mainly results in a physisorption, established by hydrogen bonding. 88 Figure 9.44 clearly shows that the 3750 cm 1 of the free hydroxyls shifts towards 3710 cm" and 3400 cm ' upon siloxane adsorption. At the same time, naturally, the characteristic bands for methyl groups appear. The 3400 cm ~ band has been assigned to a hydrogen bonding between the siloxane oxygen and the silanol hydrogen atoms (figure 9.44 a). The 3710 cm ~ band has been assigned to an adsorption complex between the oxygen atom of the silanol group and the silicon atom of the siloxane (figure 9.44 b). _YOH~ 3400 cm -1
o"t/
(a)
J, ,,,
H--
O
~Si(CH3) 3 ~~
H~
/O
Si(CH3) 3
O~
~OH 3710 c
Si ,,,
I
Si(CH3) 3 Si(CH3) 3
(b)
Figure 9.44 Assignments of the silanol stretching vibrations upon physisorption of hexamethy ldisiloxane. Degassing of the silica results in a complete desorption of the hexamethyldisiloxane. Irreversible chemisorption of alkoxysiloxanes starts at reaction temperatures above
292 673 K. Tertykh ss evidenced that complete removal of silanols is achieved on an aerosil, treated at 923 K, and reacted with hexamethyldisiloxane for 40 minutes at 773 K.
References
o
0
0
E.P. Plueddemann, Silane Coupling Agents, 2nd ed., Plenum Press, New York, 1991. S.W. Morrall, D.E. Leyden, in Silanes, Surfaces and Interfaces, D.E. Leyden ed., Gordon & Breach Science Publ., New York, 1985, 501. K.C. Vrancken, P. Van Der Voort, I. Gillis-D'Hamers, E.F. Vansant and P. Grobet, J. Chem. Soc. Faraday Trans., 1992, 88, 3197.
4.
K.M.R. Kallury, P.M. MacDonald and M. Thompson, Langmuir, 1994, 10, 492.
5.
J.P. Blitz, R.S.S. Murthy and D.E. Leyden, J. Am. Chem. Soc., 1987, 121, 63.
6.
D.G. Kurth and T. Bein, Langmuir, 1993, 9, 2965.
7.
P. Trens, Ph-D thesis, Aix-Marseilles, 1994.
8.
C. Chiang, H. Ishida and J.L. Koenig, J. Colloid Interface Sci., 1980, 74, 396.
0
E.P. Plueddemann, Silane Coupling Agents, 2nd ed., Plenum Press, New York, 1991, 64.
10.
D.J. Kelly and D.E. Leyden, J. Colloid Interface Sci., 1991, 147, 213.
11.
C.W. Chu, D.P. Kirby and P.D. Murphy, J. Adhesion Sci. Technol., 1993, 7, 417.
12.
S. Naviroj, S.R. Culler, J.L. Koenig and H.Ishida, J. Colloid Interface Sci., 1984, 97, 308.
13.
C.G. Pantano and T.N. Wittberg, Surf. Interface Anal. 1990, 15, 498.
14.
H.-J. Kang and F.D. Blum, J. Phys. Chem., 1991, 95, 9391.
15.
E.T. Vandenberg, Bertilsson, B. Liedberg, K. Uvdal, R. Erlandsson, H. Elwing and I. Lundstr6m, J. Colloid Interfac. Sci., 1991, 147, 103.
293
16.
R.S.S. Murthy and D.E. Leyden, Anal. Chem., 1986, 58, 1228.
17.
R.S.S. Murthy, J.P. Blitz and D.E. Leyden, Anal. Chem., 1986, 58, 3167.
18.
M.T. McKenzie, S.R. Culler and J.L. Koenig, Appl. Spectrosc., 1984, 38, 786.
19.
C. Chiang, N. Liu and J.L. Koenig, J. Colloid Interfac. Sci., 1982, 86, 26.
20.
J.W. De Haan, H.M. Van Den Bogaert, J.J. Ponje6 and L.J.M. Van De Ven, J. Colloid Interfac. Sci., 1986, 110, 591.
21.
G.S. Caravajal, D.E. Leyden and G.E. Maciel, in 'Silanes, Surfaces and Interfaces', D.E. Leyden ed., Gordon & Breach Sci. Publ., New York, 1985, 283.
22.
G.S. Caravajal, D.E. Leyden, G.R. Quinting and G.E. Maciel, Anal. Chem., 1988, 60, 1776.
23.
E.J.R. Sudh61ter, R. Huis, G.R. Hays and N.C.M. Alma, J. Colloid Interface Sci., 1985, 103, 554.
24.
T.L. Weeding, W.S. Veeman, L.W. Jenneskens, H. Angad Gaur, H.E.C. Schuurs, and W.G.B. Huysmans, Macromolecules, 1989, 22, 706.
25.
B. Pfleiderer, K. Albert, E. Bayer, L. Van De Ven, J. De Haan and C. Cramers, J. Phys. Chem., 1990, 94, 4189.
26.
K.C. Vrancken, K. Possemiers, P. Van Der Voort and E.F. Vansant, Coll. Surf., in print.
27.
F.D. Blum, W. Meesiri, H.-J. Kang and J.E. Gambogi, in Silanes and Other Coupling Agents, K.L. Mittal ed., VSP, Utrecht, The Netherlands, 1992, 181.
28.
K.C. Vrancken, P. Van Der Voort, K. Possemiers and E.F. Vansant, J. Colloid Interface Sci., 1995, 169; in print.
29.
T.G. Waddell, D.E. Leyden and M.T. DeBello, J. Am. Chem. Soc., 1981, 103, 5303.
30.
P. Van Der Voort, I. Gillis-D'Hamers, K.C. Vrancken, E.F. Vansant, J. Chem. Soc. Faraday Trans., 1991, 87, 3899.
31.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1976, 80, 1998.
32.
L.H. Dubois and B.R. Zegarski, J. Phys. Chem., 1993, 97, 1665.
33.
P. Van Der Voort, K.C. Vrancken, E.F. Vansant, J. Riga, J. Chem .Soc. Faraday Trans., 1993, 89(14), 2509.
294 34.
P. Fink, I. Plotski and G. Rudakoff, Wiss. Zeitschr. Friedrich-Schiller Univ. JenaNaturwiss., 1990, R-39, 217.
35.
K.C. Vrancken, P. Van Der Voort, K. Possemiers, P. Grobet and E.F. Vansant, in Chemically Modified Surfaces, J.J. Pesek and I.E. Leigh eds., The Royal Society of Chemistry, Cambridge, UK, 1994, 46.
36.
S. Kondo, T. Ishikawa, N. Yamagami, K. Yoshioka and Y. Nakahara, Bull. Chem. Soc. Jpn., 1987, 60, 95.
37.
for method see also chapter 9 paragraph 1.4.1.
38.
H.R. Anderson, F.M. Fowkes and F.H. Hielscher, J. Polym. Sci., Polym. Phys. Ed., 1976, 14, 879.
39.
P.R. Moses, L.M. Wier, J.C. Lennox, H.O. Finklea, J.R. Lenhard and R.W. Murray, Anal. Chem., 1978, 50, 576.
40.
D.E. Leyden, D.S. Kendall and T.G. Waddell, Anal. Chim. Acta, 1981, 126, 207.
41.
H. Linde, J. Appl. Polym. Sci., 1990, 40, 613.
42.
S.R. Culler, H. Ishida and J.L. Koenig, Appl. Spectrosc., 1984, 38, 1.
43.
S.R. Culler, S. Naviroj, H. Ishida and J.L. Koenig, J. Colloid Interface Sci., 1983, 96, 69.
44.
K.P. Battjes, A.M. Barolo and P. Dreyfuss, J. Adhesion Sci. Technol., 1991, 5, 785.
45.
K.C. Vrancken, L. De Coster, P. Van Der Voort, P.J. Grobet and E.F. Vansant, J. Colloid Interface Sci., 1995, in print.
46.
B.A. Morrow and A.J. McFarlan, J. Phys. Chem., 1992, 96, 1395.
47.
A.A. Chuiko, V.A. Tertykh, V.A. Khranovskii, Yu.P. Egorov and L.M. Roev, Teoreticheskaya i Eksperimental'naya Khimiya, 1966, 2, 247; Engl. transl., p. 189.
48.
M.J.D. Low, A.G. Severdia and J. Chan, J. Colloid Interface Sci., 1982, 86, 111.
49.
M.L. Hair and W. Hertl, J. Phys. Chem., 1969, 73, 2372.
50.
M.L. Hair and W. Hertl, J. Phys. Chem., 1973, 77, 2070.
51.
M.L. Hair and W. Hertl, J. Phys. Chem., 1973, 77, 1965.
52.
M.L. Hair, J. Colloid Interface Sci., 1977, 60, 154.
53.
M.L. Hair, Silica surfaces, in Silanes, Surfaces and Interfaces, ed. D.E. Leyden, Gordon and Breach science publishers, 1985.
295 54.
B. Evans and T.E. White, J. Catal., 1968, 11,336.
55.
P. Van Der Voort, K.C. Vranken and E.F. Vansant, J. Chem. Soc. Faraday Trans., 1995, in press.
56.
P.H. Emmet and S. Brunauer, J. Am. Chem. Soc., 1937, 59, 1553.
57.
B.A. Morrow, I.A. Cody and L.S.M. Lee, J. Phys. Chem., 1976, 80, 2692.
58.
G.E. Maciel and D.W. Sindorf, J. Am. Chem. Soc., 1983, 105, 3767.
59.
V. Davydov, A.T. Zhuravlev, A.V. Kiselev, Zh. Fiz. Khim., 1964, 38, 2047.
60.
K. Yoshinaga, H. Yoshida, Y. Yamamoto, K. Takakura, J. Colloid Interface Sci., 1992, 153, 207.
61.
P. Van Der Voort, I. Gillis-D'Hamers and E.F. Vansant, J. Chem. Soc. Faraday Trans., 1990, 86, 3751.
62.
R. Maoz and J. Sagiv, J. Colloid Interface Sci., 1984, 100, 465.
63.
N. Tillman, A. Ulman and T. L. Penner, Langmuir, 1989, 5, 101.
4.
S.R. Wasserman, G.M. Whitesides, I.M. Tidswell, B.M. Ocko, P.S. Pershan and J.D. Axe, J. Am. Chem. Soc., 1989, 111, 5852.
65.
G.A. Carson and S. Granick, J. Appl. Polym. Sci., 1989, 37, 2767.
66.
M.D. Porter, T.B. Bright, D.L. Allara and C.E.D. Chirdsy, J. Am. Chem. Soc., 1987, 109, 3559.
67.
C.D. Bain and G.M. Whitesides, Science, 1988, 240, 62.
68.
P. Silberzan, L. lager, D. Ausserre and J.J. Benattar, Langmuir, 1991, 7, 1647.
69.
D.L. Angst and G.W. Simmons, Langmuir, 1991, 7, 2236.
70.
C.R. Kessel and S. Granick, Langmuir, 1991, 7, 532.
71.
J. Nawrocki and B.J. Buszewski, J. Chromatogr. Rev., 1988, 449, 1.
72.
J. Sagiv, J. Am. Chem. Soc., 1980, 102, 92.
73.
J.B. Peri, J. Phys. Chem., 1966, 70, 2937.
74.
C.G. Armistead and J.A. Hockey, Trans. Faraday Soc., 1967, 63, 2549.
75.
C.P. Tripp and M.L. Hair, Langmuir, 1991, 7, 923.
296 76.
D.W. Sindorf and G.E. Maciel, J. Am. Chem. Soc., 1981, 103, 4263.
77.
R.K. Gilpin and M.E. Gangola, J. Chromatogr. Sci., 1990, 28, 277.
78.
C.P. Tripp and M.L. Hair, Langmuir, 1992, 8, 1120.
79.
H. Ishida and J.L. Koenig, Applied Spectroscopy, 1978, 32, 462.
80.
H. Ishida and J.L. Koenig, Applied Spectroscopy, 1978, 32, 469.
81.
C.P. Tripp and M.L. Hair, Langmuir, 1992, 8, 1961.
82.
C.P. Tripp and M.L. Hair, Langmuir, 1992, 8, 1961.
83.
C.P. Tripp and M.L. Hair, J. Phys. Chem., 1993, 97, 5693.
84.
J.M. Kinkel and K.K. Unger, J. Chromatogr., 1984, 316, 193.
85.
J.P. Blitz, R.S. Murthy and D.E. Leyden, J. Colloid Interface Sci., 1988, 126, 387.
86.
P. Van Der Voort, I. Gillis-D'Hamers, K.C. Vrancken and E.F. Vansant, Ceramic Industries International, 1992, 102, 17.
87.
P. Van Der Voort, E.F. Vansant and J. Riga, Ceramic Industries International, 1993, 103, 17.
88.
V.A. Tertykh and L.A. Belyakova, Chemical reactions involving the silica surface, Naukova Dumka Publishers, Kiev, 1991. (in Russian)
89.
W. Hertl and M.L. Hair, J. Phys. Chem., 1968, 72, 4676.
90.
H. Engelhardt and P. Orth, J. Liq. Chrom., 1987, 10, 1999.
91.
D. Gorski, E. Klemm, P. Fink and H. H6rhold, J. Colloid Interface Sci., 1988, 126, 445.
92.
D.W. Sindorf and G.E. Maciel, J. Am. Chem. Soc., 1983, 105, 3767.
93.
E.I.R. Sudh61ter, R. Huis, G.R. Hays and N.C.M. Alma, J. Colloid Interface Sci., 1985, 103, 554.
94.
J.M.J. Vankan, J.J. Ponjee, J.W. De Haan and L.J.M. Van de Ven, J. Colloid Interface Sei., 1988, 126, 604.
95.
R. West, L.S. Whatley and K.J. Lake, J. Am. Chem. Soc., 1961, 83, 761.
96.
W. Noll, Angew. Chem., 1963, 2, 73.
297 97.
M.G. Voronkov, Chemistry and practical applications of organosilicon compounds, 1961, 6, 136. (in Russian)
98.
H. Kriegsmann, Z. Anorg. Algem. Chem., 1959, 299, 139.
This Page Intentionally Left Blank
299
Chapter 10
Modification with boron compounds
1 Introduction Boron containing molecules, such as B2H6, BCI3, and BF3, have often been used as modifying agents. Much of these efforts have been motivated by the possibility of developing new catalytic materials, based on the properties of the electron deficient boron atom. Researchers ~'2'3'4'5noticed that silicas, containing boron at their surface, interact much more strongly with organic bases, water, ammonia and all other molecules, capable of either proton- or electron - transfer. Therefore silicas modified with boron containing molecules could perhaps yield catalysts in which the active centres exist at the surface in a discrete array and possibly with determinable structure and coordination. 2 The greater part of the modification studies of silica with boron compounds has been directed towards achieving an understanding of the surface structure of silica and silica based adsorbents, utilizing the quantitative reactions with boron containing probe molecules. Because hydroxyl type specificity occurred in some reactions, boron compounds were used to make a distinction between isolated and vicinal surface hydroxyls. Diborane was even utilized as a probe to distinguish surface silanols from hydration water. 6'7 Besides, further reaction of the chemisorbed surface species with other compounds is also possible and interesting.
300 The following overview of the modification reactions of silica with boron compounds will be focused on B2H 6 and BX 3 (X -- C1 and F).
2 Modification with boronhalides
2.1 Reaction mechanisms
The boronhalides (BX3) are volatile, highly reactive, monomeric compounds, which show no detectable tendency to dimerize. In this they resemble organoboranes, BR3, but differ sharply from diborane, B2H 6. The molecules have a planar trigonal structure with the boron atom (sp2 - hybridization) in the centre. This class of compounds also exhibits partial double bond character due to p,- p~ interactions of the empty pz-orbital of boron and the filled pz-orbital of the halogen. Because of the incomplete octetstructure of the central boron atom, borontrihalides behave as strong Lewis acids. In the discussion no subdivision will be made for BC13 and B F 3. Both molecules are strong lewis acids and follow the same reaction mechanisms. BCI 3 may be taken as the more reactive species of the two. The B-F bond is stronger, due to a higher p~- p, interaction. This overlap of the p-orbitals is lost during the reaction stage, where a pyramidal shaped transition state is found. Since the stabilisation effect of the overlap is stronger for BF3, this molecule is more reluctant to react compared to BCI 3. BX 3 notations will be used in what follows to describe the reaction mechanisms. The specific halogen forms will be used where appropriate. In the past, several
studies 1'2'3'4'5'8'9'10'11'12'13'14'15'16'17'18'19'20 w e r e
devoted to the reaction
of BX3 with silica powders, pressed silica discs and porous glasses by means of physical and chemical methods. A fast reaction of the BX3-molecules with the oxygen atoms of surface silanols, surface siloxane groups and hydration water takes place, giving rise to chemisorbed surface species. Morrow and McFarlan 2~ followed the reaction of self supported silica discs (evacuated at 353 K) with a threefold excess of BCI 3 at room temperature using fast scanning FTIR spectroscopy. They reported that the modification reaction was already completed after 30 seconds, since they observed no further changes in the infrared spectra for longer reaction times. Hair and Hertl 4
301 also stated that reaction with BC13 is very rapid at all temperatures studied. Reactions went to completion after 2 - 10 minutes respectively for 133 and 13 Pa of BC13. All surface hydroxyls were modified. After B F 3 modification times as small as 10 seconds, Morrow and Devi ~4 reported the formation of significant B-O and B-F IRvibration bands from chemisorbed B F 3 in the spectra. One can state that reaction of silica with boronhalides is very rapid due to the reactivity of the compounds, which makes it difficult to study the course of the reaction. interacts mainly with the attainable surface hydroxyl groups. The chemisorbed species exist in two forms, according to the reactions (A) and (B).
BX 3
=- S i O H
+ B X 3 --,
--- S i O B X 2 + H X
(A)
2 --SiOH + BX3 --, ( m S i O ) 2 B X + 2 HX
(B)
Both monodentate (reaction (A)) and bidentate (reaction (B)) species are formed. Reactions (A) and (B) are not the only reactions yielding this surface species, as will be evidenced further in the text (reactions (C), (D) and (G)). The formation of tridentate surface species is sterically not possible. 21 The occurrence of the two reactions is based on different observations. Especially infrared spectroscopy has been used to study the modification reactions with borontrihalides. Much of the published infrared w o r k 1'2'4'8'11'21'22'23 has been concerned with the spectral changes in the OH stretching region (3800 - 3200 cml). Some authors ~'2~
focused their attention
on the spectral regions of the adsorbed species. Figure 10.1 shows the infrared spectra of Kieselgel 60 pretreated at 473 K before (a) and after (b) reaction with excess BC13 at 293 K. Spectrum (c) is the difference spectrum of (b) and (a). The disappearance of the bands at 3747 cm -1 and 3550 cm ~ clearly demonstrates that both free and bridged hydroxyl groups have reacted. While all the free hydroxyl groups have disappeared from the surface, still a residual band is seen in the region of the bridged surface hydroxyls. Besides reacting with the hydroxylic surface groups, BX3-molecules are known to react with siloxane groups. The reaction of B X 3 with siloxane groups was first witnessed by Morrow and Devi in 1971.14 Treating totally dehydroxylated (1523 K) samples with B F 3 they found IR-bands, characteristic for the monodentate surface groups. Furthermore they reported the formation of new silica bands at 888 and 908 cm ~ due
302
t~msity(a.u.)
4000
38'00
36'00 3i00 Wavenumber (can")
32'00
3000
Figure 10.1 FTIR-PA spectrum of the hydroxyl stretching bands of silica gel, (a) pretreated at 473 K, (b) reacted with BCl3 at 293 K. Spectrum (c) is the difference spectrum of (a) and (b) .
to the formation of 'reactive (strained) siloxane groups'. 25'26'27 Titrating the totally dehydroxylated surface with small doses of BF 3 led to an increasing reduction of the 888 and 908 cm ~ bands, indicative for the reaction with the surface siloxanes. For silica surfaces carrying also free silanol groups the intensity of the free hydroxyl band did not change. Only if larger quantities of BF3 were added the free hydroxyl band did disappear, showing the higher reactivity of the siloxanes towards BF 3 compared to the residual hydroxyl groups. This is true for strongly dehydroxylated silicas, evacuated at temperatures > 973 K, ~4while for hydroxylated surfaces reaction occurs predominantly with the silanol groups. The same results were found for BC13 .14'27 Figure 10.2 presents the IR-spectrum of Kieselgel 60 (pretreated at 973 K), treated with hexamethyldisilazane (HMDS) before and after B C I 3 treatment at 293 K. HMDS removes all hydroxyl groups, creating a hydrophobic silica surface (cfr. chapter 4). After reaction with B C I 3 at room temperature and degassing at 373 K to remove physisorbed species, a B-O vibration band is formed, while the bands of the methyl groups at 2900 cm ~ remained the same in intensity. This indicates that chemical bonds are formed with the surface by reaction with the siloxane surface groups. No bands were observed that could indicate any hydrolysis of the B C I 3. Chloride analysis of the HMDS-surface after BCI 3 modification led to a value of 0.05 mmol/g of
303
siloxane sites that reacted, assuming that three chloride groups are found on the surface when one siloxane group reacts (reaction (C)). This value has to be treated with care, however, since the HMDS treated surface is chemically different from the surface of pure silica. Furthermore the trimethylsilyl-groups on the HMDS surface are bulky groups, possibly hindering the BC13 molecules to react with all available siloxanes. - S i - O - S i ~-
+
BX 3
--~
-
Si-OBX2 + X-Si-
(C)
The reaction path under mild conditions is shown in reaction (C). At higher reaction temperatures of 670 - 720 K, the silicon-oxygen framework of silica is attacked by the strong Lewis acids BF3, BCI3, or A1CI3, accompanied with the elimination of SiF 4 or SIC14. These reaction products do not react with the silica structure, being too weak Lewis acids. 13
Intensity(a.u.)
a
4000
35'00
3000 2500 Wavenumber (era t )
20()0
15100
Figure 10.2 (a) FTIR-PA spectrum of silica gel, pretreated at 973 K, modified with HMDS, (a) after reaction with excess BCL3 at 293 K.
It is interesting to note however that reaction occurs with siloxane groups at room temperature for silica samples pretreated at high temperatures. If the so formed monodentate (see reaction (C)) groups are close enough to a residual SiOH group, reaction may occur with the formation of bidentate groups on the surface.
This
304 reaction was pointed out by Morrow and Devi ~4and possibly explains, together with the bifunctional reaction with geminal hydroxyls (see chapter 9), the occurrence of bidentate species at high evacuation temperatures. Since the C1/B ratio is 3 for a reacted siloxane group, the reaction with siloxanes can also account for the high CI/B ratio's ( > 2) found for the high pretreatment regions. 4
2.2. Influencingfactors 2.2.1 The effect of the pretreatment temperature Peglar et al. 2 stated that BC13 reacts completely with the free and bridged surface hydroxyls of self supported discs. They assigned the residual band at 3650 cm 1 to intraglobular hydroxyls, since the band could be reduced in intensity after heating and rehydrating the pure silica samples. Rhee and Basila 16 reported that a BF amodification of silica, pretreated at 773 K in vacuo, removes all hydroxyl groups. A more detailed study was made by Tyler et al. s They first exchanged the surface hydroxyls for deuteroxyls treating the silica discs with D20, followed by a reaction with BC13. D20 treatment exchanges all surface hydroxyls, giving rise to new IR bands at 2755 cm ~ and 2620 cm ~ assigned to the free and bridged deuteroxyls respectively. All deuteroxyls are removed from a silica, pretreated at 973 K and consequently reacted with BC13 at room temperature. Silicas pretreated at 298 K show a residual OD-band after BC13 reaction. In the latter case, all free deuteroxyls react, while some hydrogen bonded groups remain. This study proves that there are some surface hydroxyls available for interaction with D20, although they are unavailable for interaction with a larger molecule such as BC13. The authors stated, based on IR band integration of the OH/OD-regions: 'D20 can underestimate by as much as 100% the total concentration of bulk hydroxyls that are unavailable to absorbing molecules larger than D20, even those that are still relatively small in absolute terms, such as BCI 3 which is about 0.5 - 0.6 nm in diameter. Morrow and McFarlan 21 also reported the complete disappearance of the free hydroxyls and a residual band at 3660 cm ~ after modification of aerosil (pretreated at 423 K) with excess BC13 at 293 K. If a difference in reactivity exists between free and bridged hydroxyls, this might have a chemical origin or might arise from steric
305 factors either related to the proximity of the silanols or to the size of the reactant itself. After examining the reaction with several hydrogen sequesting agents Morrow and McFarlan concluded that three factors play a role in determining the number of nonisolated hydroxyls which react with a given product: (1)
The ability of the reactant to penetrate into regions of interparticle contact, that is, to reach the inaccessible silanols. The smaller the reactant, the more groups are modified.
(2)
The ability to react bifunctionaly, i.e. the ability of a reactant to react with two surface hydroxyls. This causes a higher degree of modification.
(3)
The capability of a chemisorbed species to prevent a neighbouring silanol from reacting for steric reasons. The bigger the chemisorbed group the less silanols can react.
The residual band after BCl3-modification of D-exchanged silica is the result of the different dimensions of BC13 and D20. D20 , being a smaller molecule, can penetrate better into the interparticle regions than BC13 and reacts with more surface hydroxyls. The disappearance of the hydroxyl bands in the IR-spectra shows that BCI 3 interacts with the surface hydroxyls, but the residual OD-band indicates that BCI 3 does not react with all surface hydroxyls at room temperature. All free hydroxyls react, while some hydrogen bonded groups remain on the surface. Together with the disappearance of the OH bands in the infrared spectra, the formation of the mono- and bidentate groups, following reactions (A) and (B), was shown through the formation of different BO and BX bands. After the reaction of BF 3 with cabosil, Morrow and Devi 14'17 noticed some remarkable changes in the IRspectrum. Apart from the changes in the OH-stretch region, several new bands were found, especially in the 1400- 1500 cm -~ region. Table 10.1 lists the wavenumbers of the different bands after modification.
306 Table 10.1 IR band assignment of BF3 treated cabosil according to Morrow and Devi 14'17
Pretreatment temperature
11BF3
I~
T_>973 K
1452 1393 685
1500 1448 710
BO stretch monodentate BF asym. stretch monodentate OBF2 bending monodentate
T < 973 K
1452 1409 1393 1341 685
1500 1485 1448 1387 710
BO stretch monodentate OBO stretch bidentate BF asym. stretch monodentate BF stretch bidentate OBF2 bending monodentate
Modifications were performed with BF 3 containing 11BF3 and I~
in a ratio of 4:1.
Exact band assignment was made with the use of isotopically pure BF 3 and performing reactions on 180 exchanged silica. A splitting of the free OH-band due to the Si16OH (3749 cm -1) and SilgOH (3738 cm -1) surface silanols was found after this 180 exchange. 17 Modification of this surface with 11BF3 (1~
split the band at 1452 cm -~
(1500 cm -1) into a doublet with a new band at 1434 cm 1 (1485 cm 1) (see figure 10.3). The bands at 1393 and 1448 cm -1 did not show any splitting. Furthermore the ratio of the intensity of the 1434/1452 cm -1 (1485/1500 cm -1) bands after modification was the same as that of the 3738/3749 cm ~ bands before reaction. And this for different degrees of 180 exchange. The splitting after 180 exchange into a low wavenumber band and the shift with isotopically different BF3, led to the obvious assignment of the 1452 cm 1 (1500 cm -1) band to a B-O vibration band (uas). With this study Morrow and Devi 14'17 came to different IR-band assignments than the earlier study of Rhee and Basila. After IR-investigation of the BF 3 modifications of silica and silica/alumina, Rhee and Basila assigned the band at respectively 1390 and 1394 cm ~ to a B-O vibration. Bermudez ~ made a low wavenumber infrared study of the BC13 modification of pressed silica discs. The infrared spectra for modified silicas pretreated at different temperatures were studied comparing the intensity of the infrared modes and this before and after controlled hydrolysis. Bermudez found an intense band at 1380 cm 1 after modification that appeared under all reaction conditions studied. He stated that this band was caused by a B-O bond, since a wide variety of materials containing the SiOB group yield a strong band between 1370 - 1400 cm ~.
307
Also Vycor glass, a porous silica glass containing 2 - 3 wt% B203, has the B-O stretching vibration at 1396 cm-1. 24 The B-O vibration band for BC13 treated silica absorbs at a lower frequency than the B-O band of BF 3 treated silica. The B-O vibration bands are not caused by any possible hydrolysis reaction taking place. Together with the liberation of HC1 - gas observed, 11 the formation of the B-O bands in the infrared spectra incontestably showed the modification of the hydroxyl groups forming SiOBClx- species.
0
0
u~ u'}
IOOJ
,o~ - '~o
,~oo'
cm-I
,,'oo
doo
Figure 10.3 Infra-red spectrum obtained after chemisorption of I~ on dehydrated CabO-Sil: (a) untreated silica, (b) ISO exchanged silica. The dotted line represents the background spectrum of silica before the BF3 chemisorption. Besides the occurrence of the B-O stretch, also B-X vibrations are seen. The B-C1 stretch of chemisorbed groups was said to occur at 924 cml. 1 Bermudez I assigned several bands in his work, but caution must be taken. Band assignment based on comparing intensities is difficult when bands are overlapping, distorted or are present against more intense bands of a sloping background. Also the environment of a chemisorbed molecule is different for room temperature and high temperature degassed silica samples. This can have an effect on the peak position and intensity. The assignment of BF vibration bands is easier, since the groups absorb at higher
308 frequencies. Absorption frequencies for different B-F and B-O bands of B F 3 modified silica are listed in table 10.1. Morrow and Devi ~4witnessed differences in the infrared spectra for silicas pretreated at different temperatures. For the low temperature range (T < 973 K) extra bands appear in the spectrum in comparison with higher temperatures (table 10.1). Morrow and Devi 14 found the intensity of these extra bands to decrease with increasing pretreatment temperature of the silica. The results were explained as the formation of different surface species under the different pretreatment conditions. Morrow and Devi proposed the formation of bidentate groups at the lower pretreatment temperatures of the silica, besides the monodentate form. They came to these conclusions since the observed number of IR-bands was compatible with the proposed surface species and because the transformation of the monodentate species into bidentate species was more rapid when the concentration of SiOH groups was high. 14 This observation not only showed the formation of both mono- and bidentate species, but also showed the relative occurrence of the two groups to differ with the evacuation conditions of the silica. At high pretreatment temperature the silanol surface density is low (at 973 K OtoH = 1.10H/nm2), so that B F 3 is predominantly bound as a monodentate specie (reaction A). Decreasing the pretreatment temperature, increases the silanol density on the pure silica surface, giving rise to an increasing amount of B F 3 that reacts to the bidentate form (reaction B). The B-O band of the bidentate group showed no splitting on 180 exchange. This was explained to be due to the complex splitting pattern, since two oxygen atoms are bound in the structure, that was undetected because the infrared bands were rather broad. Based on the IR-data interpretation of BC13 modified silicas, Bermudez 1 also proposed the existence and increased occurrence of bidentate groups on the modified surface with decreasing pretreatment temperature. The infrared based results, indicating the increased occurrence of monodentate groups with increasing degassing temperature of the silica, are confirmed by the chemical analysis of the modified surfaces. Boehm 9 found a decrease in the chloride and boron content of the modified surface with increasing pretreatment temperature. This is easily explained by the lesser availability of hydroxyls at higher pretreatment temperatures. However, the chloride to boron ratio was found to increase with pretreatment temperature. Boehm et al. 9 reported that the C1/B-ratio increases from
309
1.36 at 373 K to 1.8 at 823 K for BC13 modified Aerosil (table 10.2). The aerosil was degassed at 473 K for 40 hours to remove all adsorbed gases. The C1/B-ratio is two for the monodentate form and one for the bidentate form. The increase in the C1/Bratio with temperature proofs the higher concentration of the bidentate form at lower pretreatment temperatures. Table 10.2 compares some of our own results TM with the observations of Boehm. In analogy with the equations derived in chapter 9, the number of reacted hydroxyls can be calculated if the C1 and B contents are known.
Table 10.2 Variation of the C1/B-ratio after modification of silica gel with excess BCls Temperature
Boehm 9 Aerosil
Kieselgel 60
(K) 373 473 573 623 673 723 773 823 873 973
1.36 1.35 1.60 1.56 1.67 1.81 -
1.32 1.41 1.80 1.50 1.53 1.91
Boehm 9 made use of the analytical boron and chloride data to determine the surface hydroxyl concentration and to differentiate between isolated and vicinal hydroxyls. He suggested that BC13 reacts monofunctionally with the isolated hydroxyls, and bifunctionally with the vicinal silanols. Armistead and Tyler ~1 only used the chloride contents of BCI 3 modified silicas to come to a similar conclusion.
Several other
authors z'~2'2~ also suggested that vicinal and isolated hydroxyls react differently with BX 3. This was mainly grounded upon the fact that more bidentate groups are formed when more bridged hydroxyls are present on the silica surface.
310 Peglar 2 asserted that the O-O distance of the bidentate species has to be sufficiently small to form an unstrained hydrolytically stable specie. Assuming sp2-hybridization around boron, this distance is approximately 0.25 to 0.3 nm, being of the same order as the O-O distance for bridged surface hydroxyls. The O-O distance for free hydroxyls is of the order of 0.5 nm and thus too large to form stable bidentate species. 2 Hambleton ~2proposed therefore that clearly most free hydroxyls will give monodentate groups, while bidentate groups may be expected to occur at bridged hydroxyls. Using step scan FTIR, Morrow and McFarlan 21 were able to record IR-spectra every 2 seconds during BC13 chemisorption on the silica surface. As seen in difference spectra in figure 10.4 both free and bridged groups react with BC13 after very short reaction times. This non-specificity of the modifying agent is due to the high reactivity of the molecule. Compared to TiCI4, A1Me3 and SiCI4, BC13 reacts faster and more extensive with the bridged hydroxyl sites. The faster initial reaction with bridged sites for BC13 was explained by the authors as the ability of BCI 3 to react bifunctionally with the vicinal groups. A1Me 3 has the same size as BC13, but reacts only monofunctionally. 21 Although TiC14 has the ability to react bifunctionally, fewer bridged groups are modified at the end of the reaction compared to BCI 3 because of the larger size of TiC14. Morrow and McFarlan suggested that the experimental observation of a reactant reacting initially relatively more rapidly with vicinal groups can possibly be used as evidence for a bifunctional reaction. Based upon several independent observations, 2'4'8'9'11'18'21'22BCI 3 reacts both with free and bridged hydroxyl groups. A total modification is achieved for the free groups, while residual bridged groups are found after modification of the silica surface. The bridged groups react preferentially to yield bidentate groups, the free hydroxyls reacting to monodentate species. A total modification is only obtained for pretreatment temperatures > 773 K.
311
L.
o o o
_ .
~0 i
~ . _ _ .
o !
Intervol
J
o t
o
~e.
1
2 3
Figure 10. 4 Difference spectra of the modification reaction of Aerosil with different hydrogen sequesting agents, showing the spectral changes in the first three time intervals. Spectra were taken every 2 seconds for BCI~, thus difference spectrum 1 corresponds to the 0 s minus the 2 s spectrum.
2.2.2 The effect of the reaction temperature Not only the pretreatment temperature, but also the reaction temperature plays a decisive role in the optimization process. Figure 10.5 shows the C1/B-ratio of silica pretreated at 473 K modified with excess BC13 at different temperatures. A linear decrease of the C1/B-ratio with reaction temperature is found, indicating the increased formation of bidentate and the disappearance of monodentate groups with higher temperature. TM At higher temperature, obviously the reacting species have a higher thermal agitation, enabling one BC13 molecule to react with more hydroxyls. At 273 K the C1/B-ratio equals 2, indicating only monodentate groups being present.
312 As can be inferred from figure 10.5, the reaction products after BC13 modification are strongly dependent upon the reaction temperature used. Both types of hydroxyl groups react strongly with BC13 after very short reaction times at room temperature. Based on the C1/B-ratio for this reaction temperature most of the reacted BC13 molecules are bound as monodentate groups. Therefore the formation of different surface species is mainly caused by the availability of the hydroxyls for a reacting BX3 molecule and not by a difference in reactivity of the hydroxyls towards BX 3. At higher temperatures the surface species have a higher thermal agitation, enabling more bidentate species. Indeed although the infrared and analytical results indicate that bridged groups give rise to bidentate forms, no evidence has been given until now that all bridged groups react in this way. The infrared studies were performed at room temperature and higher temperatures, enabling a BX 3 molecule to react bifunctionally and yielding bidentate species after reaction with bridged groups.
CI/B
2.25
1.75
1.25 250
I
300
I
I
350 400 Temperature (K)
I
450
500
Figure 10.5 Cl/B-ratio of the silica surface, pretreated at 473 K, treated with BCI3 at different temperatures. 2.2.3 The effect of the degassing temperature The formation of the bidentate species can occur in two ways as shown in reactions (B) and (D). 1'11'14'15'19 Reaction (B) is the direct modification reaction of a BX 3 with two surface hydroxyls, reaction (D) is a consecutive reaction of a monodentate specie
313 with a n o n - reacted hydroxyl group. Reaction (B) will occur when the hydroxyl group density on the silica surface is high enough. The occurrence of reaction (D) is seen when a B X 3 modified silica, having residual hydroxyl groups at its surface, is thermally treated after reaction. Figure 10.6 shows the results for KG 60 pretreated at 473 K and treated with e x c e s s BC13 at 293 K. TM After reaction the residual gases were frozen out and the surface was thermally treated under vacuum. Increasing the post reaction curing temperature above the modification temperature, leads to a decrease in the chloride content of the surface, while HC1 gas is formed. The boron concentration remains the same, giving rise to the decreasing C1/B-ratio. IRinvestigation shows a decrease in intensity of the hydroxyl band with curing temperature, indicating the appearance of reaction (D). The reaction proceeds faster at high temperature and reaches completion after some time. The higher the temperature, the more extra hydroxyls are removed due to a higher thermal agitation of the species on the modified surface. More bidentate groups are formed at higher temperature. TM These experimental results show the degassing temperature to be an influencing factor for the surface species to be formed. 2 --Si-OH
+ BX3
-- S i - O H - Si-O-BX 2 +
--,
+ B X 3 --,
(-Si-O)2B-X
+ 2 HX
- Si-O-BX 2 + HX
- S i - O H --,
(- Si-O)2B-X
fB) (D)
+ HX
2.2.4 The effect of the transition-state Armistead and Hockey 15 used the consecutive reaction path (D) to explain the difference in reactivity and selectivity of B X 3 and methylchlorosilanes towards hydroxyl groups. Methylchlorosilanes reacting predominantly with free hydroxyl groups, and only under more stringent reaction conditions. B X 3 reacts with the hydroxyls through a transition state with sp3-hybridisation, where three of the valencies are occupied by the chlorine groups and one is satisfied by the oxygen atom of the reacting surface hydroxyl, namely the bond formed by the lone pair of oxygen. Elimination of HX from this complex yields the reaction product. The three chlorine atoms are held away from the
314
CIIB
1.8
1.6 1.4-
,all
,d,,
1.2 1.0 0
I
I
I
I
I
20
40
60
80
100
120
Time (min) Figure 10.6 Cl/B-ratio of the modified surface ( reaction temperature 293 K) evacuated at (a) 293 K, (b) 373 K and (c) 473 K.
surface in the transition state so allowing the B X 3 molecule to react easily with the single hydroxyls and also through the same mechanism consecutively with the hydrogen bonded hydroxyls. The selectivity of chlorosilanes towards surface hydroxyls on the other hand, comes forth from the sterical hindrance of methyl groups and silanols in the transition state. In the transition state a sp3d-hybridisation is formed. Elimination of HCI of the transition complex and reversion of the methylchlorosilane silicon atom to sp 3 yields a stable product bonded to the surface. Hindering the approach towards the oxygen atom of the hydrogen bonded surface groups, the methyl groups prevent the creation of the transition state and chlorosilanes show (almost) only reaction towards free silanols. Increasing substitution of the methyl groups for smaller groups decreases selectivity. ~5 Gorlov ~3 assigned the higher reactivity of BC13 towards hydroxyls compared to chlorosilanes to the much lower deformation energy of the molecule in the transition state.
315 2.2.5 The effect of the reaction time A detailed investigation of the progression of the BX3 modification of the silica surface is difficult, because of the short reaction times involved. The reaction is known to reach completion after 2 to 10 minutes. 4'14'21 Hair and Hertl 4'23 however performed a study of the BC13 reaction in function of the reaction time with the aid of infrared spectroscopy. Silica discs pretreated at 1073 K were used. The reaction was monitored measuring the rate of disappearance of the free hydroxyl band at 3747 cm -1. Comparing the intensity of the silanol band after a certain reaction time to the original value yielded the progression of the reaction. The reaction was said to end when the SiOH band disappeared completely. The obtained reaction curves were clearly sigmoidal in shape, i.e. the initial part of the reaction is very slow, it then increases in rate until the bulk of the hydroxyls have reacted and slows down again as the reaction comes to completion. This sigmoidal shape shows the BC13 modification of the silica surface to be an autocatalytic reaction. Since the gases in the system were pumped out between two infrared measurements and fresh BC13 was added, the autocatalytic effect showed to be due to changes on the surface, rather than to the gas phase constituents produced by the reaction. 4 The autocatalytic reaction indicates that adsorbed BC13 molecules increase the reactivity of the residual hydroxyl groups. Thus introducing sites of the Lewis acid type into the surface increases the activity of the (adjacent) Br6ndsted sites, in analogy to the activity of silica/alumina cracking catalysts. 23 The autocatalytic reaction curve is only observed at low reaction pressures. In order to obtain the sigmoidal curve Hair and Hertl 4 had to react the silica at 303 K using only 10 - 130 Pa of BC13. Increasing the pressure eliminates the sigmoidal shape, since at these higher pressures an abundance of BC13 molecules is present for reaction near the silica surface, covering up the autocatalytic effect of presorbed BC13.
2.3 Stability of the modified surface
The BX 3 modified surface has amply been studied on its hydrolytic stability. On admitting 1-120 to the surface, a fast hydrolysis occurs in which the halogen groups are removed from the surface and replaced with OH-groups. Hambleton et al. ~2reported that the reaction with D20 was rapid and went to completion within a few minutes.
316 After reaction a different surface was found containing two chemically distinct types of surface hydroxyls. The extreme sensitivity of BCI 3 to even traces of water has been noted by Bermudez. ~ He reported that reaction between BC13 and water adsorbed on the alkalihalide windows of an IR-cell complicated the spectra with B(OH)3 bands. Because of the H20 sensitivity, BX 3 modifications have to be carried out in reaction vessels which have previously been degassed thoroughly. The presence of hydration water on the silica surface has to be avoided by thermally pretreating the silica. It is important to notice that if a BX 3 modified silica, containing bulk hydroxyls (these are hydroxyls that are not available for reaction with BX3) , is heated above its pretreatment temperature, the bulk hydroxyls may condense, releasing water. As was proved by Armistead et al. ~1 for BCI 3 modified silica, this water hydrolyzes the halogen containing surface products thereby producing HC1 in the gas phase. The chemisorbed surface groups react differently with H20 as shown in reactions (E) and (F). The bidentate species are hydrolytically more stable and form a (SiO)2BOH group giving rise to an IR vibration band at 3700 cm ~ ol,2,4,12,14 The band was identified by Low and Ramasubramanian 3'~~as the stretching vibration of an OH group bonded to a boron atom. The bridged hydroxyl groups, removed by treatment with BCI 3 and forming bidentate groups, are not regenerated on hydrolysis. They are replaced by the single sharp band at 3700 cm~, ~2 while the free hydroxyl band is reformed. By heating silica to 723 K in vacuum prior to BC13 treatment, thereby removing most of the closely adjacent hydroxyl groups, the final --Si-O-BX 2 + 3 H20
~
-Si-OH
+ H3BO3 + 2 HX
(-- Si-O)2B-X + H20 -> (~- Si-O)2B-OH + HX
(E) (F)
hydrolysis product exhibits a much smaller B-OH peak compared to a 298 K pretreated sample. Titrating the BX3 modified surface with micromol doses of 1-120,2'~2each of which in themselves are to small in quantity to effect total hydrolysis, some marked changes are seen in the IR-spectra. First the band is created at 3700 cm -~, followed by the reformation of the free hydroxyl band and a broad peak around 3200 cm ~ from B(OH)3. The band at 3200 cm -1 only appeared on reformation of the free hydroxyl band. The same final surface state could be achieved by the addition of an immediate excess of H20 to the BX3 treated silica. 2 These observations led to the conclusion that monodentate groups, especially formed with free hydroxyls,
317 totally hydrolyse reforming the SiOH group and leaving B(OH)3 groups at the surface (reaction (F)). It further showed the bidentate group to be more sensitive to hydrolysis. The BOH peak is also found after the modification-hydrolysis cycle on silicas pretreated at temperatures of 973 K and higher. These silicas carry no hydrogen bonded groups on their surface. After hydrolyzing a BCl3-modified sample evacuated at 1073 K, Hair and Hertl 4'22 reported the formation of a BOH band at 3703 cm -~. The silanol peak at 3747 cm -~ reappeared with about half its original intensity. Repeating the BCI3/H20 reaction cycle several times yields the formation of a BOH peak having the same intensity as the free silanol band. Hair and Hertl stated that the 3747 cm ~ band on pure silica was composed of two types of groups, one group that is truly a free hydroxyl group and another pair of hydroxyl groups which are sufficiently close together to react bifunctionally although they are not hydrogen bonded. Camara et al. 2s came to the same conclusion. Spectra of aerosil evacuated at 973 K showed only free hydroxyl absorption bands in IR. The sample after treatment with BCI3, hydrolysis and re-evacuation at 873 K showed an intense peak at 3710 cm ~ assigned to the =BOH unit. Modifying the silica surface, containing =BOH and =SiOH groups, with hydrogen sequesting agents completely changed the reaction kinetics of these products compared to the pure silica surfaces. 4'22'23 The effect of the boron atoms was to increase the reactivity of the SiOH groups. For example C12SiMe2 reacts with the pure silica surface at 623 K and at 1.3 kPa reaction pressure following 1.6 - order kinetics. The boron containing surface reaction proceeded with first order kinetics at 30 Pa and 423 K. 23 The =BOH groups were shown to be more reactive than the silanol groups. 23 Bermudez ~ studied the BC13 modification of silica with IR, comparing modified high temperatures (HT) and room temperatures (RT) pretreated silicas before and after hydrolysis. Many of the bands in RT samples were also present in HT samples with more or less the same intensity. Among the peaks in HT/H20 spectra some were clearly not due to B(OH)3. Bermudez explained this by the presence of the geminal hydroxyl groups on the surface. These geminal sites can react bifunctionally with BC13, giving rise to the observed bands in the IR-spectra (see reaction H). The occurrence of geminal hydroxylic species on the silica surface was proved with the use of Si MAS NMR. 29 The formation of a bidentate specie through reaction with geminal
318 hydroxyls was shown to be geometrically possible. 9 Bermudez used the bidentate reaction with geminal groups to explain the discrepancy between the values of the free hydroxyl concentration measured after reaction with C H 3 O H (1.70H/nm2) 19'3~ and after reaction with BC13 under the assumption that only monodentate groups were formed (1.30H/nm2).
/ OH =Si
\ OH
/ O\ +BX 3
-
=Si
B-X
+2HX
(G)
"\ 0 /
Bermudez included also the possibility that bidentate groups can be formed via a secondary reaction of a monodentate group with neighbouring siloxane sites as for AI(Me)3. 2
Summary: Based on infrared analysis of the Si-OH, B-O and B-X regions and the quantitative elemental surface analysis of the modified silica surface, the main interaction of B X 3 with the silica surface was shown to be with the surface silanols. In these reactions chemically bound mono- and bidentate groups are formed, together with the liberation of HC1 gas. Both free and bridged hydroxyls react with B X 3. Apart from the reaction with the hydroxyls B X 3 also reacts with the siloxane bridges. A total modification of the hydroxyls is however only obtained at high pretreatment temperatures of the silica, where at lower pretreatment temperatures all free groups react, but bridged groups reside. The formation of the bidentate groups can proceed via different reaction mechanisms. The relative and absolute concentrations of the mono and bidentate groups is dependant upon reaction temperature, hydroxyl density and post-reaction curing temperatures. The B X 3 modified surface is extremely sensitive to hydrolysis. Upon hydrolysis the monodentate groups are removed from the surface, where the hydrolysed bidentate groups remain chemically bound to the surface.
319 3 Modification of the silica surface with diborane 3.1 Introduction
Diborane (b.p. = 181 K) is characterized by its atypical structure, consisting of two B H 3 groups linked together by two 3-centre-2-electron B-H-B bonds. Diborane is a mildly endothermic compound at room temperature and is known to decompose slowly on prolonged storage. 3~ Furthermore diborane is electron deficient, giving rise to its very important chemical properties. Indeed, diborane must be numbered among the most reactive compounds known. It certainly is one of the most versatile c h e m i c a l s . 31'32'33 Another factor which has a profound influence on the chemistry of B2H 6 is its exceptionally high affinity for fluorine, nitrogen and especially oxygen. The boron atoms will therefore often shed attached hydrogens or inorganic groups, if they can be replaced by oxygen in the form of e.g. hydroxy or alkoxy groups. The reaction of diborane with silicas, aluminosilicates and silica-alumina catalysts has already been reported in the 1950's. 6'7'34'35'36'37Especially the modification reaction of The reaction consists in silica has found much attention. 6'7'36,38,39,4~ hydrolysis of diborane to form hydrogen gas and a surface specie that contains a = S i - O - B = bond 6,39,40,41,47,48 A large variety of gases has been used to prove the reactivity of the hydroxylic sites on silica. Reactions of boron and silicon compounds, mostly halides and hydrides, were applied to determine its hydroxyl content. Also B2H 6 has been used to probe the silica surface. 6,7,35,36,46,49,50Besides information gained on the structural properties of the surface, the modification with B2H 6 also gave in some cases an augmentation of the catalytic activity of the substrate. For example, Weiss and S h a p i r o 49 reported the modification of the active sites of silica-alumina with B2H 6 to result in a marked increase of the catalyst towards cyclization of acetylene to benzene. In addition, the reaction yielded only one end product. The authors showed the BH2 substituted A1-OH groups to be the catalytic active sites. Mc Dowellet al. 5~ reported the use of B2H 6 - activated silica-alumina to synthesize 14C6H6, which is of sufficient purity to be used directly in counting natural ~4C by liquid scintillation spectrometry. Reaction of diborane with the hydroxylic species of the silica surface results in the hydrolysis of the molecule to form hydrogen gas. This reaction is expressed by the hydrolysis ratio, R, which is the ratio of hydrogen evolved in the reaction to diborane
320 consumed. The hydrolysis ratio has been used to study the diborane modification in a quantitative way. Based on a 1" 1 relationship between the amount of hydrogen gas liberated and surface hydroxyls reacted, the modification reaction was utilized to determine the silanol number of the surface. Because the R-value appeared to be different for hydration water and surface silanols, it was assumed that the hydrolysis ratio could be used to differentiate on a quantitative base between the two surface compounds. 6'7'36'46Furthermore attempts were made to distinguish between the relative portions of free and bridged hydroxyl groups, based on the hydrolysis ratio gained after reaction and a different R-value for both species. Hydrolysis reactions of diborane in the presence of gaseous B2H6 are characterized by a primary hydrolysis ratio, Rp,~m~y,while reactions occurring after excess diborane has been removed, are expressed by their secondary hydrolysis ratio, Rs~o,~y. The amount of hydrogen, produced in the secondary reactions, depends mainly on three parameters.
(1) (2) (3)
The amount of presorbed B2H6, which defines the potential amount of releasable 1-12 The reactivity of the residual hydroxyl groups on the surface. The R-value of the primary reactions. Since the maximum attainable hydrolysis ratio is 6, the R-value of the secondary reactions can only reach a value of (6 -
P~nm~)3.2 Reaction mechanism
3.2.1 Reaction with hydroxyls The first studies of the modification reaction of silica with diborane were performed by Weiss and Shapiro. 34'35'36'49 Their studies were focused on the nature and localization of the hydroxylic species in the silica. Reaction of untreated silica gel with B2H6 at room temperature gave a hydrolysis ratio of six, indicating total hydrolysis of the B2H6 molecules. The same situation is found when reacting diborane with excess water, 34 according to the reaction 9
321
BzH 6 -k 6 HzO
--, 6 H e + 2 H3BO 3
(R=6)
(It)
Lowering the reaction temperature to 193 K, decreases the R-value to 4, again the same as for water. This lower value was explained by the authors as the restricted motion of the water molecules at this low temperature, which inhibits the total hydrolysis of diborane. So in the presence of hydration water, the hydrolysis behaviour of B2H 6 was concluded to be the same as with the bulk liquid. When the silica gel was first heated to 423 K in vacuo, thus removing all adsorbed water, and modified with B2H 6 at 250 K and 383 K, the hydrolysis ratio came to respectively 2 and 2.8. Naccache et al. 7 found a hydrolysis ratio of 2 and 2.2 respectively for reaction temperatures of 258 and 323 K. These R-values, smaller than 6, indicate that the (larger part of the) 'bound water' is present in the form of hydroxyl groups linked to surface Si atoms and that upon hydrolysis of B2H 6 the silicon atoms served as anchor points to prevent further migration of the partially hydrolysed borine. 36 The R-value equalling two was explained by Weiss and Shapiro 34'36 by reaction (I). - S i - O H + 1/2 B,H6 --, ~- Si-O-BHz + 1-12
(R=2)
(I)
The authors assumed as for the hydrolyzation reaction with free water, the symmetrical cleavage of diborane in two BH 3 species, each of them reacting with a surface hydroxyl. Performing the B2H 6 modification at reaction temperatures above 250 K, resulted in hydrolysis ratios greater than t w o . 35'36 Weiss and Shapiro 35 explained the Rprimary-values varying between 2 and 4 at higher reaction temperatures as the formation of another surface specie. The diborane molecule reacts with two surface hydroxyls forming a bidentate group. The reaction will especially take place for the low pretreatment temperatures of the silica gel, where surface hydroxyl concentration still is high. As can be seen in reaction (J), the hydrolysis ratio of the reaction equals four explaining the higher R-values. The formation of the same two surface species were also suggested by other authors. 39'4~ When silica modified with BzH 6 at 250 K was heated, the hydrolysis ratio increased with temperature by the liberation of extra H 2 gas. 36 The results are listed in table 10.3. 2 - S i - O H + 1/2 B , H 6 ~
( - S i - O ) , B - H + 2 1-12
(R=4)
(J)
322 Table 10.3 Silica gel evacuated at 423 K in vacuo and modified at 250 K with excess B2H 6 and heated after reaction to different temperatures 36
Temperature (K) 250 273 298 333 374 409 698 763
R-value = Rp~m. + R~. 2.00 2.20 3.12 3.74 3.96 4.36 6.07 6.07
Augmenting the temperature to 373 K, resulted in an increase of the R-value from 2 to 4. The formation of the extra hydrogen gas showed the presence of hydroxylic groups, unavailable for reaction at lower temperature, but accessible at the higher temperatures. Shapiro and Weiss 36 concluded that this thermally induced hydrolysis can be caused by reaction with surface hydroxyl groups or by diffusion of bulk water to the surface. The latter is created at high temperatures, when bulk hydroxyls condensate and the water formed diffundates to the surface. This bulk water accounts for the R-values equalling 6 at temperatures exceeding the pretreatment temperature of the silica gel, which presumes a total hydrolysis of the surface boronhydride groups. The formation of tridentate groups, groups formed through reaction of a borine group with three surface hydroxyls, is sterically not likely. 2~ The liberation of extra hydrogen gas when treating the 250 K modified surface at higher temperatures, can also account for the formation of extra bidentate groups. Gillis-D'Hamers as'aT showed the quantity of hydrogen gas formed in the secondary reactions to be linear with the applied reaction temperature, as can be seen in figure 10.7. The silicas were modified at 338 K for 17 hours with diborane. Irrespective of the pretreatment temperature, a high correlation was found, the positive slope reflecting the increased reactivity of the hydroxyl groups with temperature. Both reaction temperature as the post-reaction curing temperature affect the relative presence of the different reactions. Naccache and ImelikTM reported that the hydrolysis ratio decreases with increasing pretreatment temperature of the silica. The reactions were performed at 323 K and hydrolysis ratios approximating 2 were found at pretreatment temperatures of 423 to
323
H 2 evolution (mmol/g)
_,___._____-----.
0
...._....o-- - - - - - - - - - w - ~
I
323
423
I
I
I
523 623 723 Reaction temperature (K)
823
Figure 10. 7 t12 production vs. reaction temperature according to Gillis-D'Hamers ( 473K; 0 5 73 K," e2 6 73 K).
573 K. At higher pretreatment temperature the hydrolysis ratio equalled 1. The authors explained this by a variation in the nature of the hydroxylic species with increasing pretreatment temperature or by the occurrence of different reaction mechanisms. In combination with infrared analysis of the modified silica surface and comparing the boranation reaction to the behaviour of minerals containing crystallization water, Mathieu et al. put forward some possible reaction paths for the silicas treated at high and low temperatures. 6'7'46 These reactions however were contradicted by consequent studies. 39'42'48 Bavarez and Bastick39 and Fripiat et al. 42 also reported a decreasing hydrolysis ratio when raising the pretreatment temperature. Reactions were performed at room temperature. The results of Bavarez and Bastick are shown in table 10.4. Bavarez stated that the continuous variation of the R-value with the pretreatment temperature of the silica indicated that several reactions, each with their own hydrolysis ratio, occurred simultaneously. The diminishing R-value indicates that the relative occurrence of the different mechanisms changes with increasing pretreatment temperature. Since the hydrolysis ratio varied from 3.70 to 0.57 on raising the pretreatment temperature of the silica from 473 K to 1173 K (see table 10.4), Bavarez and Bastick 39 supposed at least two reactions taking place with R-values of 4 and 0. A hydrolysis ratio of zero indicates a reaction where diborane is sorbed to the silica surface, without the liberation of H2-gas. So at least one other
324
Table 10.4 Influence of pretreatment temperature of silica gel on the B2H6 chemisorption39 T
473 573 673 773 873 973 1073 1173
R avg
B2H6 consumed mmol/g
nD~xp mmol/g
OH (TGA) mmol/g
OH (B2H6) mmol/g
% OH accessible to B2H6
3.70 2.80 2.11 1.74 1.35 0.97 0.80 0.57
0.89 1.11 1.21 1.15 1.11 1.25 1.24 1.19
0.07 0.34 0.56 0.63 0.68 0.75 0.81 0.85
4.67 3.89 2.92 2.12 1.57 1.24 0.97 0.70
3.27 3.10 2.62 1.99 1.50 1.21 0.99 0.68
70 79.5 90 94 96 98 100 99
Infrared spectroscopy has often been used as a tool in studying boranation reactions. An infrared study of the B2H6 modification reaction of silica gel performed by Bavarez et al., 39 indicated the disappearance of the hydroxyl bands in the 3800 - 3200 cm ~ region. This observation proved the reaction of B2H6 with the surface hydroxylic groups. Also a B-O vibration band around 1375 cm ~ was seen. These observations were confirmed by Maschenko, 48 Morrow, 4~ and Gillis-D'Hamers. 4~ Figure 10.8 shows the FTIR spectra of silica gel, degassed at 500 K for 17 hours and treated with B2H6 at 338 K, with different amounts of diborane chemisorbed. 4~ As the number of sorbed borane groups increases, the intensity of the free hydroxyl group vibration gradually diminishes and finally disappears at a chemisorption degree greater than 1 mmol/g silica. A residual band was found after completion of the reaction, of which the intensity decreases with increasing pretreatment temperature. 39'4~ The band disappears completely around 873
K . 39
The presence of the residual band indicates
that at pretreatment temperatures < 873 K there are some hydroxylic groups not available for reaction.
Their number decreases with increasing pretreatment and
reaction temperature. The origin of the groups is either on the surface or in the bulk structure of the silica.
The isolated hydroxyl groups are totally modified for all
pretreatment temperatures, as indicated by the complete disappearance of their infrared band at 3747 cm ~ after reaction. 39'47 Besides the intensity decrease of the hydroxyl bands, the formation of B-H and B-O vibration bands of chemisorbed groups are seen in the infrared spectra, confirming the
325
•/
4000.0
..-)
3600.0
wavenumber/cm-'
3200.0
Figure 10.8 FTIR-PA spectrum of silica gel degassed at 493 K and treated with B2H6. Amount of chemisorbed B2H6 (mmol/g): (a) 0.00, (b) 0.25, (c) O.64, (d) O. 75, (e) 0.97, 09 1.00 and (g) 1.21. creation of various surface species. Morrow and McFarlan 4~ performed an infrared study utilizing isotopically pure diborane, and found several B-O bands after boranation of the silica surface. The wavenumber of the B-O groups changed with the pretreatment temperature of the silica. The authors witnessed the occurrence of B-O bands at 1356 cm -~ (~B) and 1386 cm -~ (~~ for a modified silica evacuated at 773 K, while for a pretreatment temperature of 423 K two extra bands were seen in the 1380 -1400 cm ~ region. Fripiat reported the formation of a broad band at 1375 cm -~ regardless the pretreatment temperature of the silica. 42 Since the B-O stretching vibration of BOH containing species absorbs at significantly higher wavenumbers ( _ 1450 cml), the band was assigned to the B-O stretching vibration of surface -OBHx species. In his infrared study of the B2H 6 treatment of silica gels Mashenko 48 also indicated the presence of different B-O bands. Morrow et al. 4~ assigned the extra bands observed for the lower pretreatment temperatures to the B-O vibrations of a bidentate specie. This assignment was based on the fact that (a) boroxine, a cyclic B-O containing molecule, also gives the B-O band in the 1380- 1400 cm ~ region and (b) the infrared bands of the surface group are more intense at a higher surface hydroxyl concentration. The various vibration bands thus confirm the occurrence of -OBH2 groups at all pretreatment temperatures of the silica gel, whereas -OBHOgroups are only found at low pretreatment temperatures. These infrared data therefore
326
confirm the occurrence of the two different reactions put forward by Shapiro et al. 35'36'49 based on the R-values found after reaction. Especially the B-H bonds, formed upon reaction with diborane, have been thoroughly investigated.
Because all the B-H vibrations frequencies essentially depend on the
electron structure of the corresponding compound, they can be used as a probe for the nature of compound formed. When the molecule only has the usual covalent B-H bonds, the absorption bands lie in the 2490 - 2630 cm -1 region of the spectrum. For molecules with a bridging structure, like diborane, there are three regions of adsorption: (1) 2490 - 2630 cm -~ for the terminal B-H bonds, (2) 1900 - 2200 cm ~ and (3) 1450- 1600 cm ~ for the three-centre bridging B-H-B bonds. In coordination compounds of boranes, the B-H stretching vibrations shift to lower wavenumbers and absorb in the 2270- 2440 cm -~ region. Boronhydride ions absorb only in the 2100 2300cm 1 region. So based on the observed wavenumbers, information is gained about the possible surface structures present after modification. Mathieu and I m e l i k 6'52 studied the B-H absorption bands, seen after the sorption of B2H 6 on silica, to gain more information on the formed surface species. Several bands
were reported in the 2500 cm -~ region, their wavenumbers changing with the evacuation temperature of the silica. To explain the quantitative data, the authors suggested the formation of BH2-O-BH2 groups, due to reaction with surface water, at low pretreatment temperatures. While for pretreatment temperatures >__ 673 K = Si-O- B2H5 groups were formed from reaction with surface silanols. Although the interpretation of the infrared data was questionable, different surface species were proved to occur on the surface depending on the degassing conditions of the silica. The presence of covalent B-H bonds was also seen. No bridging species were found. Bavarez and Bastick 39 were the first to make a thorough infrared analysis of the boranation of silica over the whole mid infrared region. A complex pattern of B-H stretching vibrations was observed, which the authors decided into three groups. Each group of infrared vibrations showed the same intensity variations on changing the reaction parameters and was ascribed to the same surface species. The exact band assignment was performed on comparing the spectrum with that of gaseous diborane. The different groups, together with their infrared bands are shown in table 10.5. The first group consisted of only one vibration band at 2580 cm ~. It was the only B-H vibration observed that did not show any intensity variations upon degassing the
327
modified silica. Since only one band is seen, the 2580 cm -~ mode was connected with a surface group with only one B-H bond. The band was assigned to the B-H stretch of a bidentate group formed through reaction of a BH3 radical with two surface hydroxyls. Its intensity decreased when augmenting the pretreatment temperature of the silica, due to the decreased surface concentration of the hydroxyls. The second group includes three bands, which are formed when degassing the modified sample. On admitting fresh B2H6 they disappear again. The three infrared vibrations were assigned by the authors to the B-H bands of a BH 3 group sorbed at the surface. Because of the adsorption the symmetry of the group is lost, giving rise to three different B-H vibrations. The third group showed infrared bands of both covalent B-H bonds as bridged B-H-B bonds. The structure, having the IR bands of a diborane like structure, was assigned to a coordinatively bound B2H6 molecule. The wavenumbers were thought to be too high to be caused by a simply physically adsorbed specie. The vibration bands of the third group disappeared on degassing, while forming the bands of group two. Augmenting the degassing temperature results in the faster formation of the group 2 bands. The infrared modes however are reformed when fresh B2H6 is added to the infrared cell. This complementary behaviour of the infrared bands of group two and three upon degassing, therefore was caused by the same surface group: its structure changes when degassing the sample.
328 Table 10.5
Group
IR - bands Bavate,z
Morrow
Bavate,z
Moxrow
Bastiek
MeFatlan
Bastiek
McFadan
&
1
3
Assigned bands &
&
2580
2585
B - H in
2560 2479 2452
2565 2479 2456
B - H in
2597 2520 2110 2050 1560 1535
2587 2516
&
H I
=Si- O - B - O - Si = =--Si \O / -Si
,=.1,
BH 3
Si - O - BH
B - H in
B/H"B
=S/Oi -' B2H 6
= S i - O B ~/ 5
Based on the variations observed in the B-H vibration bands, Bavarez and Bastick 39 proposed the following reactions" (1)
The reaction of B2H 6 with the hydroxyl groups resulting in the formation of hydrogen gas and a bidentate specie (reaction (J), R=4). This in agreement with the results obtained by Shapiro et al. 36
(2)
The linking of B2H 6 and B H 3 to surface oxygen atoms via a coordinative bond (R=0). Bavarez suggested the siloxane sites as the adsorption sites for the coordinatively bound diborane and borine, since, as he stated, these oxygen atoms are more negatively charged and have better donor properties. The study reported for the first time the possible interaction of the diborane molecule with the siloxane bridges of the silica surface.
These reactions, having an R-value of 0, could account for the low hydrolysis ratios that were found (table 10.4) at high pretreatment temperatures of the silica. Reaction (L) also explains the observed equilibrium existing between the two surface species upon evacuation. Following the desorption process volumetrically, Bavarez and
329
-Si -Si-O-Sis
+
B2H6
9
B2H6
(K)
-Si" -Si
-Si /
-Si
O -- B 2 H
6
,o==,,-~
/
O -- BH
3
+
BH 3
(L)
-Si
(L) also explains the observed equilibrium existing between the two surface species upon evacuation. Following the desorption process volumetrically, Bavarez and Bastick39 noticed the gas desorbing in two stages: a fast desorption is seen in the first half hour, defined as the desorption of physically adsorbed species. This phase is followed by a stage in which another amount (riD) is liberated slowly over a period of 48 hours, explained as the equilibration reaction (reaction (L)). With the use of the experimental hydrolysis ratios found, the amount of diborane able to desorb (nD) according to reaction (L) was calculated. 39 For pretreatment temperatures below 773 K a good agreement was found (within a 5 % error margin) with the experimental value (see table 10.4). At higher pretreatment temperatures the experimentally obtained results differed from the calculated amount, which the authors explained as the occurrence of yet another reaction with a hydrolysis ratio equalling 2 (reaction (I)). At higher pretreatment temperatures the surface hydroxyl concentration becomes lower, preventing a B2H 6 molecule to react with two surface species. B a v a r e z 39 stated that reaction (J) is gradually replaced by reaction (I) as the pretreatment temperature rises above 773 K. This is in agreement with the data obtained by the interpretation of the B-O bands and the quantitative volumetric study of the boranation reaction, that indicate a decreased amount of bidentate species at higher evacuation temperatures of the silica. Bavarez and Bastick39 did not investigate the modified surfaces for pretreatment temperatures above 773 K spectroscopically and could not attribute the IR-bands of the-OBH2 species.
330 Morrow and McFarlan 4~ performed an infrared study over the total pretreatment temperature region. Reactions with diborane were performed at 373 K. Morrow reported approximately the same wavenumber values as Bavarez and Bastick39 for the B-H vibrations of 11B2H 6 modified silicas. The bands were grouped into the same three classes. The infrared bands after modification with isotopically pure B2H 6 c a n be found in this publication. 4~ The same author also noticed the existence of the equilibration process. Infrared spectra taken with diborane gas still present in the reaction cell, showed the existence of B-H-B groups linked to the silica surface. The bands disappeared when degassing the sample at 373 K for 30 minutes. Based on the observed infrared vibrations, the authors suggested the existence of two reactions. Table 10.5 lists the interpretation of Morrow compared to those of Bavarez and Bastick. 39 One reaction is the formation of a bidentate group according to reaction (J). The second reaction is the formation of monodentate groups, where these surface species are equally responsible for the equilibrium process witnessed (see reactions (M) and (N). - Si-OH
+
- Si-O-B2H
BzH 6 ~
5
~
- Si-O-B2H
-- Si-O-BH2
+
s -k H 2
BH a
(M) (Nt
Morrow et al. concluded that the reactions put forward by Bavarez and Bastick were not able to explain the observed infrared bands. First of all the absorption bands of coordinated diborane and borine groups absorb near 2300 cm l, not in the 2 5 0 0 2600 cm 1 region. Secondly Morrow showed that the infrared band at 2475 cm -1 is not caused by a B-H group vibration. Bavarez and Bastick 39 ascribed the three bands observed at 2560, 2475 and 2452 cm l to a B H 3 group, having lost its symmetry by adsorption at the surface. Using ~80 exchanged silica, Morrow et al. showed that the band at 2475 cm 1 originates from a combination band of the SiOB stretching mode at 1386 cm ~ and the in plane deformation mode (1089 cm 1) of a BH2 group. Exchanging the surface with ~80 groups, destroys the Fermi resonance existing for the 160 case and only one band is observed near 2470 cm l. This indicates the assignment of the vibrations of group two by Morrow to be just. The bands of the monodentate group are predominantly formed at high pretreatment temperature of the silica. This in agreement with the conclusion of Bavarez and Bastick that the bidentates are gradually replaced by the monodentate groups.
Reaction (N) also explains the
331
increasing amount of B2H 6 being desorbed (nD) when increasing the pretreatment temperature of the silica, due to a higher relative concentration of the monodentates with respect to the bidentate form. The reaction of B2H 6 with the surface hydroxylic groups can however not explain the hydrolysis ratios smaller than 2 observed by Fripiat 42 and Bavarez and Bastick39 at high pretreatment temperatures. Apart from the reactions with the hydroxyls, Morrow 4~ showed BzH 6 to react with the siloxane sites. The reaction consists in the formation of a Si-H group and a surface Si-O-BH2 group. The reaction has a hydrolysis ratio of zero. A gradual increase of the reaction with R =0, compared to the other reactions with R > 1, therefore can cause a decreasing Rto~ value, that eventually may become smaller than one. Morrow's results 4~ matched those of Mashenko. 48 Mashenko explained the infrared spectra as being due to the vibrations of -SiOBH2 and (----SiO)2BH species, changing relatively in concentration with degassing temperature of the silica. The bidentate group is mainly formed at high hydroxyl concentrations. The author also noticed the reaction with the siloxane groups, forming Si-H bands, besides reaction with the surface hydroxylic species. He reported bands characteristic for boron coordination compounds or bridging structures being absent. The infrared spectra were taken after degassing the sample at room temperature. So interpretation of the B-H vibration bands confirmed the different reactions of diborane with the hydroxylic surface groups put forward by Shapiro and Weiss. 35'36'49 In addition the infrared analysis of the modified surface revealed (1) 9 the reaction of BzH 6 with siloxanes, explaining the low hydrolysis ratios found under certain reaction conditions and (2) the equilibration reaction existing between BzH 6 in the gas phase and the monodentate groups formed through reaction of BH3 with a surface hydroxyl. 3.2.2 Reactions with siloxanes As was pointed out above, B2H 6 reacts, apart from the hydroxyl groups also with the siloxane sites in a chemical way. The reaction consists in the formation of a Si-H group and a surface SiOBH2 group, as shown in reaction (O). Mashenko 48 was the first to report this reaction, based on infrared analysis of the modified surface. A band at 2300 cm -~ was found which was clearly no typical =B-H band.
The
332
frequency is too low for a covalent B-H bond and the intensity of the band increases greatly during thermal treatment of the modified surface, whereas the intensity of the B-H bands in the 2400 - 2600 cm ~ region decreases to zero. Since the band shifted to 1670 cm ~ treating the sample with D2 at 573 K, the mode was assigned to a hydrogen containing surface group. Gillis-D'Hamers 4~ also noticed, besides the B-H bands at 2400 - 2600 cm -~, the formation of a band at 2270 cm ~ when performing the B2H6 modification reaction. The bands were formed at all pretreatment temperatures. Comparison with IR-library data and literature of similar work, suggested the formation of a Si-H surface compound. So it was assumed that a side reaction occurred on the silica surface, resulting in the formation of the silane bond. Because no reaction of (di)borane with hydroxylic species is known to result in a silane bond, a siloxane bridge has to be the active site. Integrating the band together with the B-H band, showed that the integration of the B-H region as a function of the sorbed amount of diborane is not highly correlated with the integration of the silane band. As a result the reactive sites should be different. Morrow et al. 25'26'4~ noticed this band on silicas evacuated above 673 K. The modification reactions were performed under reaction conditions where little reaction with hydroxyls occurred, as was seen from the very weak B-H bands. Instead, a strong Si-H band at 2283 cm ~ was generated and the bands due to the siloxane species at 888 and 908 cm -~ decreased in intensity. The authors concluded that B2H 6 reacted with reactive siloxane sites following reaction (O). - Si-O-Si-
+ 1/2
B2H 6
--->
-
Si-H + B H 2 - O - S i - -
(0)
Kinetic experiments of the diborane chemisorption confirm this reaction, since the best fit is obtained for a bimolecular competitive reaction. The proof for the bimolecular competitive reaction is a linear relationship of the ratio of H2 evolved to the amount of Si-H as a function of reaction time. H2 is formed after reaction with the hydroxyls. The amount of Sill on the surface was obtained by integration of the Sill band at 2770 cm -~.
Figure 10.9 shows that the H2/SiH versus time indeed is linear.
Introducing a different number of siloxane groups, should affect the ratio H2/SiOH. The ratio decreases linearly with pretreatment temperature (figure 10.10) indicating that at higher pretreatment temperatures the boranation of siloxane sites are favoured, instead of the direct boranation of the hydroxyl groups.
333 ratio / a.u.
12 10
o0
5'
10
1'5
' 20
' 25
Time (min)
Figure 10.9 Ratio of evolved He and the integration of the silane band as a function of reaction time for diborane-treatment silica gel.
16
ratio./a.u.
12 o
o o
0
273
o
473 673 873 Temperature (K)
1073
Figure 10.10 Ratio of H 2 and the integration of the silane bands as a function of the pretreatment temperature of the diborane-treated silica gel.
334 Although the interaction of the diborane molecules with the surface siloxane bridges as proposed by Bavarez and Bastick 39 did not appear to account for the observed infrared bands, it suggested for the first time an interaction with these surface sites. However a quantitative volumetric analysis of the course of the diborane modification reaction of silica, led Fripiats 42 to the conclusion that a physical interaction between the diborane molecules and the siloxane groups must occur. Performing the modification at different reaction temperatures, he always witnessed an induction period of several minutes, after which the reaction started rapidly. The reaction was followed by measuring the evolved hydrogen gas in e x c e s s BzH 6. The induction time appeared longer for both hydrated and completely dehydrated samples, with a minimum observed for samples pretreated between 373 K and 773 K. It was as well noticed that the induction period increased markedly if the initial diborane pressure was augmented. Gillis-D'Hamers 44 however did not find a induction period in his kinetic study of the diborane chemisorption on silica gel. Fripiat et al. 42 explained the induction period as the adsorption of B2H 6 to the silica surface, thereby splitting into two B H 3 radicals. The B H 3 groups are adsorbed on the Lewis basic surface sites, the induction period being the time required to cover them with a given concentration in B H 3. After this concentration has reached an adequate value, a fast chemisorption reaction with e.g. the hydroxyl groups occurs. The reactions are shown in figure 10.11. This reaction path accounted for the higher induction time found when increasing the initial B2H 6 pressure. At higher pressure there is a greater degree of reformation of the adsorbed B H 3 radicals into B2H6, due to the reaction of the B H 3 groups with the gas phase constituents, with a possible desorption of the molecule. This slows down the covering process of the Lewis basic sites, resulting in a longer period before the adequate surface concentration is reached. The oxygen atoms of the surface siloxane groups were suggested as the Lewis basic sites. Outgassing the samples markedly generates a larger number of siloxane bridges and consequently increases the surface concentration of B H 3 groups, provoking to some extent their recombination. This explains the lengthening of the induction period when outgassing the sample at temperatures > 773 K. The higher induction time for room temperature degassed samples was explained by the presence of hydration water. The presence of water molecules increased the induction time by covering some active sites. In the study on the increasing activation energy of chemisorption of the
335
diborane molecule with the degree of surface coverage, Gillis-D'Hamers found that the symmetrical cleavage of the physisorbed diborane requires an activation energy of 15 kJ/mol. 44
Summary The combination of the infrared results of the O-H, B-O and B-H region and the quantitative analysis of the B2H 6 reaction, evidences that the overall reaction starts with the adsorption of gaseous diborane at the surface, splitting symmetrically into two B H 3 radicals. These B H 3 groups are adsorbed onto the surface siloxane bridges. When a certain critical concentration of these groups is attained, they react with the active surface groups. The reaction can occur either with surface hydroxyl groups or with siloxane sites, depending on their surface concentration. At higher pretreatment temperatures the reaction with the siloxane groups gets more important. At surfaces treated at low evacuation temperatures (to remove the hydration water but to retain still a significant hydroxyl surface concentration) there is an increased probability of a B H 3 group simultaneously interacting with two neighbouring hydroxyl groups, thus forming a bidentate group. The presence of monodentate and bidentate groups is dependent both of reaction temperature and curing temperature. The occurring reactions are summarized in figure 10.11.
3.3 Hydroxyl specificity Shapiro and W e i s s 36 pointed out that the quantitative hydrogen gas data could be utilized to determine the amount of hydroxyl groups on the silica surface. They assumed that each hydroxyl group liberates one hydrogen gas molecule on reacting with B2H 6. So based on the 1" 1 relationship, the amount of hydrogen gas found equals the amount of hydroxyls reacted. This assumption was confirmed by the formation of HD gas if the modification reaction was performed with B2D 6 instead of B2H6 .35 When using B2H 6 to determine the hydroxyl concentration, it is important to make sure that the reaction with B2H 6 has come to an end before analyzing the reaction gases. 7 Since B2H 6 is known to react with hydration water, it is evenly important to avoid the presence of these groups. Shapiro and W e i s s 36 performed their calculations for
336
B2H6 t
B2 H6
---Si
~O~
Si-
/O~ ---Si
:
Si-
BH3 ~O~
-=Si
B2 H6
Si
~O~
S i -=
t
_Si ~ O ~
Si=
+
BH 3
=si~O~si~
~O~
- S i - OH
:
-=Si
~;i = + =Si'-~ 0 "--"Si_=
:
-=Si
O~si =
"Si= + - S i - O - B H
2+H
BH 3 -=Si
~0~
O
~Si-
+
=Si-O-BH2
H
BH3
--si
~i-
+ 2 =-Si-OH
~O-B:
I
-=Si
+ -Si - O
O. "Si-
+
2H 2
Figure 10.11 Overview of the occurring reactions when silica is treated with B:d~6. different pretreatment temperatures of the silica and compared the calculated amount of reacted hydroxyls with the hydroxyl concentration gained from thermogravimetric analysis of the untreated silica gel. For every water molecule lost in the thermogravimetric experiment, two hydroxyl groups were thought to disappear, according to the reaction: -= Si-OH + H O - S i - -
--, - S i - O - S i -
+ 1-120
(P)
For pretreatment temperatures of 773 K and 1023 K the thermogravimetric and 1-12data coincide, indicating that all hydroxylic groups are accessible to diborane to react. It also shows that at these temperatures all hydroxyl groups are on the surface. Naccache et al. 7 also included the possibility of a polymerization reaction of B2H6, activated by the thermally treated silica, to account for the decreasing R-value with pretreatment temperature. The polymerization is known to occur when increasing the reaction temperature to about 373 K 31'52 and produces hydrogen gas, thus subverting Shapiro's hypothesis that all H2 was formed through reaction of B2H 6 with hydroxylic groups. Mathieu 52 reported the presence of infrared bands indicating the formation of polyborane structures on the surface, when reaction temperature was increased to
337
423 K. Bavarez and Bastick 39 however showed through the use of B2D6and analyzing the gases formed during reaction chromatographically, that no polymerization reactions occur at room temperature. Therefore surface hydroxyl concentration determination with the use of B2H6, has to be performed at temperatures lower than 373 K and with reaction times being long enough to ensure total modification. Reacting the silica gel with B2H 6 at 323 K, Mathieu and Imelik6 described that 90% of the hydroxylic groups reacted for silica pretreated at 673 K, while 100% were modified for a pretreatment temperature of 1073 K. Table 10.4 lists the results of Bavarez and Bastick, 39 following the line of results of Shapiro and Mathieu. The results were confirmed by the infrared data, where the residual OH-stretch band of the unreacted hydroxyl groups was found as long as the volumetric data indicate no complete reaction. Performing the hydrolysis reaction at 338 K, Gillis-D'Hamers 3g'47 reported that the hydrogen gas evolved decreases linearly with pretreatment temperature, until a constant value is reached at about 873 K. The total amount of hydrogen evolved was said not to coincide with the total hydroxyl density, the silanol number being significantly larger compared to the number of reacted hydroxyls, especially in the low temperature region. Since the FTIR results indicated a total disappearance of the free hydroxyl groups and kinetic data showed the chemisorption to start preferentially on these free hydroxyls, 44 Gillis-D'Hamers et al. 38 studied a possible hydroxyl specificity of the B2H 6 chemisorption on the silica surface. The reaction with the silica surface was investigated by a volumetric monitoring of the H2 liberated both in primary and secondary reactions and comparing it with the concentration of the free and bridged hydroxyl groups. The absolute concentration of the two types of hydroxyl groups was calculated from the hydroxylic shift model presented in chapter 5. 53 Figure 10.12 gives the amount of hydrogen gas evolved per surface hydroxyl group as a function of pretreatment temperature. No functional relationship was found if the bridged hydroxyl groups were included in the calculations. The free hydroxyls however exhibited a 1" 1 relationship, except in the region of low pretreatment temperature, presuming diborane to chemisorb quantitatively on the free hydroxyl groups. The deviation of the 1" 1 relationship occurs in a region where the concentration of the bridged groups exceeds that of the free hydroxyls significantly. Therefore if the order of magnitude of the additional reactivity towards the bridged groups is related to the relative amount of the two hydroxyls, the authors supposed the ratio of bridged to free
338 groups to be a good measuring tool for the overall trend of the 1-12evolved. Indeed a linear relationship was found between the (OHb,dg~)/(OHf,D ratio and the H2/(OHf,~) ratio. This confirms the relative presence of the bridged hydroxyl groups to control the degree of hydroxyl specificity. At high concentration of the bridged groups the reaction with diborane is no longer hydroxyl specific, B2H6 reacting with the two hydroxyl types. At higher pretreatment temperature of the silica, starting from around 673 K, the reaction gets specific with the free hydrogen groups, both in the primary and secondary reactions. 10
ratio
8
+
373
4~/3
5~/3 6~/3 7273 8~/3 Pretreatment temperature (K)
9~73
1073
Figure 10.12 H2/[OHJ vs. pretreatment temperature; x: free ( 0 ) or bridged () hydroxyl groups.
When looking at figure 10.7 giving the H2-gas evolution during secondary reactions, the intercept at 338 K, the temperature of the primary reaction, reflects the fundamental characteristics of the silica surface with reference to the B2H6 chemisorpfion. A comparison plot between the free hydroxyl concentration and the constant was drawn, revealing a linear relation (figure 10.13). Exception had to be made for the values at 373 K and 473 K. At those evacuation temperatures the silica exhibits a large amount of bridged groups at the surface and consequently a deviation from the exclusive specificity has to be considered. Correction for this surplus effect yields a total linear relationship. Extrapolation of the data indicates that no chemisorption will occur in the absence of free hydroxyl groups on the surface. The absolute amount of reacted bridged hydroxyls in excess of the amount of free hydroxyl groups was thus shown to be proportional to the relative surface coverage of the two
339 hydroxyl types. These results explain the occurrence of the residual hydroxyl band found at 3650 cm -1 after boranating the silica surface. constant (primary reactions)/mmol g-1
4
o
3
o
2
1 o
0
0
1
2
3
4
[ O H f ~ ] / n m -2
Figure 10.13 Constant of the linear regression [H2(sec) vs. temperature] as a function of the hydroxyl density, corrected ( 9 and uncorrected ((>). 3.4 Kinetic study of the boranation reaction Several
a u t h o r s 7'34'36'42'44 studied
the boranation reaction of silica gel as a function of
reaction time. Most of the times the reaction was screened using the amount of H2gas liberated, under the assumption of a 1" 1 relationship of H2-evolved to hydroxyls reacted. Both B2H6-chemisorption and H2-evolution curves exhibit the same trend, differing only in absolute numbers. 36 The boranation proceeds very quickly in the first few minutes of the reaction, after which a much slower progress is reported. TM To obtain completion, reaction times often exceeding 24 hours are necessary. Fripiat and Van Tongelen 42 reported the occurrence of an induction period (as discussed before), varying in length with the reaction conditions, after which reaction starts. In their studies Fripiat and Van Tongelen made use of an excess B2H6 to keep the apparent reaction order with respect to the gaseous reagent at zero. Plotting the chemisorption data in reduced coordinates, the sorption process could be described with the following equation (1): c~ is the advancement degree of the reaction, given by the ratio of the amount of H2evolved at a time t and the amount at time t----cx. 7- represents the reduced time t/to.s.
340
a =
1:+0.25
(1)
The time t was considered from the end of the induction period, while to.s is the time necessary to reach or=0.8. The advancement of the H2-evolution (or) versus the reduced time (r) was found to fit a single function (Q), irrespective of the reaction temperature or pretreatment temperature, i.e. irrespective of the water or hydroxyl surface densities. This indicated that a single kinetic process, determining the reaction rate, took place. Fripiat 42 assumed that the continuous dissociation of diborane molecules, required to obtain an adequate density of sorbed BH3-groups , was the rate limiting step. These results indicated that the boranation of the silica surface could not be used to distinguish among the surface components (hydroxyls and hydration water) .42 The same rate law was obeyed for silica gels with pore radii changing from 2.3 to 4 0 n m . 42 This similarity of the kinetic data for silicas with very different textures, showed diffusion operating mechanisms not to play a determining role. In order to correlate the reaction parameters to the hydroxyl content, Fripiat and Van Tongelen 42 used ao, the initial hydroxyl density determined through MgICH3 chemisorption. The calculated activation energy was found to increase linearly with these ao values, equalling zero for a hydroxyl content of 1 . 4 0 H / n m 2. The linear relationship between the activation energy and ao with a positive slope, was interpreted as a relationship with negative slope between activation energy and siloxane sites. The siloxane sites being the sites for BH3 adsorption. The activation energy is thus high when the probability of a sorbed BH3- molecule meeting adjacent hydroxyls is lower. 42 The ao = 1.4 was explained as a situation where the silica surface has a maximum value of active surface sites. For lower hydroxyl concentrations, thus higher siloxane contents of the silica surface, the BH3 recombination process to form diborane becomes too fast to ensure further reaction with hydroxyl sites. Samples outgassed at temperatures > 873 K, thus temperatures high enough to lower ao below 1.4, show an extremely slow Hz-evolution and the experimental results no longer correlate with the other results.
341
Gillis D'Hamers studied the diborane chemisorption on different silica gels in order to specify the role of the hydroxyl groups 38,44 Kinetic experiments performed in an excess B2H6 however showed no induction period. A significant rate of chemisorption on the hydroxyl groups was observed from the start of the reaction. The sorption data, obtained from H2-evolution, were found to fit the Elovich equation. 44 This equation describes the activated adsorption of gases on solids and has the general form (2):
dq _a, e -bq dt
(2)
where q is the amount adsorbed gas and a and b are constants. So the Elovich equation describes a situation where the adsorption rate decreases exponentionally with the increase of the amount of adsorbed gas. In the integrated form the Elovich equation transforms to equation (3):
1 ,lnt+to q =-~ to
(3)
with to = 1/ab. If a linear fit for the experimental data is found between q and the logarithm of time for large t values, the sorption is said to be described by the Elovich equation. The value of to is chosen empirically so as to obtain the best linear fit of the data. Different theoretical models exist that describe mathematically the adsorption of gases in such a way that an exponential relationship between the amount and the rate of adsorption results. These models are used to obtain from Elovichian type kinetic data information about the energetic reaction parameters. Different models can be used which differ from each other by their physical approach to the reaction process. 1. Site model : in this model a homogeneous surface is assumed, where all reactive sites have the same activation energy during the adsorption process. The number of available sites however is said to change exponentially with the amount adsorbed.
342
(4)
nq=n o , e - ~
2. Energy models : in these models a constant number of reactive sites is supposed, while the activation energy of adsorption changes with the amount adsorbed. Two models, mathematically equivalent, have been used site 9 heterogeneity and induced heterogeneity. 2a. Site heterogeneity model a9 continuous energy function is assumed, that depends linearly on the degree of coverage. Each adsorption site has its own activation energy in this model, the sites with the lowest energy being covered first. 2b. Induced heterogeneity model : in this model the active sites are gradually influenced through the adsorption on neighbouring surface sites. A linear increase of activation energy with surface coverage results.
E.~ = E o +
(5)
ol0
Where E0 is the activation energy when 0, the fraction of the surface that is covered, equals zero. c~ is the induced heterogeneity coefficient. The perturbation of the active sites is caused by already adsorbed molecules through lateral interactions with reacting molecules or by directly influencing the free surface sites. Using the isothermal rate equation of the form
v=dq=rfp),n,e dt
9
E RT
K(p) is a pressure dependant constant comprising the collision frequency of the gaseous molecules per unit of surface area, n is the number of
(63
343
sites available and E is the activation energy of the reaction. obtains in the case of the induced heterogeneity 9
dO - k , ( 1 - 0 ) , e dt
E0+g0 Rr
One
(7)
Since the variation of (1-0) is negligible compared to the exponential term, an Elovichian type of relationship is obtained. With a = k*e -B~ and b = a/RT. The parameters Eo and a are determined by measuring the adsorption process at different temperatures. For the same values of q the plot of log(dq/dt) as a function of (-l/T) yields a curve with slope (E0 + ot0)/R. Determining the slopes for different q values in this way and plotting these values against q gives a linear relationship. The slope is equal to a and the intercept gives Eo. For more background information the readers are referred to the works of Aharoni 54'55 and LOW 56. Gillis D'Hamers 44 showed the B2H 6 chemisorption to follow the Elovich equation as was proved by the linear relationship between the number of modified hydroxyl groups and the natural logarithm of time. The kinetic data are presented in figure 10.14. The plots are linear, with correlation coefficients always exceeding 0.95, regardless the type of silica gel, i.e. porosity, pretreatment temperature or the reaction temperature used. The results were studied using the induced heterogeneity model. The calculations of the activation energies were based upon the experimental coverage versus time data and resulted in the E~ = E0 +otq values, using the methods of L o w 56 and Aharoni 55. The similarity of the obtained order of magnitude of the E, with that of other authors 42 supported the use of the induced heterogeneity model. The resulting E, values correspond to the overall activation energy needed for a nucleophilic attack of a free electron pair of an oxygen atom of the hydroxyl group to the electron deficient boron atom. Table 10.6 represents the E0 and c~ values obtained for silica gels with varying pore diameter and surface hydroxyl density. The zero coverage energy, Eo, represents the activation energy required for the first surface reaction between an OH group and an adsorbed borane molecule. The induced heterogeneity coefficient, a, corresponds to the order of magnitude of the induced heterogeneity.
344
1.6 1.4 1.2 1.0 O~ 0.8 "10 0.6 0.4 0.2
0 10
i
l
!
I
! IIII
I
100
I
I
t/s
I
IIIII
I
f
I
1000
I
! III
10000
Figure 10.14 Surface coverage of silica gel vs. the natural logarithm of time. Pretreatment temperature of silica gel: Kieselgel 40 (0), Kieselgel 60 ((>) and Kieselgel 100 (D).
Table 10.6 Activation energies for the coverage of surface hydroxyl groups on silica gel Temp. (K)
373 473 573 673 773 873 973 1073
Eo (KG. 40) (kJ.mo1-1 )
22.0 22.5 22.0 24.7 23.0 21.8 20.0 19.5
Eo (KG. 60) (kJ. mo1-1)
21.1 23.3 36.2 38.3 31.9 28.9 24.2 22.9
Eo (KG. 100) (kJ. mol -~)
35.2 40.7 41.0 42.3 39.2 38.3 33.3
c~ (KG. 40) (kJ. mol -~)
ot (KG. 60) (kJ. mol -~)
c~ (KG. 100) (kJ.rnol-~)
22.3 20.1 22.9 26.7 263 343 408 678
15.1 16.4 23.3 77.8 295 348 542 597
11.1 12.4 11.9 12.7 12.7 554 595
345 Following the hydroxyl dependence of the diborane chemisorption, Gillis-D'Hamers 44 reported a linear correlation between the absolute surface concentration of the free hydroxyl groups and Eo. The results are shown in figure 10.15. No functional relationship was obtained for the absolute bridged hydroxyl concentration or for Otovalues. Based on the linear relationship Gillis-D'Hamers assumed the chemisorption of B2H 6 to start exclusively on the free hydroxyl groups. In the rate-determining step of the chemisorption process, i.e. the nucleophilic attack of oxygen on boron, the free hydroxyl group operates as a reactant as well as the adsorbed borane molecule. The proximity between the free hydroxyl groups and the borane molecules determines the probability of the reaction. Increasing the reaction temperature causes an increase of the chemisorption rate, indicating that the physisorption of B2H 6 on the siloxane sites is not the rate determining step. The increase of the activation energy with the number of free hydroxyl groups, thus for a higher proximity of the reacting molecules, was explained not to be contradictory. A higher number of hydroxyls on the surface generates a distribution of the lone-pair electrons on the oxygen atoms. This results in a lower nucleophilicity of the oxygen atoms, resulting in a weaker attack of the oxygen to the electron deficient boron and consequently in a higher activation energy. The E0 values indicated a preference of the reaction to initiate on the free hydroxyl groups, but the reactions following on this initial chemisorption also have the same preference, as was evidenced by infrared and volumetric studies, as The pore structure also has a profound influence on the required activation energy for chemisorption (figure 10.15). The slope was found to increase linearly with pore diameter. 44 Since E0 values are involved, no induced heterogeneity effects are present. According to the changing curvature of the silica pores, the different slopes originate from varying proximity of the physisorbed borane molecules to the free hydroxyl groups. In the smaller pores the reacting groups are closer together, resulting in an easier reaction, what was reported as a lower slope. In figure 10.15 there is a common extrapolated intercept of 15 kJ/mol at a free hydroxyl concentration equalling zero, independent of the pore structure. Since no hydroxyl groups are involved in this sorption process, the value was assigned to the physisorption step of the reaction. However, since physisorption proceeds without energy, the activation energy of 15 ld/mol was assigned to the symmetrical cleavage of the adsorbed diborane molecules.
346 50
E, (kJ/mol) I
/
40
30
20
10
0
0
I
I
I
I
I
0.5
1
1.5
2
2.5
3
OHfree / n m 2
Figure 10.15 Activation energies at zero coverage vs. the free hydroxyl content. Kieselgel 40 (0), Kieselgel 60 ((>), and Kieselgel 100 (E3).
The effect of chemisorbed molecules on the Ea is seen in the c~ values. The induced heterogeneity coefficients showed a small, relatively constant value in the lowpretreatment region (373 - 673 K), At temperatures above 773 K there was a dramatic increase in the magnitude of the induced heterogeneity. Furthermore a reverse relationship was observed between c~ and the average pore diameter in the low pretreatment range. Steric hindrance appears to be the main effect. For smaller pores chemisorbed molecules are closer together and perturb the free sites to a higher degree, resulting in a higher value of c~. The increase at higher temperatures was explained through the occurrence of another phenomenon that eclipses the proximity dependence, which determines the induced heterogeneity in the low temperature range. The authors ascribed the increase of the c~-values to local contractions of the surface lattice at these temperatures affecting neighbouring S i O 4 tetrahedra. Siloxane bond angles decrease and the original siloxane bridge concentration, active in diborane adsorption, diminishes in absolute numbers. The free hydroxyl concentration, decreasing by proton migration over strained siloxane bridges, was also put forward as a factor influencing proximity. Since proximity remains the rate-determining factor with the free hydroxyls and adsorbed borane molecules as the reactants, a decreasing original, active siloxane bridge and free hydroxyl concentration on the surface
347
decreases the rate of chemisorption and contributes to the increase in the slope of the activation energy function (E~ = E0+aq).
3.5 Influence of the boranation reaction on porosity of the silica
Besides specific surface area, silicas are also characterised by their porosity. Most of the silica's are made out of dense spherical amorphous particles linked together in a three dimensional network, this crosslinked network building up the porosity of the silica. Where the reactivity of diborane towards the silica surface has been profoundly investigated, little attention has been paid to the effect of those reactions on the pore structure. However different methods are developed to define the porosity and physisorption measurements to characterise the porosity parameters are well established. Adsorption isotherms give the specific surface area using the BET model, while the analysis desorption hysteresis yields the pore size distribution. Gillis-D'Hamers has made a thorough study of the effect of the B2H 6 chemisorption on the porosity of silica gel. Three silica's that differed only in their pore structures and were shown to be mesoporous solids containing no micropores, were investigated. A cylindrical shape of the pores, as is generally taken as the basic pore shape for this kind of compounds, was chosen. For a cylindrical pore shape chemisorption reactions can affect the pore structure in two ways. Altering the pore length, where a decrease is caused by pore blocking, and/or altering the pore diameter, where the deposition of a coating on the internal pore walls narrows the pores. The common pore size distributions however take only in account the pore diameter, yielding the pore volume or pore area as a function of pore diameter, and not the pore length. Gillis-D'Hamers proposed a method where both the absolute amount of pore narrowing as well as pore blocking is estimated. The effect of both primary and secondary boranation reactions on porosity were investigated.
* primary boranation reaction As can be seen in figure 10.16, the surface area was found to decrease with diborane chemisorption. The decrease was linear with respect to diborane loading for KG 100, while a convex relation was seen for the other two silica gels. The convex decrease
348
was found to be significantly different from a linear decrease, indicating that two reaction phenomena did occur. The authors supposed the existence of both pore blocking and pore narrowing effects. Using the cylindrical pore shape the pore volume (V) and pore area (A) can be described as : V = 1/4 a- LD 2
(8)
A = -x DL
(9)
with L the pore length and D the pore diameter. Since the porosity of the silica gels is created during the synthesis process, a correlation between the diameter and length of the pores can be considered. By differentiating V and A to D then gives :
dv/dD = 1/2
7rD 2
dL/dD
(10)
dA/dD = 7rD dL/dD
(11)
Taking the ratio of the two equations gives the variation of pore volume to the variation of pore area. dV/dA = D/4 (1 + 0rL/dA/dD)) in the case where 7rL > > dA/dD, equation (12) is simplyfied to dV/dA = D/4
(12) 9 (13)
The condition of 7rL > > dA/dD can be checked taking the ratio of dV/dA to D. If a value of 1/4 is obtained the simplified equations can be applied. Also equation (11) can be rewritten as
9 dL/dD = l/a" dA/dD
(14)
This relationship gives a variation of the pore length that is related to the variation of the pore area, found from the pore size distribution and can be used to study the pore blocking effect. From equation (14) different interesting data can be obtained. (1) The differential pore length distribution as a function of pore diameter. (2) The cumulative
349 pore length as a function of diameter being the value of the integrated pore length distribution between D = 0 and the current diameter. The total pore length is determined via integration over all diameters. Comparing the total pore length before and after modification with diborane yields a measure of the pore blocking.
S BET / m2/g 600 l t . . ~
..i....._
500
400
200
I
0
0.25
I
I ~
0.5 0.75 B2 H 6] mmol/g
i
1
Figure 10.16 Specific surface area vs. diborane loading on silica gel. narrowing contribution to the fail in the surface area: dashed line.
1.25
The pore
The total pore length of the different samples was calculated using equation (14), since the (1/D)(dV/dA) product yielded for all points a value of 0.25 ___ 0.03. A linear correlation was found between the total pore length and the diborane loading on the surface. Figure 10.17 gives the results for the three silica's, where normalised pore length values were used for comparison. A smaller pore length was seen for higher surface diborane loadings, indicating qualitatively the occurrence of pore blocking. The slope of the curves gives the magnitude of pore blocking and was inversely proportional to the average pore diameter of the untreated silica gel. The pore blocking appeared to end at an average pore diameter of around 12 nm as is seen in figure 10.18.
350
aoU.
1.1
1.
++ +
0.9-
0.8-
0.7
I 0.25
0
I t 0.50 0.75 B 2 H6[ mmol/g
1
1.25
Figure 10.17 Integrated normalised pore length vs. diborane loading on the silica gel surface. narrow / a.u.
block/a.u.
40
0.4
30
0.3
20
0.2
10
0.1
0
5
10
15
Porediameter (ran)
Figure 10.18 relative amount of pore narrowing (solid line), pore blocking (dashed line) respectively, with pore diameter.
KG 100 showed no pore blocking effects, as is seen by the constant pore length, although there is a decrease in surface area reported. This because the total surface
351 area loss consists of the continued effects of pore blocking and pore narrowing. If pore narrowing can be quantified, absolute values for the qualitative pore blocking values can be obtained. To make a quantification of the pore narrowing the authors performed some assumptions based on the observed pore blocking effects. First pores of diameters exceeding 15 nm show no pore blocking, the decrease in surface area in these pores being only due to pore narrowing. And secondly the procentual extent of pore narrowing in the pores with diameter > 15 nm was considered to be representative for the whole pore size distribution. This grounded upon the randomness of the boranation reaction. Calculating the procentual area loss due to pore narrowing in this way, gave a linear relationship with the diborane loading for all three silica gels. A higher surface loading resulted in a higher surface area loss. The slopes, again taken as the magnitude of the process, showed a linear relation with the average diameter of the untreated silica's as represented in figure 10.18. The higher pore diameters being more prone to pore narrowing, with the critical pore diameter for pore narrowing to occur at 2 nm. Below this value only pore blocking is observed. Absolute values for the pore narrowing were obtained comparing the calculated % surface area loss to the total surface area loss found from the adsorption isotherm after diborane modification. The results are reported in figure 10.16 and show the linear decrease of surface area due to pore narrowing with B2H 6 uptake. The slopes differ from zero, indicating the occurrence of pore narrowing. In KG 100 the line almost coincides with the experimental data in agreement with the assumption that for pores > 12 nm pore blocking effects no longer occur. The quantitative value of the pore blocking was obtained by subtracting the area loss due to pore narrowing from the total surface area loss after modification. Comparing this absolute value to the change in total pore length normalised to the untreated silica, gave a linear relationship through the origin. This indicated the suitability of the applied model for prediction and quantification of the pore blocking, besides the pore narrowing. So the primary modification of the silica has a profound influence on the porosity of the substrate. A decrease in surface area is observed, being caused by pore narrowing and pore blocking effects.
352 * secondary modification reaction The secondary reactions result in a 1-12production in the absence of gaseous diborane. The secondary reactions are induced by temperature. Although the primary and secondary reactions seem fairly compatible regarding the mechanism, their effect on porosity may differ significantly. Gillis-D'Hamers investigated also the porosity changes induced by these secondary reactions. As was the case for the primary reactions, formula (14) was used to determine the degree of pore blocking. Based on the cylindrical pore model, a value approaching 0.25 was found for every point when calculating (1/D)(dV/dA). This showed equation (14) to be applicable. As figure 10.19 shows, no variations in the total pore length were seen during the secondary reactions. So independent of the initial diborane loading and temperature of the secondary reaction, the progressive annihilation of the surface hydroxyl groups does not indicate a further pore blocking. The occurrence of a secondary reaction was evidenced by a (linear) temperature dependent release of hydrogen gas.
pore length/a.u.
[]
ILl
0.9 I
•
13
^x
o
O
1
o O
o
)
O
-
0.7 273
~ A x
~"
0
0.8
[]
I
373
I
I
473 573 Temperature (K)
I
673
773
Figure 10.19 Integrated, normalised pore lengths vs. temperature for different diborane loadings.
353
The absence of pore blocking does not exclude a priori pore narrowing. This parameter was studied using the changes in the total pore area. The evolution of the total pore area versus temperature of the secondary reaction showed also the absence of pore narrowing for all diborane loadings. This indicated that the occurrence of the secondary reactions did not alter the pore geometry in any significant way. The progressive removal of surface hydroxyl groups proceeds not with a process of pore length shortening or pore diameter shrinkage. Secondary reactions occur by the reaction/anchoring of presorbed boron hydrides solely along the pore walls and not across the pores. References
V.M. Bermudez, J. Phys. Chem., 1971, 75, 3249. R.J. Peglar, F.H. Hambleton and J.A. Hockey, J. Catal., 1971, 20, 309.
.
M.J.D. Low and N. Ramasubramanian, J. Phys. Chem., 1967, 71, 3077.
.
.
.
6. .
.
M.L. Hair and W. Hertl, J. Phys. Chem., 1973, 77, 2070. F.H. Hambleton and J.A. Hockey, J. Catal., 1971, 20, 321. M.V. Mathieu and B. Imelik, J. Chim. Phys., 1962, 59, 1189. C. Naccache, J. Franqois-Rosetti and B. Imelik, Bull. S.c. Chim. France, 1959, 404. A.J. Tyler, F.H. Hambleton and J.A Hockey, J. Catal., 1971, 13, 35.
9.
H.P. Boehm, M. Schneider and F. Arendt, Z. Anorg. Allg. Chem., 1963, 320, 43.
10.
M.J.D. Low and N. Ramasubramanian, J. Phys. Chem., 1966, 70, 2740.
11.
C.G. Armistead, A.J. Tyler, F.H. Hambleton, S.A. Mitchell and J.A. Hockey, J. Phys. Chem., 1969, 73, 3947.
12.
F.H. Hambleton and J.A. Hockey, J. Catal., 1966, 1695.
13.
Y.I. Gorlov, in React. Kinet. Catal. Lett., 1993, 50, 89.
14.
B.A. Morrow and A. Devi, J. Chem. S,c. Faraday Trans., 1972, 68, 403.
15.
C.G. Armistead and J.A Hockey, Trans. Faraday S.c., 1967, 2549.
16.
K.H. Rhee and M.R. Basila, 1968, 10, 243.
354 17.
B.A. Morrow and A. Devi, Can. J. Chem., 1970, 48, 2454.
18.
K. Possemiers, J. Chem. Soc. Faraday Trans., 1995, in press.
19.
J.B. Peri and A.L. Hensley, J. Phys. Chem., 1968, 72, 2926.
20.
R. Sheets and G. Blyholder, J. Am. Chem. Soc., 1972, 94, 1434.
21.
B.A. Morrow and A.J. McFarlan, J. Non-Cryst. Solids, 1990, 120, 61.1
22.
M.L. Hair and W. Hertl, J. Phys. Chem., 1973, 77, 1965.
23.
M.L. Hair, J. Coll; Interface Sci., 1977, 60, 154.
24.
T.A. Guiton and C.G. Pantano, Coll. Surf. A, 1993, 74, 33.
25.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1976, 80, 1995.
26.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1976, 80, 1998.
27.
B.A. Morrow and I.A. Cody, J. Phys. Chem., 1975, 79, 761.
28.
B. Camara, H. Dunken and P. Fink, Z. Chem., 1968, 8, 155.
29.
B.A. Morrow and I.D. Gay, J. Phys. Chem., 1988, 92, 5569.
30.
M.L. Hair and W. Hertl, J. Phys. Chem., 1969, 73, 4209.
31.
L.H. Long, Adv. Inorg. Chem. Radiochem., 1974, 16, 201.
32.
N.N. Greenwood and A. Earnshaw, Chemistry of the elements, Pergamom Press, 1984.
33.
F.A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, ed. J. Wiley, New York, 1972.
34.
H.G. Weiss and I. Shapiro, J. Am. Chem. Soc., 1953, 75, 1221.
35.
H.G. Weiss, J.A. Knight and I. Shapiro, J. Am. Chem. Soc., 1959, 81, 1823.
36.
I. Shapiro and H.G. Weiss, J. Phys. Chem., 1953, 57, 219.
7.
I. Shapiro and H.G. Weiss, J. Phys. Chem., 1956, 60, 325.
38.
I. Gillis-D'Hamers, K.C. Vrancken, K. Possemiers and E.F. Vansant, J. Chem. Soc. Faraday Trans., 1992, 88, 3091.
39.
M. Bavarez and J. Bastick, Bull. Soc. Chim. France, 1964, 3226.
40.
B.A. Morrow and R.A. McFarlan, Langmuir, 1986, 2, 315.
355 41.
I. Gillis-D'Hamers, J. Philippaerts, P. Van Der Voort and E. Vansant, J. Chem. Soc. Faraday Trans., 1990, 86, 3747.
42.
J.J. Fripiat and M. Van Tongelen, J. Catalysis, 1966, 5, 158.
43.
J.J. Fripiat and J. Uytterhoeven, J. Phys. Chem., 1962, 66, 800.
44.
I. Gillis-D'Hamers, P. Van Der Voort, K.C. Vrancken, G. De Roy and E.F. Vansant, J. Chem. Soc. Faraday Tans., 1992, 88, 65.
45.
I. Gillis-D'Hamers, K.C. Vrancken, P. Van Der Voort and E.F. Vansant, J. Chem. Soc. Faraday Trans., 1992, 88, 1459.
46.
C. Naccache and B. Imelik, Bull. Soc. Chim. France, 1961, 533.
47.
I. Gillis-D'Hamers, Ph-D Thesis, University of Antwerp, Antwerp, 1993.
48.
A.I. Mashenko, Kinet. Katal., 1974, 15, 903.
49.
H.G. Weiss and I. Shapiro, J. Am. Chem. Soc., 1958, 80, 3195.
50.
K.K. Unger, Porous Silica, ed. J. Wiley, New York, 1979.
51.
L.L. McDowell and M.E. Ryan, Int. J. Appl. Radiat. Isotop., 1966, 17, 175.
52.
M.V. Mathieu, Angew. Chem., 1963, 75, 728.
53.
I. Gillis-D'Hamers, K.C. Vrancken, P. Van Der Voort and E.F. Vansant, J. Chem. Soc. Faraday Trans., 1992, 88, 2047.
54.
C. Aharoni and M. Ungarish, J. Chem. Soc. Faraday Trans I, 1976, 72,400.
55.
C. Aharoni and F.C. Tompkins, Adv. Catal., 1970, 1, 21.
56.
M.J.D. Low, Chem. Rev., 1960, 60, 267.
This Page Intentionally Left Blank
357
Chapter 11
Modification with other compounds
Besides boron and silicon compounds, many other hydrogen sequestering agents have been used in the last decades to modify the silica surface. The reasons for such a surface modification vary from a purely fundamental study (gaining a better insight in the nature of the silica surface) to the creation of new materials, with outstanding properties. In this section, only the surface modification techniques will be discussed, and not the impregnation, sol-gel or co-precipitation techniques. Furthermore, it is not our aim to fully cover all chemical modifications on the silica surface. We merely want to present in introduction to and an insight in the fast expanding field of silica modifications, in order to create new catalysts, sensors and immobilizators.
1 Modification with Ti-compounds 1.1 Reaction mechanisms
The modification of the silica surface with Ti - compounds has been studied since the late 1960's. Titaniumtetrachloride (TIC14) is almost exclusively used as the reagent. The reactions of TIC14 with the silica surface proceed by the same mechanisms as their Si or B analogues. 1,2,3
358
The possible reaction mechanisms are summarized underneath (reactions (A) - (E)). --Si-OH
+
TiCI4--,
-=Si-OH
-Si-O-TiCI 3 +
~-Si-O + TiCI 4
\
CI
/
-
Ti
-=Si-OH
=Si-O
-Si \
/
(A)
HCI
+ 2 HCI \
03)
C1
-Si-C1 O
+ TIC14
-
(C)
-Si /
---Si-O-TiC13
=Si-O-TiC1
=-Si-O
3
\
-
\
/
-Si-O-TiC1 3
~-Si-C1 -=Si-O
+
\
=Si-O
_
C1
Ti
/
=Si-OH
=Si \
/
0
HC1
(13)
C1
C1
(E)
Ti
-Si / -Si-O
/
\
C1
Murray et al. 1 stated that 'treatment of either silica surface with excess TiC14 vapour at room temperature removes all the surface hydroxyl groups, both isolated and bridged'. As in the case of the chlorosilanes, this statement was soon updated. TiCI4 reacts rapidly and completely at 298 K with the isolated silanols and to some extent, i.e. within sterical restrictions, with vicinal silanols.
359 A very detailed study on the TiCI4 reaction with silica was recently performed by Haukka and co-workers, 3 following the earlier works of Kunawicz, a Armistead, 5 Damyanov, 6 Morrow, 7 Kinney, s Ellestad 9 and Kooyman. l~ Figures 11.1 (a) and (b) show the number of Ti atoms per nm 2 and the C1/Ti ratio as a function of pretreatment temperature for reaction temperatures of 448 K and 723 K, respectively. The reaction was performed in the vapour phase with excess TiCla. The curves of figure 11.1 can be interpreted in terms of effectiveness, surface coverage and stoichiometry, as discussed in chapter 9. The curves of figure 11.1 (a), reaction temperature 448 K, can be explained in a very straightforward manner. The C1/Ti ratio varies from approximately 2.5 at low pretreatment temperatures to 3 at high pretreatment temperatures. This means that the stoichiometry of the reaction is 1 at high pretreatment temperatures. Under these circumstances, the modification proceeds exclusively according to reaction (A). At lower pretreatment temperatures, the C1/Ti ratio decreases, indicating that a slight amount of secondary species is created, according to reactions (B) or (D). The decrease of the number of Ti atoms per nm 2 is a logical consequence of the decrease in the number of hydroxyl groups that are available for reaction, due to dehydroxylation processes. It is important to note that Haukka did not succeed -at any pretreatment temperatureto remove all hydroxyl groups at 448 K. In other words, the effectiveness never attained a value of 1, thereby contradicting the earlier reports. ~,2,3 The C1/Ti ratio of 2 for reaction temperatures of 723 K (figure 11.1 (b)) seems at first sight a little surprising. Haukka explained this phenomenon by the irreversible adsorption of HCI on the silica surface, although we believe that an increased occurrence of the secondary reactions (D) and (E) could also be an important contributing factor. Also the decrease in the absolute number of Ti atoms per nm 2 (in comparison to the lower reaction temperature) could be indicative for this phenomenon.
360
2.5
Clfri ratio
Number of TiCl atoms/nm2
3.5
2.0
3.0
1.5
2.5
1.0
2.0
0.5
1.5
Ca) 0 i 400 500
i 600
t i 700 800 Temperature (K)
i 900
i 1000
Number of Ti atoms/rim2
1.2
1.0 1100
Cl/Ti ratio
1.0
2.5
_._____+______+. 2.0
0.8 1.5 0.6 1.0 0.4 0.2
0.5 (b)
0
7()0
i
I
i
800
900
1000
0
1100
Temperature (K)
Figure 11.1 CI/Ti ratio and # Ti/nm 2 of a TiCl4 modification as a function of the pretreatment temperature. Reaction temperature 448 K (a) and 723 K (b).
Haukka noticed however, that upon hydrolysis of these samples mainly ---Si-OH groups reappear (and not Ti-OH groups), confirming his hypothesis of the chlorination of ---Si-OH groups (at high reaction temperatures) by liberated HC1.
361 In his paper, Haukka did not discuss explicitly the reaction (C) with strained siloxane bridges, although Kunawicz4 stated already in 1970 that these groups have 'equal, if not greater, nucleophilic reactivity than the hydroxyls'. This statement was confirmed in 1983 by Kinney et al. g Therefore, we can conclude this brief survey on the reaction mechanisms by saying that the TIC14 modification of silica is encumbered by the same practical problems as the SIC14 modification" the occurrence of at least 5 different reaction mechanisms and the absence of 5 independently measurable parameters. By a correct and careful interpretation of the available data, semi-quantitative information on the occurrence and relative contribution of these reactions can be obtained.
1.2 Analysis techniques The main spectroscopic analysis technique still is Fourier Transform Infrared Spectroscopy, although the Ti-C1 vibrations are not or badly detectable, due to the strong overlap with the Si-O-Si lattice vibrations. Infrared studies are therefore mostly restricted to the analysis of the silanol vibrations. Using the technique of photoacoustic spectroscopy (FTIR-PAS, cfr. Appendix A), Kinney et al. s were able to assign peaks at 790 cm 1 and 730 cm ~ to the TiO stretching vibrations. TiCI stretching vibrations were not observable. Malkov and Malygin assigned a band at 920 cm -~ to the Si-O-Ti stretching m o d e . ll'12 Morrow and Hardin 2 have used Raman spectroscopy to study the chemisorption. The silica discs were pretreated at 973 K and reaction occurred at 573 K. The resulting Raman spectra are shown in figure 11.2. This Raman spectrum shows two new strong features at 112 cm 1 and 400 cm -1, which are not present in the silica background, accompanied by a possible accentuation of the background near 490 cm ~. The interpretation of these Raman lines is not so straightforward, but Morrow and Hardin have presented in their publication a number of arguments that justify the assignments, summarized in table 11.1, for several chemical groups, chemisorbed on the silica surface.
362
D C
B
i
6OO
I
I
!
40O
cm
I
2O0
-I
Figure 11.2. Low frequency Raman spectra of (A) liquid TiCI4, (B) untreated silica and (C) TiC14 or 09) GeC14 on silica.
Table 11.1 Assignments of MC, MCI and SiOM stretching modes of chemisorbed species agent
TIC14 GeCI4 GeC13Me GeClzMe2 GeC1Me3 SiC1Me3 HMDS
p,s(MC) cm -~
640 620 690 690
' 440 cm -~ assigned to va, (MClz)
ps(MC) cm-~
630 595 573 600 600
~,,(SiOM) cm-~
~,,(MC1) cm-~
490 480 490 (440) a 490 490
400 410 410 400
490
363 Besides zgSi-NMR, that distinguishes between siloxane bridges, single and double silanols, especially 1H-NMR-CRAMPS (Combined Rotation and Multiple Pulse Spectroscopy) is a very powerful tool. This technique allows to differentiate the SiOH and (after hydrolysis) the Ti-OH species, yielding thereby useful information on the structure of the surface groups 3. Due to spectral overlap, this distinction is very difficult to observe by infrared spectroscopy.
The possibilities and resolution of the technique are illustrated in figure 11.3.
-- Si-OH
-Si-OH
Ti-OH
9
Ti-OH
' ' ','o ....
~ .... PPM
-io ....
,'o ....
~ ....
-Io'
PPM
Figure 11.3. 1H MAS NMR spectra of (A) silica, treated at 473 K and (B) silica treated at 833 K, reacted with TIC14, and treated with water at 473 K; taken from ref (3) with permission.
1.3 Applications
Most applications of Ti modified surfaces can be found in the field of heterogeneous catalysis, often in combination with an A1 modification to yield Ziegler-Natta catalysts. ~3'14 The research group of Malygin (St. Petersburg-Russia) have developed rather special applications, called by them 'Molecular layering precise technology'. One of the most remarkable applications was the development of pigments and fillers
364 by 'jacketing' industrial waste materials with a TiO2 layer, t2 This jacketing was performed by a variant of the Chemical Surface Coating method (see part 3). The coating of the TiO2 layer was controlled on a molecular level by cycling the TIC14 reaction and the hydrolysis" ( - Si-OH). + TiC! 4
(-- Si-O).TiCI4.. +
(4-n)HzO
~
( - Si-O).TiCI4..
--,
(--- Si-O),Ti(OH)4.. + TiC! 4
+
n HC!
( - Si-O).Ti(OH)4., + --,
(_-- Si-O).(Ti-O)mTiCl4_m +
(4-n)HCi
( - Si-O).(Ti-O),.,TiCI4_m (4-m)H20
-*
....
These samples were tested as a whitening pigment, is The TiO2 jacket was calculated to be 2 nm. This material was used as a new activator for rubber vulcanisation, substituting ZnO and establishing better mechanical characteristics.
2 Modification with Al-compounds
2.1 Reaction mechanisms
Although aluminium is not a transition metal in the strictest sense, aluminosilicates have found widespread applications as catalysts. For quite some time, it was not clear whether the catalytic cracking process of these aluminosilicates involves the concerted action of both Lewis and Br6nstedt acid sides, or only Lewis sites, or even the dissociation of the adsorbed hydrocarbons by surface metal ions (impurities). ~6'~7'18'~9 The need for oxide systems that allow evidence concerning the above catalytic theories was one of the driving forces for the efforts that have been done to coat a silica surface with aluminium compounds. One of the earlier studies, concerning both the reaction mechanisms of the aluminium
365
modification and the catalytic properties of these materials is performed by the institute of science and technology of the university of Manchester. 2~ Most studies were performed using A1CI3 or A1Me3 as a reagent. A1CI3 is a solid at ambient temperatures and it is necessary to react the silicas with A1CI3 at 473 K or higher. A1C13 is highly dimeric under solid-gas equilibrium conditions at low temperatures, but the heat of dimer dissociation ( < 65 kJ/mol) is sufficiently small for the treatment. However, it is possible that species (I) are formed on the silica surface. C1
C1
\A1
/
O Si
\
CI'
C1
A1/
/
\
(species I) O Si
In general, however, A1C13 reacts quite similarly to BC13, so mainly following reactions ((F) - (I)) occur: -Si-OH
+
AIC! 3 --,
-Si-OH
-Si-O-AIC! 2 +
=Si-O + A1C13
A1-C1 -Si-O /
-Si-OH
-Si-C1
~Si-OH
(F)
\
"-
-=Si-OH
+ A1C13
HC!
-
+ 2 HC1
(G)
(I-I) -Si-O-A1C1 2
366
=Si-OH
=Si-O -
N
AI-CI
+ 2HCI
(1)
/ =Si-O-AICl 2
--Si-O '
The reactions of A1Me3 with the silica surface have been investigated by many researchers. 1,2,4,22,23,24,25 Reactions (J) - (M) have been proposed to account for the observed CH and OH absorptions in infrared spectroscopy.
/H
/
O
+ AIMe 3
I
_
O
Si
+
0
+ AIMe 3
\
0
I
Si
'"
+
2
+
I
Me
I
Si
III
AIMe
A1Me2
0
-
Si
III
/
(J)
Ill
/
Si
CH 4
Si
Ill
/
AIMe2
Si
III
0
/
I
Si
'"
o:)
III
Me
H
~I
~Si
I
AI
~0
+
CH
4
(L)
Si-
These species give rise to absorption bands in the CH region (2940, 2900 and 2830 cm ~ for A1Me; 2960 and 2900 cm ~ for SiMe). Low 24 suggested that also reactions (N) and (O) may be important.
367
Me A1Me
/ O
I
2
A1
+
I
Si
Si III
O
/
I
Si
I
H I
I
- Si
III
!11
Me
o/
(M)
Si ~ll
Si -
AIMe2
--
I
+ A1Me 3
+ MeOH
Si
(N)
III
- Si
/ O Si III
/
AIMe2 \
Si
+ A1Me3
III
-
I Si III
Me
O +
I Si
(O)
III
These reactions were confirmed by Kinney and Staley, 8 based on the observation of a significant absorption at 2860 cm 1, which cannot be attributed to either the Si-Me or the A1-Me species. The position of this peak corresponds to one of three CH stretch peaks in the spectrum of methoxy groups, along with peaks at 3000 and 2960 cm 1. These peaks may be due either to a true surface methoxy group, or to physisorbed methanol. Low frequency absorptions are also summarized by Kinney and Staley 8 and are presented in table 11.2.
368
Table 11.2 Surface MO and MC stretch vibration frequencies
o
/
A1Me 2
I
Me
k
o/
,
Me
~o ,
-= Si
v (AIO) - 870 or 815 cm"x v(AIC) -'- 715 cm -1
v(AlO)
I
Si
III
Si =---
870 or 815 cm -1
v(A1C) -- 715 cm
v(SiC) = 680 cm "1
-1
2.2 Analysis techniques Infrared spectroscopy remains one of the most generally applied techniques. A discussion of the typical bands that can be observed after a treatment with AIMe3 was already given in the previous paragraph. Some frequencies for Raman spectra can be found in the publication of Morrow and Hardin. 2
2.3 Applications 2.3.1 Model systems for cracking catalysts One of the first issues, discussed in literature, was the creation of a model system to study the cracking properties of aluminosilicates. This naturally implies the hydrolysis of the chemisorbed A1CI3 or AIMe3 group towards a thin A1203 layer. However, the hydrolysis behaviour of these modified silicas was far more complex than originally believed. Peglar ~~ and Yates 22 studied these reactions in detail and came to the conclusion that addition of water to chemisorbed A1Me2 species yields CH 4. Furthermore, only Si-Me groups were left, which are known to be stable in the presence of water. After hydrolysis, the infrared spectrum of Si-Me was detected, but there was 'reasonable doubt' concerning the
369 presence of A1OH groups. In fact, it was suggested that, even in the absence of water, surface rearrangements occur, yielding SiMe species. Nevertheless, Hambleton and Hockey 2~ were still able to conclude that the hydrolyzed A1Me3 catalyst exhibited a cumene cracking activity that was comparable to industrial cracking catalysts. Moreover, these new catalysts showed a much less tendency for carbonization during the cumene cracking process. A decade later, Low and co-workers 24 restudied this hydrolysis and proposed a multistep mechanism. At first, the chemisorbed =A1Me group is hydrolyzed according to reaction (P)" =AI-CH 3 +
H20
~
=AI-OH
+
CH 4
(P)
Interaction of H20 with the surface species containing Si-O-A1 structure leads to the formation of Si-OH and A1-OH groups, interacting with each other and with the hydroxyls generated by reaction (P). Further adsorption then leads to 'ordinary' physisorbed water and to adsorbed water, that is coordinated to the surface Al-ions. The large amount of free silanols that is formed upon degassing suggests that the dehydroxylation involves to a large extent the coalescing of hydroxyls associated with aluminium. This coalescence would naturally explain the absence of clear-cut A1-OH bands in the infrared spectrum.
2.3.2 Ziegler-Natta catalysts The reactions between TiCI4 and A1Me3 are of particular interest in studies of olefinpolymerization catalysts as a Ziegler-Natta catalyst system. ~4 The reaction between a surface TiC1x group and A1Me3 gas is known to give an active olefin polymerization catalyst. ~ After addition of propylene to such a catalyst, Murray ~observed that the samples were covered by a 'sticky' colourless deposit, which was identified as polypropylene.
370 Kinney and Staley 8 performed a detailed infrared study on the reaction mechanisms leading to such a catalyst. They proposed the following model reaction (Q)"
CI\
/Cl
Me
Ti" / \
Cl
O
O
Si
Si
I
m
I
+ A1Me
3
Ill
Cl / \Tli/O--AI\Me / O" Me I + I
Si
(Q)
Si
III
Iil
In any case, they suggested, one of the Ti-O-Si bonds of the bridged Si-O-Ti-O-Si species reacts with A1Me3 and a Si-Me group is formed.
3 Modification with P - c o m p o u n d s
3.1 Reaction mechanisms
Following the earlier work of Kol'tsov, 26 Negievich 27 and Fink, 2g Bogatyrev and Chuiko 29 studied the reaction of PC13 with the silica surface. They concluded that two competitive reactions occur. The authors showed (table 11.3) that the relative contribution of the chlorination reaction (S) increases with increasing reaction temperature. The silica was previously degassed at 973 K. In all cases, the silanols have reacted completely. -Si-OH
+
PCI 3 ~
-Si-OPCI 2 + HCI
(R)
-Si-OH
+
P C ! 3 --,
-Si-CI
(S)
+
HOPC! 2
371
Table 11.3 Phosphorous content in the aerosil after chemisorption of PC13 T (K)
amount of phosphorous %
573 597 606 617 633 701 741
0.76 0.68 0.60 0.55 0.46 0.39 0.25
Amount of-OPCI2 groups (calculated) mmol/g of SiO2 0.25 0.22 0.19 0.18 0.15 0.13 0.08
Quite recently, Morrow, Lang and Gay3~ reevaluated the results of Bogatyrev and Chuiko, based on a combination of FTIR and 3~p-MAS-NMR. They concluded that phosphorous trichloride reacts with silica at room temperature according to reactions (R) and (S), as suggested by Bogatyrev and Chuiko. However, substantial evidence was presented to substantiate the hypothesis that the P(III) species further oxidize towards P(V) species. The exact reaction mechanisms are not known, the statement is based on the following arguments: (1)
Infrared and NMR spectra indicate, besides =SiOPC12 species, also the presence of --SiOP=O(H,OH) a and (=Si-O)2P=O(H) species on the silica surface.
-SiOP=O(X,Y) is the representation for X I
=Si
- 0
- P = 0
I
Y
372
(2)
The last two surface species are also produced from the adsorption of phosphorous acid, H3PO3, on silica at room temperature.
(3)
Phosphoryl chloride, O=PCI3, does not chemically adsorb on silica at room temperature, but it strongly hydrogen bonds with surface silanol groups.
Based on these observations, the following scheme was suggested:
-Si-OH
+ PCI3
---Si-OH
+ PC13
"
[ =Si-O-PC12
~-SiC1
]+ HC1
(R)
+ C12 POH
(s) oxydation
C12P = O(H) -Si-OH
+ C12P-- O(I-I)
-Si- O- P
--
O(H,OH)
+ 2HC1
03 =--Si-OH
-Si-O + C1 2P = O(H)
---Si-OH
\ P - O(H)
+ 2HC1
/ -Si-O
(u)
In 1991, Mutovkin and Plyuto 3~ performed a parallel study on the chemisorption of phosphorous pentachloride (PCIs) on silica. Since PCI5 is thermally unstable and disproportionates according to PCI5 ~ PCI 3+ C12 at elevated temperatures, the PCI5 chemisorption should be carried out in a solvent medium. The authors found out that the solvent has a large impact on the chemisorption behaviour of PCIs.
373 This solvent effect is double. On the one hand, the solvent exhibits a perturbing effect on the surface silanol groups. The strength of this perturbing power can be estimated by the low-frequency shift of the 3747 cm -~ free hydroxyl band. They noticed a shift of 40, 106 and 300 cm -~ for CC14,benzene and acetonitrile respectively, indicating that acetonitrile has the largest accelerating effect on the reaction. On the other hand, the solvent also influences the reactivity of the PCI5 molecule itself. CC14, in which the PC15 molecule exists in a solvated state, 32 has the minimum effect on the state of PCIs. Going from benzene to acetonitrile, PC15 dissociates increasingly according to reactions (V) and (W) 33 PCIs 2PCls
~-
~
[eCl4]
+
-I-
Or)
CI
[eEl4] + -I- [PCI6]-
(w)
So the accelerating effect of the solvent involves both the perturbation of the silanols as the perturbation (dissociation) of the reagent. Parallel with the observations of Morrow, 3~ Mutovkin and Plyuto evidenced that the PC15-chemisorption is a multistep process according to -Si-OH
+
PC! s ~
-Si-OPC! 4 +
HC!
(x)
followed by
-Si-OPC14
-
I -SiCI3 I
+ POCI3
(Y)
Kinetic experiments revealed that reaction (Y) occurs in situ during the actual chemisorption process. Upon thermal treatment of the modified silicas, the P/CI ratio did not change as a function of time.
374
3.2 Analysis techniques Fourier Transform Infrared Spectroscopy remains a very important technique in all chemisorption studies. A summary of band assignments, chemisorption, is presented in table 11.4. 30,34,35
specific for PCI 3
Table 11.4 Infrared frequencies (cm9 of PCl3and OPCI~ PC13
p (P=O) uas (PC13) us (PC13)
515 505
OPC13 (gas)
OPC13 (soln)
PC13 (SiO2)
OPC13 (Si02)
1323 594 482
1300-1310 586-594 482-487
n.o. n.o.
1290 605 n.o.
n.o. = not observed A very powerful tool for the characterization of chemisorbed P - compounds on the silica surface, is of course 31p-MAS-NMR (MAS = Magic Angle Spinning, cfr. Appendix B). The most important peak assignments are summarized in table 11.5.
Table 11.5 31P-NMR peak attributions for the reaction of PCl3 with silica3~ Chemical shift (ppm)
-16 -5 3 184 219
Assignment
( ---Si-O)2 P = O(H) ----Si-O-P = O(H,OH) OPC13 (solution) -=Si-OPC12 PC13 (solution)
375
3.3 Applications The investigation of the reaction of phosphorous compounds with the active surface centres of silica is of interest for the synthesis of new fillers for polymeric materials, catalysts and adsorbents. 4 Modification with transition metal compounds A vast amount of metalchlorides and metaloxychlorides has been used to modify the surface of silica. The mechanisms usually follow the schemes presented earlier in this section. The research group of P l y u t o 36'37 studied the modification with chromium oxychloride, according to the following main reaction" -=Si-OH + CrOzC! 2 --, --Si-O-CrOz + HC!
(Z)
Low temperature controlled hydrolysis of the -CrO2C1 surface groups is used for the synthesis of chromium oxide layers on the silica surface. The mechanisms of the chemical conversions have been studied in great detail and it has been found that grafted chromium oxide associates (clusters) are formed, based on the condensation of chromic acid molecules with the surface silanol g r o u p s . 38'39'4~ Figure 11.4 shows the monodentate and bidentate grafted structures.
o
o Vo Fo 1 ~176 /%
o/ \ o ~J Si
o-" Si
k/
%
a-2
1
n-2
Si
The a u t h o r s 36'37 developed a kinetic model to estimate the relative cluster size as a function of the initial Cr/SiOH ratio. Figure 11.5 shows the probability curves for an initial Cr/SiOH ratio of 0.05, 0.2 and 1.0, respectively.
376
relative occurrence
0.5
O!
2
3
n (association number) Figure 11.5 Relative occurrence of chromium oxide associates as a function of the association number, for an initial Cr/SiOH ratio of O.05(+), 0.2(.) and 1.0(*).
The same research group41 also studied the chemisorption of MoOC14 on the silica surface. The authors showed that the interaction of molybdenum oxy tetrachloride with the silanol groups of aerosil produces grafted --SiOMoOC13 molybdenum chloride groups by the electrophilic substitution of a proton (SEi mechanism). During vacuum heat treatment the grafted groups undergo conversions, with formation of -SIC1 groups and M002C12, which is removed from the aerosil surface (reaction mechanism (AA)).
377
H
\
O
"+ M~
U
MoOCl 2
H ~C1
\
Si
\ 0 ~M~
s
I
Si
sei
" - HCI
/\
~
Cl
I
Si
Si
SNi
(AA)
Also in the case of WC16 and WOCl4, 42'43 the chemisorption is not completed at the stage of the formation of grafted tungsten chloride groups.
The =SiOWCI5 and
- S i O W C I 3 groups can be regrouped with the formation of a ~-SiCI group and the volatile compounds WOCI4 and WO2C12, which are evolved from the silica surface during vacuum treatment. Following the earlier work of Tubis, 44 Morrow and McFarlane 45 performed a detailed study on the chemical modification of silica with trimethylgallium (GaMe3). Gallium anchored to some oxides is reported to have an unusual catalytic activity44'46 and since G a M e 3 is widely used to produce GaP and GaAs semiconductor d e v i c e s , 47'4g'49 the
reactions of a G a M e 3 modified silica surface with P H 3 a n d AsH 3 were also studied. Trimethylgallium, unlike the alkyl silicon and germanium derivatives, is extremely reactive at ambient temperatures, and, unlike its aluminium counterpart, is a monomer at room temperature or higher. It reacts with the surface silanols according to main reaction (BB).
-Si-OH
+ GaMe3--, -SiOGaMe2 + CH 4
(BB)
The reactions o f P H 3 o r AsH 3 are fairly complex. Morrow and McFarlane 45 suggested that the initial stages of the sequence could proceed according to mechanism (CC), with MH 3 = PH 3 or AsH 3.
378
MI'I3
-SiOGaMe2 +
MI'I31 ~
-SiOH
+ GaMe2
~
... (CC)
-SiOGaMe2 =SiOGa/MI'I2 \Me
+ eli 4
~
where '...' might refer to the subsequent generation, at increasing temperatures, of products such as Ga(MH2)2Me + CH 4, - S i O H + Ga(MH2)2Me or -SiOGa(MH2)2 + CH4, eventually leading to GaM + H2. Many other compounds have been used to modify the silica surface, such as VOCI3,12 Rh2(CO)4C125~ and GeC14.51
References
J. Murray, M.J. Sharp and J.A. Hockey, J. Catal., 1970, 18, 52. .
B.A. Morrow and A.H. Hardin, J. Phys. Chem., 1979, 83, 3135.
.
S. Haukka, E. Lakomaa and A. Root, J. Phys. Chem., 1993, 97, 5085.
.
J. Kunawicz, P. Jones and J.A. Hockey, Trans. Faraday Soc., 1971, 67, 848.
0
.
0
C.G. Armistead, A.J. Tyler, F.H. Hambleton, S.A. Mitchell and J.A. Hockey, J. Phys. Chem., 1969, 73, 3947. D. Damyanov, M. Velikova, Iv. Ivanov and L. Vlaev, J. Non-Cryst. Solids, 1988, 105, 107. B.A. Morrow and A.J. McFarlan, J. Non-Cryst. Solids, 1990, 120, 61. J.B. Kinney and R.H. Staley, J. Phys. Chem., 1983, 87, 3725.
0
10.
O.H. Ellestad and U. Blindheim, J. Mol. Catal., 1985, 33, 275. P.J. Kooyman, P. van der Waal, P.A. Verdaasdonk, K.C. Jansen and H. Bekkum, Catal. Lett., 1992, 13, 229.
379 11.
A.A. Chuiko, E.F. Voronin and V.A. Tertyck, Adsorbzia i Adsorbenty, 1983, 11, 22. (in Russian)
12.
A.A. Malygin, private communication.
13.
T.J. Pinnavia, J.G.S. Lee and M. Abedini, in 'Silylated surfaces', D.E. Leyden and W.T. Collins eds., Gordon and Breach, New York, 1980.
14.
J. Boor jr., Ziegler-Natta catalysts and polymerization, Academic Press, New York, 1979.
15.
E.A. Avrutina, Avtoreferat diss kandidat khim. nauka., L. 1989. (in Russian)
16.
M.R. Basila, T.R. Kantner and K.H. Rhee, J. Phys. Chem., 1964, 68, 3197.
17.
M.R. Basila and T.R. Kantner, J. Phys. Chem., 1967, 71,467.
18.
J.B. Peri, J. Phys. Chem., 1966, 70, 3168.
19.
W.K. Hall, F.E. Lutinski and H.R. Gerberich, J. Catal., 1964, 3, 512.
20.
R.J. Peglar, F.H. Hambleton and J.A Hockey, J. Catal., 1971, 20, 309.
21.
F.H. Hambleton and J.A. Hockey, J. Catal., 1971, 20, 321.
22.
D.J.C. Yates, G.W. Deinbinski, W.R. Kroll and J.J. Elliot, J. Phys. Chem., 1969, 73, 911.
23.
R.J. Peglar, J. Murray, F.H. Hambleton, M.J. Sharp, A.J. Parker and J.A. Hockey, J. Chem. Soc. A, 1970, 2170.
24.
M.J.D. Low, A.G. Severdia and J. Chan, J. Catal., 1981, 69, 384.
25.
S.R. Seyedmonir, S. Abdo and R.F. Howe, J. Phys. Chem., 1982, 86, 1233.
26.
S.I. Kol'tsov, A.I. Volkova and V.B. Aleskokii, Zhurn. Fiz. Khimii, 1970, 44, 2246. (in Russian)
27.
L.A. Negievich, A.S. Viongradova and A.A. Kachan, Ukr. Khim. Zhurn., 1976, 42, 1109 (English translation: Soviet Progress in Chemistry).
28.
P. Fink, B. Camara, F. Weltz and T.P. Ding, Z. Chem., 1971, 11,473. (in German)
29.
V.M. Bogatyrev and A.A. Chuiko, Ukr. Khim. Zhurn., 1984, 50, 831. (English translation: Soviet Progress in Chemistry, UDC 541.183, p 50).
30.
B.A. Morrow, S.J. Lang and I.D. Gay, Langmuir, 1994, 10, 756.
380 31.
P.A. Mutovkin, Yu. V. Plyuto, I.V. Babich and A.A. Chuiko, Ukr. Khim. Zhurn., 1991, 57, 367. (English translation: Soviet Progress in Chemistry, UDC 541.183, p 32)
32.
R.W. Suter, H.C. Knachel and V.P. Petro, J. Am. Chem. Soc., 1973, 95, 1474.
33.
L.M. Sergienko, G.V. Ratovskii, A.M. Dodonov and A.V. Kolabina, Zhurn. Obshch. Khimii., 1979, 49, 1982. (in Russian)
34.
K. Nakamoto, Infrared and Raman spectra of inorganic and coordination compounds, 4tb edition, John Wiley & Sons, New York, 1986.
35.
R.A. Nyquist and C.W. Puehl, Appl. Spectrosc., 1992, 46, 1552.
36.
N.V. Borisenko, Yu.V. Plyuto and A.A. Chuiko, Ukr. Khim. Zhurn., 1992, 58, 27. (English translation: Ukrainian Chem. Journal, UDC 541.183, p 24)
37.
A.V. Isarov, Yu.V. Plyuto and A.A. Chuiko, Doklady Akademii Nauk. SSSR, 1990, 311,402. (Engl. translation Soviet Progress in Chem. UDC 541.183, p 252)
38.
G. Ol'man, Izv. Khim. Bulg. Akad. Nauk., 1980, 13, 48.
39.
Yu.V. Plyuto, Yu.I. Gorlov and A.A. Chuiko, Teor. Eksp. Khim., 1983, 19, 754.
40.
A.V. Isarov, G.A. Konishevskaya, Yu.V. Plyuto and A.A. Chuiko, Zh. Fiz. Khim., 1988, 62, 2211.
41.
A.A. Gomenyuk, I.V. Babich, Yu.V. Plyuto and A.A. Chuiko, Zhurn. Fiz. Khim., 1990, 64, 1662. (English translation: Russian Journal of Physical Chemistry, 1990, 64, 891)
42.
I.V. Babich, Yu.V. Plyuto and A.A. Chuiko, Dokl. Akad. Nauk., UkrSSR, 1987, 4, 39.
43.
I.V. Babich, Yu.V. Plyuto and A.A. Chuiko, Zh. Fiz. Khim., 1988, 62, 516. (English translation: Russian Journal of Physical Chemistry, 1988 (2))
44.
R. Tubis, B. Hamlett, R. Lester, C.G. Newman and M.A. Ring, Inorganic Chemistry, 1979, 18, 3275.
45.
B.A. Morrow and R.A. McFarlane, J. Phys. Chem., 1986, 90, 3192.
46.
T.A. Gilmore and J.J. Rooney, J. Chem. Soc., Chem. Commun., 1975, 219.
47.
R.M. Biefeld, J. Cryst. Growth, 1982, 56, 382.
48.
D.J. Schlyer and M.A. Ring, J. Electrochem. Soc., 1977, 124, 569.
49.
J. Nishizawa and T. Kurabayoski, J. Electrochem. Soc., 1983, 130, 413.
381
50.
M. Capka, M. Czakoova, J. Hjortkjaer and U. Schubert, React. Kinet. Catal. Lett., 1993, 50, 71.
51.
E.F. Voronin, V.I. Zarko and E.M. Pakhlov, private communication.
This Page Intentionally Left Blank
383
Chapter 12
Ammoniation of modified silica: introduction of functional groups
1 Introduction The reaction of gaseous ammonia with silica has been investigated by several researchers. Most studies 1'2'3'4'5'6'7'8 conclude that the adsorption is reversible at low temperatures (< 500 K). Ammoniation of silica at very high temperatures leads to a nitridation of the surface. 9'1~ Nitridated silica surfaces are very important in the semi-conductor and microelectronic industries. However, direct ammoniation of silica is -so far- a very uncontrollable procedure: the required reaction temperatures are very high (> 1500 K), a lot of side products are formed (typically Si2N20, silicon-oxynitride) and the nitride diffuses into the silica matrix. Direct nitridation of silica should therefore be considered as a bulk synthesis, and not as a surface modification technique. Due to the difficulties encountered in direct ammoniation, some researchers have tried a pre-activation of the silica surface, prior to the ammoniation. This preactivation usually consists of a replacement of the surface hydroxyls groups by more reactive groups. This can be done by CC14,1 5OC12,13'14 chlorosilanes, 15'16'17'1s B2H619 or BCI3. 2~ Also titaniumchlorides and fluorides can be used. 21 In these cases, the uptake of ammonia is enhanced typically with a factor 10. In this section, the ammoniation of unmodified silica will be compared to the ammoniation of activated
silica. The relevant reaction mechanisms will be discussed in detail.
384 2 Ammoniation of unmodified silica
2.1 At room temperature The adsorption of ammonia on an unmodified silica at room temperature is believed to be reversible. ~'2'3'4'5'6'7's However, the term 'adsorption' is a rather vague one. Therefore, we would like to clearly define the terminology, used in this text.
Total adsorption: The amount of gas adsorbed on the substrate, at equilibrium pressure. This value is frequently used in adsorption isotherms; the equilibrium pressure is expressed as (P/P0); cfr. chapter 8. Irreversible adsorption: The amount of gas still captured by the substrate, after prolonged degassing at the adsorption temperature. Physisorption: The fraction of the irreversible adsorption that is physically bonded to the substrate. Molecules, attached to the surface by a hydrogen bond, will also be classified as physisorbed, since no actual chemical reaction takes place. Chemisorption: The amount of gas captured by the substrate, by a chemical bond. Since chemisorption deals in fact with a chemical surface reaction, the classical adsorption laws are no longer valid. Chemisorption curves should be modelled by expressions for reaction kinetics. In these terms, the ammonia adsorption is not entirely reversible. Figure 12.1 shows the total and irreversible ammonia adsorption (at room temperature) on Kieselgel 60, thermally pretreated at 473,673 and 973 K. The adsorption capacity is exceptionally expressed as #/nm 2. The figure clearly demonstrates that the total ammonia adsorption on silica is determined by the silanol number, in a 1" 1 relationship. This means that ammonia adsorption on silica involves an attachment (by a hydrogen bond) of 1 N H 3 molecule to 1 silanol group. Such species produce an infrared band at 3419 cm ~, assigned by Peri ~ to the ~'3 valence vibration of ammonia.
385
#/rim= k~
Silanolnumber
W'~
NI-I3 total adsorption
n:n:n
NI-I3 irreversible adsorption
iiiiii iiiiii
473
673 Pretreatment temperature (K)
iiiili 973
Figure 12.1 Total and irreversible ammonia adsorption on silica gel, thermally pretreated at 473, 673 and 973 K. In this respect, FTIR spectroscopy provides an excellent tool in distinguishing
physisorption from chemisorption. Physisorbed ammonia molecules give rise to an absorption band at 3419 cm ~, whereas chemisorption involves Si-NH2 species, giving rise to 2 absorption bands at 3510 cm ~ (asymmetric stretch) and 3425 cm -1 (symmetric stretch) respectively. Blomfield and Little 8 stated that the OH stretching vibration of a silanol, hydrogen bonded to NH 3 shifts to 3080 cm ~. Unfortunately, we have never observed a significant band in this area. Morrow 3'4's and Fink 6 postulated that irreversibly adsorbed NH 3 exhibits a weak, coordinative bond to silicon atoms. Figure 12.2 shows the infrared spectrum of a Kieselgel 60, thermally pretreated at 973 K, reacted with ammonia at 373 K for 3 h and subsequently degassed at reaction temperature. A very small band around 3419 cm -1 can be observed, proving that the remaining ammonia on the silica surface is mainly physisorbed, and not chemisorbed.
386 (a.u.)
4000
'
35~)0
' 361)0 ' Wavenumbem (era-l)
3,~00
'
3200
Figure 12.2 FTIR spectrum of Kieselge160, thermally pretreated at 973 K, and reacted with ammonia at 373 K for 3h.
2.2 At moderate temperatures
Ammonia chemisorption only occurs at reaction temperatures above 673 K , 1'6'22 and proceeds according to two reaction mechanisms. The main reaction (A) is a substitution reaction: --Si-OH + NH 3 ~
--Si-NH z + H20
(A)
Fink et al. 6 have proposed reaction scheme (A') to explain this substitution. The key feature of this reaction mechanism is the fact that a certain proton mobility is induced on the surface silanols by the interaction with ammonia. 23 The excellent leaving group, H20, is subsequently replaced by N H 3. The rate-limiting step in this reaction is the formation and desorption of the water molecule. This mechanism could explain why Fink was able to reach a much higher degree of ammoniation, using a flow system, compared to other researchers using a static system.
387
H H H \]/ N H
H
0/
o/
0
i Si
I Si
I Si
~
/1\
/1\
N -[~,
/1\
I Si
/1\
o ,~l
(-~@
/1\
/1\
H
/1\ HH
~
@ I Si
/1\
_. N /
I Si
H
I Si
H _ ~OI
~
/1\
~
,:
H!IH \0~
~@ Si
I Si
~-
/1\
H H H , ~ l e>
~H-.~O///H i
0/
H \N /
Sli
S[i
/1\
/1\
(A3
Most authors agree that also a dissociative reaction with siloxane bridges occurs: - Si-O-Si-
+
NH3 ---,
- Si-NH 2 +
-- Si-OH
(B)
This reaction only occurs at relatively high ( + 7 7 3 K) pretreatment and reaction temperatures. Fink 6 proposed a pre-adsorption of the ammonia molecule, by the interaction of the lone pair N electrons with the d-orbitals of silicon:
H H
\I
N
H
Si
/
H
--...
/
I \
'~ -
/
Si
/1\
\
/ N Si
/1\
H
/ O
H
03')
Si
/1\
This adsorption would preferentially occur on strained siloxane bridges,
since
Morrow 4 suggested that strained, asymmetric siloxanes consist of one silicon atom that is more electrondeficient than the other.
388 2.3 At high temperatures
When the reaction temperature is raised to temperatures above 873 K, the 3450 cm -~ Si-NH2 band shows a tailing effect. Figure 12.3 illustrates this phenomenon by a FTIR spectrum of Kieselgel 60, thermally pretreated at 973 K and reacted with NH3 at 923 K for 5 hours. A band centred around 3405 cm ~ can be observed, attributable to silazane (Si-NH-Si) species. Intensity (a.u.)
,
-
3460
,
,
,
3420 3380 Wave,numbers (cm- 1)
Figure 12.3 FT1R spectrum of Kieselgel 60, thermally pretreated at 973 K and reacted with NH3 at 923 K for 5 h.
The formation of silazane species is probably due to a secondary reaction of the SiNH2 species with (strained) siloxane bridges, according to reaction (C): 24
H
H
\
N
I
/I
H
/
Si /
I
-
0 "\ Si-
I\
/ --
/I
Si
N
\ /
Si
/
0
\
In analogy with reaction (A), reaction (D) is also possible: 25
/
H
(c)
389 •-- S i - N H 2
+
- Si-OH
--,
- Si-NH-Si-
+
t/20
(D)
2.4 At very high temperatures Several recent studies 7'9'1~
have reported attempts to create silicon nitride by direct
ammoniation of silica, usually as a spin-off of the integrated circuit technology research. Most of these studies agree that at temperatures about 1473 K up to 20 25 % (w/w) nitrogen can be incorporated, but silicon nitride is seldom formed. The final product of this direct nitridation method is silicon-oxynitride (Si2N20) with residual silica. The nitridation is not restricted to the surface, but the N diffuses also into the bulk structure of the silica. No adequate mechanisms were presented to explain the observed reactions.
2.5 Summary The effect of reaction temperature on the ammoniation of silica is summarized in figure 12.4. It is noticeable that for all regions an overlap of effects can be observed: at almost every reaction temperature, a mixture of several surface species is formed. Again, it is also important to note that at higher reaction temperatures, the ammoniation of silica is not longer restricted to the surface, but will also involve reaction in the bulk skeleton siloxanes, combined with a diffusion of surface species into the bulk structure. The formation of Si3N4 by a direct ammoniation of silica is not impossible. However, a number of important restrictions limit the practical use of this method:
~
.
.
4.
The ammonia uptake capacity is rather low. Even after reactions at 1473 K for 17h, less than 25 % N incorporation is obtained. Pure Si3N4 is very difficult to synthesize. In most cases, mainly Si2N20 in a matrix of residual silica will be formed. These two phases are very difficult to separate. Very high reaction temperatures are required. The formation of nitrides is not restricted to the surface. Therefore, the coating of thin neoceramic layer cannot be achieved.
390
Exclusively surface reactions
Bulk rearrangements
I Si-NH-Si groups
t!
ii
1
i
I Si-NH 2 groups
[ I
1 NH3 (coordinative) I
273 373
I
I
I
I
I
I
473
573
673
737
873
973
I
I
I
1 0 7 3 1173 1273
Reaction temperature (K) Figure 12.4 General overview of the ammoniation of silica, as a function of reaction temperature.
3 Ammoniation of halogenated silica
The reaction between silica and halogenating reagents permits the direct replacement of hydroxyl groups with halogen atoms, yielding reactive = Si-X surface groups. Chlorination of the silica surface
Reaction with SOC12 One of the most common methods for the preparation of =Si-CI groups is the treatment of the silica with thionyl chloride. In order to achieve a maximum conversion, the reaction is preferably carried out at temperatures above the boiling point of SOC12. After reaction, the physisorbed gases have to be removed by heating the product at 473 K under vacuum.
391
Deuel et al. 27 first studied the reactions of SOC12 with the silica surface. They found a chlorine concentration, varying from 2.0 to 3.2 groups per nm 2, depending on the reaction conditions. Boehm et al. 2s'29 utilized this reaction to evaluate the hydroxyl group concentration at the surface of Aerosil. They found a maximal hydroxyl coverage of 3 . 1 0 H / n m 2 for a pretreatment temperature of 450 K.
In the same period, also Uytterhoeven, 3~
Elmer 3~ and Folman ~4 have investigated the reaction of silica with SOC12. Especially researchers from the states of the former USSR have performed detailed studies on the reaction mechanism of SOC12 with the silica s u r f a c e . 32 It was suggested that the anomalously low temperature of chlorination of the silica surface is related to the initial process of electrophilic substitution of a proton of the silanol group,
the formation of intermediate compounds and their decomposition,
according to reaction scheme (E).
-Si- 0
-Si - O - H
I
C1 - S - C1
=
-HC1
I
C1- S - 0
=
- S i - C1
"S02
II
O
Reaction with
rE)
CC14
Complete chlorination of the surface hydroxyl groups can also be achieved by treatment with CC14.1'33'34'35'36 The reaction is believed to proceed according to mechanism (F):
392
C1 \
COl4
C
/
/
CI OCCI2
HC1
\
CI
CI i '
,
+
H
OH
/ O
Si /1~
/1\
.
Cl \ CI- C - C1
Si
Cl
Si
Si"
/1\
/1\
The bimolecular reaction (G) is also possible: 2-Si-OH
+
CC!4 ~
2-Si-C!
+
CO2
+
2HC!
(G)
Complete conversion is only achieved at temperatures above 650 K.
Fluorination of the silica surface Complete fluorination of hydroxyl groups is reported to be possible by the reaction of silica with 30 % NH4F solution at 973 K . 37 Treatment of Aerosil with pure HF as well as in various solvents, carefully excluding water, leads to a relatively high fluorine content of the modified product, indicating that fluorination is not limited to the surface, but also attacks the silica skeleton, as Only with the use of SF4 as a fluorating reagent, complete replacement of hydroxyls groups without side-reactions was observed. 38
Ammoniation of the halogenated silica surface Few studies have been devoted to the ammoniation of halogenated silica. Folman ~4 suggested reaction (H) for the ammoniation of silica, activated with SOC12: --Si-CI
+
N i t 3 --,
---Si-NH 2 -I- H C I
(n)
393 Peri 1 stated that NH4C1 is formed on the surface when chlorinated silica is exposed to ammonia vapour. Annealing the silica at 673 - 873 K produces sublimation of NH4C1, resulting in = Si-NH2 surface groups. Most studies of the ammoniation of halogenated silica have been performed on chlorosilylated silica. Therefore, the discussion of the reaction mechanism and kinetics is deferred to the next paragraph.
4 Ammoniation of chlorosilylated silica
4.1 Room temperature ammoniation The ammoniation of chlorosily!ated silica will be exemplified for the case of trichlorosilylated silica. 15,18 Figure 12.5 shows the ammonia uptake (irreversibly adsorbed and/or chemisorbed) on silica gel with and without trichlorosilane modification. The silica gel was thermally pretreated at 973 K and the trichlorosilylation was performed at standard conditions (623 K, l h). When the silica gel is treated with trichlorosilane, prior to the ammoniation, the ammonia uptake capacity is enhanced with a factor 5 - 10. This enhancement is effective in the entire reaction temperature region. The room temperature ammoniation of chlorinated silica surfaces is completed within 5 minutes. Obviously, all ammoniation reactions occur in the gaseous phase. Figure 12.6 shows the FTIR spectrum of Kieselgel 60, trichlorosilylated at standard conditions and ammoniated at room temperature. The most intense bands in the spectrum are the ones at 3160, 3060, 2925 and 1410 cm -~. These bands can be assigned to NH4 + vibrations. The bands at 3510, 3425 and 1550 cm -~ can be attributed to Si-NH2 species. More detailed information about the different band assignments is summarized in table 12.1.
394
2.5 -
mmol NH3 / g dry silica 1~3 unmodified silica
trichlorosilylated silica
2.01.5-
N
1.00.5-
298
473 673 973 Temperature (K) Figure 12.5 Irreversibly adsorbed and~or chemisorbed ammonia on silica gel as a function of the reaction temperature, with and without TCS modification. Intensity (a.u.)
4ooo
3~00
3~
25~
W a v e n u m b e t s (r
1)
2~
1~00
Figure 12.6 FTIR spectrum of Kieselgel 60, trichlorosilylated at standard conditions and ammoniated at room temperature.
395
Table 12.1 Infrared band assignments for surface species on silica gel Wavenumber (cm "l) 3745 3740-3500 2270 3160 3060 2925 1775 1410 3510 3425 1550 3400-3350 935
Assignment free Si-OH vibration (stretch) bridged Si-OH vibration (stretch) Si-H vibration (stretch) NH4 + vibration (v3) NH4 + vibration (v2 + v4) NH4 + vibration (2 1)4) NH4 + vibration NH4 + vibration Si-NH2- stretch (asymmetric) Si-NH2- stretch (symmetric) Si-NH2- bending (asymmetric) Si-NH-Si- stretch Si-N vibration
A remarkable feature is the perturbation of the 2270 cm 1 Si-H stretching vibration. Upon ammoniation, this band splits into at least 3 distinct bands. Actually, this band contains all the information on the concentration and the nature of the surface groups on ammoniated, trichlorosilylated silica gel. This means that knowledge of the intensity and peak position of the different shoulders in the Si-H vibration band results in a prediction of the concentration of the N and CI containing species on the surface and in a distinction of primary and secondary species. These fascinating characteristics of the silane vibration will be fully covered further in this chapter. The observations that (1) the ammonia uptake capacity is exactly the two-fold of the initial Cl-concentration; (2) the chlorine groups remain on the surface after ammoniation at room temperature and (3) the infrared spectrum shows intense bands, assigned to NH4 + and Si-NH2 species, lead to following reaction mechanism (I):
396
/ C1 -Z S i - O - S i - H
,/NI-I 2---NI-I4CI + 4 NI-I 3
~
-Si-
O- Si- H
\ Cl
f0
" ~ NH 2---NI-I4CI
The dashed line between NH2 and NH4CI means that the NH4C1 is physisorbed on the silica surface. Support for this comes follows the observation that NH 3 and NH4CI reversibly form the adduct NHaC1.3NH3 .39 The species, formed by reaction (I) will be referred to further in this work as 'ammoniated primary species'. Obviously, the secondary species will react with ammonia in a similar fashion, according to reaction
(J): Si - 0 ~
/
~" Si -Si-
O
C1
i
-Si
- 0 ~
+ 2 NH 3 H
./NI-If-
N i t 4C1
~ Si ~ -Si-
O
H
These species will be referred to as 'ammoniated secondary species'.
4.2 Ammoniation at higher temperatures When the reaction temperature is raised above 423 K, the reaction mechanism becomes more complex: NH4CI will sublime from the surface and the amine function will gradually convert towards silazane and nitride species. Some of the major conversions of surface species can be seen in figure 12.7, showing the FTIR spectra of trichlorosilylated silica gel, ammoniated at 273, 673, 823, and 1023 K respectively. A first observation is the complete disappearance of the NH4CI bands (3160, 3060, 2925 and 1410 cm -~) when the temperature is raised, due to the sublimation of ammoniumchloride at higher temperatures. In spectra (c) and (d), also the silane band has disappeared. In the same spectra (c) and (d), a band originates at 3745 cm -~, assigned to free silanols. This band is due to
397 Intensity(a.u.)
/J j
Wavenumbers (can-l)
Figure 12. 7 FTIR spectra of trichlorosilylated silica gel, ammoniated at (a) 273 K," (b) 673 K; (c) 823 K and (d) 1023 K.
reaction (K) with strained siloxane bridges, yielding amine species and a free silanol. --- Si-O-Si-
+ NH 3 --, -= Si-NH z +
--- SiOH
(K)
In spectrum (b), the symmetrical Si-NH2 stretching vibration (3425 cm ~) exhibits a tailing, which becomes a prominent shoulder in spectrum (d). This is due to a band in the region 3400 - 3350 cm ~, which can be assigned to the N-H stretching vibration of silazane species. Amine species are progressively converted to silazane species as the reaction temperature rises. Van Der Voort TM has used a combination of several analytical techniques to quantify the surface species of ammoniated, trichlorosilylated silica as a function of reaction temperature. The results are presented in figure 2.5.8, clearly showing two separate phenomena. At temperatures below the sublimation point of NH4CI, the concentrations of the different surface species are constant. Apparently, the presence of NH4CI stabilizes the adsorption of the NH2 species. At reaction temperatures above
398
1.2
mmol/ g
1.0 0.8 0.6 0.4 --4
0.2
i
273 373
i
i
473 573 673 773 Reaction temperature (K)
i
873
Figure 12.8 Distribution of the different surface species on trichlorosilylated silica gel, after reaction with ammonia at varying temperatures (+ NHx; * CI (non reacted)).
the sublimation point of NH4C1, the surface concentration of NH4C1 is 0, and the NHx species increase exponentially, with a consequent exponential decrease of the nonreacted Si-C1 species. At this point, obviously the stabilizing function of NH4C1 has vanished. The physical interaction between =Si-NHz and NH4C1 is reflected in the N-H stretching region of the amine functions. When the temperature is raised, the symmetrical Si-NH2 band shifts from 3505 cm ~ to 3530 cm -~ and the asymmetrical vibration shifts from 3430 cm ~ to 3450 cm -~. Van Der Voort has shown that the peak shift is a discrete function of temperature (figure 12.9). The sudden peak shift in the temperature region at which NH4CI sublimes strongly suggests that the physisorbed NH4C1 surface species are responsible for this phenomenon. NH4C1 stabilizes the formation of the Si-NHz species by a physical bonding. In doing so, the actual N-H vibration in the Si-NHz---NH4CI system is weakened, causing a shift towards lower wavenumbers. Therefore, the infrared assignments, presented in table 12.2 were suggested.
399
Intensity (a.u.)
673 K
473 K
3800
3600
3400
Wavenumber (cm- 1)
Figure 12.9 FTIR spectra of trichlorosilylated silica gel, reacted with ammonia at temperatures (a)- (e). Table 12.2 Infrared band assignment for the N-H stretching region 3535 cm -~ 3505 cm -I
asymmetrical Si-NH2 (stretch) asymmetrical Si-NH2---NH4CI (stretch)
3450 cm 1 3430 cm ~
symmetrical Si-NH2---NH4CI(stretch)
symmetrical Si-NH 2 (stretch)
4.2.1 Silazane functions Fink 6'~~ has studied the conversion of Si-NH2 functions towards silazane (-=Si-NH-Si-=) groups for the reaction of ammonia with untreated silica. He held following two reaction mechanisms responsible for this conversion:
400
- S i - NH 2
> 775 K
-Si \ :
- S i - NH 2
H
\/ N
NH
vacuum
H
H i
:
I
+ NI-I3
~)
-Si /
I\
\
I
(L) Si--
I\
OH
According to Fink's observations, reaction (K) only occurs at temperatures above 775 K in vacuo and above 975 K in vacuo. In a N H reaction (B).
3
a NH 3
flow. Reaction (L) occurs above 375 K in
atmosphere, reaction (L) will not occur, due to a competition with
- Si-O-Si-
+
NH 3
~
~ Si-NH
2
+
- Si-OH
(B)
Van Der Voort ~7'~g used infrared curve fitting techniques, to show that during the ammoniation of trichlorosilylated silica, the formation of silazanes already starts at 423 K in vacuum conditions and at 523 K in an ammonia atmosphere.
Since these
temperatures are much lower than the ones provided by Fink, a third mechanism must be included: /
/
- Si-O-Si-H
- Si-O-Si-H
\
\NIl 2
-si_o_sg cl
=
/
NH
+ HC1
(M)
- Si-O-Si-H \
In the next section, the actual existence of the partial chlorinated, partial ammoniated silyl species will be evidenced.
401 4.2.2 Nitride functions ~s Figure 12.10 shows the FTIR spectrum of a trichlorosilylated silica gel, ammoniated at 1023 K. This spectrum confirms the above statements. The free hydroxyl band is caused by reaction (B). The symmetric Si-NH2 band (3425 cm -~) exhibits a distinct tailing, due to the Si-NH-Si (N-H stretching) vibration (3400 - 3350 cm~). Intensity (a.u.)
3800
3700
3600
3500
3400
33'00
3200
Wavenumber (cm- 1) Figure 12.10 FTIR spectrum of trichlorosilylated silica gel, reacted with ammonia at 1023 K for 6h.
Unfortunately, the formation of nitride species cannot be studied by FTIR, since the Si-N vibration (935 cm ~) is totally disguised by the strong lattice vibrations of the silica substrate. Furthermore, the 935 cm ~ band is not a typical nitride indication, but a sum of Si-N vibration of all Si-N containing species. A very useful source of complementary information is XPS (X - ray Photo - electron Spectroscopy), which is a typical surface analysis technique. XPS is often used as a o r neoceramic valuable tool in studies of the interaction of silanes with s i l i c a 4~ coatings. 44'45 The basic principles of the XPS technique are described in appendix B.
402 Trichlorosilylated samples, reacted with ammonia at temperatures varying from 298 to 1023 K were analyzed by XPS, especially in the N(ls) region (figure 12.11). no.
of counts
300
250 200
150
100
50'
0
396
Figure 12.11 1023 K.
I
I
398
I
I
I
I
400 402 Binding Energy (eV)
I
404
XPS N(ls) core level spectrum of trichlorosilylated gel, ammoniated at
The main peak at 400.5 eV is attributed to the Si-NH-Si species. This peak also exhibits tailing. A shoulder at 401.8 eV is attributed to Si-NH2 species, whereas the shoulder at 398.5 eV originates from nitride functions. Band fitting (based on a 10/90 linear combination of Lorentzian and Gaussian curves) leads to the conclusion that about 25 % of the N-species is present on the silica surface as Si-NHz, 50% as Si-NH-Si and 25 % as nitride functions. Figure 12.12 shows the relative contributions of the different XPS bands, as a function of the temperature. At room temperature, the surface is covered with 50 % amine species and 50 % NHaCI, as could be expected from reaction (E). In the temperature region 298 - 473 K, the relative contribution of the Si-NH2 species increases, whereas NH4CI gradually sublimes from the surface. In the region 473 - 673 K the Si-NHz species decrease and at temperatures above 673 K, the surface is dominated by silazane species. At higher temperatures (around 973 K) nitrides are created. Finally, in figure 12.13, the temperatures at which the N-containing surface species are formed on a trichlorosilylated surface are compared to an unmodified silica.
403
Relative contribution (%)
100
80
60
40
20
0
,~
l..a
273
ixa
-
473 673 873 Reaction temperature (K)
1023
Figure 12.12 Relative contribution of the different XPS bands, as a function of the ammoniation temperature: = N H 4 C I . + Si-NH~" * Si-NH-Si; o Si-N.
Occurrence of species Nitrides
......
i!iiiii~iii!i!i!ii!~iii:~!i~i!~!i~!!i!~iiii:iii!i!iiiiiii!ii~iiiii~i~iiii!~iiiiiii~i~!i!iiii#~iiii~ii~ii!!!!~ii~i~i!~i~i!i!ii~ii!i:iii~!i~!~:i~i~i~!ii~!i!i!i!i!i!i~ii~!i!iiii] $i-NH 2
NH4Cl
N H 3 (coordinative) J
. I 373
.
t,.
473
I 573
. . . . l. . . . . 673
i
773
I. 873
i
I
l
973
1073
1173
1273
Reaction temperature ('K) Unmodified
silica
~
Tric~orosilylate~
silica
Figure 12.13 Comparison of the ammoniation of unmodified silica gel and trichlorosilylated silica gel.
404 A significant lowering of the reaction temperatures at which the silazanes and nitrides are created can be observed. This is not the only advantage of chlorosilylation. The reader is referred to figure 12.5, showing an ammonia uptake enhancement factor upon trichlorosilylation > 10. 4'3 Diagnostic value of the Si-H vibration ~7
In the previous section, it was concluded that the room temperature ammoniation of trichlorosilylated silica gel proceeds according to reaction (I)" / Cl - S i - O - Si- H
/ NH :---NI-I4CI + 4 NH 3
~
- S i - O - Si- H
\ C1
03
~ NI-I:---NI-I,C1
The presence of the chemisorbed Si-NH2 and the physisorbed NH4C1 causes a perturbation of the Si-H stretching vibration. This is illustrated in figure 12.14. mmol/g
ii_ (a)
_
~
0.21
-.-------
2350 2300 2 89 2200 2150
21130
Wavenumbers (cm-1)
Figure 12.14 content.
Perturbation of the Si-H stretching vibration as a function of ammonia
405 As the ammonia contents on the samples increases (indicated right in the figure), a gradual perturbation of the Si-H vibration can be observed, showing peaks at 2240cm ~ and 2190 cm -~. Apparently, the perturbation of the silane band is a function of the ammonia content on the samples. This perturbation is used as a probe to estimate the concentration of the N-containing surface species, using Partial Least Squares regression (PLS) and to characterize these different species by infrared curve fitting, combined with 29Si CP MAS NMR. 4.3.1 Quantitative determination of the surface species, using PLS A classical way to determine the concentration of the N-containing surface species consists of a combination of volumetric results, Kjeldahl and titrimetrical analysis. This procedure is very time consuming, and extreme care is needed to avoid hydrolysis of the sample. Once a calibration set is created, a faster analysis could be performed using the different stretching and bending infrared vibrations of the surface species, normalized to a reference band, attributed to an overtone lattice vibration (1933 - 1780 cml). However, the large differences in absorbtivity between the NHaCI and the Si-NH2 species, result in a poor prediction for the latter ones. On the other hand, the perturbation of the Si-H stretching vibration (2270 cm ~ ) upon ammoniation was striking and not yet satisfactorily understood. Performing PLS of this silane peak versus a calibration set results in a very good prediction model, as well as a better understanding of the mechanisms, causing this perturbation.
Principles of Partial Least Squares Regression Partial Least Squares Regression (PLS) is a multivariate calibration technique, based on the principles of Latent Variable Regression. Originated in a slightly different form in the field of econometrics, PLS has entered the spectroscopic scene. 46'47'4g It is mostly employed for quantitative analysis of mixtures with overlapping bands (e.g. mixture of glucose, fructose and sucrose). 49'5~
406
PLS looks for relations between two data matrices. The first one (X) is the spectral matrix. Each row of X is a spectrum of a calibration standard and each column is the absorbance at a given wavenumber. The other matrix (Y) is the analytical matrix. Each row is a calibration standard and each column contains the concentration of the compound in the mixture. Partial Least Squares Regression is one of the many available regression techniques. Regression techniques are used to model the relation between 2 blocks of variables, called independent or x variables and dependent or y variables (figure 12.15 a). The general regression equation is (figure 12.15 b): Y = XB + F
(1)
Where Y is an NxJ matrix containing the y variables, X is an NxK matrix containing the x variables, B is a KxJ matrix containing the regression coefficients and F is an NxJ matrix of residuals, the part of Y not modelled by the regression equation. It is assumed here that the data have been mean-centred around the variable-wise means and appropriately scaled. This eliminates the constant term in the equation and makes it easier to carry out calculations and to explain the theory. The above equation can also be written as"
K
(2)
k=l
One should be aware that in this equation Y and X contain known (measured) values and that B and F are unknown. A general solution without constraints on B or F is not possible. Least Squares methods are based on a constraint of F, namely that the sum of squares (SS) of F has to be minimized. Under this constraint, B can be calculated. The method of multiple linear regression (MLR) gives a possible solution of B, but not without problems. It requires that N is equal to or larger than K (condition 1) and the obtained values for B are sensitive to collinearity (condition 2: collinear variables are bad, orthogonal variables are good). This means that small changes in the calibration data, resulting from random noise, can introduce large changes in the values of B. Unfortunately, in spectroscopy, there are more variables than objects (a typical value of K > 100 and N usually < 10) and spectroscopic data are often very collinear.
407
Many solutions for getting rid of collinearity exist. 5~'52 Some of them make use of latent variables. This means that the K variables in X are replaced by A variables in a new matrix, called T (figure 12.15 c). The first thing that should be noticed here is that A is always smaller than or equal to N. This means that the requirements for condition 1 are fulfilled. The way of calculating the values in T also forces it to consist of orthogonal column vectors, so that also condition 2 above (no collinearity) is met. The regression equation with the use of T is expressed as follows (see also figure 12.15 d). Y = TQ' + F
(3)
Q is the matrix (JxA) of the regression coefficients between T and Y. The only thing that is needed now, is a way of transforming X into an appropriate T. Many methods for doing this only take the data in X into account. The special property of PLS is that also the data in Y are used for shaping the values in T, making them ideal for describing the relationship between X and Y. Once the transformation of X to T is known, X can be substituted back in equation (3), which then becomes: Y
=
X B p l ~ -t-
Fp~s
(4)
This is very similar to equation (1). The subscript 'pls' has to be used because the values in B and F are specific for PLS and different form those of other regression methods. There are many ways of calculating the results of equation (4). Theoretical explanations of how T and Bpls can be calculated are found in the literature 53,54,55,56,57,58,59,60,61,62 An important property of PLS is A, the number of latent variables used, also called rank. With A maximal, the PLS model is the same as the MLR model. With A very small, the model may be too simple and not able to explain the relationship between X and Y. So, the equation (4) has to be rewritten as: Y = XBpls,A + Fpls, A
(5)
The subscript A is used to identify the different models of rank A, where A decides what B and F look like. Once A is determined and found acceptable, the index A is usually left out. This relation (equation 5) can be looked upon as a multivariate
408 'calibration curve', similar to Beer's law. Putting in a spectrum (X-matrix" absorbance v s . wavenumbers) of an unknown mixture, the concentrations of the different components (Y-matrix) can be predicted. The possibility of adapting the parameter A gives a flexibility in modelling. It gives the slight disadvantage that some criterion has to be given for finding an adequate value for A. One criterion used in this section is based on considering the sum of squares of F, also expressed as II F II, since a good model would have a small disagreement between Y and its model XBp,~,A. The criterion for II F II is not the lowest value, but reaching a reasonably low plateau in the values of II F II with an as low value of A as possible. This gives a simple low rank solution that works well as a calibration solution and is easily interpreted. An advantage of PLS is that the latent variable structure can be studied. This can lead to the detection of outliers and data classes. There are also clues about which spectroscopic variables (wavelengths) influence the solution the most. A detailed view of PLS is given by the equations (6) - (8), also visualized in figures 12.15 e and f.
t. = Xw. (a= 1,...,A) X = TP' + Epts, A Y = TQ' + Fpt~, A
(63 (7) (8)
The wa in equation (6) are the PLS loading weights. They are explained in the theory in references 53 - 62. Equation (7) shows how X is decomposed bilinearly (as in principal component analysis) with its own residual Ep~,A. T is the matrix with the score vectors as columns, P is the matrix having the PLS loadings as columns. Also the vectors of P and wa can be used to construct scatter plots. These can reveal the data structure of the variable space and relations between variables or groups of variables. Since PLS mainly looks for sources of variance, it is a very good 'dirty data' technique. Random noise will not be decomposed into scores and loadings, and will be stored in the residual matrices (E and F), which contain only 'non-explained variance'. The test of the quality of a PLS model should always be done. Quality of the model is often determined by A, the rank, so a rank determination is included in the quality testing. For testing a PLS model, samples of known composition that were left outside the calibration set are used. By using the predicted value ~' for these samples,
409
and comparing them with the known values, the % error in the prediction can be found: '~r = XtestB
(9)
can be compared with the known values Ytest as follows: D =
Yt. ,
(10)
D is the matrix of deviations between 'true' and calculated. Since B is dependent on the rank A, also D will be dependent on it. A useful expression is the prediction residual error sum of squares (PRESS):
N
J
ss - E E 4
(11)
n=l j=l
where j = 1,...,J is the index of the y-variables and n= 1..... ,N is the index of the samples. Crossvalidation is carried out by leaving out selected samples from the model building phase and using them for prediction and PRESS calculation. By doing this in a cyclical manner, all samples will be left out exactly once and PRESS as a function of model quality becomes a good tool for diagnosis: the minimum in PRESS gives the rank of the best PLS model and the PRESS value at that location gives an idea of the error in the predicted concentrations. The basic principles of PLS are graphically summarized in figure 12.15.
Quantitative determination of the surface species, using PLS The perturbation pattern of the Si-H band as a function of the ammonia loading at room temperature (figure 12.14) was reasonably straightforward. When the Si-H perturbation as a function of reaction temperature is studied, the situation becomes more complicated, as illustrated in figure 12.16. As the reaction temperature increases, the perturbation seems to reverse.
Figure
12.17 shows the Si-NH 2 and NH4CI concentration (respectively indicated as + and * in the figure) on the silica surface as a function of reaction temperature.
410
spectral vat.
N[
NE] N[ ND NI
Latent variables
COliC.
J
K
J
K
A
N
J
J
N
K
known
known
unknown
unknown
B contains regression coefficients and F contains residuals. (d)
J
A
J
J
09
Figure 12.15 Visualization of the principles of PLS. (a) The data matrices containing K spectral variables for N samples and J concentrations for the N samples are known from the measurement~sample preparation. (b) The model relating the known quantities X and Y is build using the unknown matrices B and F. (c) The K spectral variables are transformed to A latent variables. (d) The latent variables are used in the regression equation. ~ e regression coefficients are in Q'. (e) The spectral matrix X is decomposed into PLS score and loading vectors (t and ~) and a residual matrix E. 09 The concentration matrix Y is decomposed into PLS score and loading vectors (t and q) and a residual matrix F.
411 Inmasity (a.u.)
2350
2300
22'50 22'00 Wavemtmber (era- 1)
21'50
Figure 12.16 Perturbation of the Si-H band as a function of reaction temperature, in excess ammonia. Reaction temperatures: (a) 273 K; (b) 473 K.
1.2 mmol/g 1.0 0.8 0.6 0.4 0.2 I
273
I
I
I
323 373 423 473 Reaction temperature (K)
523
Figure 12.17 Concentration of NH4C1 (*) and Si-NH,. (+) on the silica gel surface, together with the PLS predictions.
These concentrations are determined by volumetric ammonia uptake measurements, Kjeldahl analysis and titrimetrical CI analysis. The trichlorosilylation occurred at standard conditions and the ammoniation was performed for 5 h in excess ammonia.
412 PLS regression is performed using the data in figure 12.17 as the analytical matrix (Y) and the photoacoustic datapoints in the region 2320- 2125 cm ~ (figure 12.16) for the respective spectra as the spectral matrix (X). In order to establish the rank to be used in the PLS regression, a Cross-Validation is performed. Figure 12.18 shows a graph of the percentage of explained Y - variance as a function of the rank of the PLS regression. Apparently, the optimum number of score and loading vectors (rank) is 3. 100
explained Y-variance (%)
80 60 40 20 I
0
1
I
2 PLS rank
I
3
Figure 12.18 Explained variance for each chemical component, as a function of the rank used in the PLS prediction model: + Si-NH2; 9NH4CI.
The PLS prediction is shown in figure 12.17 (dotted lines). The Standard Error of Prediction is 0.06 mmol/g for Si-NHz and 0.03 mmol/g for NH4CI , being fairly small, considering the uncertainties in the calibration set. It can therefore be concluded that PLS is able to predict the concentration of NH4CI and Si-NHz on the silica surface, using the silane region as the only spectral dataset.
413 4.3.2 Mechanistic study of the perturbation of the silane band
Origin of the Si-H perturbation The peak position of an infrared band is not only caused by the nature of the adsorbing species, but is also affected by the electronic environment of these species. The shifting of infrared bands, caused by varying electronic environments, is often a source of supplemental information. It was mentioned earlier that the Si-H vibration of trichlorosilylated silica gel shifts from 2270 cm ~ to 2260 cm ~ upon hydrolysis, because of the replacement of the surface CI groups by OH groups. Very recently Gillis-D'Hamers 63 used the band shift of the free silanols on silica to predict the concentration of the perturbed silanols. This is a typical example of how the study of one band reveals information on other species in the neighbourhood. The perturbation of the Si-H vibration of ammoniated trichlorosilylated silica (figures 12.14 and 12.16) consists at first sight of at least 3 distinct bands, indicating that at least 3 different species with an Si-H band exist on the surface. The strong electron donating effect of the amine functions will cause a low wavenumber shift of the Si-H band. The first issue to be dealt with is whether NH4CI or Si-NH2 is causing the perturbation of the silane peak. Figure 12.18 shows that a good prediction of NHaCI requires a model of rank 3, whereas Si-NH z can be estimated very well (95 % of the variance) using only a model of rank 1. This evidences a strong 'first-order' relationship between the concentration of the Si-NH2 species and the perturbation of the silane peak. A PLS regression of rank 1 explains only about 60% of the variance in the NH4C1 concentration matrix, indicating that the relationship between the NH4CI concentration and the spectral matrix is more complex. This means that the predominant role in the silane peak perturbation is played by the Si-NH2 species, and only secondary effects can be attributed to NH4C1. Also on theoretical grounds, it can be expected that the chemically bounded, electron donating amine function would have a larger impact on the electronic environment of the Si-H species, than the physisorbed NHaCI salt. This is probably the reason why Low ~5 observed no perturbation of the silane peak, when he sublimed NH4CI on the chlorosilylated silica surface.
414
Assignment of the different Si-H species Low ~5has also observed the silane band perturbation. He performed on-line infrared transmission measurements to ascribe these bands. His conclusions can be summarized as follows"
Band 1
2270 cm "'
- Si-O-S~~
(I)
Band 2
-i 2240 cm
/ NH 2--NI'I4Cl ZSi-O-Si-H \el
0I)
Band 3
2190 em-1
/NH 2 --NH4C1 ZSi-O-Si-H 1
0/I)
Species (I) will be referred to 'non ammoniated primary species'; (II) 'partially ammoniated primary species' and (III) 'completely ammoniated primary species'. Low assigned band 2 to an intermediate phase. This partially ammoniated chlorosilyl group is indeed the major contribution to this band for room temperature ammoniation. However, the contribution of band 2 in e x c e s s N H 3 is surprisingly high. It seems for instance unlikely that the significant band at 2240 cm l in figure 12.14 (e) would only be caused by species (II), since the reaction occurred for prolonged reaction times in excess ammonia. Therefore, one should take into account another species, resulting from the secondary reaction in the chlorosilylation step: ~ S i - O. ~Si- O I
H ~ NI-I2 - - N H 4 C1
Species (IV) will be referred to 'ammoniated secondary species'. This species would absorb in the neighbourhood of band 2 (2240 cm~; species (II)), but due to the absence of a CI surface group, presumably at the low wavenumber side of this band. This hypothesis is proved, using infrared curve fitting and silicon NMR.
415 Ammoniated secondary species
A distinction between primary and secondary species can be made using 29Si CP MAS NMR. Primary species cause a chemical shift at -36 ppm (reference: TMS), whereas secondary species cause a shift at -60 ppm. Following the procedures of Maciel and Sindorf, 64'65which take into account the dependence of the CP (Cross - Polarisation) line intensifies with the CP contact time, the percentage of primary and secondary species is calculated. In figure 12.19 the contributions of the three bands in the silane stretching region (established by curve fitting software) are correlated versus the percentage of secondary species on the surface, as determined by NMR.
60%
Relative contribution to Si-H band (%)
5og 40% 30~
20%
10% l
0%
20%
I
I
40% 60% % secondary species
I
80%
100%
Figure 12.19 Relative contribution of the three bands in the Si-H stretching vibration as a function of the number of secundary species(m band 1; + band 2; * band 3).
Ammoniated secondary species (species (IV)), with one NH2---NH4CI group, can only contribute to peak 2, and not to peak 3, since this band requires two NH2---NHaCI groups. Therefore, the gradual relative decrease of peak 3 and increase of peak 2 as a function of the number of secondary species suggests the contribution of the secondary species to peak 2.
416 Figure 12.19 also shows an increase of peak 1 as a function of the number of secondary species. Since this peak represents the amount of non-reacted chlorosilyl groups (species (I) and (IV)), this rising tendency indicates that secondary species are less reactive towards NH3 at room temperature than the primary ones, which is reflected in a larger fraction of non-reacted species. Figure 12.20 shows the loading-vector ('eigenspectrum') of the first dimension, explaining 74,9 % of the variance in the spectral data set, and predicting 82,5 % of the variance in the calibration matrix. The most prominent feature to be observed is the inverse relationship between band 1 and 3, which was indeed expected to be the most important source of spectral variance. Loadings
9
.
L
2350 2300 2250 2200 2150 2100 2050 Wavenumber (cm-1) Figure 12.20 PLS spectral loading vector ('eigenspectrum') of rank 1. The behaviour of band 2 in the first eigenspectrum seems rather ambiguous, however. This band has a positive as well as a negative contribution to the loading-vector, suggesting that band (2) in fact consists of two poorly resolved bands. Figure 12.21 shows in detail the silane stretching vibration of an ammoniated trichlorosilylated silica gel, containing 45 % of secondary species. Besides the peak at 2240 cm ~, attributed to species (II), a band at 2225 cm ~ is observable. This band is found in all the silane bands of samples, containing secondary species, and can therefore be attributed to the ammoniated secondary species (IV). Correlation of the
417
Intemity (a.u.)
2325
,
,
,
2275
2225
2175
wavemtmbers (era-1)
2125
Figure 12.21 Deconvolution of the Si-H stretching vibration on an ammoniated trichlorosilylated silica sample, containing 45 % of secondary species.
integrated area of this peak vs. the percentage of secondary species as determined by NMR gives rather poor results, especially with samples, containing a small fraction of secondary species. This is probably due to the low spectral resolution of the 2240 cm l and the 2225 cm l bands. 4.3.3 Conclusion
Partial Least Squares Regression is a valuable tool in FTIR-spectroscopy, not only for (routine) quantitative analysis of mixtures, but also as a research application. Due to its ability to expose correlations in complex, multivariate data sets, PLS is gaining importance rapidly in spectroscopy-assisted-research. PLS has been used mainly for calibration purposes in analytical chemistry. In this case the determination of unknown concentrations is the most important demand. In spectroscopic research, there is also the interpretation of diagnostic plots such as the score plots and loading plots as a function of reaction mechanisms and spectroscopic background knowledge. Also the interpretation of rank as complexity of a mechanism is a valuable tool. A nice property of latent variable methods is that they do not demand advanced knowledge of the system studied, but that the measurements
418 themselves build up the model and give an indication of what is going on. It is expected that the use of methods like PLS in the study of complex systems and reaction mechanisms in physical chemistry will increase in the future.
5 Ammoniation of silica, activated with boron compounds
5.1 The boron-nitrogen chemistry The concept of isosterism between the atom groupings B-N and C-C, when introduced more than 60 years ago, implied the expectation that a relatively extensive boronnitrogen system of compounds, analogue to hydrocarbons, should be forthcoming. With reference to the principal classes of aliphatic hydrocarbons, Wiberg 66 subdivided the boron-nitrogen system into three groups of amineboranes ( B H 3 ~ NH3), aminoboranes (BH2NH2) and borazines (BHNH). Although most of the known unsubstituted boron-nitrogen compounds are indeed isosteric analogues of hydrocarbons, their number is still small and largely confined to cyclic structures. 67 The reaction of B2H 6 with N H 3 has gained particular interest, because it leads ultimately to thin films of boron nitride (BN). 68'69'70These thin films have applications in the semi-conductor industry, as well as protective coatings. In 1992, Carpenter and Ault7~ postulated an initial reaction of B2H 6 with N H 3 to form the H 3 B - N H 3 adduct, followed by the rapid elimination of H2 at elevated reaction temperatures to form H2B-NH2. In very general terms, the reaction of B2H6 with N H 3 is summarized in figure 12.22. A vast amount of recent publications is devoted to the preparation of various (BNH)x-polymers and their conversion towards B N . 72'73'74'75'76'77'78'79'80'81'82'83'84'85'86'87'88'89'90'91 Unfortunately, very little is known about the structure of the preceramic BH precursors or the mechanisms for polymer decomposition and conversion. A typical example of a borazine gel is shown in figure 12.23. Rye at al. 7z recently suggested a mechanism of subsequent NH3 losses in the conversion of a borazine polymer towards BN. A distribution of fused ring structures is formed from a series of thermally promoted borazine ring openings, followed by a B-N bond rotation that brings the newly generated, reactive B-NH fragments in the
419
2H6 I I >325K
[
478 K
I
I
'['
H I
I
H ~ N / B ~ N~H I I B N ~ B xH I-I/ ~
1' I B3N3H6t. . . . .
678 K
I
H 973 K
Figure 12.22
I
,~
I
Borazine
The reaction of B.J~6 with NH3.
H~ N /
H~ N /
H~N/ I H-..,.N/B ~ N / H H-..,.N/ ~ N / H H.,,.N/B ~ N / H I
I B
I
0I3C)3Si~ ~ B ~ N / ~ N
B
/
B
~N /
~N n
,i
Cl
Figure 12.23 Borazine gel.
vicinity of terminal, saturated borazine ring NH, or bridging NH groups. Subsequent hydrogen transfer, expulsion of N H 3 and ring closure would result in an efficient formation of fused-ring fragments (figure 12.24). Further consolidation of these fused-ring units into h-BN will be a much more demanding process since diffusion of the relatively large and probably interconnected fragments is more restricted, and the residual hydrogen resides in more sterically congested regions that should hinder dehydrogenation processes. Thus, it would
420 /
I
HN
I
I
B
H
NH
I B
H
/B
1
B--
H
HN
I B
H
B~.
HN
NH
I B
NH
I
B--
HN NH
HN
NH
N
H
NH
I
H
HN /
tJ
I
~N/ H
H
B/
N ,/
~B~N~B~
I
I
I
I
I I
Figure 12.24
HN/
~NH
H
NH
I n
I n
HN~
I n
I S H
I
~NH
I
B
~B/N.~
I
H N~B/N~
aN
~ B
s I . . /"..~"---fN..B/N'-
"~B ~
I
I
NH
I
HN
I ~'NH
HN
--4--
H
I
H
I NH
]
HN / HN
I
HN
H
I tN
tt
I
Thermal conversion of borazine polymers, according to Rye. 72
appear reasonable for diffusion and dehydrogenation of these fragments to be the kinetically limiting feature in the second stage of pyrolysis. The most characteristic infrared bands for the B-N compounds are presented in table 12.3.
421
Table 12.3 Characteristic infrared bands of B-N compounds Frequency (cml)
Attributions
Amineboranes 3312 3245 2316 2115 1597 1374 1165 1058 785
N-H stretch (asymmetric) N-H stretch (symmetric) B-H stretch (asymmetric) B-H stretch (symmetric) N-H bend (asymmetric) N-H bend (symmetric) B-H bend (asymmetric) B-H bend (symmetric) B-N stretch (symmetric)
Aminoboranes 3300 3250 2420 2300 1605 1558 1400
N-H stretch (asymmetric) N-H stretch (symmetric) B-H stretch (asymmetric) B-H stretch (symmetric) N-H bend (asymmetric) N-H bend (symmetric) B-N stretch (asymmetric)
Borazines 3490 3450 2535 2530 1465 938
N-H N-H B-H B-H B-N B-N
Boron nitride 1374 813
B-N stretch (asymmetric) B-N stretch (symmetric)
stretch stretch stretch stretch stretch stretch
(asymmetric) (symmetric) (asymmetric) (symmetric) (asymmetric) (symmetric)
422
5.2 Ammoniation of boranated silica The low temperature ammoniation of boranated silica is believed to proceed quite similar to the bulk reaction: -Si-O-BH
273 K 2 + NH 3 ~
=Si-O-BH
2.-NIl 3
(J)
> 373 K - S i - O - B H 2"- N H 3 ~
-Si-O-BH-NII
2 +
H2
(K)
The same reactions are proposed for secondary groups"
-=Si - O ~
---Si - O B-H
+
NH 3
=-Si - O /
~
BH'-NH =-Si - O /
=Si - O ~ BH .- NH 3 -Si - O /
> 373 K ~
(L) 3
-Si - O,.~ B-NH 2
+
H 2
(M)
-Si - O /
It should be noted though, that reaction (M) is not applicable to aluminosilicates (zeolites). Philippaerts 92 has shown, using deuterated ammonia, that following reaction occurs in a mordenite zeolite:
423
Si- O \
Si - O \ BH M_O
+ ND3 ~
BH'ND3 M-O
/
Si- O - B H - ND 2 ~
/
(N) M-O-D
It is obvious however that the aluminium sites in zeolites are much more acidic and more easily broken. Gillis-D'Hamers '9 used the integrated form of the Elovich equation to evaluate the activation energy and the induced heterogeneity. q - R T In I t + t o
[ to
(11)
q represents the relative coverage of surface boron sites, R the gas constant, T the reaction temperature and t the reaction time. to is an empirical constant, used to linearise the equation. ~ can be considered as a heterogeneity coefficient. As Eo is the activation energy at zero coverage, oe represents the slope of the linear increase in activation energy as a function of coverage, according to E = E o + c~q
(12)
The results, calculated according to the methods of LOW 93 and Aharoni 94 are presented in table 12.4. Obviously, the amineboranes are formed without an appreciable activation energy. The rate-limiting step in the reaction sequence will therefore be the formation of aminoboranes.
It is at first sight remarkable that the activation energy at zero
coverage (Eo) increases with an increasing surface loading of reactive borane groups. This type of behaviour was also observed by Verbiest 9~ for zeolite substrates. It can be explained by the fact that a nominal increase in active site concentration diminishes their intrinsic reactivity by dispersing the nucleophilic or electrophilic character.
424
Table 12.4 Activation energy function for amineborane and aminoborane formation on the silica gel structure amine_borane
amino_borane
diborane mmol/g
Eo kJ/mol
ct kJ/mol
Eo kJ/mol
ct kJ/mol
0.12 0.35 0.33 0.52 0.63 0.89 0.99 1.01
4.2 0.7 -0.9 3.2 1.7 1.6 1.6 2.7
0.7 1.6 0.5 1.4 1.4 1.1 0.8 0.7
44.9 48.9 46.6 49.2 50.6 56.8 58.2 57.5
451 781 361 303 214 87.3 77.7 59.9
Again, at higher reaction temperatures, the mechanism becomes more complex. Based on a quantitative analysis of consumed N H 3 v s . liberated H2, S e g e r s 96 evidenced the existence of a secondary reaction (O).
2 -Si-O-BH 2
+
NH3
~
-Si - O - B - H N NH
/
+2H 2
(O)
-=Si - O - B - H
Comparison of the integrated peak areas of the primary and secondary N-H stretching vibrations revealed that at reaction temperatures of 650 K, the secondary reaction (O) occurs for more than 50%. At reaction temperatures above 673 K, an infrared peak arises at 1480 cm ~. The phenomena causing this band are not yet fully understood at present. The B-N vibration band of cyclic borazines is situated at 1465 cm l, suggesting that possibly a
425
strained cyclic structure is formed at the silica surface. The same bands, characteristic for cyclic B-N structures, were observed by Philippaerts 92 for the ammoniation of boronated mordenite.
5.3 Ammoniation of silica, activated with B CI3 A few studies
97,98,99,100,101,102 have
described the synthesis of BN from BC13 and NH 3,
mainly by means of CVD processes. Although the detailed reaction mechanism is still quite obscure, the overall reaction mechanism proceeds according to:
(P)
BC! 3 + NH 3 ~ BN + 3HC!
The ammoniation of silica, activated by BC13, has received little to none attention in literature. One would expect that the reaction proceeds relatively similar to the ammoniation of chlorosilylated silica. There should be a difference in reactivity though, since boron is obviously more electrophilic than the Si analogue. 1~ This is evidenced in figure 12.25, showing the NHx and NH4C1 species on silica, activated as the number of N-species that have replaced the active Cl-groups. In the low reaction temperature region (273 - 423 K), there is at first sight no reactivity difference between the silylated and the boranated sample. In both cases, an equal amount of-NH2 species and NH4C1 species is formed. This is confirmed by the infrared spectrum of trichloroboranated silica, ammoniated at room temperature (figure 12.26). This spectrum is very similar to the spectrum of the ammoniated silica (figure 12.6), suggesting an identical reaction mechanism:
B/Cl -Si-O-
/ NI-I: --- NH4Cl +4NH 3 ~
\ Cl
-Si-O-B
\
--- NH4Cl
(Q)
426
1.1
N/Clorig.
1.0
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2
NH4Cl
/ (both BCI3 and HSiCI3)
0.1 0
i
273
373
i
i
473 573 673 773 Reaction temperature (K)
873
Figure 12.25 Distribution of NH~ and NH4CI species on trichlorosilylated and trichloroboranated silica as a function of reaction temperature (+ NH2 on HSiCI3 treated silica," 9NH 2 on B Cl:reated silica). Intensity (a.u.)
3500 i
3000 i
2500 i
2600
1~00
Wave.umber (cm-1)
Figure 12.26 Infrared spectrum of trichloroboranated silica, ammoniated at room temperature.
427 Large reactivity differences are found in the temperature region above 473 K. It was stated previously that -in the case of trichlorosilane- the NHaC1 has a stabilizing role, allowing maximum capacity at temperatures below the sublimation point of NHaC1. It is only above this temperature that the actual differences in reactivity between B and Si preactivation are revealed. The silica, preactivated with BCI3, has a 1:1 capacity (N/C1) towards NH3, independent of reaction temperature, whereas the trichlorosilylated silica shows a typical exponential increase in capacity as a function of temperature. The differences in reactivity are also quite obvious when the reaction kinetics are compared (figure 12.27). The experiments were performed in a large excess of gaseous NH3. mmol/g 2.5 2.0 1.5 1.0 0.5
//
7
f
I
0
50
I
100 Time (minutes)
I
150
200
Figure 12.27 Adsorption kinetics of ammonia on trichlorosilylated (=) and trichloroboranated (*) silica gel at 673 K.
In spite of the reactivity differences, the underlying reaction mechanisms are very similar, even in the higher reaction temperatures. Troubleyn and Possemiers ~~ studied the symmetric and asymmetric B-NH2 vibration to evidence that following species (V) are formed on the silica surface at reaction temperatures > 573 K:
428
~-Si - O - B / N H 2
/Nil
\
(v)
~-Si - O - B
~NH 2
The mechanisms leading to these surface species need a more profound study, but there is at first sight no obvious reason why the mechanisms (K), (L) and (H) for trichlorosilylated silica would not be applicable for trichloroboranated silica. As in the case of ammoniation of silica pretreated with B2H6, an infrared band around 1480 cm ~ arises, probably attributable to a strained cyclic B-N structure.
References
1. Q
11,
4. .
J.B. Peri, J. Phys. Chem., 1966, 70, 2937. I.D. Gay and S. Liang, J. Catal., 1976, 44, 306. B.A. Morrow, I.A. Cody and L.S.M. Lee, J. Phys. Chem., 1975, 79, 2405. B.A. Morrow and I.A. Cody, J. Phys. Chem., 1976, 80, 1998. B.A. Morrow, I.A. Cody and L.S.M. Lee, J. Phys. Chem., 1976, 80, 2761. P. Fink, I. Plotski and G. Rudakoff, Wiss. Ztschr. Friedrich Schiller Univ., 1988, 37, 911. (in German)
7.
R. Wusiraka, J. Am. Ceram. S.c., 1990, 73, 2926. G.A. Blomfield and L.H. Little, Canadian Journal of Chemistry, 1972, 51, 1771.
.
J. Sj6berg and R. Pompe, J. Am. Ceram. S.c., 1992, 75, 2189.
10.
P. Fink, B. M/iller and G. Rudakoff, J. Non-Cryst. Solids, 1992, 145, 99.
11.
J. Sj6berg, K. Rundgren, R. Pompe and J. Sj6berg, High tech Ceramics, 1987, 535.
12.
J. Sj6berg, J. Sj6berg and K. Rundgren, personal communication.
429
13.
O.E. Brust, F. Sebastian and I. Halasz, J. Chromatogr., 1973, 83, 15.
14.
M. Folman, Trans. Faraday Soc., 1961, 57, 2000.
15.
M.J.D. Low, A.G. Severdia and J. Chan, J. Catal., 1981, 71, 144.
16.
S. Koseli, Chem. Soc. Japan, 1992, 65, 3176.
17.
P. Van Der Voort, J. Swerts, K.C. Vrancken, E.F. Vansant, P. Geladi and P. Grobet, J. Chem. Soc. Faraday Trans., 1993, 89, 63.
18.
P. Van Der Voort, K.C. Vrancken, E.F. Vansant and J. Riga, J. Chem. Soc. Faraday Trans., 1993, 89, 2509.
19.
I. Gillis-D'Hamers, PhD-thesis, University of Antwerp, Antwerp, 1993.
20.
K. Possemiers, PhD-thesis, University of Antwerp, Antwerp, 1995. (in press)
21.
T. Takeuchi and T. Miwa, Anal. Chim. Acta, 1993, 282, 565.
22.
A.V. Kiselev and V.I. Lygin, Infrared spectra of surface compounds, John Wiley & Sons, New York, 1975.
23.
B. Standte, Z. Physik. Chem., 1974, 255, 158. (in German)
24.
P. Fink, I. Plotski and G. Rudakoff, Wiss. Ztschr. Friedrich Schiller Univ., 1990, 39, 217. (in German)
25.
R.K. Brow and C.G. Pantano, J. Am. Ceram. Soc., 1987, 70, 9.
26.
F.H.P.M. Habraken, A.E.T. Kuiper, Y. Tamminga and J.B. Theeter, J. Appl. Phys., 1982, 53, 6996.
27.
H. Deuel, J. Wartmann, K. Hutschneker, U. Schobinger and G. G/idel, Helv. Chim. Acta, 1959, 42, 1160.
28.
H.P. Boehm and M. Schneider, Z. Anorg. Allgem. Chem., 1959, 301,326.
29.
H.P. Boehm, M. Schneider and F. Ahrendt, Z. Anorg. Allgem. Chem., 1963, 42, 320.
30.
J. Uytterhoeven and H. Naveau, Bull. Soc. Chim. Fr., 1962, 27.
31.
T.H. Elmer, I.D. Chapman and M.E. Nordberg, J. Phys. Chem., 1963, 2219.
32.
V.A. Tertykh and L.A. Belyakova, Chemical reactions involving the surface of silica, Naukova Dumka Publishers, Kiev, 1991 (in Russian), and references therein.
33.
M.L. Hair, Infrared Spectroscopy in Surface Chemistry, Dekker, New York, 1967.
430 34.
V.A. Terkyth et al., Fiz. Khim. Mekh. Liofil'nost Dispersnykh Sist., 1973, 4, 37. (in Russian)
35.
M.L. Hair and W. Hertl, J. Phys. Chem., 1973, 77, 2070.
36.
M.P. McDaniel, J. Phys. Chem., 85,532, 1981.
37.
T.H. Elmer, U.S. Patent 2982053 (1956).
38.
K.K. Unger and E. Gallei, Kolloid-Z.Z. Polym., 1970, 237, 358.
39.
G.W.A. Fowles, in 'Progress in inorganic chemistry', ed. F.A. Cotton, volume 6, Interscience, New York, 1964.
40.
P.R. Moses, L.M. Wier, J.C. Lennox and H.O. Finklea, Anal. Chem., 1978, 50, 576.
41.
N.G. Cave and A.J. Kinloch, J. Adhesion, 1991, 34, 175.
42.
C.G. Pantano and T.N. Wittberg, Surface and Interface Analysis, 1990, 15, 498.
43.
M.L. Miller and R.W. Linton, Anal. Chem, 1985, 57, 2314.
44.
M. Ledoux, S. Hantzer, J. Guille and M. Desaneaux, J. Catal., 1988, 4097.
45.
A. Nakao, M. Iwaki and H. Sakairi, Nucl. instruments and methods in physics research, 1992, 352.
46.
W. Lindberg, J. Persson and S. Wold, Anal. Chem., 1983, 55, 643.
47.
M. Otto and W. Wegscheider, Anal. Chem., 1985, 57, 63.
48.
D. Haaland and E. Thomas, Anal. Chem., 1988, 60, 1202.
49.
P. Beauchesne, Quantitative analysis of solutions of three sugars by FTIR: a PLSplus application, Application Note, Bomem, 450 St. Jean-Baptiste, Quebec, G2E 5S5, Canada.
50.
P. Van Der Voort, Spectral Data Manipulation, ECC-Comett Program, Fourier Transform Infrared Spectroscopy, Antwerp, Belgium, 1992.
51.
N. Draper and H. Smith, Applied Regression Analysis, John Wiley & Sons, New York, 1981.
52.
H. Martens and T. Naes, Multivariate Calibration, John Wiley & Sons, Chichester, 1989.
53.
S. Wold, C. Albano, W. Dunn, K. Esbensen, S. Hellberg, E. Johansson, W. Lindberg and M. Sjbstrbm, Analusis, 1984, 12, 477.
431
54.
T. Naes and H. Martens, Comm. in statistics-simulation and computation, 1985, 14, 545.
55.
I. Frank and B. Kowalski, Anal. Chim. Acta., 1985, 167, 51.
56.
P. Geladi and B. Kowalski, Anal. Chim. Acta, 1986, 185, 1.
57.
P. Geladi and B. Kowalski, Anal. Chim. Acta, 1986, 185, 19.
58.
A. Lorber, L. Wangen and B. Kowalski, J. Chemometrics, 1987, 1, 19.
59.
R. Manne, Chemometrics and Intelligent Laboratory Systems, 1987, 2, 187.
60.
A. H6skuldsson, J. Chemometrics, 1988, 2, 211.
61.
P. Geladi, J. Chemometrics, 1988, 2, 231.
62.
D. Haaland and E. Thomas, Anal. Chim. Acta, 1988, 60, 1193.
63.
I. Gillis-D'Hamers, K.C. Vrancken, E.F. Vansant and G. De Roy, J. Chem. Soc. Faraday Trans., 1992, 88, 2047.
64.
D.W. Sindorf and G.E. Maciel, J. Am. Chem. Soc., 1983, 105, 3767.
65.
G.E. Maciel and D.W. Sindorf, J. Phys. Chem., 1983, 87, 5516.
66.
E. Wiberg, IUPAC, 1957, 16~ Conference, Paris.
67.
K.W. Bfddeker, S.G. Shore and R.K. Buntig, J. Am. Chem. Soc., 1976, 123, 1721.
68.
M. Hirayama and K. Shohno, J. Electrochem. Soc., 1975, 122, 1671.
69.
S.B. Hyder and T.O. Yep, J. Electrochem. Soc., 1976, 123, 1721.
70.
A.C. Adams and C.D. Capio, J. Electrochem. Soc., 1980, 127, 399.
71.
J.D. Carpenter and B.C. Ault, Chemical Physics Letters, 1992, 197, 171.
72.
R.R. Rye, D.R. Tallant, T.T. Borek, D.A. Lindquist and R.T. Paine, Chemistry of Materials, 1991, 3, 286.
73.
P.J. Fazen, J.S. Beck, A.T. Lynch, E.E. Remsen and L.G. Sneddon, Chemistry of Materials, 1990, 2, 96.
74.
A.T. Lynch and L.G. Sneddon, J. Am. Chem. Soc., 1989, 111, 6201.
75.
K.J.L. Paciorek, D. Harris and R. Kratzer, J. Polym. Sci., Part A: Polym. Chem., 1986, 24, 173.
432 76.
K.J.L. Paciorek and R.H. Kratzer, Ceram. Eng. Sci. Proc., 1988, 9, 993.
77.
K.J.L. Paciorek, W. Krone-Smidt, D. Harris, R. Kratzer and K.J. Wynne, ACS Symp. Ser., 1988, 362, 392.
78.
A.T. Lynch and L.G. Sneddon, J. Am. Chem. Sot., 1987, 109, 5867.
79.
M.G.L. Mirabelli and L.G. Sneddon, J. Am. Chem. Soc., 1988, 110, 3305.
80.
J.S. Beck, C.R. Albani, A.R. McGhie, J.R. Rothman and L.G. Sneddon, Chem. Mater., 1989, 1,433.
81.
M.G.L. Mirabelli and L.G. Sneddon, Inorg. Chem., 1988, 27, 3271.
82.
M.G.L. Mirabelli, A.T. Lynch and L.G. Sneddon, Solid State Ionics, 1988, 32/33, 655.
83.
W.S. Rees and D. Seyferth, J. Am. Ceram. Soc., 1988, 71, C194.
84.
C.K. Narula, R.T. Paine and R. Schaeffer, Proc. Mater. Res. Soc., 1986, 73, 383.
85.
C.K. Narula, R.T. Paine and R. Schaeffer, ACS Symp. Ser., 1988, 360, 378.
86.
C.K. Narula, R. Schaeffer, R.T. Paine, A.K. Datye and W.F. Hammetter, J. Am. Chem. Soc., 1987, 109, 5556.
87.
R.T. Paine, C.K. Narula, R. Schaeffer and A.K. Dayte, Chem. Matter, 1989, 1, 486.
88.
A.K. Dayte, R.T. Paine, C.K. Narula and L. Allard, Proc. Mater. Res. Soc., 1989, 153, 97.
89.
C.K. Narula, R. Schaeffer, A.K. Dayte and R.T. Paine, Inorg. Chem., 1989, 28, 4053.
90.
R.T. Paine and C.K. Narula, Chem. Rev., 1990, 90, 73.
91.
D.A. Lindquist, T.T. Borek, S.J. Kramer, C.K. Narula, G. Johnston, R. Schaeffer, D.M. Smith and R.T. Paine, J. Am. Ceram. Soc., 1990, 73, 757.
92.
J. Philippaerts, E.F. Vansant, G. Peeters and E. Vanderheyden, Anal. Chim. Acta., 1987, 185, 237.
93.
M.J.D. Low, Chemical Reviews, 1960, 60, 267.
94.
C. Aharoni and M. Ungarish, J. Chem. Soc. Faraday Trans. I, 1976, 72, 400.
95.
J. Verbiest, PhD thesis, University of Antwerp, Antwerp, 1988.
433 96.
W. Segers, Ms-thesis, University of Antwerp, Antwerp, 1994.
97.
T. Takahashi, H. Itoh and A. Takeuchi, J. Crystal Growth, 1979, 47, 245.
98.
K. Sugiyama and H. Itoh, Materials Science Forum, 1990, 54/55, 141.
99.
M.J. Rand and J.F. Roberts, J. Electrochem. Soc., 1968, 115, 423.
100.
K. Nakamura, Materials Science Forum, 1990, 54/55, 111.
101.
S.S. Dana, Materials Science Forum, 1990, 54/55, 229.
102.
A.W. Moore, J. Crystal Growth, 1990, 68, 6.
103.
E. Troubleyn and K. Possemiers, yet unpublished results.
This Page Intentionally Left Blank
This Page Intentionally Left Blank
437
PART 3: CHEMICAL SURFACE COATING
Chapter 13 Coating techniques
Coating techniques can be defined as all procedures which share the final aim to create a thin layer on a foreign substrate. The thickness of such a coating varies from a monomolecular layer (nm) to several millimetres. The concept of coating is very extended. Some books on coating techniques describe the art of painting. In this book, the discussion will be restricted to procedures, developed to create a stable surface coating on an inorganic substrate, by means of deposition techniques or by means of a series of chemical reactions of the substrate with various reagents. The coating techniques will be classified as Chemical Vapour Deposition (CVD), Physical Vapour Deposition (PVD), Atomic Layer Epitaxy (ALE) and Chemical Surface Coating (CSC).
1 Chemical Vapour Deposition (CVD) ~ The CVD process involves the reaction of a mixture of gases with a heated substrate. Due to the thermoshock at the substrate's interface, the gases decompose and solid decomposition products deposit at the surface. During the last decades, several alternative CVD methods have been developed, allowing the creation of a coating at lower temperatures and/or in a more localized way. Some of these alternative techniques will be discussed.
438
1.1 Conventional Chemical Vapour Deposition Conventional Chemical Vapour Deposition (also called Thermal Chemical Vapour Deposition) is the simplest CVD technique, in which a mixture of reactive gases (typically metalhalides and hydrides) are decomposed by and deposited on a heated substrate ~'2. At the moment, it is the only CVD technique that has been implemented on a commercial basis to produce large amounts of cheap coated materials. A typical configuration of a thermal CVD reactor is shown in figure 13.1. The applications of conventional CVD are far too extended to cover in this introduction. Tables 13.1 and 13.2 present the typical process parameters and the typical deposition reactions in a CVD experiment. The range of these parameters indicates the scope and versatility of CVD. The high process - temperature (1100- 2300 K) causes a strong diffusion of the coating into the substrate, resulting in a good attachment. The CVD technique does not suffer from the 'line-of-sight' effect (cfr. PVD): the substrate is coated entirely with a relatively uniform layer. At the same time, this high temperature is the biggest disadvantage of the CVD technique. The different thermal expansion coefficients of substrate and coating cause thermal cracks in the ceramic coating upon cooling. This is the main reason for the development of a number of alternative CVD techniques, allowing a significant lowering of the deposition temperature.
439
71
6
To ve,nt
ooooo
9
3
!
II 13
! /
Figure 13.1 Schematic diagram showing various components of a CVD system (1: reactor," 2: heating element; 3: reaction tube; 4: water-cooled end flanges; 5: power controller; 6."pressure indicator," 7." temperature sensor; 8,10,11: precursor gas tanks, 9." metal halide (liquid) vaporizer," 12: particulate trap," 13: gas scrubber," 14: flow meter; 15: flow meter valves; 16: gas tank regulators; 17." substrate support; 18: substrates).
440 Table 13.1 Typical process-parameters of a conventional CVD experiment Temperature Pressure Precursors
1100 K - 2300 K < 0.1 - 101300 Pa reactive gases (metal halides) reducing gases (HE) inert gases (Ar, N~) other gases (CH4, CO2, NH3)
Table 13.2 Typical deposition reactions, occurring in a conventional CVD experiment Pyrolysis Reduction Oxidation Hydrolysis Co-reduction
CH3SiC13 --, SiC + 3HC1 WF6 + 3H2--, W + 6HF Sill4 + 02 ~ SiO2 + 2H2 2A1C13 + 3H20 ~ A1203 + 6HC1 TIC14 + 2BC13 + 5H2---, TiB2 + 10HC1
1.2 Metal-organic CVD (MOCVD) This technique uses organometallic compounds with a relatively low decomposition temperature ( < 1073 K) as coating precursors. There is a lot of interest for MOCVD, especially by manufacturers of semi-conductors, who already employ it commercially for the deposition of semi-conducting films of GaAs, InAs, InP and ZnSe. The films are very thin, in the/~ngstrom range, and are usually epitaxial. Amorphous as well as crystalline films can be deposited, but the latter usually have higher defect densities, limiting their usefulness in critical applications. In microelectronics, this technique is referred to as organo-metallic vapour phase epitaxy (OMVPE). A detailed account of this technique is given by Dapkus. 3 The MOCVD technique has also been used to deposit a number of refractory compounds for a wide range of applications. One of the advantages of this technique is the lower deposition temperature, which makes it very suitable for deposition on substrates, which are thermally sensitive, such as steels. Wear - resistant tungsten carbide coatings have been deposited on steel, using WF 6 and suitable hydrocarbon gases at temperatures below 873 K. 4'5
441 1.3 Plasma-Enhanced CVD (PECVD) Plasma-enhanced chemical vapour deposition has gained importance rapidly in recent years, because this technique provides some unique advantages over conventional CVD. The important advantages include lower deposition temperatures, deposition of non-epuilibrium phases and a better control of stoichiometry and purity of deposits. In this technique, the activation energy for the breakdown of reactive species, and their subsequent interaction with other species to form a deposit, is provided by the high kinetic energy of electrons in the plasma (figure 13.2). I
I RF plate
(cathode)
(~
RF generator (or DC power source)
~"//////:~
ates
IiiiIII-'l
Source gas
Pump
Figure 13.2 Schematic representation of a radial flow plasma CVD reactor.
The plasma is created by an electrical field between both parallel plates (see figure), ionizing the gas volume inbetween. In a plasma, the energy is transferred by collisions between all particles. Due to their smaller mass, the energy of the electrons increases much faster than the energy of the heavier ions. This means that mainly the electrons are responsible for the ionization processes and the formation of reactive free radicals. The big energy difference between electrons and ions is reflected in the respective temperatures" the electron temperature of a typical plasma is about
442 10000 K, whereas the gas temperature usually does not exceed 850 K, provided that the pressure is low ( < 1300 Pa). The dissociation of gas molecules in a plasma involves the formation of intermediate, highly reactive fragments, which are very unstable under normal circumstances. Therefore, PECVD allows the deposition of quite unique materials with very unusual properties. One of such unique coatings is Diamond Like Carbon (DLC). The conventional synthesis of synthetic diamonds requires extremely high temperatures and pressures. By PECVD, Diamond Like Carbon is created under mild conditions by the decomposition of methane in H2/CH 4 mixture. The applications of DLC are numerous" coatings for cutting tools, optical fibres, electronic devices for reading magnetic tapes, or even protective coatings in chemical reactors.
1.4 Laser CVD (LCVD) This technique has also been used more and more in recent years, primarily in the microelectronics field. The activation of gaseous species, in this case, is achieved by shining a laser beam in the reactor. Even though both with plasma and laser techniques the same general result is achieved, there are some significant differences between the two techniques. The main difference is the ability of the laser to create high energy electrons in a very narrow energy band as compared to the electron energy distribution in a typical plasma. Again, as a result of such a localized activation of the gas volume, the deposition temperatures in LCVD can be considerably lower than in the conventional CVD. Basically, there are two types of laser CVD techniques, as illustrated in figure 13.3. In a photolytic or photochemical LCVD experiment, the gas absorbs the laser energy whereas the substrate is transparent. The gas molecules are ionized and fragmentation occurs, resulting in the deposition of a coating on the substrate, which is at relatively low temperature. In this technique, the wavelength of the laser can be chosen in such a way as to allow only selected gaseous species to be activated, permitting the formation and deposition of selective film compositions.
443
Volatile Products Reactant ~ ~ ~ ,,. - - "" ~ ~ ~ k
r_JAltemate !-I Substrate -]Position
t__
. _ , , ~ hv _ ~ ~ ~
[-- Substrate I/L~ ~Hot
Spot
~_~ Deposit
Substrate ~ Products
Deposit
Reactant
Figure 13.3 Schematic arrangement of Laser CVD systems. Left: photolytic LCVD; right." pyrolytic LCVD. In a pyrolytic or thermal LCVD experiment, the gas is transparent and the substrate absorbs the laser energy. This creates a so - called hot- spot on which a normal thermal CVD process occurs. Pyrolytic LCVD allows a very precise localization of the coating. In a sense, this technique may be compared to the 'cold - wall' CVD technique in which the substrate may be heated by passing an electric current through it (resistance heating), or by induction, where the substrate itself acts as a susceptor. In these cases, the gas volume is not heated significantly (hence the name 'cold - wall' CVD). The main difference between the cold-wall CVD and the pyrolytic laser CVD is that in the latter, the heated area can be localized and scanned very precisely.
1.5 Fluidized-bed CVD (FB CVD) This is a relatively special technique which combines the principles of fluidized - bed heating and CVD. It is primarily used to coat powders of very fine size with suitable films for special applications. The most prominent application of this technique is in the coating of nuclear fuel particles used in high-temperature gas-cooled reactors (HTGR). A typical fluidized-bed CVD reactor is shown schematically in figure 13.4. Considerable work is this area has been carried out at the Oak Ridge National Laboratory where coatings of high and low density graphite and of SiC have been deposited on uranium oxide and thorium oxide microspheres. 6 The purpose of these
444
SIG-4T PORT COOt.e~, wATER
I----COOUNG WATER OUTLET GRAPHITE
EXHAUST /~;).l'_~ED_~ E Ds
CHAt~IE3ER GqAP~TE I~ATfJG ELEIvqENI CARBON Ir[LT I~SULAT~Oq
CL._VEW e.OFL[
WATER-COOL[~ SIEEL JACKEl
OUTER _J JACKET
ELECTRiCAl INSULAIOR
I I | ,'-ce~..:
II t~SULAT~
WATER-C00LED COPPER ELECIROC~
COOLING1 L INNER JACKET WATER
.,i,tn-
*
COOLED GAS IN&CTOR f
ii i' (;AS ~NLET
!t
d~
Figure 13.4 Schematic representative of a fluidized bed CVD reactor.
films is to contain the fission products of the nuclear reaction in order to minimize exposure and contamination from radio- active species. The inner, low density coatings protects the outer coatings from fission - recoil damage and provides a free volume for the fission products. The outer, high - density coating acts as a pressure vessel and diffusion barrier for solid fission fragments. Another prominent application of FBCVD is in the manufacture of high purity silicon. Silicon seed particles are fluidized in a bed, in which a mixture of silane and hydrogen is introduced. Decomposition of silane on the silicon particles results in the formation of a pure silicon film. It is of importance to control the process in order to minimize homogeneous nucleation of silicon dust by appropriately designing gas distribution system and controlling process parameters. 7,s
445
1.6 Chemical Vapour Infiltration (CVI) The increasing use of light ceramic composites for high temperature and space applications has stimulated the development and optimization of the Chemical Vapour Infiltration technique. The use of conventional ceramic techniques for the fabrication of fibre-reinforced composites damages the fibres both mechanically as chemically. Also, the high process temperature causes a thermal degradation of the fibres. The substrate in a CVI experiment is a highly porous material. This porous structure is infiltrated by vapours at considerably lower temperatures and pressures (compared to a conventional CVD experiment), causing a deposition on and between the fibres. This yields a very strong composite with high density. Three possible reaction schemes can be discerned: * isothermal diffusion with a concentration gradient * diffusion with a thermal gradient * diffusion with a pressure gradient Figure 13.5 shows a schematic representation of the second possibility. 9 The main advantage of this configuration is that the reaction only occurs on the hot surface, away from the gasinlet. As the deposition proceeds, the density and thermal conductivity of the coated zone increases, lowering the hot reaction zone towards the gasinlet. In this way, the entire volume is infiltrated uniformly and progressively. CVI has already been successfully used for the fabrication of fibre-reinforced composites of A1, A1N, BN, SiC, TaC and TiB2 .7'8'9'1~
1.7 Materials and applications The types of materials deposited by CVD, range from pure metals to compounds, ceramics, powders, whiskers and composite coatings. The various types of metals and compounds which have been successfully deposited by CVD are shown in table 13.3. In most of these cases, the precursors are metal halides which are unstable in the temperature range of deposition. When suitable halides cannot be obtained, metalorganic compounds can be used.
446
Heating element
Hot zo~ 1473 K
~Retainin~
Exhaust gas
Inf'fltratcd composite N "////d "//'/,~
;'///2 "///// ~.//// z/'/.,~
N
Water-cooling holder
Fibrous preform
Coating gas
Figure 13.5 Schematic representation of chemical vapour infiltration process. Table 13.3 Materials deposited by CVD Metals: Compounds: Ceramics:
A1, As, Be, Bi, Co, Cr, Cu, Fe, Ge, Hf, Ir, Mo, Nb, Ni, Os, Pb, Pt, Re, Rh, Ru, Sb, Si, Sn, Ta, Th, Ti, U, V, W, Zr. Also carbon and boron. II-VI and III-V compounds, borides, carbides, nitrides and silicides of transition metals, as well as sulphides, phosphides, aluminides, etc. A1203, A1N, B203, BN, SiC, Si3N4, UO2, Y203, ZrOz, etc.
Table 13.4 shows some typical applications of chemical vapour deposition. This table illustrates the wide variety of applications and the versatility of the CVD technique.
447
Table 13.4 Typical applications of chemical vapour deposition * Tribological coatings * Wear-resistant coatings * High-temperature coatings for oxidation resistance * Dielectric/insulating films * Optical/reflective films * Photovoltaic films
* Decorative films * Superconducting films * Emissive coatings * Coatings for fibre composites * Free-standing structural shapes * Powders and whiskers
By far the largest areas of application of CVD include manufacture of powders (for pigments), micro - electronics (involving dielectric films, optical films, super conducting and emissive films), and tribology. Irrespective of the application of the CVD coating, one of the most important criteria in the selection of the coating process is a consideration of the substrate/gas interaction. In conventional CVD, the reaction temperatures are typically above 1073 K. Thus, those substrates which undergo phase transformation related dimensional changes at high temperature become unsuitable. Many metals also show a propensity for reaction with furnace gases at these temperatures, limiting their use as substrates. The presence of residual stresses at the interface, or in the coating, can be a cause of serious concern in some applications, while degradation of the interfacial region below the coating/substrate interface can often cause failure of the coated component due to mechanical shock before the advantages of the coating are manifested. Thus, the most suitable substrates are refractory metals (such as tungsten, molybdenum), certain ceramics (such as alumina, mullite, silicon nitride), and graphite. High - speed steels and some other high - alloy steels have been used in selected applications, but the use of steels, in generally is limited in conventional CVD. Cemented tungsten carbide is used as a substrate for depositing wear-resistant coatings of TiC and TiN for applications in metal - cutting. Graphite and molybdenum are often substrates of choice for fabricating free - standing parts by using them as shaped mandrels. Of course, in many cases, the choice of substrate is limited by the application. In such cases, a proper design of the coating chemistry is important to ensure overall enhancement of properties with a minimum compromise of substrate properties. ~
448 Coatings used for tribological applications require good adhesion, a low coefficient of friction, high hardness and wear resistance. Coatings of TiC, TiN, CrTC3 and tungsten carbide have been used in many of these applications such as forming tools, ball beating components, machine parts, gears, steel cutting tools, surgical and prosthetic implements, etc. Titanium nitride is a popular coating for decorative applications in addition to its excellent frictional characteristics. In many tribological applications, environmental degradation is also a factor. Therefore, these coatings are also required to provide improved chemical resistance. Hard coatings have been applied to many types of forming tools, such as deep-drawing punches and dies, wire - drawing dies, injection moulding dies, etc. It was shown that the application of coating resulted in an eight - fold increase in service life at 80% reduction in tool costs. Another well-known example of improvement in service life by the applications of a wear-resistant coating is that of cemented carbide cutting tool inserts. These tools are typically coated with TiC, TiN, A1203, TaC, HfN, etc. These coatings impart improved abrasion resistance, chemical resistance, frictional characteristics by deflecting the heat generated during metal cutting into the metal chips, away from the tool, thereby preventing softening of the tool due to excessive heat at the tool tip. Coatings for high temperature service require good thermal and chemical stability. Refractory compounds such as various oxides, silicides and aluminides are commonly used in these .applications. Other coatings which are also useful are SiC, Si3N4 and certain refractory metals such as iridium. Many high temperature applications involve particulate erosion ablation, corrosive environments and severe thermal cycling. Coatings in such applications are expected to withstand these conditions. Another situation involves coatings used in fusion reactor components, where erosion of wall coatings can lead to poisoning of the plasma. In such cases, resistance to sputtering erosion, and reaction with hydrogen ions is important. Thermal barrier coatings for superalloy components in high temperature turbines and engines require good adhesion, a careful match of thermal expansion coefficients through multiple layers, and good resistance to oxidation, erosion and corrosion. The CVD technique can be successfully used in depositing such multiple layers with excellent control of thickness and uniformity.
449
2 Physical Vapour Deposition (PVD) Physical Vapour Deposition (PVD) is another coating technique. The reactants (precursors) are solids, which are forced in a gaseous state. This can be done by simple heating, but mostly, this procedure involves ion - bombing in order to create a plasma. The gaseous phase deposits on the solid substrate at relatively low temperatures. The main advantages of the PVD technique are: * the purity of the coating; * the broad range of suitable precursors; * the low temperature of the substrate; * the fine structure of the coating, which makes polishing unnecessary. The main disadvantages are: * the low deposition rate; * the complexity of the process and the equipment; * the high demands concerning the purity and cleanliness of the substrate' surface; * the line - o f - sight effects. The line-of-sight effect is the biggest disadvantage. It means that the part of the substrate, which is located directly above the source, will achieve a much thicker coating that other parts of the substrate. As a consequence, only very simple geometric forms can be coated uniformly by PVD, provided that either the substrate is rotated, or that multiple sources are present in strategic places in the reactor. PVD is a time consuming process" not only the deposition rate is relatively low, also severe precautions are necessary concerning the cleanliness of the substrate. This involves long pretreatment procedures. Depending on the way the solid precursor is volatized, three PVD methods are discerned" * thermal PVD * sputtering * ion-plating
450
2.1 Thermal Physical Vapour Deposition P VD Figure 13.6 shows the basic configuration of a thermal PVD reactor.
[
]S
\
Iv lip
Figure 13.6 Schematic representation of a thermal PVD reactor (S= substrate," V= source P= pumps).
Since in a thermal PVD experiment, the solid source material is volatized by simple heating, the choice of precursors is rather limited (melting temperature < 1850 K). Conventional thermal PVD is mostly used for the deposition of metals. PVD of a ceramic layer involves the introduction of an additional gas. However, the energy of the gas is much too low to form a ceramic reaction product. For example, the formation of TiN from Ti and N2 needs much higher temperatures than the normal PVD process temperatures. This supplemental activation of the gas mixture is established by placing a positive electrode in the gas volume. This technique is called Activated Reactive Evaporation (ARE) and is illustrated in figure 13.7.
451
II +
Is
I
TE
\
~
/
V
II
P
Figure 13.7 Activated reactive evaporation (G= gas supply," S= substrate; V= source; TE= positive electrode," P= pumps).
2.2 Sputtering If a solid or liquid at any temperature is subjected to bombardment by suitable high energy atomic particles (usually ions), it is possible for individual atoms to acquire enough energy via collision processes to escape from the surface. This means of causing ejection of atoms from the surface is called sputtering. The atoms ejected from the surface can be used in depositing a coating on a substrate (figure 13.8).
I I C
I
I
IT
Is lip
Figure 13.8 Sputtering (G= gasinlet," T= target," S= substrate," P= pumps).
452 In the magnetron-sputtering technique (figure 13.9), the target is subjected to a magnetic field. This causes an additional ionization of the plasma near the target, allowing lower target temperatures.
II
~
L--
M
I
!r
i
is il
P
Figure 13.9 Magnetron sputtering (G= gasinlet," M= magnetron; T= target; S= substrate; P= pumps).
2.3 Ion plating Ion plating is a process in which a substrate is subjected to ion bombardment both before and during the time it is being coated. The coating process used in ion plating is vacuum evaporation from resistance - heated sources. The vacuum procedures in ion plating are the same as those used in sputtering in that the system is first pumped to a good vacuum and argon gas is then admitted to establish a state of dynamic epuilibrium in the chamber. The chamber pressure remains constant with argon being pumped out of the chamber at the same rate as it is flowing into the chamber. Chamber pressures in the order of 5 Pa are fairly standard in ion plating. A typical ion plating system is depicted in figure 13.10. Radio-frequency power is applied to the substrate, as in sputter cleaning, for the purposes of generating a plasma and of causing ion bombardment on the substrate surface. After sputter-cleaning has removed the contaminants from the substrate to establish atomically clean surfaces, the evaporation source is activated to begin the coating process.
453
§+
T
-I
II a I r +
T
IS II P
Figure 13.10 Ion plating (G= gasinlet; S= substrate," T= target," P= pumps). 3 Atomic Layer Epitaxy Chemical Vapour Deposition involves a very complex mixture of gases and/or ions, causing numerous uncontrollable side reactions. From a chemical point of view, these reactions are extremely difficult to monitor. A basic understanding of all reactions occurring in a CVD experiment is lacking at the present time. This was recknognized by Suntola ~ in the late 1970's, when he developed a new coating technique: Atomic Layer Epitaxy (ALE). Atomic layer epitaxy is a method for producing thin films and layers of single crystals one atomic layer at a time, utilizing a self-control obtained through saturating surface reactions. ALE is thus based on separate surface reactions between the growing surface and each of the components of the compound, one at a time. These components are supplied in the vapour phase, either as elemental vapours or as volatile compounds of the elements. ALE was originally developed to meet the needs of improved ZnS thin films and dielectric thin films for electroluminescent thin film display devices. As an illustration of the ALE process, figure 13.11 summarizes the basic sequences of ALE in two alternative ways for producing ZnS.
454
ooooo, OO,o. 9
'I ~ CI + Surplus H2S [
Zn (a)
2Zn+S
Figure 13.11 the reactants.
-. 2ZnS
, I (b)
ZnCI + H S -. ZnS + 2HCI
ALE process for ZnS. (a) Zn and S as the reactants; (b) ZnCI2 and H:S as
In each reaction there are sites for only one monolayer to make a bond with the original surface. A basic condition for a successful ALE process is that the binding energy of a monolayer chemisorbed on a surface is higher than the binding energy of subsequent layers on top of the formed monolayer. The temperature of the substrate is used as the primary controlling parameter. It is adjusted low enough to keep the monolayer on the surface until reaction with the following reactants takes place, but high enough to re - evaporate any of the subsequent layers on the top of the monolayer. The control of the monolayer can further be influenced with the aid of a laser beam or other extra energy. The greater the difference between the bond energy of a monolayer and the bond energies of the subsequent layers, the better the self- controlling characteristics of the process. Several general reaction mechanisms can be discerned in the ALE process.
The
simplest reaction sequence is an elemental reaction, which can be represented as reaction (A).
B(s) + A(g) --, BA(s)
(A)
455 The reaction of elemental Zn and S to form a ZnS layer is a typical example of such reaction. However, there might be different reasons for using compounds of the elements A and B as the reactants, rather than the elemental vapours of A and B. The most important reason is that for many metals, the vapour pressure is (too) low, which makes it very difficult to prevent condensation of the element at low substrate temperatures. In this cases, compound exchange reactions are used:
B(s) + AX(g) --, B-AX(s) B-AX(s) + B Y ( g ) - , B-AB(s) + XY(g)
(B)
The reaction ZnC12 + H2S --* ZnS + 2HC1 is a typical example. In many cases, the compound AX is a metal halide and the compound BY is a hydride. However, also organometallic compounds are attractive alternatives as reactants in an ALE process, owing to their high vapour pressure. There is a large amount of information available on many important organometallic compounds, because of the wide use in the CVD technology. Originated in the late 1970's, ALE has become a widely used synthesis route for thin epitaxial layers. Many ALE coated materials are produced commercially today. A very extended review on the achievements of the ALE technique, written by the pioneer in the field, is found in reference 11.
4 Molecular Layering In Eastern Europe, parallel to the development of ALE, a quite similar coating technique was developed, called Molecular Layering. 12'13'~4'~5 The principles of Molecular Layering are based on the irreversible chemical reactions of the substrate with a functional reagent. In this way, monolayers can be created to be used as such or to undergo further reaction with a second reagent. Its principles are visualized in figure 13.12.
4~
a) layer of given thickness
functionalgroups BB
BB
BB
pB
~
BB
/
B B B B B
~
~
\
+Ac2 + BC-
solid body b) layers with given disposition of monolayers of different chemical nature +AC 4
c
cc
cc, ,c
/
BB
\
/
BB
\
/
BB
k
+ I~4
/
v
B
C
- BC ~
CC CC C "(~ "(~/ "(1~/
+
-BC c) muldcoml~nent monolayers with given C CC CC CC C
monolayer
mixture
C CC CC CC C \
+ (xAC4 + yNC 4 )
I
\_.J
\_.../ \
I
- BC, - zAC4
B CN'C B
C~(~/C
+ yNC,
-B---F-"
C CCCC C ~A)/ ~ ~
~ - ~, \ \' ~~\ ~~\ \'~~\ \ . ~ .~%
BB
+ zMC4
~"
- BC
"
Figure 13.12 Principles of the molecular layering technique, according to Malygin; taken from ref. (22) with permission.
457 An important application of this technique is the creation of thin transition metal oxides on a substrate, by reacting consecutively with a metalchloride (TIC14) or oxychloride (CRO2C12, VOC13) and water, according to reaction (C). 16'17'~g ( - Si-OH),, + TiCl 4
(~- Si-O).TiCl,.
+
~
(4-n)H20
( - - Si-O)nTi(OH)4_n
+
( - Si-O).TiCl4..
~
+
(-= Si-O).Ti(OH),.
TiC! 4
n HC!
+
(4-n)HCl
---> ( - - Si-O).(Ti.O)mTiCl4_ m
(~- Si-O).(Ti-O)mTiCl4_ m +
(4-m)H20
-,,
....
(C)
Another typical application is the creation of sensors. Vanadium modified silica, according to reaction (D), yields a very useful compound for the visual control of humidity in gas media. 19 (-- SiOH). + VOCI 3 ~
(---Si-O).VOCI3_ n + nHC!
(D)
The different colour changes, compared to commercial cobalt containing silica, are presented in table 13.5.
Table 13.5 Comparison of the sensitivity (colour changes) of V-modified silica (ML-method) and commercial Co-containing silica Relative humidity % Dew point K
0.5 233
1.6-4.6 6-10 243-253 255-260
10-13 261-263
15-45 264-279
48-95 281-292
colour V-silica of the external layer commerc, Co-silica
lemon
bright yellow
light orange
dark orange
from red brown to dark
blue
blue
light blue
light rosy
rosy
458
5 Chemical Surface Coating In the early 90's, a new technique, 'Chemical Surface Coating (CSC)' was developed by Vansant, Gillis-D'Hamers, Van Der Voort and Vrancken, in order to create thin ceramic layers on a substrate by successive chemical modifications. The principles and developments of this technique are the subject of a separate section of this book (cfr. chapter 14).
References
D.G. Bhat, A review of chemical vapour deposition-techniques, materials and applications; surface modification technologies, eds. T.S. Sudarshan and D.G. Bhat, 1988.
.
D
C.F. Powell, in Vapour Deposition, C.F. Powell, J.H. Oxley and J.M. Blocher jr. eds., John Wiley and Sons, New York, 1966, pp. 249-276.
3.
D.P. Dapkus, Ann. Rev. Mater. Sci., 1982, 12, 243.
4.
R.A. Holtz, R.E. Benander and R.D. Davis, U.S. Patent 4 427 445, 1984.
5.
N.J. Archer and K.K. Yee, Wear, 1978, 48, 237.
6.
R.B. Pratt et al., Nucl. Appl., 1969, 6, 241.
0
S. Lai, M.P. Dudokovic and P.A. Ramachandran, Chem. Eng. Sci., 1986, 41, 633.
8.
G.C. Hsu et al., NASA Tech. Brief., 1985, 9, 104.
9.
D.P. Stinton, A.J. Caputo and R.A. Lowden, Ceramic Bulletin, 1986, 65, 347.
10.
C. Hollabaugh, in Proceedings of sixth international conference on chemical vapour deposition, The Electrochemical Society Inc., Princeton, New Jersey, 1977, pp. 419-429.
11.
T. Suntola, Materials Science Reports, 1989, 4, 261 - 321 and references therein.
12.
A.A. Chuiko, Teoret. i. Eksperim. Khimiya, 1987, 5, 597.
13.
V.B. Aleskovskiy, Stehiometriya i sinteztverdyh soedinery, L/Nauka, 1976.
459 15.
S.I. Koltsov and V.B. Aleskovskiy, Zhurnal Prilk. Khimii, 1967, 40, 207.
16.
M.N. Tzetkova, I.M. Yur'evskaya and A.A. Malygin, Zhurn. Prikl. Khimii., 1982, 2, 256.
17.
S.D. Dubrovenskiy, A.E. Emelyanov and A.V. Zimin, Zhurn. Prikl. Khimii, 1992,
65, 2259.
18.
E.A. Avrutina, Avtoreferat diss. kandidat khim. nauka., Leningrad, 1989.
19.
A.A. Malygin, private communication.
This Page Intentionally Left Blank
461
Chapter 14
Chemical Surface Coating
1 Principles of Chemical Surface Coating The development of Atomic Layer Epitaxy was the first step in the simplification of the complex CVD process, in order to achieve a better understanding and thus a better control of the reaction mechanisms involved. However, the interaction between the coating and the substrate still has Van Der Waals character. Although the homogeneity of the ALE layer has improved enormously, the problem of (in)compatibility between substrate and coating remains. Therefore, in the early 1990's a new coating technique has been developed by GillisD'Hamers, Van Der Voort, Vrancken and Vansant. ~'2'3 This new procedure is called Chemical Surface Coating (CSC) and is presented in figure 14.1. In the gas-phase modification of Chemical Surface Coating, the inorganic substrate is subjected to subsequent, single step, one component reactions. This process is repeated in a cyclic way. The reaction temperatures are very low, usually room temperature. In this way, the ceramic precursor is built. Its thickness is a function of the number of CSC cycles involved. A CSC cycle is a subsequent modification of the substrate with 2 gases. Finally, the ceramic precursor is converted towards a ceramic coating by a thermal treatment.
462
CSC Chemical Surface Coating
[iiiiiiiiiilliiiiii!lliiiiiii ii"ii'iii'i"iii i
iiiiii"i
I Modification I
GAS 1
I
LIQUID Organosilane
GAS 2 iii}|174174
|
[~iiiii~ii!iii~ii!iiii!i~i!iiiii~iii~i~i|
Ceramic ~onversion
Iiiiiiiiii iliii!il :!!! i~