Inorganic and Organometallic Oligomers and Polymers: Proceedings of the 33-rd IUPAC Symposium on Macromolecules 9789401054171

Although, carbon is only one of one hundred plus elements, the polymer science lit­ erature consists primarily of studie

224 97 5MB

English Pages 327 [328] Year 1991

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Cover
Half Title
Inorganic and Organometallic Oligomers and Polymers: Proceedings of the 33rd IUPAC Symposium on Macromolecules
Copyright
Contents
Preface
Introduction
List of Authors
Polysilanes and Polystannanes
Catalytic Dehydrogenative Polymerization of Silanes to Polysilanes by Zirconocene and Hafnocene Catalysts. A New Polymerization Mechanism
1. Introduction
2. Mechanistic Considerations
3. d0 Silyl Complexes as Catalyst Precursors
4. Mechanism Based on σ-Bond Metathesis
5. Summary
6. Acknowledgement
7. References
Silane Dehydrocoupling Reactions Catalyzed by the Late Transition Metals
1. Introduction
2. Results
3. Discussion
4. Acknowledgements
5. References and Notes
Synthesis of Poly(methylsilylene) by Catalytic Dehydrocoupling with Cp2MM2 ( M = Ti, Zr) Catalysts
1. Introduction
2. Experimental
2.1 General Procedures
2.2 Synthesis of Methylsilane
2.3 Polymerization of Methylsilane
3. Results
3.1 The Polymerization Reaction
3.2 IR and NMR Spectra of Polymethylsilane
3.3 Molecular Weiht Properties
4. Discussion
5. Conclusions
6. Acknowledgements
7. References
The Preparation of Polysilylenes
Introduction
Experimental
Results
1. Chlorination
2. Preparation of cyclic material
3. Polymerization of tetracyclosilanes
Discussion
References
Characterization of Polysilanes by UV, GPC and Light-Scattering
1. Introduction
2. Experimental
2.1. Synthesis and Fractionation
2.2. Gel Permeation Chromatography
2.3. Light-Scattering
3. Results
3.1. Syntheses and Fractionation
3.2 GPC Calibration
3.3. Hydrodynamic Properties in Solution
3.4. UV Absorption Maxima Wavelengths
4. Conclusion
5. Bibliography
Effect of Thermally Induced Transitions on Electronic Transport in Aliphatic Polysilylenes
Linear and Nonlinear Optics in Substituted Polysilanes
Electronic Structures and Physical Properties of Sigma-Conjugated Polymers
The Molecular and Electronic Structure of Polycyclic Polystannanes
Non-oxide Ceramic Precursors
Polymethylsilane as a Precursor to High Purity Silicon Carbide
Silicon Carbide Preceramic Polymers as Binders for Ceramic Powders
Pyrolytic Characteristics of Polysilazanes
Silicon and Boron Containing Oligomers: Potential Precursors for Ceramics
Preparation and Microstructure of Organometallic Polymer Derived AIN-BN Composites
Polymeric Precursors to Boron Nitride Ceramics
Ceramic Materials via Derivatization Reactions of Polymers
Oxide Ceramic Precursors
Some Aspects of the Chemistry of Transition Metal Oxide Gels
Kinetics and Structure of Silicate Sol-Gels
Mechanistic Aspects of the Pyrolytic Transformation of Metal Alkoxides to Oxides
Novel Organometallic Polymers
Poly(alkylene phosphates): Synthetic Strategies
New Metal-Chalcogen Compounds with Polymeric Structures
The Study of Plasma Stability of Poly(Organophosphazene) Films Prepared on Silicon Wafer
Oxygen-Carrying Polychelates Derived from Bisphenolic Complexes
Coordination Polymers Derived from Bisphenolic Complexes
List of Key Words, Names
Recommend Papers

Inorganic and Organometallic Oligomers and Polymers: Proceedings of the 33-rd IUPAC Symposium on Macromolecules
 9789401054171

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

INORGANIC AND ORGANOMETALLIC OLIGOMERS AND POLYMERS

INORGANIC AND ORGANOMETALLIC OLIGOMERS AND POLYMERS Proceedings ofthe 33rd IUPAC Symposium on Macromolecules

Edited by

JOHN F. HARROD Department of Chemistry, McGilI University, Montreal, Canada

and

RICHARD M. LAINE Department of Materials Science and Engineering, University of Michigan, Ann Arbor, MI, U.S.A .

..

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

Library of Congress Cataloging-in-Publication Data Inorganic and organometall ic oligomers and polymers proceedings of the 33rd IUPAC Sy~posium on Macromolecules ! edited by John F. Harrod. Richard M. Laine. p. cm. "Papers [whichl orig,nated in the symposium on Inorganic Polymers and Oligomers held as part of Macro 90 at the 33rd IUPAC International Symposium on Macromolecules in Montreal, Canada"-Introd. Includes bibl iographical references and index. ISBN 978-94-010-5417-1 ISBN 978-94-011-3214-5 (eBook) DOI 10.1007/978-94-011-3214-5 I. Harrod, John F. (John 1. Inorganic pOlymers--Congresses.

III. IUPAC Symposium on Frank) II. Laine, Richard M., 1947Macromolecules (33rd 1990 Montreal, Quebec) QD196.I42 1991 91-614 547.7--dc20

ISBN 978-94-010-5417-1

Printed on acid-free paper

AU Rights Reserved © 1991 Springer Science+Business Media Dordrecht Originally published by Kluwer Academic Publishers in 1991 Softcover reprint of the hardcover 1st edition 1991

No part of the material protected by this copyright notice may be reproduced Of utilized in any form or by any means, electronic or mechanical, inc\uding photocopying, recording or by any information storage and retrieval system, without written permission from the copyright owner.

CONTENTS Preface

vii

Introduction

ix

List of Authors

xiii

Polysilanes and Polystannanes Catalytic Dehydrogenative Polymerization of Silanes to Polysilanes by Zirconocene and Hafnocene Catalysts. A New Polymerization Mechanism. T. Don Tilley and Hee-Gweon Woo Silane Dehydrocoupling Reactions Catalyzed by the Late Transition Metals Claire A. Tessiers, Vance O. Kennedy, and Eugene A. Zarate. Synthesis of Poly(methylsilylene) by Catalytic Dehydrocoupling with

3 J3

Cp2MM~

( M = Ti, Zr) Catalysts. Ying Mu and John Frank Harrod

23

The Preparation of Polysilylenes S. Gauthier and D. J. Worsfold

37

Characterization of Polysilanes by UV, GPC and Light-Scattering J. Devaux, D. Daoust, A.-F. de Mahieu, C. Strazielle

49

Effect of Thermally Induced Transitions on Electronic Transport in Aliphatic Polysilylenes M. Stolka, M. A. Abkowitz, F. E. Knier, K. M. McGrane, R. J. Weagley, and J. M. Zeigler

61

Linear and Nonlinear Optics in Substituted Polysilanes F. M. Schellenberg, R. L. Byer, R. D. Miller, R. H. French, S.S. Kano, Y. Takahashi, Y. Shiraki, R. Ito

73

Electronic Structures and Physical Properties of Sigma-Conjugated Polymers N. Matsumoto

97

The Molecular and Electronic Structure of Polycyclic Polystannanes Lawrence R. Sita and lsamu Kinoshita

J 15

Non-oxide Cframic Precursors Polymethylsilane as a Precursor to High Purity Silicon Carbide Z.-F. Zhang, Y. Mu, F. Babonneau, R. M. Laine, J. F. Harrod, and J. A. Rahn

127

Silicon Carbide Preceramic Polymers as Binders for Ceramic Powders William H. Atwell, Gary T. Burns, and Gregg A. Zank

147

vi Pyrolytic Characteristics of Polysilazanes Yigal D. Blum,Gregory A. McDermott, and Albert S. Hirschon

161

Silicon and Boron Containing Oligomers: Potential Precursors for Ceramics Kenneth E. Gonsalves

177

Preparation and Microstructure of Organometallic Polymer Derived AIN -BN Composites D. Kwon, W. R. Schmidt, and L. V. Interrante

191

Polymeric Precursors to Boron Nitride Ceramics Larry G. Sneddon, Kai Su, Paul J. Fazen, Anne T. Lynch, Edward E. Remsen, and Jeffrey S. Beck

199

Ceramic Materials via Derivatization Reactions of Polymers L. Maya

209

Oxjde Ceramjc Precursors Some Aspects of the Chemistry of Transition Metal Oxide Gels J. Livage, F. Babonneau, C. Sanchez

217

Kinetics and Structure of Silicate Sol-Gels Roger A. Assink, C. Jeffrey Brinker, and Bruce D. Kay

229

Mechanistic Aspects of the Pyrolytic Transformation of Metal Alkoxides to Oxides Ayusman Sen, Manish Nandi, Hilmar A. Stecher, and Doug Rhubright

235

Noyel Organometa!ljc Polymers Poly(alkylene phosphates): Synthetic Strategies Gary M. Gray, Keith E. Branham, Lung-Hua Ho, Jimmy W. Mays, Prakash C.Bharara, Andreas Hajipetrou, and James B. Beal

249

New Metal-Chalcogen Compounds with Polymeric Structures Y. Park, J.-H. Liao, K.-W. Kim, and M. G. Kanatzidis

263

The Study of Plasma Stability ofPoly(Organophosphazene) Films Prepared on Silicon Wafer M. Kajiwara, and Y. Yamashita

277

Oxygen-Carrying Polychelates Derived from Bisphenolic Complexes M. Spiratos, A. Airinei, and N. Voiculescu

295

Coordination Polymers Derived from Bisphenolic Complexes A. Airinei and M. Spiratos

30 I

List of Key Words, Names

313

PREFACE Although, carbon is only one of one hundred plus elements, the polymer science literature consists primarily of studies on carbon based polymers. In part, this reflects the varied feedstock sources and in part, the type of bonds and bond forming reactions available to form organic polymers that are not available to the inorganic and organometallic chemist. However, recent intense interest in polymers with novel optical, electronic or magnetic properties or polymers that can serve as precursors to ceramic, semiconductor, metallic or superconductor materials has served as a driver for the development of novel synthetic routes and characterization techniques that have launched many new inorganic and organometallic oligomers and polymer systems. The following chapters represent an effort to provide an overview of several new and continuing areas of development in inorganic and organometallic polymer science. This book represents the second in a series of books we have edited on inorganic and organometallic polymer chemistry (1. Transformation of Organo-metallics into Common and Exotic Materials, NATO ASI Series Vol 141. 3. Inorganic and Organometallic Polymers with Special Properties, NATO ASI Series in press). In this series, we attempt to develop, for the reader, an understanding of the breadth, depth and potential of inorganic and organometallic polymer science. We would like to thank the IUPAC committee for the Internat-ional Symposium on Macromolecules for permission to publish this symposium. We would also like to thank our co-organizer, Jacques Livage and, the many authors and reviewers for their timely efforts. We would also like to thank Dr. Kenneth Wynne of the U. S. Office of Naval Research for his participation in the meeting and for writing the Introduction. February, 1991

John F. Harrod Department of Chemistry McGill University Montreal, Canada

Richard M. Laine Department of Materials Science and Engineering University of Michigan Ann Arbor, MI

vii

INTRODUCTION This Volume of papers originated in the symposium on Inorganic Polymers and Oligomers held as part of Macro 90 at the 33rd IUPAC International Symposium on Macromolecules in Montreal Canada. The symposium reflects continued interest in non-carbon backbone polymers which stems from unusual properties and the promise of unique applications. This symposium was not comprehensive in nature, but focussed on two topical areas: po/ysi/ylenes and, oxide- and non-oxide preceramic polymers. The symposium gained breadth through covering related polymers in a session on novel organometallic polymers. The polysilylenes continue to be of interest because of the novel electronic and optical properties of these macromolecules. These properties stem from the presence of a o:-delocalized -Si-Si- polymer backbone. The interplay between conformational states, sidechain order/disorder transitions, and electronic states creates interesting temperature and pressure dependent optical phenomena. These novel properties have stimulated interest in improved synthetic methods. Thus, the first three papers in this symposium focus on understanding and improving catalytic methods for generating silylene chains. These efforts are followed by Worsfold's research on improving the traditional Wurtz coupling method of synthesis. The interesting properties of the poly(silylenes) are then described in the remaining papers. Included are efforts on the nonlinear optical behavior, photoconducting properties, and phase and solution behavior. A paper on polycyclic polystannanes completes this topic. The principal emphasis of the remainder of the symposium is on the synthesis and characterization of preceramic polymers. The overall goal of preceramic polymer research is the creation of polymers with noncarbon backbones having physical and mechanical properties associated with typical organic polymers, e.g., facile synthesis and ease of processing at low temperatures. In this approach, the polymer is processed into the final shape by standard solution or melt processing methods. This strategy would give fibers, films, coatings, and even bulk bodies in complex shapes, for example by injection molding. Only in the final stage does thermal treatment give the required ceramic material. This approach provides a route to compositions, morphologies, and phases difficult or impossible to ix

x

obtain by traditional ceramic synthesis and processing methods. Near net shape injection molding followed by thermal treatment would obviate costly machining. To reach these goals, much research has been carried out in preceramic polymer synthesis and processing. Preceramic polymers represent a challenging area of research and a good deal of the work has been ad hoc in nature. Some of the difficulties include: (1) the lack of catalyst specificity for the creation of polymer chains of desired homo- or heterocatenated atoms, (2) characterization difficulties associated with preceramic polymers which are usually amorphous, reactive, and may contain branched structures and a variety of repeat units randomly distributed, (3) amorphous intermediates in the pyrolysis process which have intermediate properties between ceramics and thermosetting polymers, (4) difficulty in obtaining elemental analyses on polymers which have highly refractory oxidation and/or reduction products, (5) control of shrinkage due to the higher density of ceramic product versus polymer precursor, and (6) ceramic products which are themselves difficult to characterize (e.g., amorphous to x-rays unless crystallized at high temperatures). Fortunately, these difficulties have been seen by the many talented participants in this symposium as opportunities, and one or more of the items listed have been attacked with vigor by research groups whose results are described in this volume. It is convenient to divide the area of preceramic polymers into ~ and non-oxide polymers and the papers are organized accordingly. This distinction is useful, but important work on "hybrid" systems must be recognized, such as the work of Lasocki1 on silicon oxygen-nitrogen containing polymers. In the non-oxide area, research on precursors to silicon carbide, silicon nitride, and related ceramic materials continues to receive impetus from the pioneering work of Seichi Yajima in the 70's.2 This well known work formed the basis for Nippon Carbon "NicalonR" silicon carbide fibers, currently distributed in the United States by the Dow Corning Corporation. 1 Z. Lasocki, M. Witekowa, J. Organornetal. Chern. 1986, .lll..17. 2 S. Yajima, Y. Hasigawa, J. Hayashi, Iirnura, J. Mater. Sci.", 1978, 13, 2569.

xi

The potential applications for research in carbide, nitride, and other non-oxide precursor polymers have not changed greatly since this work was reviewed some years ago,3 and include fibers, binders and protective coatings. A unique property of the ceramic often drives interest in utilizing the preceramic polymer approach. Thus, the high thermal conductivity of aluminum nitride makes this ceramic attractive for reinforcing fibers for circuit boards or other applications where thermal sinking is important. Silicon carbide has been a target material for much research. The first two papers in the section on non-oxide ceramic precursors address progress in the synthesis of new SiC preceramic polymers. Silicon nitride has also been of great interest because of its low dielectric constant and optical properties. Blum describes a useful route to Si3N4 in this section. Aluminum nitride and boron precursors are addressed in the remaining papers in this section. Boron nitride fibers have the layered hexagonal sheet structure similar to that found in carbon fibers, but localization of 1t-electron density results in BN being an insulator with good dielectric properties. Oxide-ceramic precursors are covered in the next section. In the sol-gel approach, monomers are usually metal or metalloid alkoxides, M(OR)n' Hydrolysis and condensation of these alkoxides leads to formation of oxo-polymers in solution or suspension. The kinetics of these polymerization reactions and the polymer structure are strongly dependent on the nature of M, the pH of the solution, and the presence of additional groups attached to M. This interesting complexity is described in the three papers on oxide ceramic precursors. The control of structure and size of oxo-polymers through moderation of reactivity by the presence of certain chelating groups brings about concomitant control of ceramic particle size. The important consequences of this control are described by Livage. In the last section, new coordination polymer chemistry related to sol-gel polymers is presented. In addition, two papers address phosphorus containing polymers and two address traditional coordination polymers used for a wide variety of applications. Overall, the papers presented in this symposium give a representative view of the vitality of research on poly(silylenes) and prece3 K. J. Wynne, R. W. Rice, Ann. Rev. Mater. Sci., 1984 14,297.

xii

ramic polymers. The interesting progress described promises continued advances for new materials with new capabilities. Kenneth J. Wynne, Program Manager Organic and Polymeric Materials Office of Naval Research

LIST OF AUrnORS Affiliation Abkowitz, M. A. Airinei, A. Assink, R. A. Atwell,W. H.

Xerox Webster Research Center, Webster, N. Y. P. Poni Institute of Macromolecular Chemistry, lasi, Romania Sandia National Laboratories, Albuquerque, NM Advanced Ceramics Program, Dow Coming Co., Midland, MI

Babonneau, F.

Chimie de la Matiere Condensee, Universite Pierre et Marie Curie, Paris, France Dept. of Chemistry, University of Montevallo, Montevallo, AL Dept. of Chemistry, University of Pennsylvania, Philadelphia, PA Dept. of Chemistry, University of Montevallo, Montevallo, AL SRI International, Menlo Park, CA Dept. of Chemistry, University of Alabama, Birmingham, Al Sandia National Laboratories, Albuquerque, NM Advanced Ceramics Program, Dow Coming Co., Midland, MI E. L. Ginzton Laboratory, Stanford University, Stanford, CA

Beal, J. B Beck, J. S. Bharara, P. C. Blum, Y. D. Branham, K. E. Brinker, C. J. Burns, G. T. Byer, R. L. Daoust, D. Devaux, J. de Mahieu, A.-F.

Laboratoire des Hauts Polymeres, Universite Catholique de Louvain, Louvain-Ia-Neuve, Belgium Laboratoire des Hauts Polymeres, Universite Catholique de Louvain, Louvain-Ia-Neuve, Belgium Laboratoire des Hauts Polymeres, Universite Catholique de Louvain, Louvain-Ia-Neuve, Belgium

Fazen, P. J. French, R. H.

Dept. of Chemistry, University of Pennsylvania, Philadelphia, PA E. I. DuPont Experimental Station, Wilmington, DE

Gauthier, S. Gonsalves, K. E. Gray, G. M.

National Research Council of Canada, Ottawa, Canada Dept. of Chemistry, University of Connecticut, Storrs, CT Dept. of Chemistry, University of Alabama, Birmingham, Al

Harrod, J. F. Hajipetrou, A. Hirschon, A.S. Ho, L-H.

Chemistry Dept., McGill University, Montreal, Canada Dept. of Chemistry, University of Montevallo, Montevallo, AL SRI International, Menlo Park, CA Dept. of Chemistry, University of Alabama, Birmingham, Al

Interrante, L. V. Ito, R.

Dept. of Chemistry, Rensselaer Polytechnic Institute, Troy, N. Y. Research Center for Advanced Science and Tech., Tokyo, Japan xiii

xiv

Kajiwara, M. Kanatzidis, M. G. Kano, S.S. Kay, B. D. Kennedy, V. O. Kim, K.-W Kinoshita, I. Knier, F. E. Kwon, D.

Dept. of Applied Chemistry, Nagoya University, Nagoya, Japan Dept. of Chemistry, Michigan State University, East Lansing, MI IBM Tokyo Research Laboratory, Tokyo, Japan Sandia National Laboratories, Albuquerque, NM Dept. of Chemistry, University of Akron, Akron, OR Dept. of Chemistry, Michigan State University, East Lansing, MI Beckman Institute Molecular Materials Resource Center, Caltech, Pasadena, CA Xerox Webster Research Center, Webster, N. Y Dept. of Chemistry, Rensselaer Polytechnic Institute, Troy, N. Y

Lynch, A. T.

Dept. of Mater. Sci. and Eng., University of Michigan, MI Dept. of Chemistry, Michigan State University, East Lansing, MI Chimie de la Matiere Condensee, Universite Pierre et Marie Curie, Paris, France Dept. of Chemistry, University of Pennsylvania, Philadelphia, PA

Maciel, G. Marchetti, P. Matsumoto, N. Maya, L. Mays, J. McGrane, K. M. McDermott, G. A. Miller, R. D. Mu,Y

Dept. of Chemistry, Colorado State University, Fort Collins, CO Dept. of Chemistry, Colorado State University, Fort Collins, CO NTT Basic Research Laboratories, Tokyo, Japan Oak Ridge National Laboratory, Oak Ridge, TN Dept. of Chemistry, University of Alabama, Birmingham, Al Xerox Webster Research Center, Webster, N. Y SRI International, Menlo Park, CA IBM Almaden Research Center, San Jose, CA Chemistry Dept., McGill University, Montreal, Canada

Nandi, M.

Dept. of Chemistry, Pennsylvania State Univ., State College, PA

Park, Y

Dept. of Chemistry, Michigan State University, East Lansing, MI

Rahn, J. A. Rhubright, D. Remsen, E. E.

Dept. of Mater. Sci. and Eng., University of Michigan, MI Dept. of Chemistry, Pennsylvania State Univ., State College, PA Dept. of Chemistry, University of Pennsylvania, Philadelphia, PA

Sanchez, C.

Chimie de la Matiere Condensee, Universite Pierre et Marie Curie, Paris, France E. L. Ginzton Laboratory, Stanford University, Stanford, CA Dept. of Chemistry, Rensselaer Polytechnic Institute, Troy, N. Y Dept. of Chemistry, Pennsylvania State Univ., State College, PA Research Center for Advanced Science and Tech., Tokyo, Japan

Laine, R. M. Liao, J.-R Livage, J.

Schellenberg, F. M. Schmidt, W. R. Sen, A. Shiraki, Y

xv

Sita, L. R. Sneddon, L. G. Spiratos, M. Stecher, H. A. Stolka, M. Strazielle, C. Su, K.

Beckman Institute Molecular Materials Resource Center, Caltech, Pasadena, CA Dept. of Chemistry, University of Pennsylvania, Philadelphia, PA P. Poni Institute of Macromolecular Chemistry, Iasi, Romania Dept. of Chemistry, Pennsylvania State Univ., State College, PA Xerox Webster Research Center, Webster, N. Y. Institut Charles Sadron, Strasbourg, France Dept. of Chemistry, University of Pennsylvania, Philadelphia, PA

Takahashi, Y. Tessier,C. A. Tilley, T. D

IBM Tokyo Research Laboratory, Tokyo, Japan Dept. of Chemistry, University of Akron, Akron, OH Dept. of Chemistry, University of California, San Diego, CA

\biculescu, N.

P. Poni Institute of Macromolecular Chemistry, Iasi, Romania

Weagley, R. J. Woo, H.-G. Worsfold, D. J. Wynne, Kenneth J.

Xerox Webster Research Center, Webster, N. Y. Dept. of Chemistry, University of California, San Diego, CA National Research Council of Canada, Ottawa, Canada Office of Naval Research, Arlington, VA

Yamashita, Y.

Dept. of Applied Chemistry, Nagoya University, Nagoya, Japan

Zarate, E. A. Zank, G. A. Ziegler, J. M. Zhang, Z.-F.

Dept. of Chemistry, University of Akron, Akron, OR Advanced Ceramics Program, Dow Coming Co., Midland, MI Silchemy Co., Albuquerque, NM Dept. of Mater. Sci. and Eng., University of Michigan, MI

POLYSILANES

AND POLYSTANNANES

CATALYTIC DEHYDROGENATIVE POLYMERIZATION OF SILANES TO POL YSILANES BY ZIRCONOCENE AND HAFNOCENE CATALYSTS. A NEW POLYMERIZATION MECHANISM.

T. Don Tilley and Hee-Gweon Woo Chemistry Department, D-006 University of California at San Diego La Jolla, California 92093-0506 U.S.A. ABSTRACT. Polysilanes, (-SiRR'-)n, represent a class of inorganic polymers that have unusual chemical properties and a number of potential applications. Currently the most practical synthesis is the Wurtz-type coupling of a dihalosilane with an alkali metal, which suffers from a number of limitations that discourage commercial development. A coordination polymerization route to polysilanes based on a transition metal catalyst offers a number of potential advantages. Both late and early metal dehydrogenative coupling catalysts have been reported, but the best to date appear to be based on titanocene and zirconocene derivatives. Our studies with transition metal silicon complexes have uncovered a number of observations that are relevant to this reaction chemistry, and hopefully important with respect to development of better catalysts. We have determined that many early transition metal silyl complexes are active catalysts for polysilane synthesis from monosilanes. A number of structure-reactivity correlations have been established, and reactivity studies have implicated a new metal-mediated polymerization mechanism. This mechanism, based on step growth of the polymer, has been tested in a number of ways. All proposed intermediates have now been observed in model reactions.

1. Introduction Recent advances in the synthesis, characterization, and study of inorganic polymers has directed considerable attention toward this growing area of polymer chemistry [1]. Much of this interest has focused on polysilanes, (-SiRR'-)n, which possess a backbone consisting entirely of Si-Si bonds. Though polysilanes have been known for some time, recent synthetic and physical studies have revealed that these polymers have unusual properties that suggest a number of applications. These include uses as photoresist materials, preceramic fibers, photoconductors, dopable semiconductors, nonlinear optical materials, and photoinitiators for radical polymerizations. Currently the most viable method for preparing polysilanes is the Wurtz-type coupling of dichlorosilanes by an alkali metal, which suffers from the limitations imposed by severely reducing conditions and non selectivity [2]. A feasible alternative to the Wurtz-coupling method is a coordination polymerization based on transition-metal catalysis, especially given the broad range of reactions that have been identified for transition metal-silicon systems [3]. The problem involves design of a catalyst that will extrude a silylene (SiR2) or disilene (R2SiSiR2) unit from a small molecule and transfer it to a growing polymer chain. So far most attention has focused on the dehydrocoupling ofhydrosilanes as shown in equation 1 [4]. 3 J. F. Harrod and R. M. Laine (eds.).lnorganic and Organometallic Oligomers and Polymers. 3-11. © 1991 Kluwer Academic Publishers.

T. D. TILLEY AND HEE-GWEON WOO

4

catalyst

n RR'SiH2

(n-1)H2 +

H(SiRR')nH

(1)

Two classes of transition-metal catalysts have been reported for this reaction. The first, discovered 20 years ago [5], is based on a platinum-group metal such as rhodium, palladium, or platinum. Of these the most active catalyst appears to be (Ph3P)3RhCI [6], which produces low to moderate yields of di- and trisilanes from secondary silanes (RR'SiH2) [7]. A second class of catalysts was discovered by Harrod and coworkers [8], who showed that titanocene and zirconocene derivatives dehydrogenate primary silanes (RSiH3) to polysilanes with 10-20 monomer units, in high yield. These catalysts therefore appear to be more promising with respect to production of polymers, and have revived interest in coordination polymerization routes to polysilanes. 2. Mechanistic Considerations

Progress toward development of useful transition-metal mediated syntheses of polysilanes probably depends on an understanding of how the catalysts operate. Two main mechanistic proposals have been put forth to explain how transition metal-silicon chemistry can disproportionate Si-H bonds to H-H and Si-Si bonds. The older proposal is based on a metal silylene intermediate (LnM=SiR2) [9]. In this scheme (Figure 1), a silane is converted to a silylene ligand at a metal center. This intermediate then undergoes coupling reactions with more silanes via Si-Si bond formation. The second mechanism is based simply on oxidative addition-reductive elimination cycles [10] (Figure 2). Although both mechanisms are feasible for platinum-metal catalysts, they seem less likely for the Harrod-type catalysts based on early transition metals. For these catalysts, arguments against mechanisms involving a silylene complex can be based on the fact that secondary silanes are oligomerized [11,12], and because only one a-bond of a CP2MRR' (M = Ti, Zr; Cp = 115_C5H5) complex is necessary for the polymerization reaction [13]. In addition, the mechanism of Figure 2 is not very attractive for the group 4 metal catalysts, since for these metals Mn+/M(n+2)+ oxidation state cycles are in general not facile. R2HSiSiHR2

--1---

LuM-H I SiR2SiHR2 H

LuM

/-H2

I

LuM-SiHR2 II

SiR2 Figure 1. A mechanism for dehydrogenative coupling of silanes, based on a transitionmetal silylene complex as an intermediate. When CP2MMe2 (M = Ti, Zr) complexes are used as catalysts for the dehydrocoupling of a primary silane, complex reaction mixtures are obtained and the

CATALYTIC DEHYDROGENATIVE POLYMERIZATION OF SILANES TO POLYSILANES

5

catalyst precursors are converted to a number of new species. Typically, an induction period is observed during which the CP2MMe2 complexes are converted to the real catalysts. This process is accompanied by loss of the methyl ligands, and some of the metal complexes that were isolated from the reaction mixture suggested that primary silyl derivatives SiH3/SiH2 > SiH2/SiH2 > SiH2/SiH > SiH/SiH. As the reaction proceeds, concentration effects lead to a diminishing importance of the reactions to the left of this series, and an increasing importance of the reactions to the right.

5. Conclusions. From the present study, it is clear that methyl silane can be easily polymerized to poly(methylsilane), using either DMT or DMZ as catalyst. By proper choice of conditions, a polymer completely soluble in common organic solvents can be obtained, but prolonged reaction eventually leads to insoluble gel. As will be described elsewhere, this cross-linking reaction is of some advantage when the polymers are used as precursors for synthesis of silicon carbide.

6.

Acknowledgements.

Financial support for this work from the Natural Sciences and Engineering Research Council of Canada, the Fonds FCAR de Quebec

SYNTHESIS OF POL Y(METHYLSIL YLENE) WITH CP2MMe2

35

and the US Office of Naval Research (contract no. N00014-K-0305) is gratefully acknowledged.

7.

References.

1. Hiyashi,J.; Omori,M.; Yajima,S. U.S.Patent 4,159,259, 1979; Yajima,S.; Hayashi,J.; Omori,M. Chem.Lett. (1975) 931; Yajima,S. Ceram.Bull. (1985) 62, 993; Hasegawa,Y.; Okamura,K. J.Mater.Sci. (1985) 20, 321. 2. Wesson,J.P.; Williams,T.C. J.Polym.Sci., Polym.Chem.Ed. (1981) 19, 65; Kumar,K.; Litt, M.H. J.Polym.Sci., Polym.Chem.Ed. (1988) 26, 25. 3. For a recent review see Miller,A.D.; Michl, J. Chem.Rev. (1989) 89, 1359. 4. Harrod,J.F. Transformation of Organometallics into Common and Exotic Materials. R.M.Laine ed.; NATO ASI Series E: Appl.Sci. no.141; Martinus Nijhoff, Amsterdam, 1988. p.103. 5. Claus,K.; Bestian,K. Justus Liebigs Ann.Chem. (1962) 654, 8. 6. Wailes,P.C.; Weigold,H.; Bell,A.P.; J.Organometal Chem. (1972) 43, C29. 7. Xin,S.; Aitken,C.A.; Harrod,J.F.; Mu,Y.; Samuel,E. Can.J.Chem. (1990) 68, 471. 8.Harrod,J.F.; Yun, S.S. Organometallics, (1987) 6,1381. 9. Walsh, A. Accts.Chem.Res. (1981) 14, 246 10. Kanabus-Kaminska,J.M.; Hawari,J.A.; Griller,D.; Chatgilialoglu, C. J.Am.Chem.Soc. (1987) 109,5267 11. Mu,Y.; Aitken,C.; Cote,B.; Harrod,J.F.; Samuel,E. Can.J.Chem. submitted for publication. 12. W.H.Campbell, T.K.Hilty and L.Yurga, Organometallics, (1989) 8, 2615. 13. Aitken,C.; Harrod,J.F.; Malek,A.; Samuel,E. J.OrganometaI.Chem. (1988) 349, 285. 14. Laine,R.F.; Zhang,Z.-F.; Rahn,J.A.; Mu,Y.; Harrod,J.F.; Babboneau,F. This volume.

THE PREPARATION OF POLYSILYLENES

S. GAUTIIIER and D.J. WORSFOLD National Research Council of Canada Ottawa, Canada, KIA OR6.

ABSTRACT. The common preparation of polydialkylsilylenes is by the reaction of the corresponding dichlorides with sodium in refluxing toluene. The products are complex, appearing to give three types of polysilylene with different molecular weight regions. It has been shown in the past that the reaction appears to follow addition polymerization lines rather than those of a conventional polycondensation. It is found here, by the reaction of chlorine ended chains, that normal condensations of chlorine ended oligomers play little role in the polymerization. A major product of the polymerization is cyclic polysilanes. It is shown that certain cyclic polysilanes may be polymerized, and the yield of these cyclics may be optimized. Introduction. The preparation of soluble polysilylenes by the condensation of dialkyldichlorosilanes with sodium was fIrst described by West et al. (1). The polymer was found to have many interesting properties, and to be of potential commercial interest (2-4). This method of preparation is of general utility, although it has some shortcomings particularly in regard to the variety of polymeric species produced. There have been several studies on the reaction (5-11), with several suggestions to improve the yield. Alternate routes to polysilylenes have also been described (12,13), but these will not be considered here, except for one. This exception is the polymerization of some cyclic polysilanes, which was found to occur in the course of this work. The reaction mechanism of polysilylene synthesis has been the subject of several studies, and although much has been learned of the route of the reaction, many details of the mechanism still remain unknown. The reaction is normally carried out in a comparatively high boiling solvent such as toluene, above the melting point of sodium. Under these conditions, analysis of the reaction product by size exclusion 37

1. F. Harrod and R. M. Laine (eds.). Inorganic and Organometallic Oligomers and Polymers. 37-47. © 1991 Kluwer Academic Publishers.

S. GAUTHIER AND D. J. WORSFOLD

38

chromatography (SEC) shows that three different molecular weight polysilylenes were formed, giving three peaks (1Ia) in the chromatograph. The two higher molecular weight fractions have molecular weights near Ht and 106• The low molecular weight fraction is essentially cyclic material. The SEC peaks frequently overlap because of the broadness of the distribution. One of the higher molecular weight peaks can be eliminated by special conditions, such as the use of lower temperatures and an ultrasonic probe (9,10), or by the addition of phase transfer reagents in catalytic amounts (11 c). Although the polymer forms through the condensation of dichlorosilanes , the reaction does not follow the normal course of a polycondensation reaction. High polymer is formed from n RR'SiCI2 + 2n Na

~

-(SiRR')n- + 2n NaCI .... (i)

the start of the reaction, with the absence of appreciable dimers and trimers. A polymerization stopped at 50% reaction of the dichloride showed only the cyclic material as the low molecular weight fraction (Ila). Moreover high molecular weight polymer (MW 106) forms even with non-stoicometric concentration of the reactants. The polymerization has more the characteristics of a chain reaction with an initiation by the reaction of the initial dichloride with sodium, followed by a stepwise addition of further silyl groups to give the higher molecular weight material. If this is so, the active centres probably have a lifetime comparable in length to the reaction time because the molecular weight of the highest polymer tends to increase during the reaction (lIa). Also it is possible to form blockcopolymers by sequential polymerization of two different dichlorides (lIb). Nevertheless, because of the complexity of the products the possibility that two concurrent mechanisms are operative must be considered. Two of the problems that remain are firstly, whether there is in fact some element of a condensation reaction between shorter polymer chains with terminal chlorines which coexists with a straightforward addition polymerization. Secondly there is the problem of the source of the cyclic material, whether it is formed from chains by cyclization when the chain reaches a suitable length, or if it is formed by an endbiting reaction as found in for example the polymerization of some cyclic oxides. The first of these problems was attacked by forming short polymer chains with terminal chlorines by cleaving high molecular weight polymer with chlorine, and then reacting this. The second problem of the cyclic poly silane is important as it can constitute over half the products. This was studied in more detail for dialkyldichlorosilanes.

THE PREPARATION OF POLYSILYLENES

39

Experimental High molecular weight poly(hexylmethylsilylene), on a 1 gram scale, was reacted with several equivalents of chlorine per chain, in CC~ solution at -20°C under vacuum. The chlorine colour disappeared instantly. The reaction mixture was allowed to come to room temperature, and the solvent pumped off. To remove the residual CCI4 , toluene was condensed in, the product dissolved, and the toluene pumped off and the cycle repeated three times. The product was finally dissolved in toluene and stirred overnight at 110°C under vacuum with freshly distilled sodium. The reaction product was finally terminated with MeLi in diethylether to remove any remaining chlorine, and analyzed by SEC. The polymerizations of hexyl- and propylmethyldichlorosilane were performed as before (llc) in the presence and absence IS-crown-S. The cyclic tetra silane was polymerized by heating under vacuum. The tetracyclosilanes used were isolated from the products of the reaction of the dialkyl-dichlorosilanes with sodium in the presence of IS-crown-S. This was near 35% of the final product, and contained up to 10% of the pentacyclosilane. 29Si NMR spectra were measured on a Bruker spectrometer at 79.5 MHz. Chemical shifts are quoted in ppm from TMS. The SEC measurements were made on a 5 column set, using a polystyrene calibration. Hence molecular weights quoted can only be considered approximate. A smaller set of columns specifically for low molecular weights gave imperfect separation of the cyclic materials from which only qualitative assessments could be made.

1. Chlorination West and others have described the action of chlorine on dodecamethylcyclohexasilane to give dichloropolysilanes (l4). The reaction cleaves SiSi bonds to give a terminal SiCI group. The same reaction appears to occur with high molecular weight polysilylenes. On chlorination the molecular weight of poly(hexylmethylsilylene) was found to decrease (Figure 1). Greater ratios of chlorine gave lower molecular weights. Due to the difficulties of transferring these small amounts of chlorine quantitatively, the final molecular weight was only approximate. This was compounded by the uncertainty in the molecular weight calibration. If the product polysilylene is of sufficiently low molecular weight, a peak appears at 27 ppm in the 29Si NMR spectrum near that expected of a SiSiRR'Cl.

S. GAUTHIER AND D. J. WORSFOLD

40

Chlorinated

Original

MW Figure 1 SEC traces for poly(hexylmethylsilylene) before and after chlorination in CCI4 • This chlorinated polysilylene after reaction with sodium in toluene at 110°C remained unchanged in molecular weight. The SEC trace of the sodium treated chlorinated polymer coincided with that of the original chlorinated polymer. The 29Si NMR spectrum of the solution before it was deactivated was also unchanged. The same chlorination process was also carried out using the corresponding pentacyclosilane as the starting material. The 29Si NMR spectrum showed a strong peak at 27 ppm, together with the peak of the starting cyclosilane. Again treatment with sodium at 110°C in toluene had no effect. However, if the reaction with sodium was done in THF solution, the product was all converted to the cyclosilane.

THE PREPARATION OF POLYSILYLENES

41

2. Preparation of cyclic material The cyclic polysilane formed in the polycondensation of the dichlorides was examined in greater detail. It was claimed previously (lla) that the lowest molecular weight material found from the polymerization was principally cyclic, and primarily pentacyclic. There have been found circumstances, however, when the major component of this material is tetracyclic. When the condensation of hexylmethyldichlorosilane by sodium in toluene solution is done in the presence of the phase transfer reagent 15crown-5 (llc), the products of reaction are only bimodal in the SEC analysis. The two products are polymer, molecular weight 105, and a low molecular weight material, principally cyclic. The 29Si NMR spectra, however, showed that 90% of this cyclic material has a peak at -24.6 ppm. The product from the condensation in THF, which is mainly the cyclic material, has a 29Si peak at -39.1 ppm. Cyclic material produced by the depolymerization of high molecular polymer by potassium in THF has a peak at 39.1 ppm. The -39.1 ppm peak is also the primary peak in the spectrum in the lowest molecular weight fraction isolated from the polymerization in toluene solution in the absence of crown ether, when conditions are such as to give a good yield of the highest molecular weight polymer (lla). However, in the last case the 29Si NMR spectrum of the final product also contains a substantial peak at -24.6 ppm, Figure 2. Examination of the of the cyclic product from the polymerization in the presence of 15-crown-5, after short path very high vacuum distillation, showed that it was mostly (90%) a cyclotetrasilane. It shows a MW of 512 by mass spectrum, and the corresponding cyclic material prepared from propylmethyldichlorosilane, a molecular weight of 344. The latter material also gave a molecular weight of 355 (±15) by vapour phase osmometry. It has been shown in the past that the ratio of the products in the polymerization of hexylmethyldichlorosilane was dependent on the stirring rate. The stirring rate controls the molten sodium particle size and hence the sodium surface area. The effect on the stirring rate on the ratio of tetracyclic to pentacyclic in the products was determined. The results are shown in Table 1. The stirring is only a qualitative assessment because different type stirrers were used to vary the sodium particle size. However, it is evident that although the yield of pentamer was constant, the tetramer yield varied inversely with that of the highest molecular weight material.

42

S. GAUTHIER AND D. J. WORSFOLD

-20

(a)

(b)

(c)

(d)

-20

-40

-40

ppm Figure 2 29Si NMR spectra of lowest molecular weight fraction of product of reaction of hexylmethyldichlorosilane with sodium in toluene. a, slow stirring; b, medium stirring; c, fast stirring; d, in presence of 15-crown-5 (0.25%) Table 1

Polymerization of Hexylmethyldichlorosilane Variation of % yield with stirring

Stirring

High MW

Fast Medium Slow Medium*

19 27 38 53

* In the presence of 15-crown-5

Medium MW 19 18 17 9

Tetramer

Pentamer

42 35 22 34

20 20 23 4

THE PREPARATION OF POLYSILYLENES

43

Propylmethyldichlorosilane condensation is less dependant on the stirring rate. In a similar series the cyclic yield varied little from 23%, the ratio of the two cyclosilanes was 1:1.1 tetramer:pentamer (NMR 29Si -26.0, -40.2 ppm respectively). Moreover, appreciable tetramer was formed by the reaction in THF, 30% of the total cyclic. Also followed by SEC was the ratio of the cyclic products during the course of the polycondensation, although only a qualitative assessment could be made. At the start of the reaction of both dialkyldichlorides, only the pentameric material was formed. It was not until only about 10% of the dichloride remained that the tetracyclic material began to build up. At this point there appeared insufficient monomer to provide the final ratio of tetracyclics. In two reactions cyclic material was added at the start of the condensation, one rich in tetramer (90%) and the other rich in pentamer (70%). In both these reactions, after the reaction of part of the dichloride only pentacyclic material appeared to be present, and the tetracyclic reappeared towards the end of the reaction. However, the final amounts of cyclic material approximated that expected as if the initial cyclic added was present on top of the normal yield of cyclics. 3. Polymerization of tetracyclosilanes The tetracyclosilane could be expected to have a somewhat strained ring, and hence susceptible to polymerization. It was found that this cyclic material from the polymerization of either hexylmethyl- or propylmethyldichlorosilane in the presence of IS-crown-S could be polymerized by heating. Other methods of initiation were attempted at lower temperatures, but with no success (Table 2).

Table 2

Initiators in attempted polymerization of tetrahexyltetramethyl cyclotetrasilane

Initiator Heat Bz peroxide Na UV

TOC >100 60 110 20

Polymer

+

Initiator AIBN BF3.Ether HPtCl6 BuLi/THF

rc

Polymer

60 20 20 0

Whether this was because the methods of initiation were ineffective, or because the temperature was below that at which polymerization is possible is uncertain. It was found that the yield in the thermal polymerization appeared to drop off at lower temperatures (Table 3). This is unusual in a polymerization of a strained ring that could

S. GAUTHIER AND D. J. WORSFOLD

44

be expected to show a ceiling temperature and hence give greater yields at lower temperatures. At the highest temperature the yield did fall, but this was due to the conversion of the tetracyclic poly silane to the pentacyclic. Above 180°C the unpolymerized cyclic was all pentamer, including the ca. 9% in the starting material.

Table 3.

Polymerization of tetracyclo(propylmethylsilane)

Temperature °C 280 240 180 160 140 100

Yield %

*

MW

X

10.6

1.4 1.9 1.8 1.1 1.9 .13

50 92

92 91 87 20

* The starting cyclic material contained up to 9% of the pentacyclic, the yield is based on the initial total cyclics. Diluting the cyclic material with toluene reduced the yield (Table 4).

Table 4. Effect of dilution on polymerization Yield %

Conc. Wt. fract. .314 .181

180 180

78 66

The cyclic materials from condensation reactions on other substituted dichlorides were heated to determine whether they would form polymer. The cyclic material formed in the condensation of dihexyldichlorosilane was found by 29Si NMR to have an absorption at -25.1 ppm and was probably tetracyclic. It would not polymerize by heat, but according to Watanabe et al. (15) the two large substituents would probably make this the most stable ring. The phenylmethyl- cyclic material appears to be only one ring, which is probably pentacyclic CZ9Si NMR -39.7 ppm), and not polymerizable. However, the cyclic compound from the copolymerization of dimethyl- and phenylmethyldichlorosilane CZ9Si NMR, -40 ppm) containing 44 mole% dimethyl gave 19% polymer. From the position of the NMR absorption this cyclic material appears

THE PREPARATION OF POLYSILYLENES

45

pentacyclic but West (16) has shown that the stable ring for dimethylcyclopolysilanes is the six membered ring. Discussion The reactions with chlorine ended polymers suggest that in toluene solution the condensation of these dichlorides with sodium contains no element of a condensation polymerization. Chain growth seems to be restricted to the addition of monomer units to the chain end. From the observation that the 27 ppm peak in the 29Si spectrum remained unchanged at the end of the reaction of chlorine ended polymers with sodium, it must be concluded that sodium does not react rapidly with a chlorine ended chain to form a sodium ended active chain. This was suggested earlier (llc) as part of the polymerization mechanism to account for the low dependence of the rate on the sodium surface area. This low dependence was first shown for the hexylmethyl monomer (lla), but has been confirmed with the propylmethyldichlorosilane. Low order dependences of the rate of monomer consumption have been found in ionic polymerizations, and attributed to the aggregation of ionic chain ends in non-polar solvents (17). The reactive species is then a small concentration of monomeric chain ends in equilibrium with the aggregate. These conditions could well exist here in toluene solution. However, reaction with the sodium surface is an essential part of the reaction. In THF solution, however, the chlorine is removed from the chain end in the reaction of chlorine ended polymer with sodium. Evidently condensation would be possible in this solvent, but a concurrent backbiting reaction to give the cyclic pentamer leaves little polymer. Two major cyclic polysilanes are formed in the reaction of hexylmethyl- or propylmethyldichlorosilane. There is the pentacyclic polysilane, and because it is also formed by degrading the polymer in THF solution with sodium, it must be assumed that it is more thermodynamically stable than the polymer. The tetracyclic material, however, may be polymerized by heat to give the polymer. If this reaction follows the normal route of polymerization of cyclic monomers, there would be initiation of a chain followed by attack of the active chain end on the cyclic monomer in an addition type polymerization. If that were the case the monomer would have to be less thermodynamically stable than the polymer. Under these circumstances when the tetracyclic material is formed in the condensation of dichlorides in toluene, it would be unlikely to have been formed from the polymer in a concurrent endbiting side reaction. However, a variety of common radical and ionic polymerization initiators were The only method of incapable of polymerizing this tetracyclic compound.

S. GAUTHIER AND D. J. WORSFOLD

46

polymerization found was heat, and perhaps it should be considered that the ring is sufficiently unstable that, on heating, it opens and the product acts as the active monomer. The thermal opening of the ring is also in accord with the absence of ceiling temperature effects found in the polymerization of modestly strained rings. If this is the case, taking into account the entropy loss on polymerization, this ring compound itself could still be thermodynamically favoured over polymer, and its formation still be the result of an endbiting reaction in the polymerization of the chlorides. It was found that on heating poly(propylmethylsilylene) to over 3500 in vacuum thermal degradation to distillable products occurred, but the products contained little cyclic material. This is different from dodecamethylcyclohexasilane which under comparable conditions gives large yields of cyclics, principally pentacyclic. Under these conditions at least the propylmethyl cyclics are not particularly stable. The experiments, however, where cyclic material was added at the start of the reaction, suggest that under these conditions the tetracyclic material is either incorporated into the polymer or isomerized to the pentacyclic, and only reforms at the The most reasonable route for isomerization is the end of the reaction. polymerizationldepolymerization route, with the proportions governed more by kinetic effects connected with the ultimate loss of chain end activity. The polymerization of the cyclic material by heat is a good route for the formation of high polymer which is essentially free from impurities except for residual cyclic. With care, the starting cyclic material may be distilled and freed from any salts left from its preparation, if this is from the chloro compounds. The mechanism of the reaction is still uncertain, although the thermal opening of the ring does fit most of the evidence. Presumably the active intermediate is a radical. The molecular weight of the polymer is high, close to that found for the highest polymer fraction in the normal polymerization of dichlorides, and it is possible that the two routes share a common intermediate. References 1a

West, R., David, L.D., Djurovich, P.I., Stearly, KL., Srinivasan, KS.V., Yu, R., J. Amer. Chern. Soc., 103, 7352 (1981)

1b

Mazoyasni, KS., West, R., David, L.D., J. Am. Ceram. Soc., 61, 504 (1978)

2

West, R., Maxha J., ACS. Symp. Ser., 360, 6 (1988)

THE PREPARATION OF POL YSILYLENES

47

3.

Miller, R.D., Rabolt, I.F., Sooriyakumaran, R, Fleming, W., Fickes, G.N., Farmer, B.L., Kuzamy, H., ACS. Symp. Ser., 360, 43 (1988)

4.

Abkowitz, M.A., Stolka, M., Weagley, R.I., McGrane, K.M., Knier, E., Adv. in Chern. Ser., 224, 468 (1990)

5.

Trujillo, R.E., I. Organomet. Chern., 198, C27 (1980)

6.

Wesson, I.P., Williams, T.C., I. Polym. Sci., Polym. Chern. Ed., 17,2833 (1979)

7.

Zeigler, I.M., Polymer Preprints, 27(1), 109 (1986) Zeigler, I.M., Harrah, L.A., Iohnson, A.W., Polymer Preprints, 28 (1), 424 (1987)

8.

Miller, RD., Hofer, D., McKean, D.R, Willson e.G., West, R, Trefonas, P.T., ACS Symp. Ser., 266, 293 (1984)

9.

Matyjaszewski, K., Chen, Y.L., Kim, H.K., ACS. Symp. Ser., 360, 78 (1988)

10. Bianconi, P.A., Schilling, F.C., Weidman, T.W., Macromolecules, 22, 1697 (1989) lla lIb lIe

Worsfold, D.I., ACS. Symp. Ser., 360, 101 (1988) Gauthier, S., Worsfold, D.I., Adv. Chern. Ser., 224, 299 (1990) Gauthier, S., Worsfold, D.I., Macromolecules, 22, 2213 (1989)

12.

Harrod, I.F., ACS. Symp. Ser., 360, 89 (1988)

13.

Nagai, Y., Watanabe, H., Matsumoto, H., Naoi, Y., Sutou, N., Adv. Chern. Ser., 224, 505 (1990)

14.

Wojnowsky, W., Hurt, C.I., West, R., I. Organometal Chern., 124,271 (1977)

15.

Watanabe, H., Muraoka, T., Kageyama, M., Yoshizumi, K., Nagai, Y., Organometallics, 2, 141 (1984)

16.

Brough, L.F., West R, I. Organometal. Chern., 194, 139 (1980)

17.

Bywater, S., Comprehensive Chemical Kinetics, Ed. Bamford C.H., Tipper e.F.H.l1, 1 (1976)

CHARACTERIZATION OF POLYSILANES BY UV, GPC AND LIGHT-SCATTERING.

1. DEVAUX, D. DAOUST, A.-F. DE MAHIEU

Laboratoire des Hauts Polymeres, Universite Catholique de Louvain Place Croix du Sud,l B-1348 LOUVAIN-LA-NEUVE, BELGIUM C. STRAZIELLE

Institut Charles Sadron, Centre de Recherche sur les Macromolecules Rue Boussingault, 6 F-67083 STRASBOURG, FRANCE ABSTRACT. This contribution concerns poly(methylphenyl)silane and, involves three topics: the first one is the calibration of the GPC technique by LS measurements. The second one is the study of the solution properties of that polysilane by LS technique and the last one, the calibration of a UV -Mw relationship. The PSi GPC calibration is close to the PS one despite its behavior in solution can be described using a model based on rigid rod-like elements involving about 40 Si atoms.

1. Introduction Since their discovery in the early 80's (1) soluble polysilanes became the subject of numerous papers in differents fields : synthesis, physical properties, optical properties and applications. Among them very few are devoted to the actual molecular weight determination by cross-analyses. The first objective of this contribution will be to elucidate the problem of true molecular weight determination of an homopolymer: the poly(methylphenyl)silane by the combined use of LS and GPC techniques. As LS analyses are concerned, the second topic aims to report some LS observations leading from viscosity calculation to chain behavior in solution. A third topic will be the molecular weight dependance of the UV absorption maxima wavelengths.

2. Experimental 2.1. SYNTHESIS AND FRACTIONATION Methylphenyldichlorosilane was purchased from ALFA PRODUCTS and fractionally distilled before use. The reaction was performed in brown glassware under argon. In the three 49 1. F. Harrod and R. M. Laine (eds.J. Inorganic and Organometallic Oligomers and Polymers. 49-60. © 1991 Kluwer Academic Publishers.

50

J. DEVAUX ET AL.

necked round-bottomed 500 ml flask titted with a magnetic stirrer, toluene (100 ml) (freshly distilled on Na) was first introduced together with sodium chunks (20% weight excess). When the system reached refluxing temperature, the monomer (0,1632 mole) was carefully introduced from a lateral pressure-equalized flask. The reaction was then continued for 6h. Thereafter, the system was cooled and 10 ml of methanol was introduced to neutralize the excess of sodium. Finally 200 ml of deionized water were introduced in the vessel. The organic phase was recovered by decantation. The crude polymer was then obtained by precipitation into methanol. This crude polymer was redissolved in toluene (10% WI vol). Then the fractionation was performed by addition of small fractions of isopropanol. The fractions were recovered by filtration and dried overnight at 120°C under vacuum. 2.2. GEL PERMEATION CHROMATOGRAPHY The GPC chomatography was constituted of a HPLC 590 WATERS pump, 6 ULTRASTYRAGEL WATERS columns (ld', 105, 104, 10 3, 500 and 100 A), a variable wavelength UV detector (Waters 484) set at 254 nm. Data were processed on a TRIVECTOR TRILAB 200 computer with a horne-made GPC program. The solvent was unstabilized THF at 25 DC. The tlow rate was 1 mV min and the sample concentration 0,1 grll. 2.3. LIGHT-SCATTERING The LS measurements (static method) were performed on a FICA 42000 photogoniometer fitted with a He-Ne laser source of 4 mW at 632 run wavelength, vertically polarized. Before these measurements the samples were centrifuged for 2 h at 14000 T/min. The refractive index increments were separately determined at the same wavelength using a BRICE-PHOENIX refractometer. The dynamic light-scattering measurements were performed on a CHROMATIXKMX-6 (LALLS) detector fitted on a classical GPC device equipped with two SHOW A DENKO mixed bed columns. In that device, the LALLS detector was followed by a refractive index detector (SHIMADZU RID 6A).

3. Results 3.1. SYNTHESES AND FRACTIONATION The Wurtz coupling of methylphenyldichlorosilane by a molten sodium dispersion in toluene obeys the following reaction:

CHARACTERIZATION OFPOLYSILANES BIJ UV, GPC AND LIGHT-SCATTERING

I.

CH3 2 n Na

+

n

51

CH3 110

CI-Si-CI

0



C

Toluene

I

-Si-

0

+

2 n NaCl

n

The resulting polymer is characterized by a multimodal molecular weight distribution (figure 1). In order to obtain narrower molecular weight distribution samples, the fractionation of the crude polymer has been performed, following a procedure described in the experimental part.

3.2. GPC CAUBRATION Five fractions were characterized by three methods : GPC with PS calibration, lightscattering (static method) and GPC with light-scattering detector (dynamic method). The values of weight-average molecular weights obtained by the three methods are reported in table 1. It has to be noticed that the figures in the first column of the table 1 were obtained with the set of six ULTRASTYRAGEL columns while the Dynamic LS values (third column) were obtained by using another GPC device including two SHOWA DENKO columns. Table 1 : Weight-average molecular weights obtained by GPC with PS calibration, light-scattering (static method) and GPC with light-scattering detector (dynamic method).

1

GPC Mw * 7700

Static LS Mw 9530

DynamicLS ~w 8000

2

11500

13800

11250

3

28600

30200

28110

4

137200

110000

113600

5

462700

420000

436500

Reference

* PS equivalents

52

J. DEVAUXET AL.

o

18

Tg) is extremely low, practically approaching zero. The change of activation near Tg is similar to that observed in amorphous charge-transporting polymers and other glasses including selenium and its alloys. Polymers such as poly(di-n-hexylsilylene) (PDHS) which undergo conformational transitions at temperatures TTr due to side chain crystallization and melting, also experience a change in mobility and its activation energy at TTr. The change in the transport behavior and the association with the bathochromic shift in the uv spectra is demonstrated. 1. INTRODUCTION

Linear silicon and germanium backbone polymers, polysilylenes and polygermylenes, with a bonded Si or Ge atoms in the chain, display behavior that resembles more closely 7l" conjugated polymers such as polyacetylene or its derivatives l ; 2,3 rather than analogous a bonded carbon backbone polymers. Spectra ofpolysilylenes show a catenation dependent red shift4 with a saturation value near 300-325 nm as the molecular weight increases 5-7 . The molecular weight dependent red shift indicates a significant delocalization of electrons along the a bonded backbone. When doped with strong oxidants, polysilylenes exhibit high DC conductivity8. In the undoped pristine state, polysilylenes and polygermylenes are insulators. In the presence of an electric 61 J. F. Harrod and R. M. Laine (eds.), Inorganic and Organometallic Oligomers and Polymers, 61-72.

© 1991 Kluwer Academic Publishers.

62

M. STOLKA ET AL.

field, however, these polymers are capable of transporting injected or uv photogenerated holes (electron transport has not been observed). The drift mobilities J1 (velocities per unit electric field) observed in polysilylenes and pollgermylenes are quite high for disordered amorphous organic solids, about 10- cm2N.s at room temperature and an electric field E - 105 V/cm. Hole transport in these polymers was found to be electric field and temperature dependent.9-11 The field and temperature dependence of J1 is similar to that observed in other organic systems where the charge transport mechanism is hopping among discrete sites. The values of activation energies, the first order insensitivity of J1 to molecular weight, and the fact that all-aliphatic polysilylenes and polygermylenes show the same transport efficiency as polymers with pendant aromatic groups, suggest that transport proceeds by hopping among backbone derived states 12 . It is also assumed that the localization is associated with the domain -like suborganization of the chain, presumably all trans segments separated by gauche links or some other regions of disorder13. Since transport involves backbone derived states, it is expected to be sensitive to thermally induced changes of backbone conformation, or changes of the wavefunction overlap among neighboring backbone derived hopping si tes. Of particular interest are phenomena associated with two thermally induced transitions: 1) glass transition, and 2) melting of side group aggregates observed for example in PDHS at 315K14-17 and other poly(nalkylsilylenes)18. In this chapter we describe results ofTOF measurements on aliphatic polysilylenes and poly(di-n-butylgermylene) carried out over a broad range of temperature encompassing the glass transition temperature Tg, and the phase I - phase II (order - disorder) transition caused by side chain ordering and melting, which is accompanied by a blue shift in the uv spectra of the polymers. The effect of main chain crystallization was not studied because of our inability to prepare coherent thin films of highly crystalline polysilylenes such as polydimethylsilylene or polydiethylsilylene. 2. EXPERIMENTAL

2.1 Materials. Polysilylenes (Fig. 1) were synthesized by the Wurtz coupling reaction of dichlorodialkylsilanes with molten sodium by a procedure similar to that described by Zeigler 19 . For example, PDHS was prepared by slow but steady addition (over 25 min) of100 gof30% sodiumdipersion (-0.84 mole Na) in CH3

I

+fii C3 H7 Poly(methyl -n-propylsilylene) (PMPrS)

C4H9

I

+fii C4H9 Poly(di-n-butyl silylene) (PDBS)

C6H13

I

+fii C6H13 Poly(di-n-hexyl silylene) (PDHS)

C4H9

I +yei C4H9 Poly(di-n-butyl germylene) (PDBG)

Fig. 1. Structures of investigated silicon and germanium polymers.

ELECTRONIC TRANSPORT IN ALIPHATIC POLYSIL YLENES

63

toluene to a mixtute of 0.4 mole of dichlorodi-n-hexylsilane in 800ml of a 85 /15 toluene / n-octane mixture at - 115 DC (inverse addition). The total reaction timE' was 4 hrs. The hot reaction slurry was then diluted with 300 ml toluene to reduce the viscosity, and filtered on glass filters to separate the insolubles (sodium chloride, residual sodium and traces of crosslinked polymer) from the polymers in solution. The polymer was isolated by precipitation with isopropanol, washed with water and ethanol and purified by repeated precipitation from hexane into isopropanol. The weight average molecular weight ofPDHS was - 1,1 X 106 and the molecular weight distribution factor was 2.1. Poly(di-n-butylgermylene) (PDBG) was synthesized by a "normal addition" process, by the addition of 0.17 mole (34 ml) ofdichlorodi-n-butylgermane into 80 ml toluene containing 8 g (0.35 mole) N a at - 115 DC with vigorous stirring. After 3 hours of reaction at reflux, the polymer was isolated and purified by a similar technique. Films of all polymers were cast from toluene or toluenecyclohexane mixture.

2.2 Transport Measurements. Drift mobilities were measured on - 10 - 15 p thick films deposited on aluminum substrate by the small-signal current mode Time-Of-Flight (TOF) technique. The current transients were produced by intrinsic photoexcitation using attenuated 337 nm nanosecond pulses from a nitrogen laser. 2.3 Film Thickness Measurements. Thicknesses of films of the polymers were determined by two methods: 1) mechanically, using a high precision micrometer with electronic readout and 2) by measuring film capacitance C on a multi frequency LCR meter. The required dielectric constant was itself calculated by an iterative procedure from an average of multiple measurements of capacitance on free standing films whose thickness had been determined mechanically. 2.4 Phase II to Phase I Conversion. The conversion of the "disordered ,. phase II of film specimens ofPDHS to the more ordered phase I was accomplished by rapid quenching of the melt from 350K to dry ice temperature (-195K). The conversion to the ordered phase I is virtually complete as demonstrated by the absence of the absorption peak corresponding to phase II. Slow cooling of the mel t to temperatures below TTr (315K) or even rapid quenching to ice temperature (273K) always resulted in incomplete and rather irreproducible degree of conversion to phase I as manifested by the presence of the optical absorption associated with the disordered phase II. The same rapid quench procedure was used for other polymers. 3.

RESULTS AND DISCUSSION

3.1 Transport Behavior at T < Tg: At temperatures below the glass transition Tg, the investigated polysilylenes

64

M. STOLKA ET AL.

and PDBG (Fig. 1) exhibit a familiar pattern of the temperature and electric field dependence of mobility, similar to that observed previously on a prototypical silicon backbone polymer, poly(methylphenylsilylene) (PMPS)13,20-23. In all these polysilylenes including poly(methylcyclohexylsilylene)22 etc, drift mobilities were found to be field dependent, following the linear log 11 vs El/2 relationship. The gradient of this dependence decreases with temperature as the temperature approaches Tg from below. The linearity of the log 11 vs lIT plots suggests that mobilities are activated. The activation decreases with increasing electric field and exhibits a square root dependence. 23 The extrapolated Arrhenius plots of mobility parametric in field intersect at a specific temperature To.13 The overall behavior suggests that hole drift mobility in polysilylenes can be represented by an empirical equation first proposed by Gill24 for PVK namely 11 = 110 exp {-(Eo-J3E1!2) / KTeff} (2) where llTeff= lIT-lITo, and Eo is the zero field activation. At T = To the field dependence of mobility vanishes. The parameter To is sometimes associated with the glass transition temperature Tg as in studies of transport in poly(n-vinylcarbazole) and the phenylcarbazole / polycarbonate solid solutions 26 , even though To and Tg are not coincident. The data on PMPS also indicated 13 that extrapolated Arrhenius plots intersect at a To, close to but above the glass transition temperature Tg. The values of activation E observed in aliphatic polysilylenes at T ..... .....

12.0

U

Q)

c:;::::

~

8.0

4.0

0.0

o

5

10

15

20

25

30

35

40

45

Energy (eV) Figure 2: Reflectivity of poly(di-n-hexylsilane), adjusted for second surface reflection in transparent regions and normalized to match the refractive index in the visible.

LINEAR AND NONLINEAR OPTICS IN SUBSTITUTED POL YSILANES

77

detector signals are collected and normalized for the reflectivity of the iridium reference mirror by computer. The resulting reflectance measurements are then compared to transmission measurements, and adjusted to compensate for possible second surface reflections. Finally, the resulting reflectivity spectrum is normalized by a constant to agree with known values of refractive index at visible wavelengths (n,..1.58 at 633 nm) [21]. The reflectivity R generated by this measurement for a supported film is shown in figure 2. The previously observed UV transitions at 3.32 eV and 3.91 eV are clearly present in the spectrum, along with many stronger transitions occurring at higher energies. The theoretical (0-4 dependence of the reflectivity at high energy is well matched by our measured (O-3.8±0.2 dependence, indicating that these data represent a complete 1000 300 200 150

100

80

60

50

40

run

5.0~~~~~~++~~--~-+---~--~~------~--,

Poly-(di-n-hexy lsilane) Dielectric constant

4.0

5

10

15

20

25

30

35

40 run 1000 300 200 150 100 80 60 50 4.0 ~1QBIl~~I+t+------+--~ 3.0

w S ......

2.0

1.0

o.ot-J~--------'-:==::===:::::::::.::::::::::::.-~ -1. 0Or:-.........L.-'-'--!:5,-J-............,-J-l,..1;;0.....L.-l.....L..-'-:I~S............,-J--'--;f.20;;-'-............-'--;26S,-L-'---'--'-;:;3!r;0............,-J-~35

Energy (eV) Figure 3: Real and imaginary parts of the dielectric function for poly(di-nhexylsilane) calculated from the reflectivity data shown in figure 2. The negative values for £2 just below the first UV transition are artifacts of the extrapolations used in the Kramers-Kronig analysis.

78

F. M. SCHELLENBERG ET AL.

1.2

Poly(di-n-hexylsilane) 1.0 ...... ~

(1)

'u S (1)

0

U

0.8

0.6

~

0 ..... ......

u 0.4

.S ...... x

~

0.2

0.0

o

2

4

6

8

10

12

Energy (eV) Figure 4:

Extinction coefficient for poly(di-n-hexylsilane), calculated from the reflectivity measurements shown in figure 2, along with the extinction coefficient for randomly oriented planar-zigzag poly(di-methylsilane), as calculated from the band structure theory ofMintmire [14].

measurement of the joint density of states for the valence and conduction bands. A Kramers-Kronig transform of the reflectance data will therefore yield an accurate calculation of the optical properties of the film. Using routines described elsewhere [22], the phase shift e corresponding to the reflectance amplitude r=R 1/2 was generated, and the optical properties calculated. The real and imaginary parts of the dielectric function El and E2 (related to dispersion and absorption, respectively) calculated in this way are shown in figure 3. The slightly negative values of E2just before the first UV transition are an artifact of the extrapolations used in the Kramers-Kronig analysis. These extrapolation parameters were chosen to give the best numerical agreement with absorption and refractive index data measured independently in the visible and UV. Theoretical predictions of the optical properties of polysilanes have been recently published. From his LDA band structure calculations, Mintmire has calculated the extinction coefficient expected for various orientations of planar-zigzag poly(dimethyl silane) [14]. His results for a randomly oriented film are shown in figure 4, along with the extinction coefficient for poly(di-n-hexylsilane) from our experiment. With the assignment of the one-dimensional direct gap of the band structure calculation to the ~ UV absorption peak at 3.91 eV, the band structure calculations predict both the

LINEAR AND NONLINEAR OPTICS IN SUBSTITUTED POL YSILANES

79

relative energy difference between the three broad features of the measured extinction coefficient and the overalliineshape fairly well. For this figure, Mintmire's calculations have been shifted to higher energy by 0.95 eV, so that the first UV peak in his calculations corresponds to the measured 3.91 eV absorption maximum. Such a shift is common for LDA band structure calculations, which can routinely underestimate gap energies by 20-50% [14,23]. The width of the first UV transition at 3.32 eV is extremely narrow (:=0.4 eV) when compared to transitions at higher energy. This is also much narrower that the direct gap predicted by Mintmire's calculations, and corresponds to that expected for an exciton. Because excitonic properties have been associated with this frrst feature, which disappears above the phase transition associated with the transition from a planar-zigzag to a highly gauche conformation, it is quite reasonable to assign the sharp transition to be a purely excitonic transition associated with the planar-zigzag one-dimensional silicon backbone chain. The sharp peak at 3.96 eV which appears in the high temperature phase of the polymer might then correspond to a similar excitonic transition for the helical phase. Since this helical phase is known to be present to some degree in the polymer sample at room temperature, our observations may contain transitions associated with this phase. However, the correspondence between the experiment and band structure theory with the assignment of the second peak to the direct gap suggests that similarity of Amax for the frrst, sharp high temperature absorption and the broader room temperature feature may be coincidental, and that the relative amplitudes of the absorption peaks in the room temperature spectrum are not indicative of the relative proportion of the phases of the polymer. Further experiments to characterize the electronic structure of the more disordered high temperature phase must be carried out to address this issue. The other feature not well predicted by Mintmire's calculations is the shoulder at higher energy. It should be recognized that the polymer used in these experiments is not planar-zigzag poly(di-methylsilane) but has instead two linear planar-zigzag hexyl groups attached to each Si atom. We therefore expect that the high energy transitions may correspond to the electronic transitions in the hexyl sidechains. The spectrum of the absorption coefficient for poly(di-n-hexylsilane) at high energy, shown in figure 5a, is in fact similar to VUV spectrum measured by others for linear, planar-zigzag carbon backbones [24], as shown in figure 5b. Previous excitonic models have proposed that the direct gap of these polymers lies much higher, around 6-7 eV. The frrst UV absorptions would therefore arise from very tightly bound (i.e. Frenkel) excitons, with binding energies of several eV. However, our assignment of the frrst UV transition to an exciton and the second transition to a 1-D direct gap yields an energy difference on the order of 600 meV. This can be compared with the predictions of the excitonic binding energy for infinite semiconductor quantum wires [25]. In this case, the 'wire' diameter is reduced to :=0.4 nm to correspond to the diameter of the polysilane backbone, and the effective masses mh:= 0.045 and me:= 0.061 are calculated from the curvature of Mintmire's published band diagrams [14]. The result of the calculation is a loosely bound exciton with binding energy of =660 meV, delocalized over =4 nm (corresponding to :=22 Si atoms). This is in good correspondence to that observed with our assignment of the direct gap to the second absorption maximum, and the extent of delocalization corresponds to the extent of delocalization (20-30 Si atoms) deduced for excitations in polysilanes through other means [9,26].

F. M. SCHELLENBERG ET AL.

80 l.4x 106

.:'

Poly(di-n-hexylsilane)

a)

1.2 x 106

1.0x 106

E 8.0K lOS

u -.......-

tI

6.0x lOS 4.0x lOs 2.0x lOs O.Ox 10° 3.50

~

u

c:::

.,;

,J::l

..... 0

,J::l '"

~

-

0.01 .~~

;':'

.:: Q.>

f\.

==

Q.>

;Z

.. .. . . ....

0.003

0.001

I'

....

.......

.. .

.

.... ..1

....

0.OOO3L,.~""""..L.-~""""''''''1~~'''''''''~2-'-'-~'"-:;3-'-'''''''''''''''''4~~~105 10- 1 10 10 10 10

Incident Exposure (J/cm 2) Figure 6:

Birefringence (iln) vs Incident Exposure in (J/cm2) for polarized UV (A) and polarized pulsed laser (.) exposure. The continuous 325 nm UV power density was 0.2 W/cm2 , while the 575 nm pulsed laser intensity (using 8 ns pulses at 10 Hz) was 1.6 GW/cm2.

LINEAR AND NONLINEAR OPTICS IN SUBSTITUTED POL YSILANES

83

initially proportional to exposure (slope ",1 on a log-log plot), but saturates quickly at ",0.03 and begins to decay slowly with continued photoexposure. This corresponds to rapid photodegradation for chains segments aligned with the laser polarization, followed by slower degradation for chains of other orientations excited either directly or by energy transfer. A simple model of birefringence growth, assuming that chains break: in proportion to the projection of the electric field vector of the polarized laser onto the backbone, yields a growth curve approximated by a modified Bessel function of the first kind, 1 .-1n(t)oc 11(t)=i

f" 0

e

t cas(S)

cos(e)de

(2)

This function is shown in figure 7, scaled on both axes for comparison to the birefringence growth induced by two-photon absorption. The lineshape is in general quite similar to that observed for birefringence growth and decay in the two-photon absorption case, and the discrepancy at higher exposures may be accounted for by realizing the film thickness changes somewhat under two-photon exposure (thickness was assumed constant for the theoretical model). However, the rollover and subsequent rapid decay of the birefringence with UV exposure prevents such a simple model from adequately explaining the growth and decay behavior of birefringence for single photon exposure. The observed behavior is consistent with some mechanism of energy transfer for these 0.05

Data

0.04

Model

QJ

= .-= "'" ~

QJ

OIl

~

QJ

.-"'"

0.03

0.02

..

~

0.Q1

0.00

'l.fl

0

100

200

300

400

500

600

700

Absorbed Dose (J/cm2) Figure 7:

Birefringence growth curve from figure 6, along with a scaled plot of the Modified Bessel Function of the first kind 11 (t).

84

F. M. SCHELLENBERG ET AL.

excitations to chains of various orientations, as has been observed in low temperature glasses containing polysilanes [33]. We therefore conclude that rapid energy transfer to chains of other orientations occurs for single-photon excitation, but does not seem to occur directly from the two-photon excited state. At low values of birefringence, ~ is proportional to the number of aligned polymer backbones segments undergoing scission Psc per unit area in time t. This in turn is related to multi-photon absorption by An

oc

Psc

oc

qpabs

oc

q[l-e-~(A) I Z] It "" qf3(A) I zIt

(3)

where q is the scission quantum yield (generally ",,0.01 in the solid state), Pabs is the number of photons absorbed per unit area, I is the laser intensity in W/cm2, 13 is the twophoton absorption coefficient in cm/W, t is the exposure time in sec, and z is the film thickness in cm. The exposure time (t) needed to induce a fixed birefringence (i.e. fixed Pabs) for a given f3(A) is inversely proportional to the square of the exposure intensity for a two photon process. Similarly, for fixed Pabs and intensity, the exposure dose is inversely proportional to f3(A), the nonlinear absorption coefficient. Relative values of 13 calculated from the exposure needed to induce a fixed birefringence can therefore be normalized by comparing the absolute change in UV absorption (also proportional to the number of bonds broken) for both UV and two-photon exposure to infer the number of excitations. This treatment assumes similar scission quantum yields for both two-photon and UV excitation. A spectrum of 13 for poly(di-n-hexylsilane) calculated by the method outlined above is shown on a logarithmic scale as a function of two-photon excitation energy in figure 8. The squares are data, normalized to the previously measured value of XP)=3.2xlO- 1O esu. (corresponding to 13=0.45 cm/MW) at 560 nm. The spectrum shows a sharp peak at 579 nm (corresponding to a two-photon energy of 4.3 eV), on top of a broad background centered at 570 nm. The solid line is a least-squares fit of the sum of two gaussians to the data, with the linewidths of 33 meV and 410 meV for the sharp and broad features, respectively. The width of the broad band is comparable to that observed for single photon absorption, shown as the dotted line in the figure. The sharp spike, however, seems to bear little resemblance to any feature observed in the single photon absorption case. Although it is not unusual for two-photon absorption spectra to show transitions normally forbidden for linear absorptions, the narrow linewidth of the multi-photon transition inducing the birefringence suggests a more localized or site-selective excitation, without inhomogeneous broadening by the distribution of chain segment lengths. Further evidence that the sharp spike may arise from site selectivity comes from closer examination of the linear absorption spectrum for various samples of poly(di-nhexylsilane). Shown in figure 9 is an absorption spectrum from a low molecular weight sample ofpoly(di-n-hexylsilane) (Mw ~O,OOO by GPe). Such spectra, although not well understood at this time, show structure and sharp features which can have widths comparable to those observed for the two-photon absorption spectrum. We have observed similar results for high molecular weight samples after photo scission and quenching at dry ice temperatures [34], and suspect that the narrow features could be associated with highly oriented long chain segments, which can 'untangle' as the molecular weight is reduced, or with the formation low energy conformal traps in the chain. These explanations are,

LINEAR AND NONLINEAR OPTICS IN SUBSTITUTED POL YSILANES

85

Wavelength (nm) One Photon

400

350

300

275

250

Two-Photon

800

700

600

550

500

107

-S ..........

.....

-3



106

~ Q

I:S

-

"-'

c: .Q.,

"Q

...

,: I ,,, ,

til

c: Q Q

~

'

,,

,,

, ,.

.Q

-
-[Me2Si]x[MeHSiCH2]y-

(8)

Yajima discovered that heating polydimethylsilane at 4OO+ o C leads to pyrolytic cleavage of Si-Si bonds and subsequent reaction of the generated radicals, as seen in Scheme I:

POLYMETHYLSILANE AS A PRECURSOR TO HIGH PURITY SILICON CARBIDE

Me

I

Me

I

"""""Si-Si...-v" I I Me Me

Me

I

Me 400°C I - - - - - - ~Si· +

I

.....-vv"S i-CH2 -S i...-v"

I

Me

I

Me

I

·Si...-v"

I

I

Me

H

Me

H I H-C

Me ~.---

I

.....-vv"S i-H

Me

I

Me

131

+

~

/

Siv'V"

Me

Scheme I The end result is the transfonnation of a poly silane moiety into a polycarbosilane moiety. These structural changes are readily evident in the 29Si and 13C spectra of the resulting polymers as discussed below. They can also be detected by IR, because the Si-CH2-Si moiety exhibits a weak but noticeable out-of-plane bend at"" 1355-1360 cm- 1 [6-10,28]. The original-[Me2Si]x[MeHSiCH2]y- copolymers produced at 400°C had Mn "" 10002000 D. However, recent efforts show that the introduction of Lewis acid catalysts can provide copolymers with Mn "" 4000-5000 D [29]. One of the better studied copolymers, PC-470, is produced by pyrolysis at 470°C rather than 400°C. In this instance, -[Me2Silx[MeHSiCH2]y-, has a 1: 1 x:y ratio. The higher heating temperature is directly related to the higher [MeHSiCH2] content. The resultant polymers are spun and then cured in air by heating at 0.2-0.5°/min to temperatures approaching 200°e. The presence of significant quantities of [MeHSiCH2] units is necessary (for the Yajima Process) to obtain effective curing as these units oxidize to fonn [(Me)(CH2)Si-0-Si(Me)(CH2)] crosslinks. The composition of an uncured PC470 fiber is SiC1.88H4.0200.03' After curing, the new composition, SiC1.63H3.3400.61' reflects the considerable oxygen incorporation. Pyrolysis of the cured polymer to 1300°C gives a ceramic yield of"" 80% and a chemical composition of SiC1.46HO.0300.36' The carbon content (assuming a Si:C ratio of 1:1 is desirable) is 50% high. Much of the oxygen is retained as Si02. The XRD of PC-470 heated to 900°C shows slight, very broad patterns reminiscent of I3-SiC. As heating is continued, XRD spectra recorded in 100°C increments show corresponding incremental line narrowing associated with the crystallization of I3-Sie. Crystallization occurs coincident with reaction of the Si02 with the excess carbon to fonn CO and SiO, which volatilize (vide supra). The crystallization and volatilization processes

Z.-F. ZHANG ET AL.

132

are directly associated with a rapid downturn in fiber strength at temperatures> 1100°C.Sl The uncured bulk polymer can also be converted to a ceramic material. On heating PC-470 to 13000C, without curing, the ceramic product obtained has the following composition: SiC1.4oH.osO.04' The 1300°C ceramic yield for the uncured material is

R2NH.BH 2CN + Hz + NaCI

Reaction of the hydrochloride derivative of the PEl fraction of a MW of 1800 liberated 94% of the anticipated hydrogen within a few hours. In contrast with this the 10000 MW fraction produced only 86% of the theoretical amount of H2 . These two materials were obtained as free flowing powders after solvent evaporation which retained the sodium chloride by-product. Attempts to separate the salt by ultrafiltration or dialysis were partially effective because the polymers behaved as ion exchangers and retained chloride ions. An alternate synthetic approach to the cyanoborane derivatives which would not produce salt as a byproduct was explored. In this case, previously formed pyridine.BHzCN was used in a transamination reaction with neutral PEl producing a derivative that showed 59% derivatization. Attempts to derivatize polyallylamine hydrochloride, PAA, by reaction with NaBH3CN failed even in boiling water under reflux. Apparently this marked difference with PEl stems from differences in pKa values of the two hydrochlorides which was measured to be about 5.0 for PEl and 7.0 for PAA. 3.3.

CYANOBORANE REACTION WITH DIMETHYLSULFIDE BORANE

Additional experiments were conducted to increase the boron content of the cyanoborane derivatives of PEl through the reaction of the cyano moiety in the derivative with Me 2S. BH 3 . The reactions proceeded as expected, to the extent of 60% conversion, as established from the chemical analysis and material balance. The IR spectrum of these derivatives showed a significant decrease in the cyano stretching band intensity as compared with the starting material. 3.4.

SILYL DERIVATIVES OF POLYETHYLENEIMINE

Attachment of a silyl moiety to the PEl backbone could afford precursors to silicon nitride and/or silicon carbide. Thus, the derivatization of PEl with hexamethyldisilazane, HMSA, and N,N-Dimethyltrimethylsilylamine, DMSA, were attempted. The transamination of PEl with HMSA in a sealed ampule at 100°C produced silyl incorporation to the extent of 22% of the N. An interesting observation was the formation of ammonia instead of trimethylsilylamine as the reSUlting fragment of the transamination reaction. Evidently, Me3SiNH2 is unstable under the conditions of the reaction and transaminates further to liberate NH 3 . Repeating the reaction in an open system, also at 100°C, with the expectation that ammonia loss would drive the reaction to higher conversion proved to yield the same results. Similarly, silylation with DMSA produced a

CERAMIC MATERIALS VIA DERIVATIZATION REACTIONS OF POLYMERS

213

conversion limited to 20%. Apparently, the derivatization with these bulkier moieties may be limited by steric effects. 3.5.

TITANIUM INCORPORATION INTO PEl

An attempt was made to attach titanium moieties to the PEl backbone through a transamination reaction between PEl and Ti(NMe2)4' An ampule loaded with the reaction mixture produced very little, ::; 5%, dimethylamine as a product being displaced from the titanium alkylamide. On the other hand there was evidence of a reaction since the mixture changed color and didn't undergo hydrolysis upon exposure to air. It seems that the titanium compound reacted by expanding its coordination sphere. This reaction product may have some value but its relatively low titanium content, 15%, detracts from its usefulness as a ceramic precursor. 3.6.

PYROLYSES

The boron containing derivatives were pyrolyzed to 800°C under vacuum. The residues were characterized by chemical and spectroscopic means finding that in all cases boron nitride was produced as an amorphous material containing a variable amount of residual carbon ranging between 18 and 38%. The sodium chloride present in some of the samples sublimed to colder portions of the pyrolysis apparatus. The ceramic yield of the pyrolyses ranged between 30 to 60%. The borane derivatives were also examined as to their ability to form coatings. Solutions of the borane derivatives, about 30% in DMF, were spin coated unto a variety of substrates such as fused quartz, calcium fluoride or silicon wafers. The coatings, approximately 1 ;.tm in thickne,ss were pyrolyzed in stages, increasing the temperature between each treatment by 200°C. The coatings were examined by FTIR after each stage in order to observe the transformation of the precursor into boron nitride. Initially, up to 200°C, residual solvent is eliminated and the BH stretching decreased in intensity as well as being shifted to higher energies. There was also a marked decrease in the intensity of the NH stretching band. Treatment at 400°C eliminated the BH vibrations altogether while treatment at 600°C showed the onset of BN formation by the presence of bands at 1380 and 800 cm- 1 characteristic of this compound. In addition of these, some residual CH was still present at that temperature. Finally, treatment at 800°C under a stream of ammonia produced a very clean and sharp spectrum of boron ni tride. The ammonia treatment was found in a parallel effort dealing with the chemical vapor deposition of boron nitride films [9] to be very effective in elimininating residual carbon in the films. In that particular case the carbon content was reduced from 18 to 2%. The silyl derivatives were not adequate as ceramic precursors since complete volatilization was observed under the conditions of the pyrolysis. The original aim of the present study was to produce a variety of ceramic precursors through derivatization reactions; however, the results were

L.MAYA

214

encouraging only for the preparation of boron nitride precursors. Alternative synthetic approaches to the production of boron nitride precursors, usually through condensation reactions are well represented in the literature; for a recent review see ref. [10]. 4.

Conclusions

The present study demonstrated the feasibility of using preformed polymers as starting materials for the preparation of ceramic precursors. It also showed the limitations of such an approach as illustrated by the fact that derivatization reactions can not be carried to completion, utilizing all the available reaction sites. However, through a judicious choice of starting materials and reaction conditions it might be possible to obtain useful precursors. Acknowledgment Research sponsored by the Division of Materials Sciences, Office of Basic Energy Sciences, U.S. Department of Energy, under contract DE-AC05840R2l400 with Martin Marietta Energy Systems, Inc. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Rice, R.W. (1983) "Ceramic from Polymer Pyrolysis, Opportunities and Needs-A Materials Perspective" Am. Ceram. Soc. Bull., 62, 889-92. Yajima, S. (1980) "Development of Ceramics, Especially Silicon Carbide Fibers, From Organosilicon Polymers by Heat Treatment" Philos. Trans. R. Soc. London, A294, 419-26. Yajima, S., Shishido, T., and Hamano, M. (1977) "SiC and Si3N4 Sintered Bodies with New Borodiphenylsiloxanre Polymers as Binder" Nature (London), 266, 522-24. Maya, L., and Angelini, P. (1990) "Preparation and Use of a Boron Nitride Precursor as Binder for the Densification of Boron Nitride" J. Am. Ceram. Soc., 73, 297-302. Gonsalves, K.E., and Agarwal, R. (1988) "modification of PAN by Titanium Dia1kylamides: Precursors for Titanium Carbide" J. Appl. Po1ym. Sci., 36, 1659-1665. Graybill, B.M., Ruff, J .K., and Hawthorne, M.F. (1961) "A Novel Synthesis of the Triborohydride Anion B3Hs -" J. Amer. Chern. Soc., 83, 2669-70. Wisian-Neilson, P., Das, M.K., and Spielvogel, B.F. (1978) "A General Synthesis of Amine-Cyanoboranes" Inor. Chern. 17, 2327-29. Brown, H.C., Choi, Y.M., and Narasimhan S. (1981) "Improved Procedure for Borane-Dimethy1 Sulfide Reduction of Nitri1es" Synthesis 605. Maya, L., To be published. Paine, R.T. and Naru1a, C.K. (1990) "Synthetic Routes to Boron Nitride" Chern. Rev. 90, 73-91.

OXIDE CERAMIC PRECURSORS

SOME ASPECTS OF THE CHEMISTRY OF TRANSITION METAL OXIDE GELS J. LIV AGE, F. BABONNEAU, C. SANCHEZ Chimie de la Matiere Condensee

Universite Pierre et Marie Curie 4, place Jussieu, 75252 Paris France

ABSTRACT. Sol-gel chemistry is based on the polymerization of molecular precursors such as metal alkoxides M(OR)n. Hydrolysis and condensation of these alkoxides lead to the formation of oxo-polymers which are then transformed into an oxide network. Sol-gel chemistry offers the possibility to design molecular precursors in order to control the polymerization process so that tailor made materials can be obtained. Monodispersed colloids or long chain polymers can be synthesized via the chemical modification of metal alkoxides by nucleophilic reagents. Organic and inorganic polymerization reactions can be performed during the sol-gel synthesis leading to the formation of new materials made of an organic-inorganic hybrid polymeric network.

1. Introduction The sol-gel synthesis of glasses and ceramics received a great amount of scientific and technological interest during the last decade [1]. Sol-gel chemistry is based on the polymerization of molecular precursors. Metal alkoxides M(OR)n where R is an alkyl group appear to be very versatile precursors. They are rather easy to make. Some of them are commercially available. Polymerization can be described as a two steps reaction [2]. i) Initiation is performed via the hydrolysis of alkoxy ligands. It leads to the formation of hydroxylated M-OH groups: M(OR)n + H20 ---> [M(OR)n-l (OH)] + ROH

(1)

ii) Propagation then occurs via the polycondensation of these hydroxylated species giving rise to oxo-polymers. Depending on the experimental conditions olation or oxolation can compete. Olation leads to the formation of OH bridges while coordinated solvent molecules are removed: M-OH + M-XOH ---> M-OH-M + XOH

(X = HorR)

(2)

Oxolation leads to the formation of oxygen bridges and XOH species which have to be removed: M-OH + M-OX ---> M-O-M + XOH

(X =H or R)

(3)

An oxide network is finally obtained upon drying and calcination. One of the main advantage of the sol-gel route is to allow a chemical control of the whole process, from the molecular precursor to the final material. Molecular precursors can be designed at the very early stages of the synthesis so that crystalline phases are 217 J. F. Harrod and R. M. Laine (eds.), Inorganic and Organometallic Oligomers and Polymers, 217-228. © 1991 Kluwer Academic Publishers.

1. LIVAGE ET AL.

218

obtained at low temperature. A new "molecular engineering" is growing up. Titanium oxide based materials for instance can be made from molecular octahedral [Ti06] species. Dimers can then be obtained either via comer or edge sharing octahedra. In the first case a perovskite structure is obtained as in BaTi03 while the second one leads to rutile or anatase phases. The morphology of condensed species can be tailored in order to control the viscosity so that fibres can be drawn or films deposited by such techniques as spindrawing or dip-coating [3]. Metal alkoxides are very prone to nucleophilic reagents. Most chemical reactions involved in the sol-gel process can be described as follows:

(4)

M(OR)n + rnXOH ---> [M(OR)n-m(OX)m] + m ROH

where X stands for H (hydrolysis), M (condensation) or L (complexation), L being an organic or inorganic ligand. These reactions can be divided into three steps [4]. H

'p +

X

H H M-OR --(1)--> ~O - M-OR --(2)--> XO-M - d --(3)--> XO-M-OH + ROH

R

X

1) nucleophilic addition of a negatively charged OX group onto the positively charged metal atom. 2) proton transfer within the transition state toward the leaving group (OR). 3) departure of the positively charged protonated leaving group ROH. The chemical reactivity of metal alkoxides toward hydrolysis and condensation mainly depends on the positive charge of the metal atom oeM) and its ability to increase its coordination number N. For a given oxidation state Z, it increases when going down the periodic table as shown in Table 1. Silicon alkoxides are not very reactive. Gelation occurs within several days after water has been added. Cerium alkoxides are very sensitive to moisture. They must be handle with great care in a dry glove box. Precipitation occurs as soon as some water is present. TABLE 1 partial charge oeM) and maximum coordination number N of some tetravalent metal alkoxides (Z=4). alkoxide Si(OPri)4 Ti(OPri)4 Zr(OPri)4 Ce(OPri)4

oeM)

N

Z

N-Z

+ 0.32 +0.60 +0.64 +0.75

4 6 7 8

+4 +4 +4 +4

0 2 3 4

Sol-gel chemistry of silicon alkoxides is rather simple. Molecular precursors are always monomeric tetrahedral Si(OR)4 species. They are not very reactive so that hydrolysis and condensation reactions are mainly controlled via acid or base catalysis. Long chain

CHEMISTRY OF TRANSITION METAL OXIDE GELS

219

polymers are obtained at low pH while monodispered spherical particles are formed at higher pH [5]. Sol-gel chemistry of transition metal alkoxides is more complicated. Metal atoms can exhibit several coordination states. Oligomeric alkoxides are rather usual. Molecular precursors are highly reactive so that the control of hydrolysis-condensation reactions is usually performed via the chemical modification of alkoxides. This paper addresses the molecular engineering of transition metal alkoxides. It shows how the molecular structure, the functionality and the chemical reactivity of these alkoxides can be controlled via chemical parameters such as the steric hindrance of alkoxy groups, the nature of solvents or the presence of nucleophilic reagents. A wide range of Ti02 based materials for instance can be obtained via the hydrolysis-condensation of titanium iso-propoxide Ti(OPri)4. Monodispersed powders are rapidly precipitated when the alkoxide is dissolved into PriOH while polymeric gels are formed in AmtQH. Gelation rates decrease significantly when acetic acid is added to the alkoxide leading to clear transparent gels. Stable colloidal solutions are obtained in the presence of acetylacetone and ionic ally conducting gels with glycerol [4]. This paper shows that some chemical modification of the alkoxide occurs at a molecular level leading to the formation of new precursors that can be used in order to control the formation of oxo-polymers and oxide materials.

2. Molecular Structure of Transition Metal Alkoxides Transition metals have rather low electronegativities so that their partial charge oeM) in alkoxides is much higher than for silicon. Moreover, their oxidation state Z is often smaller than their usual coordination state N in an oxide network. Therefore coordination expansion is a general tendency for M(ORh alkoxides [6]. Metal atoms tend to increase their coordination number by using their vacant d orbitals to accept electrons from nucleophilic ligands. This usually occurs via solvation or alkoxy bridging. The molecular complexity of metal alkoxides then depends on the solvent and the nature of alkoxy groups. Within a particular group, it increases with the atomic size of the metal atom [7]. Molecular complexities used to be estimated from molecular weight measurements. Nowadays the structure of alkoxides can be studied in solution by NMR or X-ray absorption. This technique was extensively used for titanium alkoxides. The shape and intensity of the prepeak are related to the coordination state of titanium, while EXAFS analysis give information on Ti-O and Ti-Ti distances. As a general rule steric hindrance appears to be the predominant parameter. Primary alkoxy groups lead to the formation of oligomeric species while more bulky secondary or tertiary groups prevent oligomerization. Trimeric species [Ti3(OEt)12] have been evidenced for titanium ethoxide while monomeric Ti(OPri)4 or Ti(OAm t )4 are observed [8]. The chemical reactivity of these alkoxides depends on their molecular structure so that the choice of the precursor is very important. Spherical monodispersed Ti02 powders are obtained via the controlled hydrolysis of dilute solutions of titanium ethoxide while polydispersed powders of irregular shapes are obtained from titanium iso-propoxide [9]. Hydrolysis rates depend on molecular complexity. As a general rule, for a given metal atom, they decrease when complexity increases. The fourfold coordinated monomeric Ti(OPri)4 is much more reactive than the fivefold coordinated [Ti3(OEt)12] trimer. Therefore hydrolysis is fast. It cannot be controlled and condensation occurs as soon as one OPri group is hydrolyzed. Nucleation and growth take place simultaneously so that precipitation leads to polydispersed powders.

220

J. LIVAGE ET AL.

Hydrolysis of trimers is not as fast. Therefore nucleation and growth can be separated. The first nuclei grow rapidly while the following ones are not yet formed leading to monodispersed particles [l 0].

6t::

t

pr;~OPr; OPr

c

b

a

i

EtotEtO

EtO~

Ti

OEt

EtO'

~

~OEt

Ti

EtO

.. OEt

Ti

'OEt

OEt

I

Pr OH

oEt

i

pr~

Pr

~O~

I

Ce

ii

OPr

~OPr

i

I

Ce

Pc'o~; \f,,~ \~i Pr 0

HOPr

Figure 1. Molecular structure of metal alkoxides. 1a. Bulky alkoxy groups prevent oligomerization of Ti(OPri)4 1b. Coordination expansion occurs via alkoxy bridging in [Ti(OEt)413 Ic. Coordination expansion via alkoxy bridging and solvation in [Ce(OPri )4,Pri OH]2 Coordination expansion via solvate formation may occur when the alkoxide is dissolved into the parent alcohol. The stability of such solvates increases with the size and the electropositive character of the metal. Dimeric [Ce(OPri)4,PriOH)z single crystals have been isolated from a solution of cerium iso-propoxide into iso-propanol. X-ray diffraction shows that, in the solid state cerium is octahedrally coordinated. Coordination expansion occurs via both alkoxy bridging and solvation. X-ray absorption and NMR experiments show that such dimeric species are also observed in the solution [11]. Alkoxy bridges are more stable toward hydrolysis than coordinated solvent molecules. It is possible to taylor the kinetics and therefore the resulting structure of oxide particles by an appropriate choice of solvents. Oligomeric [Zr(OPrn)4]n species are formed in non polar solvents such as cyclohexane allowing slow hydrolysis rates and the formation of clear gels. The molecular complexity decreases while solvates are formed in n-propanol. Hydrolysis becomes much faster leading to precipitation [4]. Alcohol interchange reactions are known to occur when alkoxides are dissolved in alcohols other than the parent one. Alcoholysis is rather slow for silicon alkoxides. It is much faster in the case of transition metal alkoxides. 47,49Ti NMR experiments show that alcoholysis of Ti(OPri )4 occurs as soon as the alkoxide is dissolved into tertio-amyl alcohol (Fig.2). Five different species corresponding to x ranging from 0 to 4 are formed as follows:

(5)

221

CHEMISTRY OF TRANSITION METAL OXIDE GELS

Again it is possible to control the fonnation of condensed species by varying the value of x. Precipitation occurs rapidly for x=O. Stable colloids are obtained for x=4. Polymeric gels are fonned for intennediate values of x close to 2.

b

a - 900

-1200

ppm

-600

-900

Figure 2 - Alcoholysis of titanium alkoxides : 47,49Ti NMR spectra of Ti(OPri)4 dissolved into iso-PrOH (2a) and ter-AmOH (2b). 3. Complexation of Metal Alkoxides Transition metal alkoxides are very reactive toward nucleophilic reagents leading to [M(OR)n-m(OX)m] species as shown in eq.(4). Such reactions lead to important modifications of the molecular structure of the alkoxide [7]. A new precursor is fonned the reactivity of which can be completely different. Nucleophilic chemical additives such as carboxylic acids or p-diketones are commonly employed in sol-gel chemistry in order to stabilize highly reactive alkoxides and control the condensation pathway of the evolving oxo-polymer. Two examples will be used in order to illustrate this point. Carboxylates R-COO- are known to behave as bidentate ligands. Acetic acid reacts readily with most metal alkoxides giving complexes in which acetate bridges have been evidenced. A large variety of oxo-polymers is thus fonned depending on the amount of acetic acid. Small oligomers [Ti(OPri)3(OAc)]n (n=2 or 3) are first obtained when acetic acid reacts with titanium iso-propoxide in a one to one ratio [12]. Upon hydrolysis alkoxy groups are removed faster than acetates so that the hydrolysis rate decreases. Moreover the functionality of the precursor becomes smaller than four, leading to the anisotropic growth of oxo-polymers rather than the precipitation of monodispersed oxide particles. Single crystals of hexameric oligomers have been isolated (Fig.3). They do not lead to further condensation but provide structural data on the first condensation steps [13]. Esterification occurs when more acetic acid is added (m> I) so that water is slowly released. Partial hydrolysis and condensation lead to oxo-polymers that still contain organic ligands. Acetate groups then behave both as network formers (bridging) or network modifiers (chelating). The viscosity of the solution increases so that films or fibres can be made. Fibrous oxo-acetate polymers are obtained in the presence of a large excess of acetic acid.

222

J. LIVAGEET AL.

Figure 3. Molecular structure of [Ti6012-0hCIl3-0hCIl2-0Ac)4CIl2-0Pri)6COPri)6] from X-ray diffraction data [13].

Figure 4. Molecular structure of hexameric [Ce6(1l3-0)4(1l3-0H)4(Acac)12] from X-ray diffraction data [11]. Acetylacetone is a well known cheiating ligand. Its enoiic form contains OH groups which react readily with metal alkoxides so that it is often employed as a stabilizing additive in sol-gel chemistry. Strongly chelating acetylacetonate groups behave as network modifiers. They do not form bridges and are not removed easily upon hydrolysis. Therefore oligomeric species are linked either via alkoxy bridging or via the hydrolysis of

CHEMISTRY OF TRANSITION METAL OXIDE GELS

223

alkoxy groupS. As a consequence oligomers are made of a M-OR-M or M-O-M network while acac groups remain outside the core of the molecule. Such strongly bonded acac ligands prevent further condensation so that small colloidal particles are obtained when hydrolysis is performed in the presence of acetylacetone. The size of these particles depends on the amount of chelating modifier. It usually decreases when more acetylacetone is added. Hydrolysis of cerium iso-propoxide with an excess of water leads to the rapid precipitation of cerium hydrous oxides. Only solute molecular species are obtained in the same conditions after acetylacetone has been added ('t=acac/Ce> 1). Colloidal solutions are formed when the amount of acetylacetone ranges between 0.1 [- 0 - Si - 0 - Si - 0 - Si - 0 - ]n I I I CH3 0 CH3

226

J. LIVAGEET AL.

Two sets of peaks can be seen by 29Si MAS-NMR corresponding to D (SiC202) and Q (Si04) species. Almost all Si-OH groups are removed so that no water or solvent molecules remain trapped into the gel once it has been dried under ambient conditions. Monolithic xerogels remain remarkably stable during calcination. Organics are removed around 400°C (under argon), no mass variation is observed below that temperature. Monolithic materials, without cracks, are still obtained at lOOO°C. Si-C bonds are rather covalent so that they are not broken upon hydrolysis. This is no more valid with transition metals for which M-C bonds are more ionic. Complexing organic ligands such as acetates or acetylacetonates have then to be used as shown previously. However such groups behave as network modifiers rather than network formers therefore decreasing the mechanical properties of the resulting material. They are usually removed upon calcination in order to obtain oxide particles of controlled size. Hybrid inorganic-organic networks can also be obtained using polyhydroxylated species such as glycerol. Adding glycerol, CH20H-CHOH-CH20H, to titanium alkoxides gives more or less viscous solutions. It seems likely that some condensation occurs involving OH groups of glycerol leading to polymeric species in which glycerol forms bridges between metal precursors. Hydrolysis then leads to the formation of bridging oxygen atoms. An hybrid material is obtained in which both organic (-Ti-OCH2-CHOHOCH2-Ti-) and inorganic (-Ti-O-Ti-) bridges contribute to the gel network. Hydrolysis can also be performed with aqueous solutions of alcaline salts such as LiCI04. The resulting gel exhibits a high ionic conductivity arising from Li+ diffusion through the gel. Conductivities up to 5.10-4 Q-1cm- 1 have been obtained at room temperature making these gels good candidates as electolytes for micro-ionic devices [22]. Hybrid titanium oxide gels in which organic polymers are part of the gel network have been obtained with vinyl acetylacetone derivatives such as [CH3-CO-CH(CH2-CH=CH2)CO-CH3]. Chelation of Ti(OBu n)4 occurs via the acetylacetonate end of the organic molecule while organic polymerization of the vinyl group is initiated via a radical mechanism. Alkoxy groups are removed upon hydrolysis. This leads to a viscous pale yellow material that turns blue upon UV irradiation. The blue color arises from photoreduced Ti3+ ions. Acrylic acid can also be used as a chelating ligand. This was reported recently for the sol-gel synthesis of high Tc YBa2Cu307-x superconducting ceramics [23]. An aqueous solution of copper and baryum acetates with yttrium nitrate is mixed with an acrylic acid solution in which a radical organic polymerization process is initiated. The viscosity increases as polymerization proceeds so that fibres can be drawn from the solution. The organic binder is burnt upon heating so that the mechanical properties of the fibres are not good. However the superconducting phase is formed around 850°C. It exhibits a very narrow superconducting transition (f1Tc=IK).

5. Conclusion Hydrolysis and condensation of metal alkoxides are commonly used for the sol-gel synthesis of metal oxides. Oxo-polymers are first formed. They then aggregate and give rise to an oxide network after drying and densification. The chemical modification of alkoxide precursors leads to the formation of new molecular species with completely different structures, functionalities and reactivities. The sol-gel route therefore offers good

CHEMISTRY OF TRANSITION METAL OXIDE GELS

227

opportunities for the processing of "tailor-made materials". However the polymerization of inorganic species is far from beeing well understood, even in aqueous solutions. Much work has to be done before a real mastery of the sol-gel chemistry could be reached. Both chemical and physical processes are involved in the formation of solid phases from solutions so that sol-gel science appears to be really an interdisciplinary field of research.

References 1. Brinker C.J. and Scherer G.W. (1990) "Sol-gel science", Academic press, San Diego. 2. Brinker C.J. (1988) "hydrolysis and condensation of silicates: effects on structure" J.of Non-Cryst. Solids 100,31-50. 3. Klein L.C. (1988) "Sol-gel technology", Noyes Publications, Park Ridge, USA. 4. Livage J., Henry M. and Sanchez C. (1988) "sol-gel chemistry of transition metal oxides" Progress in Solid State Chemistry 18,259-342. 5. Klein L.C. (1985) "Sol-gel processing of silicates" Ann. Rev. Mater. Sci. 15,227-48. 6. Hubert-Pfalzgraf L.G. (1987) "Alkoxides as molecular precursors for oxide-based inorganic materials" New Journal of Chern. 11,663-75. 7. Bradley D.e., Mehrotra R.C. and Gaur D.P. (1978) "Metal Alkoxides", academic press, London. 8. Babonneau F., Doeuff S., Leaustic A., Sanchez e., Cartier e. and Verdaguer M. (1988) "XANES and EXAFS study of titanium alkoxides" Inorg. Chern. 27, 3166-72. 9. Barringer E.A. and Bowen H.K. (1985) "High purity monodispersed Ti02 powders by hydrolysis of titanium tetraethoxide" Langmuir 1, 414-20. 10. Livage J., Henry M., Jolivet J.P. and Sanchez C. (1990) "Chemical synthesis of fine powders" MRS Bulletin XV, 18-25. 11. Sanchez e., Toledano P. and Ribot F. (1990) "Molecular structure of metal alkoxide precursors" in Better ceramics through chemistry, MRS symp. (in the press). 12. Sanchez C., Babonneau F., Doeuff S. and Leaustic A. (1988) "Chemical modification of titanium alkoxide precursors" Ultrastructure processing of advanced ceramics 77 -87. 13. Doeuff S., Dromzee, Taulelle F. and Sanchez C. (1989) "Synthesis, solid and liquid state characterization of an hexameric cluster of titanium (IV)" Inorg. Chern. 28, 4439-45. 14. Avnir D., Levy D. and Reisfeld R. (1984) "The nature of silica cage as reflected by spectral changes and enhanced photostability of trapped Rhodamine 6G" J. Phys. Chern. 88, 5956-59. 15. Pouxviel J.C., Parvaneh S., Knobbe E.T. and Dunn B. (1989) "Interactions between organic dyes and sol-gel matrices" Solid State lonics, 32-33, 646-54. 16. Aldebert P., Baffier N., Legendre J.J. and Livage J. (1982) "V20S gels: a versatile host structure for intercalation" Revue de Chimie Minerale, 19,485-95. 17. Bouhaouss A. and Aldebert P. (1983) "Intercalation d'ions alkyl ammonium et d'alkylamines a longues chaines dans les gels de V20S" Mat. res. Bull. 18, 1247-56. 18. Kanatzidis M.G. and Wu e.G. (1989) "Conductive polymer bronzes. Intercalated polyaniline in V20S xerogels" J. Am. Chern. Soc. 111,4139-41. 19. Kanatzidis, Wu C.G., Marcy H.O., DeGroot D.C. and Kannewurf C.R. (1990) "Conductive polymer/oxide bronze nanocomposites.Intercalated poly thiophene in V20S xerogels" Chern. Mater. 2,222-24. 20. Schmidt H., Scholze H. and Kaiser A. (1984) "Principles of hydrolysis and condensation reaction of alkoxysilanes" J; of Non-Cryst. Solids, 1-11. 21. Babonneau F., Thorne K. and Mackenzie J.D. (1989) "Dimethyldiethoxysilane/ tetraethoxysilane copolymers: precursors for the Si-C-O system" Chemistry of Materials 1, 554-8.

228

J. LIVAGE ET AL.

22.Iudeinstein P., Livage J., Zarudiansky A. and Rose R. (1988) "An all gel electrochromic device" Solid State Ionies 28-30, 1722-25. 23. Valente L, Sanchez C., Henry M. and Livage J. (1989) "Synthese sol-gel d'un supraconducteur YBaZCu3Ch-x par polymerisation mixte organique-inorganique" Industrie Ceramique, 836, 193-96.

KINETICS AND STRUCTURE OF SILICATE SOL-GELS'"

Roger A. Assink, C. Jeffrey Brinker and Bruce D. Kay Sandia National Laboratories Albuquerque. New Mexico, USA 87185

ABSlRACT. The structure of a silicate sol-gel derived material depends on the nature of its reaction kinetics. The chemical state of the silicate sol-gel is characterized by both the functional group concentrations and the distribution of the functional groups about a single silicon atom. 29Si nuclear magnetic resonance (NMR) spectroscopy provides a way to quantitatively determine these concentrations and distributions as a function of time during the reaction. During the early stages of the sol-gel reaction, the distribution of hydrolyzed species enables one to calculate the relative rates of hydrolysis. During the intermediate stages of the reaction, the rate of formation of various condensed species enables one to determine the reaction rate constants for both water-producing and alcohol-producing condensation. The chemical bonding of sol-gel derived solid materials can be determined by direct polarization NMR combined with magic angle spinning techniques. These capabilities provide a valuable tool for studying the relationships between the reaction conditions, the chemical kinetics and the resulting structure of the sol-gel derived material.

1. Introduction The chemical synthesis of ceramics and glasses by sol-gel processes has become an area of significant importance in the field of materials science [1-4]. Sol-gel techniques offer the potential advantage of precisely controlling the structure of chemically pure materials. While a variety of approaches have been employed to synthesize sol-gel derived materials, a detailed description of the relationship between the reaction kinetics and final chemical structure has been lacking. We have been exploring that relationship for silicate sol-gels by employing 29Si NMR spectroscopy and theoretical modeling techniques.

2. Chemical Description of a Silicate Sol-Gel 2.1 FUNCTIONAL GROUP CONCENlRATIONS

The simplest description of the sol-gel system considers only the concentrations of the various functional groups (alkoxy, silanol and Si-O-Si bond) without reference to how they are distributed about the individual silicon atoms. The three possible reactions between these functional groups are [5]: *This work performed at Sandia National Laboratories supported by the U. S. Department of Energy under contract number DE-AC04-76POO789. 229 J. F. Harrod and R. M. Laine (eds.).lnorganic and Organometallic Oligomers and Polymers. 229-233.

© 1991 Kluwer Academic Publishers.

230

R. A. ASSINK ET AL.

SiOR + H iJ

~

SiOH + ROH

(1 )

+ H iJ

(2)

SiOR + SiOH ~ Si-O-Si + ROH

(3)

SiOH + SiOH

~Si-O-Si

These reactions are referred to as the hydrolysis, water-fOlTIling condensation and alcoholfonning condensation reaction, respectively. 2.2 FUNCfIONAL GROUP DIS1RIBUTIONS

The next level of detail for the chemical description of a silicate sol-gel specifically treats the distribution of the various functional groups about a single silicon atom. Each silicon atom retains a coordination number of 4 and can bond only to the -OR, -OH and -OSi functional groups. Thus at the nearest functional group level, there are 15 distinguishable local silicon chemical environments. These species are designated by the ordered triplet (XYZ) where the numerical values of X, Y, and Z denote the number of -OR, -OH, and -OSi functional groups attached to the central silicon atom, respectively [6]. The 15 local silicon chemical environments can be represented by the matrix shown in Figure 1. 29Si NMR provides a way to detennine quantitatively the concentrations of most of these species as a function of time during the reaction.

400

301

202

103

XYZ = Si(OR)x (OH) y(OSi) z 004

Figure 1. The 15 possible nearest neighbor functional group distributions in a silicate sol-gel.

231

KINETICS AND STRUCTURE OF SILICATE SOL-GELS

3. Time Regimes Accessible by 29Si NMR 3.1 HYDROLYSIS REACTION KINETICS

The first reactions which take place in the sol-gel solution are the hydrolysis of alkoxide groups as shown by Eq(1). Figure 2 shows the 29Si NMR spectrum of an acid-catalyzed tetramethoxysilane (TMOS) sol-gel which has reacted for 5 minutes. The resonance at -78.5ppm is attributable to the TMOS monomer while the resonances at -76.2 and -74.6ppm are attributable to singly and doubly hydrolyzed monomeric species respectively. According to the formalism of the matrix for nearest neighbor speciation (Figure 1), the peaks correspond to the (400), (310) and (220) species. The hydrolysis reaction has consumed essentially all the available water at this

0 .... • "'.0- 51-0H

o. ~ .

OH

"'.0- $1-01-.

9"'·

"'.0-51-0"'.

Jt' 0....

0...-:-'

I ' -70

-75

iii

iii

-80

-85

, I -90

Figure 2. The 29Si NMR spectrum of an acid-catalyzed tetramethoxysilane (TMOS) sol-gel which has reacted for 5 minutes. point in the reaction. Because mixing the reactants, placing the tube in the spectrometer, and accumulating the spectrum takes several minutes, we cannot directly observe the kinetics of this rapidly occurring hydrolysis reaction. We can, however, deduce the relative rates of the various hydrolysis reactions by numerically modeling the concentrations of the various hydrolyzed species. We find that the hydrolysis reactions of the TMOS sol-gel can be accurately modeled by assuming that the hydrolysis rate of each monomeric species is proportional to the number of methoxy groups in that species [7]. 3.2 CONDENSATION REACTION KINETICS

Once hydrolyzed species are formed, they can condense by the water-forming and alcoholforming condensation reactions shown by Eqs. (2) and (3). The 29Si NMR spectrum of an acidcatalyzed TMOS sol-gel which has reacted for 1 hour is shown in Figure 3. In addition to the monomeric starting material and its hydrolysis products, there are several resonances in the -82 to -86ppm region. These resonances correspond to species in the second row of the functional group distribution matrix. We can readily observe the (301), (211) and (121) species. The (202) species, which represents a silicon bonded to two other silicon atoms by Si-O-Si bonds, resonates near -94 ppm. By observing the rate of formation of these various condensed species as a function of water to silicon ratio of the sol-gel, we are able to calculate the condensation rate constants for both water-producing and alcohol-producing condensation and show how the relative importance of these two reaction pathways depends on the initial water to silicon ratio [5].

232

R. A. ASSINK ET AL.

9H

1.4.0-51-051 01.4..,.

91.4· 1.4.0-51-051 /01.4. 051 I 1.4.0-51-051

OH I

101.4.

U.0-~1-051

I I -70

I

1I

-75

0H)

I I -80 I

1. I

I

I I -85 I

I I -90 I

L

I -95 I

I

Figure 3. The 29Si NMR spectrum of an acid-catalyzed tetramethoxysilane (TMOS) sol-gel which has reacted for 1 hour. 3.3 STRUCTURE OF SOL-GEL DERIVED MATERIAL

The condensation reactions continue until most of the silanol groups have been consumed. Some silanols remain which are unable to react because of steric and/or mobility restrictions. For example, silanols on a silicate surface are not expected to react with other surface silanols if they are immobilized on silicon atoms separated by more than a few Si-O-Si bond lengths. The 29Si NMR spectrum of an ammonia-catalyzed tetraethoxysilane (TEOS) sol-gel reacted for 24 hours is shown in Figure 4 [8]. Magic angle spinning techniques were used for this solid material. The spectrum consists of three broad resonances corresponding to Q2, Q3 and Q4 where the "n" in Qn refers to the number of Si-O-Si bonds surrounding the observed silicon. Thus the three

experimental spectrum

simulated spectrum

simulation components

Figure 4. The 29Si NMR spectrum of an ammonia catalyzed TEOS sol-gel reacted for 24 hours.

KINETICS AND STRUCTURE OF SILICATE SOL-GELS

233

resonances correlate with the bottom three rows of the functional group distribution matrix of Figure 1 (z = 2,3 and 4). Since the resonances are very broad, their positions within each row cannot be determined. The experimental spectrum was simulated using a Nicolet 1280 data system. The composite simulation is shown in the center spectrum and the three Gaussian components of the simulated spectrum are shown on the lower portion of Figure 4. Since the experimental spectrum was accumulated using direct polarization techniques and pulse delay times of 240 seconds (approximately 6 times the spin-lattice relaxation time), the intensities of the three components are directly related to the concentration of each component. Thus, the spectrum is a direct measure of the chemical bonding of this material.

4.

Conclusions

29Si NMR is able to measure not only the functional group concentrations of a sol-gel but also the distribution of functional groups about the observed silicon atoms. Quantitative observation of the various species in the sol-gel liquid during the early and intermediate stages of the reaction permit one to analyze the chemical reaction kinetics of the system in great detail. Direct polarization experiments of the product enable one to measure the chemical bonding of the sol-gel derived material. These capabilities provide a valuable tool for studying the relationships between the reaction conditions, the sol-gel kinetics and the structure of the sol-gel derived material. References [1] Brinker, C. J., Clark, D. E. and Ulrich, D. R. eds., (1984) Better Ceramics Through Chemistry, Elsevier, Amsterdam. [2] Brinker, C. J., Clark, D. E. and Ulrich, D. R. eds., (1986) Better Ceramics Through Chemistry II, Mat. Soc., Boston. [3] Brinker, C. J., Clark, D. E. and Ulrich, D. R. eds., (1988) Better Ceramics Through Chemistry III, Mat. Soc., Boston. [4] Brinker, C. J. and Scherer, G. W., (1990) Sol-Gel Science, Academic Press, San Diego. [5] Assink, R. A. and Kay, B. D. (1988) 'Sol-Gel Kinetics: I. Functional Group Kinetics', J. NonCryst. Solids 99, 359-370. [6] Kay, B. D. and Assink, R. A. (1988) 'Sol-Gel Kinetics: II. Chemical Speciation Modeling', J. NonCryst. Solids 104, 112-122. [7] Assink, R. A. and Kay, B. D. (1988) 'Sol-Gel Kinetics: III. Test of the Statistical Reaction Model', J. Non-Cryst. Solids 107,35-40. [8] Badley, R. D., Ford, W. T., McEnroe, F. J., Assink, R. A., (1990) 'Surface Modification of Colloidal Silica', Langmuir 6, 792-80l.

MECHANISTIC ASPECTS OF THE PYROLYTIC TRANSFORMATION OF METAL ALKOXIDES TO OXIDES

AYUSMAN SEN*, MANISH NANDI, HILMAR A. STECHER AND DOUG RHUBRIGHT Department of Chemistry, The Pennsylvania State University University Park, Pennsylvania 16802 U.S.A. ABSTRACT: The pyrolysis of a series cerium(llI), lithium(I) and titanium(IV) alkoxides was studied. Depending on the alkoxide, the pyrolysis was carried out in the solid state, in solution or in the gas phase. The thermolysis of Ce(OCtBu3h, 1, in the solid state and in solution led to the formation of [Ce(OCHtBu2hh. 3, together with isobutylene and a small amount of lBU2CO. An analogous reaction was observed for [LiOCtBu3]n, 2. Mechanistic studies indicate that the decomposition was initiated by a homolytic cleavage of the MOC(tBu2)-CMe3 bond 0 ( = Ce, Li) which was followed by H-abstraction to yield isobutylene and, principally, the MOCHtBu2 fragment. The structure of 3, as determined by X-ray diffraction, revealed that it was a dimer with a pseudotetrahedral geometry about each Ce atom. The titanium alkoxides, Ti(ORMR = CH2CH3, 4; CH(CH3h. 5; C(CH3h, 6; CH2C(CH3h, 7; CH2CHCH2CH2 ' 8) were subjected to flash vacuum pyrolysis at 550°C and 700°C. The formation of substantial amounts of allylcarbinyl derivatives from 8 indicated at least the transient formation of radicals. However, free radicals do not appear to be involved as intermediates in the product formation since pyrolysis of 7 did not yield free radical derived products. The principal decomposition pathway appears to involve the nucleophilic attack by an incipient alkoxide ion on either the a-carbon of a neighboring alkyl group (to form an ether) or the corresponding ~-hydrogen (to generate an alcohol and an olefin). The contribution from the former pathway decreases dramatically with increasing substitution at the a-carbon. In 7, the attack, however, occurs at the y-hydrogen. Significant quantities of carbonyl compounds formed by a ~-hydrogen abstraction step were also observed for 4 and 5. Homoleptic metal alkoxides are of considerable inrerest as precursors to metal oxides. Both pyrolytic and hydrolytic methods have been .:mployed for the conversion to the corresponding oxides [1]. As an exampk of the former procedure, oxide films have been created by the exposure of metal alkoxide vapors to hot surfaces [2]. Little, however, is known concerning the mechanistic steps involved in the thermal decomposition to metal alkoxides [3]. This contrasts with the extensive mechanistic information that is available for the decomposition of the corresponding metal alkyl species [4]. Herein, we report the results of our study of the pyrolysis of a series of cerium(llI), lithium(I) and titanium(IV) alkoxides [5]. Depending on the 235 J. F. Harrod and R. M. Laine (eds.), Inorganic and Organometallic Oligomers and Polymers, 235-246.

© 1991 Kluwer Academic Publishers.

236

A. SEN ET AL.

alkoxide, the pyrolysis was carried out in the solid state, in solution or in the gas phase. The results provide the first understanding of how the decomposition mode varies with the nature of the alkoxy ligand. In addition, it provides a glimpse of the nucleation process that occurs during the formation of a metal oxide from the corresponding alkoxide through the loss of hydrocarbon fragments.

Cerium(llI) and Lithium(I) AIkoxides. In this section, we discuss the thermal decomposition of Ce(OClJ3u3h, 1, and the related [LiOCtBu3]n, 2, to the corresponding M(OCHtBu2)n derivatives (M = Ce, Li) [6]. Ce(OCtBu3h, 1, was synthesized by the reaction of Ce[N(SiMe3hh with HOCtBu3 in pentane, and was isolated as an 0T and H 20. ';ensitive yellow solid. 1 is monomeric in solution and, although its crystal structure has not been determined, it is safe to assume that it is also monomeric in the solid state. We have previously shown that the compound, Ce(2,6-lBu2-C6H30h, is monomeric in the solid state [7] and models indicate that OCtBu3 is significantly more bulky than 2,6-tBu2-C6H30. The lithium analog, 2, was obtained from tBu3COH and nBuLi in alkane solvent. It was assumed to be oligomeric based on the fact that the THF complex, [(tBu3CO)Li(I!-THF)h, was shown to be a dimer in the solid state. [6b] Thermolysis of solid 1 at 150°C under vacuum resulted in its decomposition to [Ce(OC~tB1!2h~2' 3 [8], and i~obutylene~ In addit~on, a small quantity of tBu2CO was detected mdicatmg the formatIOn of a Ce(IIl)-hydride. These results are summarized in eq. 1

(1 )

Hydrolr.sis of a C6D6 solution of the solid residue left after decomposition gave HOCH BU2 as the only organic product (>99%). The absence of structurally characterized dimeric lanthanide alkoxides [9] prompted us to determine the crystal-structure of 3 by single-crystal X-ray diffraction [10]. As Figure 1 indicates, 3 is a dimer with a pseudotetrahedral geometry ab~ut . each Ce atom. A crystallographic inversion center constrains the Ce and the bndgmg atoms to a single plane and a dihedral angle of 84.1 ° relates this plane to the O( terminal)-Ce-O( terminal) plane. Scheme 1 outlines the plausible mechanistic pathways for the thermolysis, and encompasses three rate-limiting de~omposition modes ~steps A, ~ ~d C). The H-migration pathway (step B) was Judged to make a mmor contnbutIOn at best, based on a kH/kD value of 1.5 observed for isobutylene in the decomposition of Ce[OC(tBu)(lBu-d9hh [11]. The corresponding.kH/kD value for .. . {Li[OC(tBu)(lBu-d9)2] In' 2, which decomposed ill an analogous fashIOn m the solId state, was 1.8 [11].

°

PYROLYTIC TRANSFORMATION OF METAL ALKOXIDES TO OXIDES

237

It is harder to differentiate between the ~-a1kyl migration and homolytic cleavage pathways (steps A and C, respectively). The C-C bonds between the quarternary carbons of the OCtBu3ligand in metal complexes [6,12], as well as other tri-t-butylmethyl systems [13], are relatively long (1.60-166A) and, therefore, are relatively weak [14], but this would favor both steps A and C. However, the following experimental results, are more consistent with homolysis of the C-C bond (step C). Thermolysis of 1, 2 and tBu3COH were carried out in either benzene or toluene. While no detailed kinetic analysis was performed, the following are the approximate half-life (t J) values obtained [15]: [LiOClJ3u3]n, 80 min at 95°C; Ce(OClJ3u3h, 60 h at 9::>°C; HOClJ3u3' 150 h at 140°C. Since free-radicals are electron-defficient species, increased negative charge on the neighboring oxygen atom which can be donated to the carbon participating in bond homolysis should facilitate this process. The ionic nature of the OClBu3 group would be expected to increase on going from the alcohol to the Ce compound to the Li compound and, hence, the trend in the homolysis rate would appear to be consistent with step C (although not necessarily inconsistent with step A). The addition of slightly less than one equivalent of 12-crown-4 (which is known to complex strongly with Li+ [16]) to 2 prior to thermolysis resulted in a reduction of the half-life to under 1 min at 95°C [17]. This is consistent with step C since the complexation ofLi+ by 12-crown-4 would lead to increased charge separation and, therefore, more negative charge on the oxygen of OClBu3. However, it is inconsistent with step A since distancing the metal from the OClBu3 moiety should lead to a decrease in ~-alkyl migration rate. The observation is also inconsistent with step B for a similar reason. An experiment was performed to trap the lBu. radical formed from the OClBu3 ligand during thermolysis. Earlier IH-NMR titrations had shown that Ph 2CO forms a 1: 1 complex with the Ce center in 1 [18]. When a benzenf', solution of 1 was thermolyzed in the presence of 11 equiv. of Ph2CO, only a small fraction of the expected isobutylene was observed. When the reaction mixture was hydrolyzed, the nonvolatile organic products observed were: Ph2lBuCOH, Ph2CHOH, tBu2CO, along with excess Ph 2CO and HOClBu3. Significantly, no HOCHIJ3u2 was found. The fIrst product was clearly formed by trapping of the tBu. radical by coordinated Ph 2CO. The absence of HOCHlBu2 indicated that (a) the above trapping process was faster than H-abstraction by the Ce-OClBu2. fragment and/or, (b) the larger lBu2CO ligand was rapidly displaced by the smaller Ph2CO from the Ce(H)(tBu2CO) intermediate and this resulted in the eventual formation of the CeOCHPh2 moiety rather than the CeOCHlBu2 moiety.

Titanium(IV) Alkoxides. In this section, we describe the gas phase thermal decomposition of a series of titanium alkoxides, Ti(OR)4(R = CH2CH3, 4; CH(CH3h. 5; C(CH3h; 6; CH2C(CH3h, 7; CH2CH2tHCHiH2 , 8). While the lower titanium alkoxides are known to associate in solution [3c], all are presumably monomeric ill the vapor phase like the corresponding zirconium analogs [19]. . The decomposition of Ti(OR)4 in the vapor phase was performed using flash vacuum pyrolysis technique [20] at a pressure of approx. 10-2 rum of Hg. Pyrolysis at 550°C and 700°C were carried out in glass and quartz tubes, respectively. The inorganic product obtained was amorphous Ti02 containing small amounts of carbon

238

A.SENETAL.

and hydrogen (typical analysis: C, 4.0%, H, = M-O \.:\

• .tBu

H~

, ,JC~IBu

H2C- C-l Me

I O-C··-'8u - ~IBu

H-migration

-)=

8

,

M

•.tBIi

O=C~tBu

H

Me Insertion

->= M-O \ • .lBu ~~IBU H

H~C-C··M ~

e

Me

Homolytic cleavage

H-abstraction by C

c

->=

M-H

243

PYROLYTIC TRANSFORMATION OF METAL ALKOXIDES TO OXIDES

Table 1. Products Formed Upon Flash Vacuum Pyrolysis of Ti(OR)4 COMPOUND

PRODUCTS a

TEMP.

Ti(OEt)4

C2H4 CHaCHO (10%) ; CH aCH 20H (58%) ; (CHaCH2l20 (32%) 700°C

C2H4 CHaCHO (33%) ; CH aCH 20H (38%); (CH aCH 2hO(29%)

Ti(O-iPr)4

CH30i=CH 2 CHaCOCHa (11%) ; (CHa)2CHOH (87%) ; [(CHa)2CH120 (2%) 700°C

Ti(O-tBu)4

CHaOi=CHl! ; (CHahCHOH (100%) (CH ahC=CH2

; (CHalaCOH (100%)

CHaCH 2C(CHal=CH 2 ; (CHahC=CHCHa (CHalaCHO (trace) ; (CH alaCCH2OH (100%)

~+~ ~CHO (trace) ~C~Oi (23%) ; [>-CH20CHz-CH20CH~

(18%)

a Aldehyde + Alcohol + Ether", 100%. Because of high volatility. the yield of oletins could not be precisely determined

A. SENET AL.

244

References 1.

2.

3.

4.

5. 6.

7. 8.

9.

10.

Recent reviews: (a) Brinker, C. J.; Scherer, G. W. Sol-Gel Science; Academic: San Diego, 1990. (b) Bradley, D. C. Chem. Rev. 1989,89, 1317. (c) Brinker, C. J.; Clark, D. E.; Ulrich, D. R Better Ceramics Through Chemistry II. Mater. Res. Soc. Symp. Proc. 1986, 73 (d) Science of Ceramic Chemical Processing; Hench, L. L.; Ulrich, D. REds.; Wiley: New York, 1986. (e) Ultrastructure Processing of Ceramics, Glasses and Composites; Hench, L. L.; Ulrich, D. R, Eds.; Wiley: New York, 1984 (f) Mazdiyasni, K. S. Ceram. Int. 1982,8,42. Recent references: (a) Jeffries, P. M.; Girolami, G. S. Chem. Mater. 1989, I, 8. (b) Takahashi, Y.; Kanamori, M.; Hashimoto, H.; Moritani, Y.; Masuda, Y. J. Mater. Sci. 1989,24, 192. (c) Dhanavantri, C.; Karekar, R. N. Thin Solid Films 1989,169,271. (d) Niwi, M.; Kawashima, Y.; Hibino, T.; Murakami, Y. J. Chem. Soc., Faraday Trans 1,1988,84,4327. (e) Yamane, H.; Kurosawa, H.; Hirai, T. Chem. Lett. 1988, 1515. (f) Yamane, H.; Kurosawa, H.; Hirai, T.; Iwasaki, H.; Kobayashi, N.; Muto, Y. Jpn. J. Appl. Phys., 21988,27, L1495. (g) Masanobu, Y. Jpn. Kokai Tokkyo Koho JP 63 89,667 (1988). CA: 10924220215h. (h) Hokari, Y. Jpn. Kokai Tokkyo Koho JP 6372,883 (1988). CA: 10916139748n. (i) Matsuyama, I. Jpn. Kokai Tokkyo Koho JP 61 35,847 (1986) CA: 10512106232d. Specific references on vapor deposition of Ti0 2: (j) Siefering K. L.; Griffin, G. L. J. Electrochem. Soc. 1990,137,814. (k) Takahashi, Y.; Suzuki, H.; Nasu, M. J. Chem. Soc., Faraday Trans 1,1985,81, 3117. (1) Komiyama, H.; Kanai, T.; Inoue, H. Chem. Lett. 1984, 1283. (a) Reference 1b,f. (b) Bryndza, H. E.; Tam, W. Chem. Rev. 1988,88, 1163. (c) Bradley, D. C.; Mehrotra, R. C.; Gaur, D. P. Metal Alkoxides; Academic; New York, 1978. (d) Bradley, D. C.; Faktor, M. M. Trans. Faraday Soc. 1959, 55,2117. Recent rviews: (a) Crabtree, R. H. The Organometallic Chemistry of the Transition Metals; Wiley: New York, 1988; p. 39. (b) Collman, J. P.; Hegedus, L. S.; Norton, J. R; Finke, R. G. Principles and Applications of Organotransition Metal Chemistry; University Science Books: Mill Valley, California, 1987; p. 94. Previous publications: (a) Nandi, M.; Rhubright, D.; Sen, A Inorg. Chem. 1990, 29,0000. (b) Stecher, H. A; Sen, A; Rheingold, A L. Inorg. Chem. 1989,28, 3280. While a similar reaction was reported for a bimetallic system, the mechanism of decomposition was not examined, see: (a) Hvoslef, J.; Hope, H.; Murray, B. D.; Power, P. P. J. Chem. Soc., Chem. Commun. 1983, 1438. (b) Murray, B. D.; Hope, H.; Power, P. P. J. Am. Chem. Soc. 1985,107,169. Stecher, H. A; Sen, A; Rheingold, A L. Inorg. Chem. 1987,27, 1130. IH-NMR(C 6D6)(25°C)(ppm): 31.5 (br, terminal OCHlJ3u2); 7.0 (br, terminal OCHlJ3u2); -18.0 (br, bridging OCHtBu2)' The resonance due to bridging OCHtBu2 could not be located. The broadness of the spectunn was due to the paramagnetic Ce(III) ion. This compound was also independently synthesized by the reaction of Ce[N(SiMe3hh with 3 equiv. of HOCHtBu2 in pentane. Structurally characterized monomeric lanthanide complexes: (a) Reference 7. (b) Hitchcock, P. B.; Lappert, M. F.; Singh, A J. Chem. Soc., Chem. Commun. 1983, 1499. (b) Hitchcock, P. B.; Lappert, M. F.; Smith R G. Inorg. Chim. Acta 1987, 139, 183. Crystal data for 3, CS4 Hl1406Ce.z (obtained by recrystallization from toluene): triclinic, £1 (bar), ~ = 11.611(7) A, Q= 12.497(6)A, ~ = 12.611(7) A,!:! =

PYROLYTIC TRANSFORMATION OF METAL ALKOXIDES TO OXIDES

11.

12.

13.

14.

15.

16. 17.

18. 19. 20. 21.

63.78(4)°, ft = 70.69(4)°, '1 = 79.36(4)°, V = 1542(1) A3! Z = 1, g = 15.03 cm- I , D(calc) = 1.22 g cm-3, MoKa (A. = 0.71073 A), 23°C, NIcolet R3rn11l dIffractometer. Of 5679 reflections collected (4° ~ 29 ~ 50°),5409 were unique (R(int) = 2.24%) and 4650 were observed CEo ~ 5a(Eo». An empirical absorption correction was applied (Imax = 0.744, Imin = 0.654). A Patterson map located the Ce atom, nonhydrogen atoms anisotropic, H atoms isotropic (calculated, gC-H = 0.960 A): RF = 2.87%yRwF = 3.12%, data/parameter = 16.6, GOF = 0.913, highest peak = 0.81 e/AT (Ce noise). All software: SHELXTL(5.1), G. Sheldrick, Nicolet XRD, Madison, Wi. The decomposition was carried out under vacuum at 150°C for 3h. Kinetic isotope effects were calculated form intensity of do vs. d s isobutylene (rnle = 56 and 64) and d 9 vs. dIS tBu2CO (rnle = 151 and 160), correcting for the 2:1 statistical preference for the heavier group. For 1, isobutylene gave 1.50 and tBu2CO gave 1.23; for 2, isobuty1ene gave 1.84 and tBu2CO gave 1.49. At present, we are unable to explain why the ketone-derived value is lower; we believe it suggests that the formation of tBu2CO occurs after isobutylene formation with its own independent rate limiting step and kH/kD' (a) Lubben, T. V.; Wolczanski, P. T. 1. Am. Chern. Soc. 1987, 109, 424. (b) Lubben, T. V.; Wolczanski, P. T.; Van Duyne, G. D. Organornetallics 1984,3, 977. (c) Murray, B. D.; Power, P. P. 1. Am. Chern. Soc. 1984, 106,7011. (d) Olmstead, M. M.; Power, P. P.; Sigel, G. Inorg. Chern. 1986,25,1027. (e) Stecher, H. A; Sen, A; Rheingold, A. L. Manuscript in preparation. (a) Cheng, P-T.; Nyburg, S. C.; Thankachan, C.; Tidwell, T. T. Angew. Chern., Int. Ed. Eng. 1977,16,654. (b) Cheng, P.-T.; Nyburg, S. C. Acta Crystollogr. B. 1978, B34, 3001. (c) Hagler, A T.; Stern, P. S.; Lifson, S.; Ariel, S. 1. Am. Chern. Soc. 1979,101,813. (d) Ruchardt, C.; Weiner, S. Tetrahedron Lett. 1979, 1311. (e) Wong-Ng, W.; Chen, P.-T.; Nyburg, S. C. Acta Crystollogr. C. 1984, C40, 92. Previous experimental and theoretical results on the thermolysis of bulky hydrocarbons indicated that the activation energy for C-C bond cleavage was inversely proportional to the ground state strain enthalpy which in turn directly correlated with the C-C bond length, see: Ruchardt, C.; Beckhaus, H.-D. Angew. Chern., Int. Ed. Engl. 1980,19,429. The decomposition of 2 and tBu3COH were followed by IH-NMR spectroscopy. The decomposition of 1 in solution was not clean as in the solid state, perhaps due to further reactions of the unstable intermediates. Hence, the reaction was followed by hydrolyzing aliquots removed from the reaction mixture and monitoring the amount of tBu3COH present by GC. Pacey, G. E. In Lithium, Current Applications in SC:.ence, Medicine, and Technology; Bach, R. 0., Ed.; Wiley: New York, 1985; p. 35. In the presence of 12-crown-4, isobutane rather than isobutylene was the predominant product and, in addition, a significant amount of tBu2CO was formed. We attribute the change in product distribution to 12-crown-4 acting as a good hydrogen donor for the tBu. radical. In support of this assumption, we observed extensive degradation of the crown ether in the course of the reaction. For a description of a similar titration,· see reference 7. Bradley, D.C.; Mehrotra, R. C.; Wardlaw, W. 1. Chern. Soc. 1952,4204. Brown, R. F. C. Pyrolytic Methods in Organic Chemistry; Academic: New York,1980. Kochi, J. K. In Free Radicals; Kochi, J. K., Ed.; Wiley: New York, 1973; p. 683.

245

246

22. 23. 24. 25.

A. SEN ET AL.

The cycloproplcarbinyl radical undergoes fast ring-opening to the allylcarbinyl radical, see: Griller, D.; Ingold, K. U. Acc. Chern. Res. 1980, 13, 317. Lowry, T. H.; Richardson, K. S. Mechanism and Theory in Organic Chemistry; Harper & Row: New York, 1987; p. 454. Reference 20, p. 55. Pines, H.; Manassen, J. Adv. Catal. 1966, 16,49.

NOVEL ORGANOMETALLIC POLYMERS

POLY(ALKYLENE PHOSPHATES): SYNTHETIC STRATEGIES

Gary M. Gray*, Keith E. Branham, Lung-Hua Ho and Jimmy w. Mays Department of Chemistry University of Alabama at Birmingham 219-2 UAB Station Birmingham, AL 35294 Prakash C. Bharara, Andreas Hajipetrou and James B. Beal Department of Chemistry University of Montevallo Montevallo, AL 35115

ABSTRACT. Poly(alkylene phosphate)s are an interesting class of polymers because both the polymer chain and the polymer substituents can be readily varied. Several synthetic methods are available for the preparation of these polymers. The most versatile involve poly(alkylene phosphonate)s as intermediates. Two of the most promising methods for the preparation of poly(alkylene phosphonate)s are the ring opening polymerizations of cyclic phosphonates and the condensation polymerizations of dimethyl phosphite and dialcohols. The ring opening polymerizations of the cyclic phosphonates catalyzed by triisobutylaluminum are very sentitive to the nature of the alkylene group. With R = -(CH2)3-' the polymerization gives a monomer:polymer mole ratio of 1:1. The polymer can be readily separated from the monomer using fractional precipitation, and the polymer only very slowly reequilibrates to form monomer. In contrast, with R = -(CH2)2- the polymerization occurs without catalyst to give a monomer: polymer mole ratio of 1:9. The polymer is of very low molecular weight and cannot be separated from the monomer by fractional crystallization due to rapid reequilibration. with R = -(CH2CH(CH20Me))-, no polymerization occurs in the absence of a catalyst, and the polymerization gives a monomer:polymer mole ratio of 1:1. Attempts to purify the polymer by fractional crystallization gave poor yields of the polymer. Reaction of the - (CH2) 2- or -(CH2CH(CH20Me))- monomer-polymer mixtures with chlorine and then with either imidazole and methanol or with excess diethylamine gives poly(alkylene phosphate)s that are readily separated from the monomeric impurities by fractional precipitation. The condensation polymerizations of dimethyl phosphite with triethylene glycol and l,12-dodecanedi01 have been carried out and followed by 1H, 13c and 31p NMR spectroscopy. These studies indicate that, under the reaction conditions used in these polymerizations, the polymers form at a much lower temperatures than are reported in the literature, and that the first step in the polymerization appears to be displacement of one of the methoxy groups from the phosphite by one mole of the diol. Attempts to generate higher molecular weight polymers by heating the reaction mixture to 180 - 200 °c for long periods of time caused decomposition of the polymers. 249 J. F. Harrod and R. M. Laine (eds.), Inorganic and Organometallic Oligomers and Polymers, 249-262. © 1991 Kluwer Academic Publishers.

G. M.GRAYET AL.

250

INTRODUCTION

High molecular weight, linear polymers with inorganic elements in the polymer chain exhibit properties that are quite different from the better-known organic polymers [1]. These properties include antiflammability, antioxidant ability, biocompatibility and flexibility. However, only a few classes of these inorganic polymers have been studied in detail. One class of these polymers that are only beginning to be studied are the poly(alkylene phosphate)s, shown in Figure 1 [2-13]. These polymers are of interest because they are one of the few classes of polymers in which both the polymer chain and the polymer substituents can be readliy varied. This suggests that these materials may have a variety of applications in such diverse areas as the controlled release of bioactive materials and electronics. Figure 1.

Poly(alkylene phosphate)s

°t -(P-O-R-O) n ZR'

I

R =

-( CH 2)1-' -(CH2CH20)m-' -(CH2CH(CH20Rn»-;

R'=

H, alkyl; R n = H, Ac;

Z =

NR' , 0; 1 = m = 2, 3

Several methods have been reported for the preparations of these polymers. The most versatile involves poly(alkylene phosphonate)s as intermediates. The polymers can be prepared by a variety of methods. Two of the most promising appear to be the ring opening polymerizations of cyclic phosphonate monomers [2-8,13] and the condensation polymerizations of dimethyl phosphite and dia1coho1s [9,11]. These methods yield polymers with number-average molecular weights (Rn) ranging from 10 3 to 10 5 daltons. There are several aspects of these polymerizations that are not well understood. In addition, the literature reports do not include details on the yields of the polymers and on the purification methods that were used. In this paper, we report the results of our studies on both the ring opening and condensation polymerizations. Our studies of the ring opening polymerizations focus on the effects of the alkylene groups on the polymerizations of the cyclic phosphonates and on the best methods for converting the poly(alkylene phosphonate)s to poly(alkylene phosphate)s. Our studies on the condensation polymerizations are more preliminary in nature and focus on the mechanism by which the polymerizations occur. EXPERIMENTAL

TECHNIQUES Preparation and Purification of Phosphorus Precursors. Cyclic phosphonate monomers were prepared using standard techniques [13]. Dimethyl phosphite was purified by the method of Penczek et al [11].

POLY(ALKYLENE PHOSPHATES)

251

Handling of the Polymer Solutions. All polymer solutions were handled under an atmosphere of high-purity nitrogen using either standard Schlenk techniques or in a N2 filled glove box. High purity N2 (99.999% pure) was used in all experiments. Purification of Solvents. Reagent grade dichloromethane, diethylamine, diethyl ether, tetrahydrofuran (THF), toluene and triethylamine were dried using standard procedures. 1,2-Ethanediol, 1,3-propanediol and 3(±)-methoxy-1,2-propanedio1, triisobutylaluminum, imidazole, d1chloroform, and HPLC grade methanol were used as received. All diols that were used in the polycondensation reactions were vacuum distilled prior to use. Dry chlorine gas was obtained by passing the gas sequentially through aqueous potassium permanganate, dilute sulfuric acid and concentrated sulfuric acid solutions and then through solid magnesium perchlorate. Multinuclear NMR Spectroscopy. The IH NMR spectra were taken on a Varian EM-360A 1H NMR; 13c and 31p NMR spectra were taken on a Nicolet 300 MHz, wide-bore, multinuclear NMR spectrometer. The 1H and 13C NMR spectra were referenced to internal tetramethylsilane, and the 3l p NMR spectra were referenced to external 40% phosphoric acid. Each NMR solution was prepared by dissolving the sample in d1-chloroform under N2 and then filtering the solution through a 1 cm column of Celite in a Pasteur pipet under N2 into a N2-filled NMR tube. The tube was then capped, and parafilm was wrapped around the cap. Size Exclusion Chromatography and Vapor Pressure Osmometry. Two SEC instruments were used to determine molecular weights. Polymers soluble in THF were run on a system incorporating a waters model 510 pump, a waters 410 differential refractometer and two Waters "linear ultrastyragel" columns (continuous porosity range of 10 2 - 10 6 ). THF insoluble, dichloromethane soluble polymers were run on a system incorporating a Waters model 6000A solvent delivery system, a waters model R-400 differential refractometer, and a Polymer Laboratories 'mixed PL-gel' column (continuous porosity range from 10 2 - 10 6 ). Apparent peak molecular weights are reported for the THF-insoluble polymers because similarities of the refractive indices of the polymers and dichloromethane did not allow high quality chromatograms to be obtained. Polydispersities of the THF-soluble polymers are reported. All calculations are based on a polystyrene calibration using well-characterized standards. Vapor pressure osmometry experiments were carried out on HPLC grade chloroform solutions of the polymers at 30 °c using a UIC, Inc. vapor pressure osmometer. Calibration of the osmometer was conducted using solutions of benzil that had been recrystallized and was confirmed by measurements of Rn for several low Mn polystyrene standards. SYNTHESES OF THE POLYMERS Poly(1,3-propylene phosphonate), 1. A solution of 60.9 g (498 romol) of 2-oxo-1,3,2-dioxaphosphorinane in 71.21 mL of methylene chloride was stirred at ambient temperature under N2 as 1.26 mL (4.98 romol) of triisobutylaluminum was added. This solution was stirred at ambient temperature for 24 hours and then was added dropwise to 1.6 L of toluene. The polymer was further purified by three additional precipitations from

252

G. M. GRAY ET AL.

toluene to yield 13.4 g (22.1%) of pure 1. The toluene solutions from all of the precipitations were combined and evaporated to dryness to yield the unreacted monomer. The polymerization and purification procedures described above were repeated with this material to yield an additional 6.60 g (10.8 %) of pure 1. The toluene residues from the purification of this material were then evaporated to dryness to yield the unreacted monomer. The polymerization and purification procedures were repeated with this material to yield an additional 3.60 g (5.91 %) of pure 1. The total yield of pure 1 was 38.8%. Poly(1,2-ethylene phosphonate), 2. A solution of 42.3 g (392 mmol) of 2-oxo-1,3,2-dioxophospholane in 56 mL of dichloromethane was stirred at ambient temperature under N2 as 0.98 mL (3.9 mmol) of triisobutylaluminum was added dropwise. This solution was stirred at ambient temperature for 24 hr, and then was added dropwise to 1.2 L of toluene to yield 39.1 g (92.5 %) of impure 2 as a sticky gel. Poly(1,2-{1-(methoxymethyl)ethylene} phosphonate), 3. A solution of 43.45 g (286 mmo1) of freshly distilled 4-methoxymethyl-2-oxo-1,3,2dioxaphospholane in 41 mL of dichloromethane was stirred under N2 as 0.14 mL (0.57 mmol) of triisobutylaluminum was added. This solution was stirred at ambient temperature for 24 hr. An additional 100 mL of dich1oromethane was then added. This solution was immediately used in the preparation of poly(N,N-diethyl 1,2-{1-(methoxymethyl)ethylene} aminophosphate), 6, and poly(methyl 1,2- {l-(methoxymethyl)ethylene} phosphate), 9. Poly(N,N-diethyl 1,3-propylene aminophosphate), 4. Chlorine was bubbled into a solution of 1.90 g (156 mmol) of 1 in 40 mL of dichloromethane at ambient temperature until a yellow color persisted. The solution was then placed under vacuum until the yellow color disappeared. Dry diethylamine (3.38 mL, 328 mmol) was added, and the solution was stirred for 5 h. The brown reaction mixture was cooled to 0 °c and then filtered. The filtrate was collected and evaporated to dryness. The residue was washed with several portions of THF and dried to yield 1.90 g (60.0%) of 4 as a glassy solid. Poly(N,N-diethyl 1,2-ethylene aminophosphate), 5. Chlorine was bubbled into a solution of 3.75 g (34.7 mmol) of 1 in 300 mL of dichloromethane at ambient temperature until a yellow color persisted. The solution was then placed under vacuum until the yellow color disappeared. The colorless solution was filtered and the filtrate was stirred at ambient temperature as 9.84 mL (95.1 mmol) of dry diethylamine was added. This mixture was stirred for 2 hours and then was cooled to -78 °c and filtered. The filtrate was evaporated to dryness to yield 6.4 g of 5 as a brown oil. Poly(N,N-diethyl 1,2{1-(methoxymethyl)ethylene} aminophosphate), 6. Chlorine was bubbled into one half of the dichloromethane solution from preparation of 3 (containing a total of 21.73 g (142.9 mmol) of 2-oxo1,3,2-dioxa-4(methoxymethyl)phospholane and 3) at -78 °c until a yellow color persisted. Diethylamine (44.3 mL, 428.2 mmol) was then added and, after 30 min, the solution was allowed to warm to room temperature. After 4.5 hours, the solution was cooled to - 78 °c again, and then was filtered to remove the diethyl amine hydrochloride precipitate. Anhydrous sodium carbonate was added to the filtrate, and the mixture was

POLY(ALKYLENE PHOSPHATES)

253

stirred for 19 h before being filtered 'through a layer of Celite. The mixture was filtered, and the filtrate was evaporated to dryness. The residue was dissolved in a minimum amount of dichloromethane, and the solution was filtered. The filtrate was added dropwise to 1 L of diethyl ether. The precipitation was repeated a second time, and then the residue was dried to yield 3.6 g (11%) of 6 as a glassy, yellow solid. Poly(methyl 1,3-propylene phosphate) 7. Chlorine was bubbled into a solution of 1.90 g (15.6 rnrnol) of 1 in 10 mL of dichloromethane at room temperature as chlorine was bubbled into the solution until a yellow color persisted. The solution was place under vacuum until the color disappeared, and then 1. 06 g (15.6 rnrnol)' of imidazole and 1. 26 mL (31. 3 rnrnol) of methanol were added. The reaction mixture was stirred for 5 hours under N2, and then was cooled to 0 °c and filtered. The filtrate was evaporated to dryness to yield the crude product. This material was dissolved in a minimum amount of dichloromethane, and the solution was cooled to -78 °c and filtered. The filtrate was evaporated to dryness to yield 1.83 g (77.0%) of the glassy product which still contained a small amount of imidazole (by IH NMR spectroscopy). Poly(methyl 1,2-ethylene phosphate), 8. Chlorine was bubbled into a solution of 3.75 g (34.7 mmol) of 2 in 300 mL of dichloromethane at ambient temperature until the yellow color persisted, and then the solution was placed under vacuum until the yellow color disappeared. This solution was filtered to remove the small amount of precipitate that formed during the chlorination, and the filtrate was stirred at ambient temperature as 2.16 g (31.7 rnrnol) of imidazole and 2.57 mL (63.4 rnrnol) of methanol were added. The reaction mixture was stirred for 2 hours and then was cooled to -78 °c and filtered. The filtrate was evaporated to dryness to yield 5.55 g of pale yellow oily product. Poly(methyl 1 ,2-{ 1- (methoxymethyl) ethylene} phosphate), 9. Chlorine was bubbled into one quarter of the dichloromethane solution from the preparation of 3 (containing a total of 10.87 g of 3 and 4-methoxymethyl-2-oxo-I,3,2-dioxaphospholane) at -78 °c until a yellow color persisted. Then 4.87 g (71.5 mmol) of imidazole and 5.79 mL (143 mmol) of methanol were added. The reaction mixture was allowed to warm to room temperature and was stirred for 9.5 hours. This solution was cooled to -78 °c and filtered to remove the imidazole hydrochloride precipitate. The filtrate was evaporated to dryness, and the residue was precipitated twice from dichloromethane with ether and twice from dichloromethane with acetone to yield 7.8 g (66%) 9 as a pale yellow, glassy solid. Poly(triethylenedioxy phosphonate), 10. A mixture of 5.40 mol (58.9 rnrnol) of dimethyl phosphite and 7.70 mL (57.7 rnrnol) of triethyleneglycol was heated at 80 to 120 °c at atmospheric pressure for 5 hr, then at 120 °c at aspirator vacuum for 17 hr and finally at 120 °c at 0.5 rnrn of Hg for 3.5 hr. The reaction mixture was then cooled to ambient temperature, and 10 was collected as a glassy white solid. The yield of the polymer was 100% based upon the amount of triethyleneglycol used. Poly(1,12-dodecadiene phosphonite), 11. A mixture of 4.60 mL (50.3 rnrnol) of dimethyl phosphite and 9.70 g (47.9 rnrnol) of 1,12-dodecanediol was heated at 80 °c at atmospheric pressure for 24 h, then at 120 °c at atmospheric pressure for 4 h and finally at 120 °c at 0.5 rnrn of Hg for 4 h. The reaction mixture was then cooled to ambient temperature, and 11

254

G. M. GRAY ET AL.

was collected as a hard, white solid. The yield of the polymer was 100% based upon the amount of l,12-dodecanediol used. RESULTS AND DISCUSSION

RING-OPENING POLYMERIZATIONS The scheme for the preparation of the poly(alkylene phosphonate)s and poly(alkylene phosphate)s via ring-opening polymerizations is given in Figure 2. This can be separated into two phases: 1) the ring-opening polymerizations of the cyclic phosphites to form the poly(alkylene phosphonate)s and 2) the conversion of the poly(alkylene phosphonate)s to poly(alkylene phosphate)s. These procedures are modifications of those reported in the literature and are discussed below. Figure 2.

R

o / \ \ / o

P(O)H

Scheme for the Preparations of the Poly(alkylene phosphate)s

BU~Al CH2C12

0 >

l'

-(r-0-R-O)n-

C12 --->

H 1 R 2 R 3 R

CH2CH2CH2

10 R

CH2CH CH20CH3)

12 R

CH2CH~

1

Et2NH

0

l'

-(r-0-R-O)nNEt2 4 R 5 R 6 R

llR

MeOH

1

1m1d. .01.

o l'

- Pj'-O-R-O) n6Me

CH2CH2CH2

= CH2CH~ = CH2CH CH20CH3)

The ring-opening polymerizations of three different cyclic phosphonates have been studied. The polymerization of 2-oxo-l,3,2dioxaphosphorinane (R = -(CH2)3-) is approximately 50% complete at ambient temperature after two hours with triisobutylaluminum as a catalyst. These results are consistent with those previously reported for this polymerization when carried out under slightly different conditions [14]. The presence of the equilibrium is confirmed by the fact that the addition of more initiator to the polymerization solution does not affect the monomer-polymer ratio. The polymer, 3, can be separated completely from the monomer by fractional precipitation, and the reequilibration of the precipitated polymer is slow in the solid state (only a small amount of monomer is observed in the polymer after several months at ambient temperature).

POL Y(ALKYLENE PHOSPHATES)

255

In contrast, the polymerization of 2-oxo-1,3,2-dioxaphospholane (R = -(CH2)2-) occurs much more rapidly, and no initiator is needed. This polymerization is approximately 85% complete at 24 °C. Unlike the previous equilibrium, this is quite rapid, and the polymer cannot be purified by fractional precipitation. A variable temperature 31p NMR spectroscopic study of a 0.50 M d1-chloroform solution of the monomerpolymer mixture has been carried out, and a plot of 1n (Me) vs T- 1 is shown in Figure 3. Both ~H and ~S for the equilibrium have been calculated from this data and are -10.7 ± 0.6 kJ/mo1 and -14.2 ± 0.2 J/mo1oK, respectively. These results are consistent with previous studies that suggest that the ring-opening polymerizations of 1,3,2- dioxaphospholanes should be exothermic [14,15] Figure 3. Plot of In(Me) vs. T- 1 for the Polymerization of 2-0xo-1,3,2-dioxaphospholane

-2.20

,...... -2.40 ~

c

-2.60

- 2.80 -tr"""",..,rrrrrnrrnrrnrrnrrnrrnrrnTrTTrT-rrT-rrT-rrT..",..",." 2.90 3.00 3.30 3.40 The polymerization of 4-methoxymethyl-2-oxo-1,3,2-dioxaphospho1ane (R -CH2CH(CH20CH3)-) is different from either of the above polymerizations. The monomer does not self polymerize, and the polymerization goes to approximately 60 % completion at ambient temperature with triisobutylaluminum as an initiator. The polymer cannot be separated from the monomer by fractional precipitation. It was not possible to carry out variable temperature 3l p NMR spectroscopic studies of this polymerization, because of the the variety of phosphorus environments in the polymer cause the 31p NMR resonance of the polymer to be very complex.

256

G. M. GRAY ET AL.

These studies indicate that both the alkylene chain length and the degree of substitution have a significant effect upon the ring-opening polymerizations of cyclic phosphonates. As expected, decreasing the length of the alkylene group from 3 to 2 pushes the equilibrium towards the polymer. Unfortunately, in the case of 2-oxo-1,3,2-dioxaphospholane, self polymerization occurs and only low molecular weight species are obtained. A second effect of decreasing the chain length is that the rate at which the equilibrium is reached is also increased. Thus, it is not possible to isolate pure poly(ethylene phosphonate). Substitution on the alkylene groups pushes the polymerization towards the monomer, and also decreases the reactivity sufficiently that selfpolymerization does not occur at ambient temperature. However, it is still not possible to purify these polymers by fractional precipitation. The poly(alkylene phosphonate)s were converted into the corresponding poly(alkylene chlorophosphate)s by bubbling dry chlorine into dichloromethane solutions of the polymers until the solution turned yellow [14]. The poly(alkylene chlorophosphate)s were not particularly stable and were immediately converted into either poly(N,N-diethyl alkylene aminophosphate)s through reaction with two equivalents of diethylamine or poly(methyl alkylene phosphate)s through reaction with imidazole and methanol. The reactions are outlined in Figure 2. The reactions of the poly(alkylene chlorophosphate)s with diethylamine were completed after two hours. Most of the diethylammonium chloride byproduct could be removed from the polymers by cooling their dichloromethane solutions to -78 °c and f il ter ing. The remaining diethylammonium chloride was removed by stirring a dichloromethane solution of the dry polymer over anhydrous potassium carbonate for 24 h to convert the diethylammonium chloride to diethyl amine, and then evaporating the solution to dryness. The conversions of the poly(1,3-alkylene chlorophosphate)s to poly(methyl alkylene phosphate)s were more difficult. The method of Klosinski and Penczek, involving the reaction of the polymer with 2 moles of imidazole per mol of repeat unit followed by the reaction of this polymer with the alcohol [15) was first attempted. However, this procedure yields product that is contaminated with significant amounts of imidazole. The solubilities of imidazole and the polymers were very similar, and it was not possible to remove all of the imidazole by fractional precipitation. Several attempts were made to avoid the problems encountered with imidazole by using either triethylamine, N,Ndimethylaminopyridine or DBU as HCl sinks. However, these did not give complete reactions. The best method was found to be the reaction of the polymer with imidazole and methanol in a 1:1:2 mole ratio. This method uses only half as much imidazole as that of Klosinski and Penczek, and nearly all of the imidazole is converted to imidazole hydrochloride which can be removed by filtering the cold reaction mixture (-78 oC). POLYCONDENSATIONS The scheme for the polycondensations of dimethylphosphite with diols to form poly(alkylene phoshonate)s is shown in Figure 4. This route is complementary to the ring-opening polymerization route because

257

POLY(ALKYLENE PHOSPHATES)

it can only be carried out with diols that cannot form cyclic phosphonates; whereas, the ring-opening polymerization route can only be carried out with diols that can form cyclic phophonates. This route is simpler than the ring-opening polymerization route because the poly(alkylene phosphonate)s are formed in a single step from commercially available materials. In addition, the reaction conditions force the condensation polymerizations to completion, and thus no separation of monomer and polymer is necessary. Figure 4. Scheme for the Condensation Polymerization of Dimethyl Phosphite and Dialcohols

o l'

Meo-r- OMe H

+

HOROH

80-120

°c

------->

o l'

-(Y-0-R-O)nH

Polycondensations of dimethylphosphite and a variety of diols have been reported by Penczek and coworkers [9,11]. These workers reported that the po1ycondensations only go to completion under relatively severe reaction conditions (180 oC, 31 h). We have attempted to repeat these experiments with the diols shown in Figure 3, and have found that these conditions cause the poly(alkylene phosphonate)s to decompose. With triethyleneglycol, a single 3l p NMR resonance for the polymer is observed after the reaction mixture is heated at 80 to 120 °c at atmospheric pressure for 5 hr and then at 120 °c under aspirator vacuum overnight. The presence of only three resonances in the 13C NMR spectrum of this material and the fact that two of the resonances are coupled to phosphorus (P-O-£H2: 5 13 C = 64.64 ppm, d, 1 1 J(PC) I = 6 Hz; P-O-CH2-£H2-: 5 13 c = 70.13 ppm, d, 1 2 J(PC) I = 5 Hz; CH2-0-£H2-£H2-0CH2: 5 13 c = 70.46 ppm, s) confirm that both ends of the triethylene glycol have been substituted by phosphites and that the material is a polymer. The polycondensation of dimethylphosphite with 1,12-dodecanediol has been followed by multinuclear NMR spectroscopy, and, after significant amounts of polymer are formed, by SEC. The 31 p NMR spectra of the reaction mixture at various times and temperatures are shown in Figure 5, and suggest that displacement of the second methoxy group from the dimethyl phosphite is significantly more difficult than is displacement of the first group. The SEC data, shown in Table I, indicate that the Rw for the polymer increases until no end groups are observed in the 31p NMR spectrum and then slowly decreases as the heating is continued. The data clearly demonstrate that long heating times have a detrimental effect on the Rws of the polymers. Attempts to heat this polymer at the high temperatures suggested by Penczek result in the total decomposition of the polymer.

2'; "

30 min

~O

I •

i



,

10

I

i

'--'--

r-ro

~~I~~

",JI

80 °CI

3 hr

I 10

J/VL\L

80 °CI

20

b

10

80 °CI 6

20

hr

o

80 °CI

10

20

o

10

20

24 hr

I------.-I-----.--'-~

10.5 hr

80 °CI

0

I

120

a

10

C,

Figure 5. Polycondensation of Dimethyl Phosphite and l,12-Dodecanediol

20

--~~r-,

o

10

~"I

5 hr

i

\)

120 °CI 5 hr, vacuum

~

~

...::

~

Cl

~

p

tv VI 00

259

POLY(ALKYLENE PHOSPHATES)

Table 1. 31p NMR and SEC Data for the Reaction of Dimethyl Phosphite and 1,2-Dodecanediol rxn. time a hr 6.5 10.0 15.0 22.0

MeOP{OI {HIO- b -OP(O) (H)O7.1 7.0