Precursor-Derived Ceramics: Synthesis, Structure and High-Temperature Mechanical Properties

Precursor-Derived Ceramics Joachim Bill, Fumihiro Wakai, Fritz Aldinger (Eds.) 8WILEY-VCH This Page Intentionally L...

Author: Joachim Bill | Fumihiro Wakai | Fritz Aldinger

51 downloads 927 Views 6MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Precursor-Derived Ceramics

Joachim Bill, Fumihiro Wakai, Fritz Aldinger (Eds.)

8WILEY-VCH

This Page Intentionally Left Blank

Joachim Bill, Fumihiro Wakai, Fritz Aldinger (Eds.)

Precursor-Derived Ceramics Synthesis, Structures and High Temperature Mechanical Properties

8WILEY-VCH Weinheim - New York - Chichester Brisbane * Singapore - Toronto

Dr. Joachim Bill Universitat Stuttgart lnstitut fur Nichtmetallische Anorganische Materialien HeisenbergstraBe 5 D-70569 Stuttgart Germany

Prof. Dr. Fumihiro Wakai Japan Science and Technology Corporation ,,Ceramics Superplasticity" Project 2-4- 1, Mutsuno Nagoya, 456 Japan

Prof. Dr. Fritz Aldinger Max-Planck-Institut fur Metallforschung HeisenbergstraBe 5 D-70569 Stuttgart Germany

This book was carefully produced. Nevertheless, authors, editors and publisher do not warrant the information contained therein to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Reprints of papers presented during the international workshop Grain Boundary Dynamics of Precursor-Derived Covalent Ceramics, held at SchloB Ringsberg, November 10 to 16,

1996.

Library of Congress Card No. applied for. A catalogue record for this book is available from the British Library. Deutsche Bibliothek Cataloguing-in-Publication Data: Precursor-DerivedCeramics : synthesis, structures and high temperature mechanical properties. - Weinheim ;New York ; Chichester ; Brisbane ; Singapore ;Toronto : Wiley-VCH, 1999 ISBN 3-527-29814-2 0 WILEY-VCH Verlag GmbH. D-69469 Weinheim (Federal Republic of Germany), 1999

Printed on acid-free and chlorine-free paper. All rights reserved (including those of translation in other languages). No part of this book may be reproduced in any form - by photoprinting, microfilm, or any other means - nor transmitted or translated into machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Printing: betz-druck gmbh, D-64291 Darmstadt. Bookbinding: J. Schaffer GmbH & Co. KG., D-67269 Griinstadt Printed in the Federal Republic of Germany.

Preface

The preparation of inorganic materials by thermolysis of preceramic compounds has gained substantial interest during recent years. The general idea of this process route is that elementorganic precursor molecules already contain structural units of the residual inorganics formed by thermolysis, thus providing novel paths of controlling composition, atomic array and microstructure of materials. Of special interest is the manufacture of amorphous or nanocrystalline covalently bonded inorganics on the basis of silicon, boron, carbon and nitrogen, revealing a potential of properties not known from conventionally prepared materials. In order to develop a sound basis for a successful use of such materials and to judge their potential for application, there is a need for profound basic research in the field of their synthesis and processing and in the characterization of their structure and properties. It was with this intention, when in 1995 the Japanese Science and Technology Corporation (JST) and the Max-Planck-Institut fur Metallforschung (MPI-MF) signed a five-year contract for the cooperation on the synthesis of nanostructured materials by precursor thermolysis and the investigation of their superplastic behavior. It is the aim of this international cooperation to combine studies of elementorganic chemistry, materials processing and materials characterization in order to tailor the microstructure and thus the properties of covalent materials. Of special interest are nano-grained materials and their mechanical properties at high temperatures and to study superplasticity and other dynamic grain boundary phenomena. This cooperation is the most recent item in a long lasting and fruitful cooperation of the Max-Planck-Institut fur Metallforschung and its representatives together with Japanese scientists and institutions active in materials research. In the light of this cooperation an international workshop on Grain Boundary Dynamics of Precursor-Derived Covalent Ceramics was arranged in order to present the status quo in the field of precursor-derived materials and to discuss the mid-term results in this project with the worldwide leading scientists in this research field. As directors of this cooperation project on the German and Japanese side respectively, we would like to take the opportunity to thank the Max-Planck-Gesellschaft zur Forderung der Wissenschaften e.V. and the Japanese Science and Technology Corporation for their generous support of the joint research programm. We also would like to thank all the scientists who participated in the workshop for their contributions to the scientific presentations and discussions. We would like to express our special thanks to Dr. W. Hasenclever, former Secretary General of the Max-Planck-Gesellschaft, Dr. J. Roemer-Mahler (Bundesministerium fur Bildung, Forschung und Technologie, BMBF) and Professor Dr. Dr. h.c. mult. Gunter Petzow, Emeritus of the

VI

Preface

Max-Planck-Institut fur Metallforschung for their introductory contributions on the globalization of research and Japanese-German Research Cooperations. The organizers of the workshop gratefully acknowledge the financial support given by Deutsche Forschungsgemeinschaft (DFG), Bonn/Germany; Dr. Ernst-RudolfSchloeDmann Stiftung, Germany; Japan Science and Technology Corporation (JST), Japan; Max-Planck-Gesellschaft zur Forderung der Wissenschaften e. V. (MPG), MunchedGermany .

Prof

Dr. Fritz Aldinger Max-Planck-Institut fur Metallforschung Stuttgart, Germany

Prof Dr. Fumihiro Wakai Japan Science and Technology Corporation Nagoya, Japan

Contents

List of Participants

. . . . . . . . . . . . . . . . . . . XI

1. Globalization of Research

. . . . . . . . . . . . . . . .

1

Research between Nationality and Internationality . . . . . . . . . . 3 W. Hasenclever The Implications of Globalization for Research Funding by the Federal Ministry of Education, Science, Research and Technology . . . . . . . . . . I I J. Roemer-Mahler Excitements in Japanese-German Research-Cooperation . . . . . . . . 18 G. Petzow

11. Keynote Lectures . . . . . . . . . . . . . . . . . . . 3 1 Precursor-Derived Covalent Ceramics J . Bill, F. Aldinger Ceramics Superplasticity F. Wakai

. . . . . . . .

33

. . . . . . . . . . . . . . . . . .52

111. Ceramics from Organoelement Compounds . . . . . . . . . Synthesis and Processing of Sic Based Materials Using Polymethylsilane R.M. k i n e , Z.-F. Zhang, 'J.A. Rahn, K.W. Chew, M . Kannisto, C. Scotto Silicon Carbide Fibers from Highly Reactive Poly(methy1chlorosilane)s . G. Roewer, H.-P. Martin, R. Richter, E. Miiller

59

. . 61

,

. 73

VIII

Contents

. . . . .

High Temperature Stable Ceramics from Inorganic Polymers M. Weinmann, F. Aldinger

Processing and Mechanical Properties of SiC-based Ceramics from Organosilicon Polymers . . . . . . . . . . . . . . G.D. Sorarh Polymeric Precursors for Bn and SiNCB Ceramic Fibers Th. Wideman, E.E. Remsen, G.A. Zank, L.G. Sneddon

. . .

. . . . . .

83

93

103

Microstructure Evolution and Crystallization Behavior of Polymer-Derived Si-C-N Monoliths; A TEM Study . . . . . . . . . . . . . H.-J. Kleebe, D. Suttor, G. Ziegler

.

113

Synthesis Principles and Processing of Oxidic Ceramic Materials Derived from Metal Organic Compounds . . . . . . . . . . . . . G . Miiller, D. Sporn

.

132

Synthesis of Oxide and Non-Oxide Inorganic Materials at Organic Surfaces . 143 M.R. De Guire, Th.P. Niesen, J. WolfJ;S. Supothina, J. Bill, F. Aldinger, M. Riihle

IV. Characterization

. . . . . . . . . . . . . . . . . .

163

. .

165

. .

175

. .

188

. . . . . . .

197

Thermodynamic Calculations in the System Si-B-C-N-0 . . . . . H . J . Seifert, F. Aldinger Photoelectron Spectroscopy as a Tool for Studying Ceramic Interfaces: A Tutorial . . . . . . . . . . . . . . . . . . . . F.C. Jentoft, G. Weinberg, U. Wild, R. Schlogl Corrosion: No Problem for Precursor-Derived Covalent Ceramics? K. G. Nickel Solid State NMR Studies for Ceramic Characterization K. Miiller

Solid State NMR Studies of Organically Modified Ceramics M. Templin, U. Friedrich, U. Wiesner, H.W. Spiess

.

,

Characterization of Amorphous Materials by Diffraction Methods P. Lamparter

.

. . .

205

.

. .

214

IX

Contents

X-ray and Neutron Diffraction Investigation on Amorphous Silicon Carbonitrides . . . . . . . . . . . . . . . . . . . J . Diirr, S. Schempp, P. Lamparter, J . Bill, S. Steeb, F. Aldinger

. .

224

V. High Temperature Mechanical Properties and Characterization of Grain Boundaries . . . . . . . . . .

235

Compression Creep Behaviour of Precursor -Derived Ceramics . . . . . G. Thurn, F. Aldinger

237

. .

246

.

260

Correlation Between Deformation Mechanism and Microstructural Evolution in Silicon Nitride Ceramics . . . . . . . . . . . . . . . . J.A. Schneider, A.K. Mukherjee

270

. . . . . . . . . . .

280

Nano-, Micro- and Milliboundaries in Silicon Nitride Based Ceramics P. Sajgalik Plastic Deformation of Covalent Bonded Ceramics at High Temperatures S. Tsurekawa, K. Kawahara, H. Nakashima

Fabrication of Fine-Grained Si3N4 and Sic T. Nishimura, M . Mitomo, H. Emoto

Appendix

. . . . . . . . . . . . . . . . . . . . . .

Program of the workshop on Grain Boundary Dynamics of Precursor-Derived Covalent Ceramics . . . . . . . . . . . . . . . . . . .

29 1

29 1


List of Participants Workshop on Grain Boundary Dynamics of Precursor-Derived Covalent Ceramics

Prof. Dr. Fritz Aldinger, Max-Planck-Institut fur Metallforschung, Heisenbergstr. 5 , D-70569 Stuttgart, GERMANY Dr. Florence Babonneau, Universite Pierre et Marie Curie Tour 54 E5, Chiemi de la Matiere CondensCe, 4 place Jussieu, F-75252 Paris Cedex 05, FRANCE Dr. Pietro Ballone, Max-Planck-Institut fur Festkorperforschung, Heisenbergstr. I , D-70569 Stuttgart, GERMANY Dr. Bernd Baufeld, Universitat Erlangen-Numberg Inst. f. Werkstoffwissenschaften, MartensstraSe 5, D-9 1058 Erlangen, GERMANY Prof. Dr. Gerd Becker, Universitat Stuttgart, Institut fur Anorganische Chemie, Pfaffenwaldring 55, D-70569 Stuttgart, GERMANY Dr. rer. nat. Joachim Bill, Max-Planck-Institut fur Metallforschung, Heisenbergstr. 5, D-70569 Stuttgart, GERMANY Dr. Horst Boder, Robert Bosch GmbH , Abteilung FVFLW, Postfach 10 60 50, D-70049 Stuttgart, GERMANY Dr. Dieter Brunner, Max-Planck-Institut fur Metallforschung, SeestraSe 92, D-70174 Stuttgart, GERMANY Prof. I-Wei Chen, School of Engineering and Applied Science, Dept. of Materials Science and Engineering, Philadelphia, PA 19104-6272, USA Prof. Dr. Mark De Guire, Department of Materials Science & Engenineering, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, Ohio 44 106-7204, USA

XI1


Dr.rer.nat. Wolfgang DreBler, Robert Bosch GmbH, Abt. FV/FLW, Postfach 10 60 50, D-70049 Stuttgart, GERMANY Dr. Johannes Durr WildermuthstraSe 40, D-72076 Tubingen, GERMANY Dr. Wolfgang Hasenclever, Max-Planck-Gesellschaft zur Fordemng der Wissenschaften e.V., Postfach 10 10 62, D-80084 Munchen, GERMANY DipLIng. Rainer Haug, Max-Planck-Institut fur Metallforschung, HeisenbergstraSe 5 , D-70569 Stuttgart, GERMANY Klaus Hrastnik, Robert Bosch GmbH , Abteilung FV/FLW, Postfach 10 60 50, D-70049 Stuttgart, GERMANY Prof. Dr. Leonard V. Interrante, Rensselaer Polytechnic Institute, Dept. of Chemistry, Cogswell Laboratory, Troy, NY 12180-3590, USA Prof. Dr. Martin Jansen, Max-Planck-Institut, fur Festkorperforschung, Heisenbergstralje I , D-70569 Stuttgart, GERMANY Dr. Friederike Jentoft, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Abt. Anorganische Chemie, Faradayweg 4-6, D-14195 Berlin, GERMANY Dr. Ken-ichi Kakimoto, Max-Planck-Institut fur Metallforschung, HeisenbergstraSe 5 , D-70569 Stuttgart, GERMANY Mr. Masahiro Kawasaki, Japan Science and Technology Corporation, Kawaguchi Center Building, 4-1-8, Hon-cho, Kawaguchi-shi, Saitama 332, JAPAN Dr. Hans Joachim Kleebe, Universitat Bayreuth Institut fur Materialforschung (IMA), Lehrstuhl fur Keramik und Verbundwerkstoffe, D-95440 Bayreuth, GERMANY Dr. Peter Kroll, Technische Hochschule Darmstadt, Fachbereich 2 1, Materialwissenschaft, Fachgebiet Disperse Feststoffe, PetersenstraSe 23, D-64287 Darmstadt, GERMANY Prof. Dr. Richard M. Laine, University of Michigan , Dept. of M.S.E., Room 21 14 Dow Building, 2300 Hayward, Ann Arbor, MI 48109-2136, USA


XI11

Dr. Hannes-Peter Lamparter, Max-Planck-Institut fur Metallforschung, Seestrase 92, D-70174 Stuttgart, GERMANY Dr. Jan Lucke, CFI-Marketing Polymerkeramik, Crerner Forschungsinstut GmbH, Oeslauer Strase 35, D-96472 Rodental, GERMANY Dr. Joachim Mayer, Max-Planck-Institut fur Metallforschung, Seestrase 92, D-70174 Stuttgart, GERMANY Prof. Amiya K. Mukherjee, Department of Chemical Engineering and Materials Science, University of California, Davis, CA 95616, USA Prof. Dr. Gerd Muller, Fraunhofer-Institut fur Silicatforschung, Neunerplatz 2, D-97082 Wurzburg, GERMANY Priv.-Doz. Dr.rer.nat.habi1. Klaus Muller, Universitat Stuttgart Institut fur Physikalische Chernie, Pfaffenwaldring 55, D-70569 Stuttgart, GERMANY Dr. Bertold Neizert, Max-Planck-Gesellschaft zur Forderung der Wissenschaften e.V., Generalverwaltung, Abt. I, Postfach 10 10 62, D-80084 Munchen, GERMANY Prof. Dr. Klaus G. Nickel, Eberhard-Karls-Universitat Tubingen, Institut fur Mineralogie, Petrologie, und Geochemie, - Angewandte Mineralogie Wilhelmstralje 56, D-72074 Tubingen, GERMANY Dr. Thomas Niesen, Max-Planck-Institut fur Metallforschung, Heisenbergstr. 5, D-70569 Stuttgart, GERMANY Dr. Toshiyuki Nishimura, National Institute for Research in Inorganic Materials, 1-1 Namiki, Tsukuba, Ibaraki 305, JAPAN Prof. Kiyohito Okamura, Osaka Prefecture University College of Engineering, Department of Metallurgy and Materials Science, 1- 1 , Gakuen-cho, Sakai, Osaka, 593 JapanJAPAN Dr. Gerd Passing, Bayer AG, ZF-MFA, Q 18 Gebaude Q 18, D - 5 I368 Leverkusen, GERMANY

Prof. David G. Pettifor, University of Oxford, Department of Materials, Parks Road, UK-Oxford OX1 3PH, UNITED KINGDOM Prof. Dr.Dr.h.c.mult. Gunter Petzow, Max-Planck-Institut fur Metallforschung, Heisenbergstralje 5, D-70569 Stuttgart, GERMANY

XIV


Prof. Rishi Raj, University of Colorado, Dept. of Mechanical Engeneering, Boulder, Co, 80309-0427, USA Dr. Robin Richter, TU Bergakademie Freiberg, Institut fur Anorganische Chemie, Leipziger Strarje 29, D-09599 FreibergBachsen, GERMANY Prof. Dr. Ralf Riedel, Technische Hochschule Darmstadt, Fachbereich 2 1 Materialwissenschaft, Fachgebiet Disperse Feststoffe, FTZ Gebaude D Petersenstrde 23, D-64287 Darmstadt, GERMANY Dr. Jurgen Roemer-Mahler, Bundesministerium fur Bildung und Forschung BMBF, Heinemannstrarje 2, D-53 170 Bonn, GERMANY Prof. Dr. Gerhard Roewer, TU Bergakademie Freiberg,Fak. 2 Inst. f. Anorg. Chemie, Leipziger Strarje 29, D-09599 FreiberdSachsen, GERMANY Dr. Tanguy Rouxel, Laboratoire de Materiaux Ceramiques et Traitements de Surface, Equipe: Materiaux Ceramiques, ENSCI 47 avenue Albert Thomas, F - 87065 Limoges, Cedex, FRANCE Prof. Dr. Manfred Ruhle, Max-Planck-Institutfur Metallforschung, SeestraBe 92, D-70174 Stuttgart, GERMANY Dr. Pavol Sajgalik, Institut of Inorganic Chemistry, Slovac Academy of Sciences, Dubravska Cesta 9, SK 84236 Bratislava, SLOVAKIA Prof. Dr. Makoto Sasaki, Department of Materials Science and Engineering, Faculty of Engineering, Muroran Institute of Technology 27- 1 Mizumoto, Muroran, Hokkaido 050, JAPAN Prof.Dr. Eiichi Sato, The Institute of Space and Astronautical Science, 3-1-1, Yoshinodai, Sagamihara-Shi, Kanagawa 229, Japan JAPAN Prof. Dr. Robert Schlogl, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Abteilung Anorganische Chemie, Faradayweg 4 - 6, D - 14195 Berlin, GERMANY Dr. Hans Jurgen Seifert, Max-Planck-Institutfur Metallforschung, HeisenbergstraSe 5, D-70569 Stuttgart, GERMANY Prof. Dr. Larry G. Sneddon, University of Pennsylvania, School of Arts and Sciences, Department of Chemistry, 23 1 South 34th Street, Philadelphia, PA 19104-6323, USA


XV

Prof. Gian Domenico Soraru’, Universita’ di Trento, Dipartimento di Ingegneria dei Materialia, Via Mesiano, 777, I - 38050 Trento, ITALY Dr. Daniel Suttor, Universitat Bayreuth, Institut fur Materialforschung, Lehrstuhl fur Keramik und Verbundwerkstoffe, Postfach 10 12 5 1D 95440 Bayreuth, GERMANY Teruaki Takeuchi, Japan Science and Technology Corporation, Department of International Affairs, Kawaguchi Center Building, 4- 1-8, Hon-cho, Kawaguchi, Saitama, 332, JAPAN Markus Templin, Max-Planck-Institut fur Polymerforschung, Ackermannweg 10, D-55128 Mainz, GERMANY Dr. Gunter Thurn, Max-Planck-Institut fur Metallforschung, HeisenbergstraDe 5, D-70569 Stuttgart, GERMANY Dr. Sadahiro Tsurekawa, Japan Science and Technology Corp., Ceramics Superplasticity Project, International Joint Research Programm, 2-4- 1, Mutsuno, Nagoya 456, JAPAN Dr. Manfred Unger, Hoechst Aktiengesellschaft Forschungsleitung, Postfach 80 03 20, D-65926 Frankfurt a.M., GERMANY Dr. Walter D. Vogel, Dornier GmbH , Dept. F4T/K, D 88039 Friedrichshafen, GERMANY Prof. Dr. Fumihiro Wakai, Japan Science and Technology Corporation, “Ceramics Superplasticity” Project, 2-4- 1 , Mutsuno, Nagoya, 456, JAPAN Dr.rer.nat. Manfred Weller, Max-Planck-Institut fur Metallforschung, Seestraae 92, D-70174 Stuttgart, GERMANY Prof. Dr. Sheldon M. Wiederhorn, National Institute of Standards & Technology, Building 223;Room B307, Gaithersburg, MD 20899, USA Prof. Dr. Peter Woditsch, Bayer AG Werk Uerdingen,AI-IP-LeitungGebaudeR 54 RheinuferstraDe 7-9,D-47829 Krefeld-Uerdingen, GERMANY


I. Globalisation of Research


Research between Nationality and Internationality Dr. Wolfgang Hasenclever Max-Planck-Gesellschaft zur Forderung der Wissenschaften e.V. Postfach 10 10 62,80084 Munchen

Globalization of research is the title, which the organizers of this workshop have given to this morning's opening session. Globalization has become a very fashionable term: 0 globalization of production, globalization of learning, 0 globalization of trade, of criminality, of pollution, of communication etc. etc. And this morning we are to discuss globalization of research! Globalization of research, is this really a controversal issue'? Hasn't research, hasn't science, used in the German sense of ,Wissenschaft" i.e. including social sciences and humanities - hasn't research always been universal, global, international? Is there really any room for making it so, to universalize, to globalize, to internationalize research? The spontaneous answers of most of us will be negative! And yet, the relationship between national, regional, transnational, multinational, universal research has been a subject of discussion for quite a time, not only among politicians or in the media, also among scientists. Globalization is just a new, a fashionable word which draws our attention to the fact that new, modern means of transportation and communication give a new intensity, a new quality to the wellknown worldwide interaction among people and institutions of research - normally regardless of borders and without reference to expectations of particular loyalties for political, institutional or economic reasons. But research is - like most if not all human activities - a social phenomenon, i.e. something that happens not in the famous isolated ivory tower, but in interaction with its environment. Pure thought, of course, the creative process, knowledge as such might have little social relevance, is neither national nor international. Research, however, is quick to transcend its bounds of utter intellectuality and individuality. Research takes place within a social context. It is conducted by individuals, but it occurs on the basis of knowledge that has been acquired and passed on by others. Science is communication, it needs collaboration and discussions with others. Knowledge transfer, competition, application of research results, also the contrary: the dissemination restrictions by patents or for military reasons reveal the societal context of research. Research as a social phenomenon is not only exchanging information with its environment, it also needs encouragement, support and protection by the society. Research, at least research in the 20th century, depends on favorable institutional structurs, on appropriate labs, big facilities, a good educational system etc., i.e. it depends on access to ressources and research activities need legal boundary conditions

4

Globalization of Research

and arbitration in case of conflict with the interests, goods, values of individuals or of the general public surrounding the research activity. With respect to these interactions of research with its social environment it is only natural and justified to distinguish between national and international and to discuss the consequences of shifting responsibilities and adressing expectations from one community to another. This is not a theoretical question and we are not free to choose one or the other option. The trend to more internationalization and globalization is irreversible, is a fact. It is obvious for industry as a result of hardly controllable market forces. Globalization of industry research, the subject of the next presentation, is confirmed by relevant statistics (of Physics World 10196: "Germany looks abroad Since 1990, German companies have increased their investments in R&D by 50% - in overseas R&D, that is. But domestic investment has declined rapidly from 2.1% of the gross domestic product in 1989 to 1.5% in 1995. Over the last five years, more than 30000 jobs in R&D have been cut or moved abroad, and about 32000 scientists and engineers are unemployed. 15% of R&D spending by German companies now goes overseas, most of it to the US." But coming back to the field of non application oriented, of curiosity driven research it has always been understood, that it is only quality of results, that it is only the world wide recognition of the validity of results that count. 0

0

In so far so-called fundamental research has always been international and global. The international and global character of fundamental research is not controversial, and when I have been invited by Professor Aldinger to talk to you this morning on Research between nationality and internationality he wanted me to report on a pertinent discussion in Germany which centers around the question wether the reality, the way in which nationality and internationality of research correlate today is the best reality possible.The Max-Planck-Gesellschaft dedicated two 2 .5 days long colloquia here at Ringberg Castle to this issue. In 1995 MPG invited representatives of its most important partners in the very pluralistic landscape of the German system of research institutions and research supporting agencies: representatives of the Hermann von Helmholtz-Gemeinschaft deutscher Forschungszentren HGF the association of the big national labs working in the field of new energies, health, environment etc.,

Research between Nationaliry and Internationality

5

managers of the Fraunhofer Gesellschaft FhG, which concentrates on applied research and technology transfer, the president of the rectors' conference of the German universities, the Chairman of the Wissenschaftsrat, a forum for the discussion of science policy questions among politicians, administrators and scientitsts and also a platform for the adjustment of interests among the federal gorvernment and the 16 Laender governments which according to our constitution share the political responsibility for research promotion and the educational system. As part and a very important part of internationalization of research happens through the exchange of people (students, postdocs, guest scientists), 0 the Alexander von Humboldt-Foundation, AvH, 0 and the Deutsche Akademische Austauschdienst, DAAD, the German student exchange organizations, were given opportunity to present their views. And last not least our ,national science foundation'' the Forschungsgemeinschaft, DFG, actively participated in the discussion.

Deutsche

MPG and all these institutions presented their experiences and views and participated in this exercise to gain more consciousness of the consequences of this permanent and accelerated evolution to more internationality in research and education. 0 0 0

What are the advantages? What are the disadvantages of this development? How should the scientific system react or rather not only react, but try to participate actively in coping with the challenge of inevitably growing internationalization?

Origin for the actuality of this discussion in our country and in Europe has been the Maastricht treaty, which stipulates new competences for the Brussels Commission for resesarch issues and science policy questions.

I cannot deny, that the concern of the German research organizations is consciously or rather unconsciously part of the general feeling of uncertainty with respect to the consequences of the politically decided, partly welcomed, partly feared, partly supported and partly resented evolution to more European unity. There is a general distrust about the growing bureaucracy in Brussels, there is fear that too much Deutschmark becomes European ECU or EURO and there is serious concern that in distributing the money polititcal criteria like juste retour", ,cohesion" etc. replace criteria of quality and competition. There is a also general complaint about too little transparency in procedures, lack of impartial reliable peer review. In short: the research policy of the European Community is percepted as having too much top down approach rather than bottom up inspiration.

6


In this respect the German scientific community is particularly sensible because we realize that - compared to the situation in most countries around us - we are still in a very favorable position. In the context of the world wide and again and again upcoming discussion of the eternal" question concerning an appropriate balance between application oriented or curiousity driven research the German system allows still much room for the latter element, considering fundamental research and the definition of its programmes by the scientists themselves as a long term investment and an indispensable condition for real innovations not for tomorrow but for the time after tomorrow. This situation is, of course - also in our country - not undisputed and we see the risk that the unequivocal orientation of European research policy toward economic targets might influence and strengthen corresponding tendencies in Germany. As we know that our neighbours in Europe and practically all our essential partners in the world have the same or rather similar problems, MPG decided to have a second colloquium in order to include in this discussion experience from abroad. In 1996 MPG invited its foreign partners, not only from Europe, like 0 Guy Aubert and Francois Kourilsky, president and past-president of the CNRS in France, 0 Hubert Curien, former French minster of science and member of the MPG senate, 0 Richard Brook from the British Research Councils (by the way, for some years in the eighties director in this institute MPI fuer Metallforschung in Stuttgart), 0 Jan Borgman, chairman of the European Science and Technology Assemly (ESTA), but also 0 Neal Lane from the U.S. National Science Foundation (NSF) and 0 Minoru Oda from Japan. All of our illustrious guests were aware of the growing globalization and they discussed with scientists and managers of the German institutions, which had met the year before, many aspects of this situtation: 0 advantages and disadvantages, 0 expectations and risks, 0 obstacles amd supporting mechanisms, 0 national experience and future trends, 0 set backs and challenges of the internationalization of research. It would go beyond the scope of an introductory presentation if I would try to report on this exchange of ideas, hopes, fears and experiences in detail. The proceedings of the meeting will be published soon and can be made available to anybody of you who is interested. But I'll try to give a short resumee in saying that from the two aims of these colloquia: 0 to reach a common understanding how nationality and internationality of research interact 0 and to find a science oriented common strategy to optimize this relationship,

Research between Nationality and Internationality

7

only the first has been attained. Diagnosis is always easier to agree upon than on therapy. During these discussions at Ringberg in 1995 and 1996 nobody expressed the faintest doubt, that methods and contents of research do not possess a national character - that they are international. There was, also, a general understanding that it is the researcher's competence and not his or her nationality which counts. Internationality is in this respect the only possible and undispensable boundary condition for permanent competition, and it is internationality that provides the only valid standards and criteria to discover and guarantee quality. And yet, in spite of this, it is in the correlation between research and society, between science and political authority where we encounter a wide array of national elements. In this relationship between research and its social environment, the predominant representatives of society are still the national governments which - and here I quote Hans F. Zacher, the former president of the MPG and spiritus rector of these two Ringberg colloquia - there are the national governments, the national legislator which relate to research in three different ways: 0 either in that they make research possible 0 or put research into their service 0 or restrict research. Government does so by establishing the structurs of research, by entrusting the system with ressources and in cases where freedom of research conflicts with other interests, goods and values by defining its scope of freedom. It is also the state which in general defines the expectations concerning the return on these investments. Such expectations may be of a general cultural nature, conceived as contributions to more and better understanding and knowledge. But in general expectations take on very concrete forms directed towards the solution of specific problems, problems of economic; industrial, educational, in general problems of a socio-political nature.

In spite of the international character of science as such it is national predominance that coins the day to day reality of research. Features of a particular scientific system, 0 idiosyncrasies rooted in the culture and the civilization of a country, the geographic situtation, 0 common language, all these factors are apt to establish common bonds which enhance the significance of national frontiers to science. A state may promote or obstruct the communication and mobility of reseachers. And then there are unintentional obstacles deriving from national legislation e.g. in the dornaine of fiscality, of social security, last not least legislation on the protection of intellectual property, which - though they do not eliminate the mobility and the communication of researchers , nonetheless demand often a considerable compensation to be paid for.

8

Globalizationof Research

Let me repeat: As long as research is viewed as a quest for new and more knowledge, i.e. as much as research per se is international in nature, research as a social phenomenon is predominantly national. There was a general agreement without much discussion that more internationality in research as such would always bring more benefits than disadvantages. Internationality of research in this sense is rather an indispensable boundary condition than an obstacle for excellence in science. In economics, in the industrial world it has already become evident that globalization leads inevitably to a redistribution of work, of employment and of wealth. Industry has been realizing this for some time and even politicians have become aware that the notion of a national economy as an object of government control and influence has become more and more obsolete. Competition among global players reduces the role of governments to petitioners, offering all kinds of bribes, called favorable site conditions, in order to convince potential investors to choose their country for the creation of new employment. In this context the reputation of a country's research and educational system is valued as an essential asset and it is in this meaning of research as a social phenomenon where the growing internationalization leads to a considerable potential of conflict, worth to be watched and influenced by the scientific community on a national as well as on an international level. During the discussion of these questions in Germany, a discussion which has been called ,,Standortdiskussion", the contribution of science to improve the attractiveness of a German site for industry has concentrated very much on the aspect of a good research capacity as the potential for technology transfer. This question of more or less successful technology transfer is unfortunately and wrongly considered as the most obvious manifestation of return on investment for research as a social phenomenon.

I fully agree on the importance of this issue, but it is futile to reduce this discussion on the question wether there should be more or less application orientation in research and it is to my mind even contraproductive to try to cope with the problem by subordinating research to short term economic aims within programmes defined by politicians and industry without due and appropriate input from the scientists. If you reduce the meaning of the national research system for economic competitiveness on its success in techology transfer another considerable danger is that such an understanding will strengthen tendencies to more seclusion and counterproductive if not deadly isolation. We have witnessed extreme cases in totalitarian systems. But the danger also exists in our liberal democratic societies: I only

Research between Nationality and Internationality

9

remember of the information restriction policy of the Reagen administration at the time of the SDI effort. It was obvious in all presentations during the last Ringberg colloquium that in all countries which were represented and reported from and especially in the framework programmes of the European Community the top-down approach has gained considerable weight and has become the dominant factor for Research promotion. All national science communities were concerned and engaged to defend the diminishing room for science driven elements within those programmes.

As I mentioned before, the Ringberg colloquia have not led to a well formulated common strategy how to optimize the interaction between nationality and internationality in research and science. But I dare to point to some generally accepted and confirmed essentials, which constitute important elements for a strategy in a national as well as on an international level: 1. Globalization of science has always been considered by the scientific community as an indispensable stimulus for competition. 2. The best return in investment in research and science is excellence. Competition is a strong driving force; it is essential to be considered among the best worldwide. 3. Standards of quality only count if they are formulated and recognized on a global level. 4. Interaction between nationality and internationality in research as such poses no problem. It should be promoted by a continuous exchange of information and people bilaterally or within multilateral networks. 5 . Of general concern of the scientific community is and should be the growing tendency of government as well as of international institutions to overemphasize expectations of short term results and define expected results in place of unresolved problems as domaine of research. 6. The contribution of a successful and recognized national science community to the competitiveness of a national economy should not be judged on the merits of direct technology transfer successes only. The quality of research has its most important impact as integral part of the countries educational system. This simple truth, by the way, has been known and formulated ever since the beginning of the last century as the so-called Humboldt principle which has guided the formation of the university sytems in Germany and in many other countries around the world. The real return on investment in research consists of its contribution to create the potential for solving problems of the sponsoring environment, it consists in producing imaginative and creative people, in fostering their talents for the solution of societal problems of all kinds. To solve these problems innovations are needed in technology as well as in the fields of health, environment, agriculture, education, international relations etc. etc.

I think, we must realize that the trend to more and more globalization also includes a trend to more and more globalization of problems, which need global solutions. This insight can be taken as a favorable basis to reconcile the advantages and disadvantages

10


of a stronger internationalization of research in its social relevance for the still predominantly national environment of research promotion. At Ringberg Castle in May 1996 one of the last contributions of the debate came from Neil Lane, Director of the U.S.National Science Foundation NSF, under the title ,,Erasing the boundaries but retaining the identities." ,,By erasing the boundaries", he said ,,I mean recognising the perspective to define those unsolved problems larger than scientific disciplins that are common to all peoples that scientific cooperation can and should adress." And by retaining the identities I acknowledge a recognition that most things get done by local and regional collaborations which take advantage of the unique cultural qualities of all participants. By means of this delicate balancing we can hopefully define a global agenda for scientific research with clear overarching direction in which all of us have a role and a responsibility.

The Implications of Globalization for Research Funding by the Federal Ministry of Education, Science ,Research and Technology Dr. Jurgen Roemer-Mahler Bundesministerium fur Bildung und Forschung - BMBF Heinemannstralje 2,53170 Bonn, GERMANY

Globalization means worldwide interdependence, division of labour and exchange of information within the business and science communities. The globalization process requires both individuals and institutions to develop a high degree of internationality, that is the ability to cooperate across national and cultural borders and to set up new efficient organizations.

I. The German Government's position In this report to the Cabinet on globalization, the Federal Minister of Education, Science, Research and Technology stated on 2 September 1996 that within the close-knit world economic system of the 21st century only those countries will be able to hold their ground which, on account of their openness and the competence of their human resources, have become centres of information, communication and application of knowledge. Germany's high performance in the fields of science and research as well as its ability to develop and apply advanced technologies are its most important resources, which it should use wisely and strive to improve continuously. A new government policy with regard to science and technology should, therefore, not be confined to setting up an efficient science base and providing funds to encourage the development of new technologies, but must endeavour to ensure the necessary feedback between research, development and innovation as well as the integration of various policy fields which influence the innovation process.

12


Consequently, cooperation with scientific and business experts abroad is important for a successful science and technology policy. Exchanges of students in higher education and scientists is an important basis for such cooperation, think, it is alarming news to find that German universities, while still enjoying a worldwide reputation for scientific excellence, have lost some of their international attractiveness, with more students now going to other countries with highly renowned universities. As a country with high export figures and a science-based economy,

Germany must ensure international access to and the attractiveness of is education and research systems. The Federal Government therefore promotes cooperation with other countries. The German Government or the research establishments and project management agencies it supports are members of more than 30 multilateral research institutions, and have concluded bilateral agreements on scientific and technological cooperation with more than 50 countries. The number of cooperation agreements signed by German and foreign institutions of higher education by far exceed 6,000. In addition, there are numerous European, and other bodies, e.g. those set up by OECD, UNESCO and the Council of Europe, in which scientists as well as government officials jointly discuss important issues, moving forward to solve relevant problems. Together, all these bodies form a close network of cooperative relations in science, research and education, which has, 'in Europe in particular', resulted in a collaboration environment which may well be unique in the world in terms of intensity, strength and variety of subjects covered. 11. Strategies

With the slogan "Internationalization of German science, education and training", BME3F has decided to provide impulses to strengthen Germany's ability to cooperate and compete in the world against a background on increasing globalization.

The Implications of Globalizationfor Research Funding

13

Three priority goals should be mentioned here: - to develop excellence and international presence for German research centres, - to develop new approaches for national funding of technology - fair competition with the world's industrial nations and strong R & D partnerships. I should like to explain in some more detail the second focus, which concerns new approaches to national hnding of technology development. I will take as an example the project funding under the programme entitled "Advanced materials for key technologies of the 2 1st century".

At a time when companies, anxious to enhance their competitiveness and technological competence, are increasingly free to organize their business activities regardless of national borders or economic areas, government must make every effort to ensure that the tax money allocated to specific companies to strengthen their competitiveness, in particular as regards the hnding of research, technology and innovation will have an effect in its own territory. 1. Encouraging innovation in the whole chain from research to marketing The German Government is shifting the focus of ist research funding policy from aiming at the solution of detailed technological problems to encouraging the development of entire innovation chains without, however, losing sight of the potential for scientific and technological breakthroughs. Interacion between these two elements must be improved in German science. Innovative networks of enterprises and research centres must be encouraged. 2. Setting-up of competence networks - leading new markets Lead projects jointly developed by the scientific and business communities at the national and European levels will enable also small and medium-sized companies, which normally lack the means to develop and implement their own globalization strategies, to participate in creating a critical mass of highly qualified manpower and promising competence networks.

14


Lead projects will combine ambitious tasks with a specific application prospect while bringing together different disciplines and applications. Topics for such projects might include low-pollution transport, reducing C 0 2 emmissions from power plants, introducing more closed-cycle processes and creating multimedia infrastructure services. The realization of projects like these can help to create internationally leading new markets in Germany, thus providing strong impulses for employment, in the same way jobs were created, for instance, through the introduction of cellular phones in Europe with major participation by the German Government. 3. Developing the R & D potential of foreign companies through German und European technology funding policies The attitudes of governments vis-his R & D investments in their countries by foreign companies differ widely. Some countries, including the United Kingdom, Canada, and Singapore, try to attract foreign investors by offering them specific incentives. Like all other industrial nations, Germany has a strong interest in seeing an unrestricted flow of technical knowhow and research results while emphasizing the need for a fair give and take. In keeping with this attitude, BMBF, under its research and technology funding programmes, awards government fbnds also to foreign-owned companies operating in Germany. As a country open to the world - and unlike some other big industrial nations, Germany does not first require the foreign investors' country of origin to grant German enterprises the same opportunities. The only prerequisites to be met by such companies in order to receive fbnding are having an affiliate in Germany with a real R & D capacity, implementation to make use of the R & D results predominantly in Germany. This means that foreign companies, foreign affiliates of German enterprises and German domestic firms are, in fact, given equal treatment. Existing hnding opportunities and the rules governing fbnding indeed provide incentives for foreign companies to set up research and production plants in Germany. This may not have been general knowledge in the past.

The Implications of Globalization for Research Funding

15

Establishing private foreign-owned research centres in Germany is a different matter and will be authorized only in exceptional cases, where the German government considers such establishment to be in the national interest in view of a visible gain in competence expected to benefit the German innovation system. 111. Scientific and technological cooperation

Hence economic and scientific cooperation will be fostered on a longterm basis, and BMBF's concepts for collaboration with different world regions will be expanded. Scientific and technical cooperation as well as cooperation as well as cooperation in the field of education have for decades been an integral part of Germany's bilateral relations. BMBF and its predecessors have established a very large number of cooperative relations; the resulting experience is a sound basis for specific activities to strengthen the internationality of education, science, and research in Germany. In a view of the fact that scientific and technological cooperation projects often have major economic implications, the Federal Government will continue to intensify its scientific and technological relations with those nonEuropean countries with enterprises wishing to make investments in Europe in order to build production or research facilities in countries with a strong R & D base. In addition, it is hoped that closer cooperation in science, research and technology will open up opportunities for exporting German high technology and other advanced products. At present countries in South-East Asia and Latin America offer interesting prospects for extending cooperation. Many of the countries in these two regions have achieved dynamic economic development, which opens up possibilities of intensifying economic relations. The Federal Government has responded to these developments by drawing up coherent and coordinated policy concepts for ist relations with Asia and Latin America. BMBF's important contributions to these policies were published on 20 October 1995 in the form of a concept for scientific and technical cooperation with countries in the Asian-Pacific region. A concept for cooperation with Latin America is being elaborated. The common goal of these two concepts is to match Germany's industrial, economic and technological capability with the needs and resources of

16


countries in these two regions so that Germany can play an adequate role in developing their potentials. Other important challenges in this context include the opening-up of regional markets, progress towards technological and economic solutions as well as an increase in the common awareness of existing ecological risks. There is a high demand in these countries for quality training, which the Federal Government offers to meet through appropriate courses in Germany. This is regarded as absolutely necessary for the long-term fostering of relations in the field of culture and science as well as business. Research institutions in Germany: Excellence and international presence Speaking on the globalization of science, I also want to make a few remarks on research institutions in Germany, which, in particular in the field of materials science, are crystallization nuclei, as it were, fiom which international scientific cooperation can grow. In order to be able to offer R & D services on a worldwide scale to those companies competing on world markets, research establishments must continue to develop both existing and new international networks. While some research centres - e.g. Max Planck Institutes or national research laboratories - still have great international attractiveness, it is also true that top research centres in Germany, unlike their American counterparts such as MIT or Stanford, have made only inadequate efforts to gain a reputation for excellence with big multinational companies. Also with regard to enterprises located abroad, they do little in the way of advertising themselves, trying to win contracts, etc. In the long term they run the risk of getting only a small piece of the cake because of the concentration process going on worldwide. Even world-class research institutions should not ignore the need for marketing. BMBF last July presented its "Guidelines for the strategic orientation of the German science structure". It is one of the important goals of these guidelines to establish appropriate conditions for successful participation by German research institutions in competition and cooperation with

The Implications of Globalizationfor Research Funding

17

other centres of competence. The Federal Government will create the necessary prerequisites to enable German research institutions to markedly increase their flexibility and their ability to set up innovative research networks ar home and abroad. The hnding of Fraunhofer institutes in the United States is another strategic move made by the Fraunhofer Society with support from BMBF. By setting up research centres for laser technology, production engineering and computer graphics in America, the Fraunhofer Society will gain access to new clients while at the same time acquiring new skills for cooperation in projects funded by government or private agencies. By being present there, the Fraunhofer Society will be an important partner for German companies interested in investing in the United States. At the same time, the Fraunhofer Society demonstrates the high quality of the German scientific infrastructure. As a result of scientific progress or of emerging problems in society,new areas of research develop which require an institutional framework. In particular for highly complex and long-term problems, cooperation within a European, international or bilateral framework seems to be the obvious method in order to pool the best human resources available and to agree on a reasonable sharing of costs. There are a number of countries which have also expressed an interest in such cooperation. When founding new institutes, BMBF will endeavour to link up with partner countries which can be expected to make an equal, long-term commitment provided that the sharing of tasks and financial responsibilities is compatible with the research interests of the German science community and, not least, with our concern for the German laguage and culture being visible abroad.

EXCITEMENTS IN JAPANESE-GERMAN RESEARCH-COOPERATION Giinter Petzow Max-Planck-Institut f i r Metallforschung Heisenbergstralje 5, D-70569 Stuttgart, Germany

I gladly took up the suggestion of the organizers of this workshop to report in the opening session on my impressions of the collaboration with Japanese colleagues. My impressions were all very pleasant, stimulating, enjoyable and exciting; I have had not bad or negative experiences. I made up the title of the presentation spontaneously, without knowing that there is no equivalent for the word "excitement" in Japanese. At least, this is what a Japanese guest at our institute explained to me, who chose "joy" as a replacement (ya kudo suru nihon to doitsu no kyo do kenkyu); an acceptable alternative. The honest give and take", the requirement for every good collaboration, was always adhered to. I have never found confirmation of the general opinion that the Japanese are more the takers. Both sides are also making important contributions to this workshop. And we remember: it was S. Yajima in Sendai who in 1975 started systematic precursor research, resulting in an important technical product, the Nicalon fiber. The first Si-B-C-N fibers, made by Takamizawa, also originate from the Sendai school. And we haven't forgotten that it was Fumihiro Wakai who, through his pioneering investigations on the superplasticity of nanosized ceramics who provided the decisive stimulus for intensive studies of the grain boundary dynamics of ionic and covalent ceramics. I'

Of course it is extremely important to gain a personal confidence and familiarity between the partners besides the exchange of scientific results and theories. This is not as difficult as it looks at first glance. The common scientific striving is stronger than the different mentalities caused by the differing historical development of our two nations.

Excitements in Japanese-Gemn Research-Cooperation

1 .O

19

Professional Contacts

The first Japanese scientist I met was the late Uichi Hashimoto (Figure 1-l), an uncle of Ryutaro Hashimoto, now Prime Minister of Japan. He was a guest scientist at the Kaiser-Wilhelm-Institute for Iron Research, Dusseldorf, from 1926-1929 and visited Germany later on regularly, contributing immensely to the scientific and technical exchange between our countries. The Republic of Germany honored him with the first class order for merits and the German Society for Metals made him a honorary member and distinguished him with their highest award, the Heyn medal.

Figure 1-1. Professor Uichi Hashimoto (1 897-1 986)

20


Professor Hashimoto taught at the Tokyo Institute of Technology (TIT), one of the high leveled traditional, imperialistic universities of Japan. In 1956 he founded the National Research Institute for Metals (NRIM), and he was the first director for a long time. The NRIM, in which fundamental research of applied materials is persuade, has developed itself into a worldwide leading institute in the field of materials science, from which fruitful impulses have radiated. Uichi Hashimoto committed himself strongly to the exchange of scientists between the NRIM and our institute. One of his co-workers, who he sent to Stuttgart as a post-doc, was Dr. Kagushi Nii. Dr. Nii worked for more than a year scientifically at our institute, and integrated himself socially as well. Figure 1-2 shows him as a post-doc presenting a Samurai song on the occasion of an institute's party. In the background, second from left, Dr. Fritz Aldinger can be seen, who was also at the institute as post-doc at the time. Today, both Nii and Aldinger are directors of their institutes, the National Research Institute for Metals in Tsukuba and the Max-Planck-Institute for Metals Research in Stuttgart.

Figure 1-2. Dr. Kagushi Nii, presenting a Japanese song during a festive night on the occasion of 10th anniversary of the Max-Planck-Institute for Special Metals (1969)

Excitements in Japanese-German Research-Cooperation

21

The contacts made by Professor Uichi Hashimoto have deepened over the years, which is emphasized by the many mutual guest residencies, visits and joint events. At present, Siegfried Hofmann from our institute is working for three years in a directors position in NRIM and Manfred Riihle and I are members on various advisory boards of the Institute in Tsukuba. It is also worth mentioning that Dr. Yoshizo Inomata, the present director general of the National Institute for Research of Inorganic Materials (NIRIM), another renowned research institute in Tsukuba Science City, was also one in our institute as a guest scientist, as well as several of his colleagues. Similarly close connections as to these two institutes (NRIM and NIRIM) were and still are present to many other scientific institutes, universities and companies, from whom the Japan Fine Ceramic Center (JFCC), the Tokyo Institute of Technology (TIT) with its research institutes on the Negatsuta campus, the universities of Tokyo, Kyoto, Osaka and Sendai, and the Kanagawa Academy of Science and Technology (KAST) at Yokohama should be mentioned here. The collaboration with Germany is especially fostered according to tradition at the TIT. The Tokyo Institute of Technology has it's roots in 1881 as one of the eight imperial colleges of Japan. Among the founding members was Gottfried Wagner, who came from Gottingen and, in 1868 at the age of 37 years, came to Japan. He is said to be the father of the modern Japanese ceramics industry. And still today, most of the ceramic engineers and scientists come from the world-renowned TIT. Gottfried Wagner built Japan's first coal-fired kiln and introduced the use of cobalt oxide for coloring porcelain. He died in 1892 and was buried in the Aoyama Cemetery in Tokyo. The Gottried Wagner monument on the Ooyama campus of the Tokyo Institute of Technology and the Gottfried Wagner memorial lecture remind us of his fruitful achievements as a really exciting example of the Japanese-German cooperation in research and technology. None less than Albert Einstein held lectures at the TIT in 1922 as guest professor. He was at that time director of the institute of physics of the Kaiser-Wilhelm-Society, the predecessor of our Max-PlanckSociety, One of the colleagues who invited Einstein was Professor Kotaro Honda, who is regarded as the founder of Japanese physical

22


metallurgy and was the tutor of Uishi Hashimoto. The Japanese call him the "pioneer of the metal jungle". The magnetic steels developed by him gained the top values for magnetic performance for many years. The cartoon by Ippei Okamoto from the year 1922 shows how Einstein solved the "problem" of the unfamiliar seating in traditional Japanese restaurants (Figure 1-3). Without wanting to compare myself in the least with Einstein, I appreciated that solution when I was invited to lecture at the TIT, more than 50 years after him. In 1975 I was invited by the late Professor Shinrokii Saito (1 92 1*-1994-f), later president of TIT, to lecture on high temperature materials. Professor Saito became chairman of the Kanagawa Academy of Science and Technology (KAST) after his retirement from TIT, a well known center of excellence, promoting innovative research in close relation with industry; typical research topics are the ceramic methanol engine, ultimate molecular spectroscopy, giant magnetic materials, human protein and many others. KAST at Yokohama is an institution of the Kanagawa prefecture which is the partner prefecture of Baden-Wurttemberg and there are of course close Japanese-German relations.

Figure 1-3. Albert Einstein in a restaurant in Matsushima together with Japanese physicists (1922). Cartoon by Ippei Okamoto.

Excitements in Japanese-GermanResearch-Cooperation

23

One of my supervisors in TIT was Professor Shigeyuki Sbmiya who became the director of the laboratory for Hydrothermal Synthesis at TIT and dean at the Teikyo University of Science and Technology. He is one of those Japanese scientists who built many international bridges and contributed greatly to the Japanese-German research cooperation. He sent many colleagues and students to our institute and was often a guest in Stuttgart, also over periods of months. For his contributions to the Japanese-German science cooperation he was awarded with the cross of merits (Bundesverdienstkreuz), a high decoration of the Federal Republic of Germany awarded for service to the community. Many sensational developments in materials research come from the laboratories of Japanese companies. So it was inevitable that also with these placements an often close collaboration arose, to the mutual good. Although the application relevance naturally always played a roll, the tenor of the fundamental research was always decisive. Many of our Japanese guest scientists and PhD students are financially supported by Japanese companies and many of our students and colleagues are taken on by Japanese companies for study placements and work-experience and furthermore, in several cases, as real company employees.

2.0

Scientific Cooperation

There are hardly any facets of the numerous fields of research of our institute on which Japanese colleagues haven't worked. Be it investigations on structure and bonding in the solid state, constitution and thermodynamics of alloys, high-purity materials, surfaces and interfaces, solidification and phase transformation, microstructure mechanics, powder metallurgy, advanced ceramics and high-temperature materials, composite materials, micromaterials, high resolution microscopy, amorphous solids, liquid metals and others. A great number of highly regarded publications in renowned scientific journals and many good lectures have arisen from these joint studies. I didn't go to the trouble of determining them all; there are certainly many hundreds of examples of Japanese-German joint works. I would like to mention just three of them as examples:

24


The paper "Stability of the Perovskite Phase L a 0 3 (B = V, Cr, Mn, Fe, Co, Ni) in Reducing Atmosphere"' by Tetsuro Nakamura describes the result of his work done at our Powder Metallurgical Laboratory. His measurements are based on isothermal gravimetry measurements at varying values of oxygen partial pressure, with the aids of synthesis and X-ray powder diffraction measurements. The order of stability of these oxides found was not simply based on the order of ionic radii of the B ions, but was understandable only by considering the chemical reactions. The paper has found an astonishingly high resonance. Tetsuro Nakamura was as a guest scientist at our institute for more than one year, on leave from the Tokyo Institute of Technology where he was Professor and director of an institute on the Nagatsuta campus. Besides his great abilities as a scientist he was a very good singer and a choirmaster. We enjoyed hearing his well-sounding presentation of Japanese songs at many events and social gatherings. Another contribution of fundamental significance is the paper "Analysis of Particle Growth by Coalescence during Liquid Phase Sintering'" by Shigeaki Takajo. In the investigation, a statistical approach has been applied to particle coarsening during liquid phase sintering assuming direct particle coalescence as the basic growth mechanism instead of Ostwald ripening. The coalescence process, controlled by diffusion through the melt, results in an increase of the average particle size proportional to the cubed root of sinteiing time. After a short initial sintering interval, the particle size distribution approaches a unique normalized form, which is considerably broader then forms predicted by Ostwald ripening theory. The effect of preferred coalescence possibilities for definite particle size ranges and the effect of concurrent coalescence and Ostwald ripening are examined and discussed. Shigeaki Takajo was a PhD candidate on leave from the Kawasaki Steel Corporation. He is an excellent pianist and he and his wife, who is an ikebana teacher, have performed several times on German television. The last example of the Japanese-German cooperation is the contribution from Tadatomo Suga, "Nature and Structure of Metal-


25

Ceramic Interfaces'". This topic is highly relevant in many fields of technology as ambient and high-temperature structural application of ceramics often involves special ceramic interfaces. Tadatomo Suga used microstructural examinations and bond strength determinations to gain information on the nature and structure of the transition region between metal and ceramic components. A fracture mechanism testing concept was developed, which enables the characterization of the adherence by fracture energy and fracture resistance data. The selected examples of micrographs aim to demonstrate the versatility of metallographic methods in determining the structure and quality of metal to ceramic transitions. Tadatomo Suga carried out his work for his PhD thesis. Now, as a professor at Tokyo University, he is still involved with studies of the internal boundaries between metals and ceramics. Unusual and therefore especially remarkable is the achievement of Professor Gentaro Matsumura, who spent his sabbatical in the Powder Metallurgical Laboratory as a master class student of George Kuczynski, University of Notre Dame, USA, one of the founders of the sintering theory. Next to his lecturing and seminar program, he managed to write a two-volumed book on the history of Japan: "The Emperor's Islands - A Modern View of Japan's Hi~tory"~. This work is not only a history book, but conveys also an understanding for the Japanese mentality and the influence that confrontation with the westetn world has had on it. A book, written in English, that also provides a deep look into the soul of the Japanese.

3.0

Personal Relations

Scientific interests and goal setting stand, of course, at the center of the Japanese-Getman collaboration. Just as naturally, there also nearly always arises a more or less extensive interest for the way of life of the colleagues from the other country. The mutual understanding for the specialties of the other and the respect for the tradition and culture of his land deepen quickly. A real hobby arises, not seldom, out of an at first surface interest, with an intensive wish for a deeper understanding to the other country. I know German colleagues who are great experts on the art of forging Japanese swords, or the Noh-dramas, the classical theater of Japan, or on the national sport Sumowrestling, or on the temples or

26


gardens or other specialties that distinguish Japan. And I know Japanese colleagues who are excellent connoisseurs of Swabian wines or lovers of German folk music, others who follow the soccer league in Germany with enthusiasm or are great admirers of the Stuttgart ballet.

I was very impressed, for example, by the enthusiasm by which Franz Puckert, a former PhD student who stayed for a year's research placement at the University of Tokyo and worked there on hard metals involved himself with Japanese poetry, especially the short poems. The Japanese short poems, Tanka and Haiku, are little, non-rhyming poems of unique charm, considered as the shortest form of lyric poetry in world's literature. They describe the landscape, with the mountains, countryside and sea, wind and rain, plants and blossoms. Franz Puckert translated Tankas and Haikus into German and illustrated them. He used for his pictures only green tea (cha) and soja sauce (shogu) as "paints". His "tea-soja-pictures", the Cha-Sho-Ga, are motives of his observations in Japan. He compiled them in a small book, "Chanpon Cha-Sho-Ga" (Figure 3-1), in not particular order, as the word "Chanpon", freely translated, means as much as "muddle". This book certainly produces, in its own way, a small contribution to the strengthening of the friendship between Japanese and Germans.

Figure 3-1. Cover picture of Franz Pucken's brochure (1983)

Excitements in Japanese-Germun Research-Cooperation

27

Personally, the Japanese gardens impressed me greatly, next to certain other things. You may regard the Japanese garden as a medium to strengthen the contact between life and nature. Japanese gardening differs a great deal from that of Europe. The Japanese garden, once it is made, must become "nature" itself, although it is first laid out by human hands and intent. To each material used in Japanese gardening, such as water, stone and trees, is given a meaning essential to its existence. Japanese gardens are gardens of silence, refuges in which one can draw back to himself. Without the supervisory hand of the gardener, the garden would run wild and loose its character. The art of garden care is to attain the highest form of naturalness. I liked to use the opportunity during my stays in Japan to spend time in the Japanese gardens and to enjoy myself in the peace and harmony between the plants, stones and water. And my enjoyment of this became so great that one day I began to convert my rather unimaginative lawn and flower garden behind my house into a Japanese garden, for which the art of gardening of the Middle Ages (the Kamakura and Muromachi periods) was my guideline.

And last but not least, the not seldom occurring marriages between our Japanese and German colleagues should be mentioned, certainly the most stable bridges between our borders.

4.0

Concluding Remarks

The Japanese-German research cooperation has a long-lasting tradition and its roots are go back generations before us. It has grown to be mutually highly valued, despite the many differences in mentality and way of thinking. The scientific collaboration of our institute with Japanese partners was without exception successful and enjoyable. More than two hundred guest scientists from Japan have worked at the MaxPlanck-Institut fur Metallforschung since its foundation, and have delivered valuable contributions to the various projects of our subject areas. They were all representatives of the high level and the excellence of Japanese science and technology.

28


It must be mentioned that those Japanese-German cooperation exist also between nearly all seventy institutes of the Max-PlanckSociety and relevant Japanese partners, not only with our institute. They are more or less similar organized and pleasant in their effectiveness. These intensive and fruitful cooperation have experienced a very high distinction. During his state visit to Germany in 1993 the Japanese Emperor gave a reception for about twenty scientists from our society at the Bavarian Academy and discussed with them problems of different scientific fields (Figure 4-1). We are therefore very pleased about the intensive cooperation we have with our colleagues from Japan. It is our sincere wish that the bridge of understanding and friendship will continue. All of us are well aware the creative research is based on interdisciplinary and internationality. Now as before, words of Meiji-Tenno are effective; about a hundred years ago he wrote:

"In my garden indigenous and exotic plants are side by side; they all grow up jointly". In full accordance with these words I wish for continued successful and exciting Japanese-German research cooperation.


Figure 4-1. The Heisi-Tenno (Kaiser Akihito) discussing with scientists of the Max-Planck-Society during his visit to Munich on September 18, 1993.

References 1. 2. 3.

4.

T. Nakamura, G . Petzow, L. J. Gauckler, Mat. Res. Bull., 1979, 14,649-659 S. Takajo, W. A. Kaysser, G . Petzow, Actametall., 1984, 32, 107-113 G . Petzow, T. Suga, G . Bsner, M. Turwitt in Sintered MetalCeramic Composites (Ed. G.S. Upadtryaya), Elsevier Science Publishers B.V., Amsterdam, 1984 G. Matsumura, The Emperor's Islands - The Story of Japan, Lotus Press, Tokyo, 1977

29


11. Keynote Lectures


Joachim Bill and Fritz Aldinger Max-Planck-Institut fur Metallforschung and Universitat Stuttgart, Institut f i r NichtmetallischeAnorganische Materialien, PulvermetallurgischesLaboratorium, Heisenbergstr. 5 , D-70569 Stuttgart, Germany

Abstract Novel ceramic materials and microstructures can be built up from molecular units (precursors). The transformation of these precursors into the corresponding ceramics can be achieved by solid state thermolysis (SST) or by the deposition of the molecular units via the vapor (CVD, Chemical Vapor Deposition) or liquid (CLD, Chemical Liquid Deposition) state. According to this concept amorphous ceramics that exhibit unique high temperature properties can be obtained. The in-situ crystallization of these amorphous solids permits the preparation of nanocrystalline materials by a completely powder-free process. The structure and composition of the grain boundaries that originate from these crystallization processes strongly depend on the molecular structure of the initially used precursors. As a consequence, the precursor route can be applied for the architecture of grain boundaries.

1 Introduction The structure and the properties of advanced ceramic materials are determined by characteristics of different size which extend from the architecture on atomic scale via the arrangement of the microstructure to the macroscopic dimensions of the ceramic component:

Fig. 1-1. Architecture of ceramic materials by powder technology and from molecular units (precursor route)

34

Keynote Lectures

Thermolysis SST

Deposltlon CVD

Deposition CLD

One possibility is the transformation of molecular precursors into the corresponding ceramic materials via solid state thermolysis (SST) [ I , 21. Precursors which are appreciably volatile can be deposited via the vapor phase which can be used for instance for the preparation of superhard materials like c-BN [3]. In analogy to this Chemical Vapor Deposition (CVD) technique ceramics can also be obtained by the deposition of precursors from the liquid state. This route can be designated as Chemical Liquid Deposition (CLD). The inspiration for this route is the synthesis of inorganic materials in biological systems. This topic is treated in this volume in [4]. SST processes are applied for the preparation of non-oxide as well as for oxide ceramics. Examples for the synthesis of oxide-based materials by this route are described in [S]. This paper describes the preparation of non-oxide covalent ceramics from preceramic compounds via solid state thermolysis and provides a survey about this interdisciplinary field of research based on chemistry and materials science.

Precursor-Derived Covalent Ceramics

35

2 Solid State Thermolysis The concept of solid state thermolysis of precursors for the preparation of covalent ceramics was initially suggested by Popper [ 11 who mentions in a research report for the period between 1955 and 1966 the preparation of silicon nitride-based ceramic monoliths by shaping of polysilazanes and subsequent thermolysis. Moreover, the preparation of precursor-derived aluminum nitride is described. However, it took another 20 years till the potential of precursors for material science was recognized and Veerbeck and Winter from Bayer Company in Germany [MI and Yajima and co-workers in Japan [9-111 prepared silicon nitride- and silicon carbidebased ceramic fibers. The flow diagram in Fig. 2-1 describes the preparation of ceramics from precursors by SST and it is one main aim of the contributions summarized in this volume to provide an overview about materials synthesis, processing as well as about the techniques of characterization which are necessary to carry out and to understand this whole process.

Fig. 2-1. Preparation of ceramics by solid state thermolysis of preceramic compounds

At first, polymer precursors are synthesized from monomer units. The precursors are then transformed into amorphous covalent ceramics by solid state thermolysis. If the temperature is raised further the amorphous materials convert into the corresponding crystalline ceramics. In order to correlate the structure and the properties of the amorphous and crystalline ceramic with the molecular structure of the initially used

36

Keynote Lectures

precursors research has to be focused on the reaction mechanisms which occur during the synthesis and the cross-linking of the polymer precursors as well as on the ceramization mechanisms during the thermolysis step. Additionally, the kinetics of crystallization and the diffusion mechanisms connected with the transformation of the amorphous into the crystalline state as well as the resulting microstructures have to be investigated. The main techniques for the characterization of these steps are listed on the left side of Fig. 2- 1. With respect to the synthesis and thermolysis step spectroscopic techniques have to be applied [12, 131. The rheology of the precursors which can be directly influenced by the synthesis and the cross-linking of the polymers is of great importance with respect to precursor processing to ceramic products like ceramic fibers [14, 151. Additionally, thermal analysis is needed to characterize the ceramization of the precursors and the high temperature [ 16, 171 and oxidation [ 181 behavior of the resulting ceramics. Moreover, electron microscopy [ 191, diffraction [20, 2 13 and spectroscopic techniques [ 12, 13,221 can be applied for the investigation of the obtained ceramics. During the last years great efforts have been made to synthesize precursors for nonoxide ceramic systems like AIN [23,24], W2C,Ti,N,C, [25] and TiN [26] as well as for TiN/BN and AlN/BN composite materials [27]. Additionally, several precursors for the synthesis of hexagonal BN and ternary boron carbide nitrides have been developed [28, 291. Boron carbide nitrides are graphite-like materials which show semiconducting electrical properties as well as high thermoshock an chemical resistance. They can be synthesized by thermolysis of amine borane-derived polymers [30]: \

,

C--8

/

-B

\

c - HbT,'p /

C-N

/

H

c-c,

110°C

\

-CH, -Ha

'7

/

C-C

1

/

C-C

\

-C

\

\

\

C-C

\

I

N-C

/

\

I

C-B,

\

C-

I

C-B

B-N

/

\

/

/

\

c-c \

I

c-

C--8

8-N \

N-

\

/

C-C

\

/

N-

-C, C-C, /N-B, /C-N, ,B-C, ,N-

-B

\

c-c,

/

B-C

\

I

,c-c\

C--8

\

,c-

Beside this research on silicon-free systems research also focus on the synthesis of silicon-containing carbide- and nitride-based ceramic systems. This topic will be treated in the next chapters in more detail.

Precursor-Derived Covalent Ceramics

37

2.1 Organosilicon Polymers In Fig. 2-2 silicon-containing polymers for the preparation of silicon nitride- and carbide-based ceramics are shown. These polymers can be built up from Si-, C-, and Ncontaining monomer units. Additionally , polymers which contain hrther elements like B, P ,Ti or A1 can be synthesized. The precursors shown in Fig. 2-2 can be derived from three main polymer systems, namely polysilanes, polysilazanes and polysilylcarbodiimides. In the following the main principles of the precursor route are described. Polysilazanes and polysilylcarbodiimides serve as examples. Polysilanes are also discussed in this volume [ 14, 151.

[-@@In Polysilane (M)

1-8-1,

[-@@In

Polysilane

Polycarbosilane

Monomer

r-@@0@, Polysilazane

Polycarbosllazane

Units

r-0-001, Polysilasilazane

r-@

8

N 1, Polysilazane (M) (M=B, P, Ti, A1...)

@ -[IPolysllylcarbodiimide

Fig. 2-2. Formation of organosilicon polymers from monomer units

Keynote Lectures

38

2.1.1 Polysilazanes

Ternary Si-C-N ceramics can be obtained by the thermolysis of polysilazanes [31]. These preceramic polymers with backbones containing alternating silicon and nitrogen atoms can be synthesized by the reaction of organochlorosilaneswith ammonia: r

1

The introduction of further elements into polysilazanes can be achieved by two ways which is shown in Fig. 2-3 with respect to the preparation of boron-containing polysilazanes [32, 171.

I

/B3.1 g/cc, tensile strengths of 3.0-3.5 GPa and elastic moduli of 400-470GPa.7.9.11 These values are the same as those found for dense, pure, monolithic Sic produced via standard ceramic processing methods. Control of microstructure and densification. As mentioned above, fully dense ceramic products are required to obtain superior mechanical properties. Unfortunately, conversion of a precursor to phase pure, fully dense materials is not always easy. For example, precursor-derived phase pure S i c will crystallize and undergo grain growth on heating to 1800°C; however, grain

Synthesis and Processing of Sic Based Materials Using Polymethylsilane

63

growth occurs without coincidental sintering (densification) leading to porous material^.^ This problem can be solved by adding small amounts of boron (0.30.5 wt. %) which promotes densification without much grain growth. Thus, boron must be incorporated in the precursor synthesis or processing strategies to achieve the correct microstructure." In this instance, microstructure requirements drive precursor design. Cost. Perhaps the most critical part of precursor processing is cost. Expensive synthetic and/or processing routes reduce the utility of a specific precursor. In ceramic fiber processing, the fiber spinning and pyrolysis steps are typically more costly than the chemistry. Thus, the precursor chemistry is typically not cost limiting. The above criteria provide a guide for developing a wide variety of precursors for fiber processing as well as for other applications. In the following sections, we will show how they have been used to develop PMS based S i c fibers. Results and Discussion Svnthesis of Spinnable PMS Our initial work in this area is based on the discovery of the Harrod group that it is possible to polymerize methylsilane, MeSiH,, in one step in cyclohexene to produce a PMS polymer:'* MeSiH, + cyclohexene V w 2 > -[MeSiH],,,[MeSi],,,,- +cyclohexane The actual polymer structure that can be proposed probably looks something like:

MeSiH, is a commercial product, and the PMS is spinnable simply by concentrating the reaction solution. However, the as-spun fibers melt and pyrolysis gives a Si rich ceramic, despite the initial 1: 1 Si:C ratio.13.14The polymer must be modified to obtain pure Sic (see below). In addition, the polymer is highly flammable and must be spun in a drybox. Figure 1 shows just how air-sensitive modified (see below) PMS is:

64

Ceramics from Organoelement Compounds

'O0O

200

400

600 800 1000 1200 1400 Time (min)

Figure 1. Isothermal TGA of 10 wt. % TVS modified-PMS fibers (30°C/air/24 h).

To overcome the melting problem and the off-stoichiometry, tetravinyl silane (TVS) was added to the cyclohexane solution in reaction (1).The zirconium catalyst then promotes hydrosilylation of the TVS vinyl groups with PMS to form a star-branched polymer. The intent was to: (1) increase molecular weight to reduce the likelihood of melting during processing; (2) enhance spinnability by providing better viscoelastic properties though improved chain entanglement, and (3) add additional carbon to offset the excess Si formed during pyrolysis of simple PMS. This was found to be quite successful and it was determined that adding between 5 and 20 wt. % TVS to reaction (1) gave a TVS-PMS polymer that exhibited better spinnability. Moreover, we found that it was possible to control the stoichiometry of the resulting pyrolyzed material such that chemical composition could be varied from carbon rich to silicon rich. This is important because in related work on making powder particulate composites (e.g. with Sic, Tic, B4C,etc.),lSwe determined that some oxide coatings need to be gettered by the excess carbon to give clean interfaces between the TVS-PMS derived Sic and the particulates. Thus, 5 wt. % TVS gives a ceramic materials that is 5 % rich in Si; a 10 wt. 9% TVS gives a ceramic materials that is stoichiometric S i c and 15 wt. % TVS gives a ceramic materials that is 5 % rich in C. A proposed structure for TVS-PMS is:

Synthesis and Processing of S i c Based Materials Using Polymethylsilane

65

Surprisingly, heating these materials to selected temperatures in Ar shows no effect of chemical composition on grain size: 50

2 40 E

W

Q)

N

30 . I

3 20 .-*

Q :)

c,

8w 10 -

0; 0

I

I

I

I

5

10

15

20

TVS wt % Figure 2. Grain size as a function of added TVS and pyrolysis temperatures.

66


We believe that the reason for this is that no oxygen is present that can form liquid phases that would promote grain growth. Fibers of the TVS-PMS 10 wt. % polymer were extruded using the simple system shown in Figure 3. The as-spun fibers are 20-40 pm in diameter and irregularly shaped because of uneven solvent drying. The thinnest fibers can be heated at up to 20"C/min to any temperature desired without deformation. The only problem is that they are mildly air sensitive until pyrolyzed above = 800°C. Once we had succeeded in producing fibers that were heated to 1000°C, we then did post-pyrolysis heat treatments to 1800°C to improve their densities. It was discovered at this point that these stoichiometrically pure Sic fibers do not sinter to full density as expected. Instead, they sinter without densifying. This results in phase pure S i c fibers with visible pores as shown below in Figure 4. Argon Pressure Supply

t

Swagelok 62.5 mm Inlet Brass Tube, 22 mm OD..

0

Spinnable Polymer

7-

-With

Standard Swagelok Cap 140 p m Dia. Orifice

Precursor fiber Figure 3. Simple pressure extruder used to produce TVS-PMS polymer fibers.

Figure 4. TVS-PMS fiber heated to 1800°C in Ar.


67

Based on literature studies, which demonstrate that the addition of small amounts of boron aid sintering and simultaneously reduce the grain size,16we redesigned our precursor to include boron. The method of synthesis was simply to add TVS together with sufficient boron (0.1 to 0.3 wt.%) as H,BSMe, into reaction (1) so that one vinyl group on each TVS was hydroborated. It was assumed that the remaining vinyl groups on the TVS-B species were catalytically hydrosilylated simultaneously. The resulting polymer was even easier to spin than TVS-PMS and was easily pyrolyzed. We propose a model of the resulting B-TVS-PMS as shown below (Where P = each PMS monomer):

Q

Figure 5. Each P represents a PMS unit with an approximate MW = 1200 Da.

The addition of the boron creates what is essentially a hyperbranched polymer. This polymer offers ceramic yields that range from 78-85 wt %. The 10 wt. % B-TVS-PMS polymer was very easy to spin and fibers that were obtained could be sintered at the same temperatures as before with essentially no change in properties except that they became fully dense without pores. Figure 6 shows an example of a fiber pyrolyzed to 1800°C in air. The densities of these fibers were = 3.1 g/cc which is close to the01-y.~~~~~ Furthermore, the grain sizes after heating to 1800OC were 0.1-0.3 pm which is quite good for creep resistance. Finally, these fibers are much more stable than Nicalon S i c fibers. We have conducted preliminary bend strain studies on these fibers and find that they offer mechanical properties that appear to be very similar to those expected for fully dense, monolithic Sic, as suggested by the Figure 7 data.

68

Ceramicsfrom Organoelement Compounds

Figure 6. B-TVS-PMS fiber heated to 1800°C in Ar.

5 .O

0 Tensile strength of 0

n

4.0-

Nicalon fibers

0 Bending strength of

SCS-6 fibers

0

r3 v

0

A

22 3.0-

3;

'0

3 2.0-

3

6 a

1.0-

&

0

8

8 !

0

: a 0 0

0

0

0 0

0 0

0

m

!

8

8

0

0

0

i 0

0

0

0 0.0

!

800

I

1000

I

I

I

1200 1400 1600 Temperature ("C)

I

1800

2 100

Figure 7. B-TVS-PMS fiber bend strengths (2OOC) heated to selected temperatures.

Synthesis and Processing of S i c Based Materials Using Polymethylsilane

69

In principle, the creation of B-TVS-PMS, solved all of the problems with producing phase pure, Sic fibers with properties close to those expected based on the properties of monolithic Sic. However, we still had a problem with considerable air sensitivity. We have now solved this problem by exploring the original method whereby PMS was made, by dehaloc~upling.~~*'~~'~ We have now found a way to produce a highly branchedlpolycyclic PMS exemplified by the compound shown below, for which we have mass spectral evidence:Iy

-Si

I

Si'

H

H'

\

\

H' I

Si-SiIH

\

C19H70Si19

832.4 Da SiC&:Si€J = 4.4:1 This monomer can be hydrosilyatively coupled to preformed TVS-B to form the reverse engineered precursor, PMS-TVS-B. This polymer has the same attributes as the original polymer but is not particularly air sensitive as shown in Figure 8, which indicates that the air sensitivity has been reduced by an order of magnitude. Finally, because this material shows so little sensitivity to air, we can analyze the formation of the hyperbranched polymer by gel permeation chromatography (GPC). Figure 9, shows the GPC of the above new PMS and that of PMS-TVS-B which shows the addition of PMS to the multiply branched TVS-B. In principle, TVS-B has nine vinyl groups that can be hydrosilylated by PMS. In practice not all are hydrosilylated and not all PMS oligomers hydrosilylate once--hence the polymodal behavior.

70


2

1

0

4

3

5

Time (hr) Figure 8. Isothermal TGA of 10 wt. % PMS-TVS-B polymer in static, ambient air.

850

3

Mn

----

PMS PMS-TVS-B

800 2000

PDI 1.5 7.1

\

I \ \

9

52,000

'*

\ 1 9

\ \

l " " I '

2

2.5

"

3

' I '

" ' I '

" ' I "

3.5 4 4.5 log mlecular weight

" I

-. -' "

5

Figure 9. GPC Trace of Reverse Engineered PMS and PMS-TVS-B (reference 19).

I

5.5


71

These results suggest that we have solved all of the problems related to making phase pure, processable precursors and no work remains to be done. In fact, we (and Bill et al) have recently determined that crystalline Sic can from polymer precursors. This discovery seed growth of epitaxial SiC15.19.2n,21 suggests that it may now be possible to seed the growth of single S i c crystals which may be of considerable use for electronic applications. Thus, rather than an end to a long research effort, we may be at the beginning of a still new and important area of research.

Acknowledgments. We would like to thank Dr. Martha Fletcher of the U.S. Army for continuous and enthusiastic support of our Sic programs. We would also like to thank Lockheed, Martin Marrieta for generous support of part of the work described here. References

1. F. W. Ainger and J. M. Herbert, “The preparation of phosphorusnitrogen compounds as non-porous solids,” in SDecial Ceramics, E. P. Popper Ed.; N.Y.; Academic Press (1960) p.168. 2. P.G. Chantrell and E.P. Popper, in Special Ceramics, E. P. Popper, ed.; N.Y.; Academic Press (1964) pp. 87-102. 3. a. B.J. Aylett, J.M. Campbell, J. Chem. SOC.A. (1969) 1910-16. b. B.J. Aylett, J.M. Campbell, SDecial Ceramics, (1972) 5, 71 and references therein. 4. a. R.R. Wills, R.A. Mark, S.A. Mukherjee, Cer. Bull., (1983) 62, 904. b. R. R. Rice, Am. Cer. SOC. Bull., (1983) 62, 889. c. K. J. Wynne and R. W. Rice, Ann. Rev. Mater. Sci., (1984) 14, 297. d. G. Pouskouleli, Ceram. Int. (1989) 15,213-29. e. M. Peukert, T. Vaahs, M. Briick, Adv. Mater. (1990) 2, 398-404. f. J. Lipowitz, J. Inorg. Organomet. Poly., (1991) 1, 277. 5. a. Better Ceramics Through Chemistry 1, Mat. Res. SOC.Symp. Proc., C. J. Brinker, D.E. Clark, D.R. Ulrich Eds., Elsevier Sci. Pub., New York (1984). b. Better Ceramics Through Chemis- 11, Mat. Res. SOC. Symp. Proc., C.J. Brinker, D.E. Clark, D.R. Ulrich eds., Mat. Res. SOC., Pittsburgh, PA (1986) vol 73. c. Chemistry 111, Mat. Res. SOC.Symp. Proc., C.J. Brinker, D.E. Clark, D.R. Ulrich, eds., Mat. Res. SOC.,Pittsburgh, PA (1988) vol 121. 6. a. C. K. Narula, d , Marcel Dekker; N.Y., N.Y., 1995. b. Inorganic and Organometallic Polym., Am. Chem. SOC.Adv. Chem. Ser. Vol. 360, K. Wynne, M. Zeldin and H. Allcock, eds., (1988). 7. Transformation of Organometallics into Common and Exotic Materials: Design - and Activation, NATO AS1 Ser. E: Appl. Sci.-No. 141, R. M. Laine, ed.; Kluwer Publ., Dordrecht, 1988. 8. T. F. Cooke, “Inorganic fibers-a literature review”, J. Am. Cer. SOC. (1991) 74, 2959-78.

12


9. R.M. Laine, F. Babonneau, “preceramic polymer routes to Sic,” Chem. Mat. (1993) 5,260-279. 10. a. T. Wideman, E. Cortez, E.E. Remsen, G.A. Zank, P.J. Carrol, L.G. Sneddon, Chem. Mater. 1997 9, 2218-30. b. R. Riedel, W. Dressler, Ceramics Intemat. 1996 22, 233-9. c. R. Riedel, A. Kienzle, W. Dressler, L.M. Ruwisch, J. Bill, F. Aldinger, Nature 1996 382, 796-8. d. H.-P. Baldus, 0. Wagner, M. Jansen,Better Ceramics Through Chemistry VI. Mater. Res. SOC.Symp. Proc. vol. 271, Pittsburgh, PA., 1992, pp. 821-6. e. H.-P. Baldus, G. Passing Mater. Res. Suc. Symp. Proc. vol. 346, Mater. Res. SOC.Pittsburgh, PA., 1994, pp. 617-22. f. R. Riedel, A. Greiner, G. Miehe, W. Dressler, H. Fuess, J. Bill, F. Aldinger, Angew. Chem. Int. Ed. Engl. 1997 36 , 603-6. 11. R. M. Laine, A. Sellinger, in The Chemistry of Organic Silicon Compounds Vol. 2, Zvi Rappoport and Y. Apeloig, eds., in press. 12. a. Y. Mu, J.F. Harrod in Inorganic and Ormnometallic Oligomers and Polvmers, IUPAC 33d Symp. on Macromolecules, Harrod, J. F.; Laine, R. M. eds.; Kluwer Publ., Dordrecht, Netherlands, 1991, pp 23-36. b. T.D. Tilley, Acc. Chem. Res. 1993 26, 22-29. c. E. Hengge, M. Weinberger, Ch. Jammegg, J. Organomet. Chem. 1991 410,C1-4. 13. a. Z-F. Zhang, S. Scotto, R.M. Laine, Covalent Ceramics 11: Nonoxides, Mater. Res. SOC.Symp. Proc. in press. b. Z-F. Zhang, S. Scotto, R.M. Laine, Ceram. Eng. Sci. Proc.,1994 15, 152-61. 14. Z-F. Zhang, C.S. Scotto and R.M. Laine, submitted to Chem. Mater., 1998. 15. M. Nechanicky and R.M. Laine to be submitted for publication. 16. a. S. Prochazka, in Ceramics for High-Performance Applications, Edited by J. J. Burke, A. E. Gorum, R. N. Katz, pp. 239, Book Hill, Chestnut Hill, MA (1974). b. K.M. Friederich, R. L. Coble, J. Am. Ceram. Soc.1983 66 C-141-C-142. 17. a. D. Seyferth, T.G. Wood, H.J. Tracy, J.L. Robison, J. Am. Ceram. SOC.1992 75, 1300. b. D. Seyferth, H.J. Tracy, J. L. Robison, U.S. Patent No. 5,204,380, 1993. c. D. Seyferth, C.A. Sobon, J. Borm, New J. Chem. 1990 14,545-7. d. D. Seyferth, YF. Yu. in Design of New Materials,” D.L. Cocke, A. Clearfield, eds; Plenum Press, N.Y. 1987 pp. 79-94. 18. Y. Mu, R.M.Laine, J. F. Harrod, Appl. Organomet. Chem. 1994 8, 95- 100. 19. a.R.M. Laine, A. Sellinger, K.W. Chew, U.S. Patent pending. b. A. Sellinger, Ph.D. dissertation March, 1997. c.A. Sellinger and R.M. Laine, manuscript in preparation. 20. K.W. Chew, A. Sellinger, R.M. Laine, J. Am. Ceram. SOC.in press. 21. D. Heiman, T. Wagner, J. Bill, F. Aldinger, F.F. Lange, J. Mater. Res. 1997 12, 3099-101.

Silicon carbide fibers from highly reactive poly(methylchlorosi1ane)s G. Roewer*, H.-P. Martin**, R. Richter***, E. Miiller**, Freiberg University of Mining and Technology, *Dep. Inorg. Chem., **Dep. Ceram. Mat., ***Belchem GmbH Freiberg

1. Introduction

In the last 20 years intense research has been focussed onto the elaboration of Sic, Si3N4 or SiC/N based materials which are derived from organosilicon precursors [ 13. The polymer route is quite adaquate especially for the manufacture of fibers or films. The advantage of controlled viscosity of preceramic polymers includes the ability to prepare shaped ceramic products which are difficult to obtain by conventional powder processing methods. Metal containing silicon carbide or silicides in silicon carbide can be prepared by filling reactive siliconorganic polymers with metals or metal compounds as it was shown in [2, 31. This way offers an access to new materials with interesting outstanding properties. Polysilanes, polycarbo-silanes or even organopolysilanes containing multicarbon sequences should be generally regarded to as preceramic polymers for Sic manufacture. 2. Synthesis of preceramic polymers The requirements on polymer precursors for silicon carbide fibers are extremely high. Such polymers must possess a controllable rheology which is an essential property for spinnability from the melt. A latent reactivity which allows the polymer green fiber to be rendered infusible by crosslinking in a so called curing step and thus to maintain the fiber integrity in the pyrolysis step is strongly desired. The polymer has to designed in such a way to succeed a controllable pyrolytic degradation. Indeed the fulfilment of such criteria is mainly governed by the polymer architecture associated with the applied chemical synthesis method.

74

Ceramics from Org~noelementCrimpounds

R

X = CI, Br, triflate

A

lei

C

4 1

H3C -CI

LiAIH4

H-di-ii-H CH3 CH3 CAT, I

Jn

ti.:;? n

E 4

Figure 2-1 synthesis routes of polysilanes/carbosilanes [4], A- metal condensation, B- introductionof halogen, vinyl or ethinyl groups, C- linkage of organornonosilanes, D- disilane route, E- conversion of chlorornethyldisilanes,El-copolymerisation of E with olefine

Silicon Cm-hidr Fibers from Highlv Reacrive P~~l~(meth~vlchl~~rosiklne)s75

Some examples of published polymer methods are shown in Fig.2-1 [4]. Metal condensation causes an almost complete dehalogenation of the parent halosilanes or halosilanehalohydrocarbon mixtures. For this reason the as formed polysilanes or polycarbosilanes exhibit a very low reactivity at room temperature. Consequently the curing of shaped bodies is difficult to achieve. There are some strategies to increase the polymer reactivity. For instance halogen or vinyl or ethinyl groups can be introduced into the polymer after the metal condensation. This demands a second chemical treatment as shown in the top Fig. 2-1. Harrod's discovery [ 5 ] that organomonosilanes like phenylsilane PhSiH3 can be linked to the formation of siliconsilicon bonds in the coordination sphere of transition metal complexes leads imediately to more reactive polysilanes. Hengge et al. [ 6 ] have proved a general route for disilanes.They postulated a mechanism involving silylene intermediates. Our workgroup has turned to the metal free direct conversion of chloromethyldisilanes into oligo and polysilanes by electron donor catalysed disproportionation. As known such disilanes are byproducts of the industrial direct methylchlorosilane sythesis. The NMR- and GCMS investigations have shown that trisilane {C12CH3(SiCICH3)SiCH3Cl2} and the isotetrasilane {CH,Si(SiCHjC12)3}are formed directly on the surface of the catalyst. Furthermore penta-,hexa- and heptasilanes are identified after the reaction temperature was gone up to 175°C. The trisilane and isotetrasilane, which are free of the catalyst, undergo thermally induced branching reactions giving higher oligomers and methyltrichlorosilane CH3SiC1, at reaction temperatures in the region 160-175°C. By slight further increase of the temperature up to 180°C - 185OC level the branching is accompanied with crosslinking and liberation of methyltrichlorosilane. The thermogravimetric analysis coupled with mass spectrometry of methylchloropolysilane detects the evolution of methyltrichlorosilane, methane,hydrogen and hydrogenchloride.

16

Ceramics .fmm Orgonoelement Compounds

The total weight loss of the investigated polysilane is 80%. Evolution of hydrogenchloride clearly occurs in two stages from 200°C up to 300°C and from 300-450°C. CH,

29SiCP-MAS-NMR

'3C CP-MAS-NMR

Figure 2-2 solid state NMR-results of poly(methylch1oro)silanes

The chloromethyloligosilanes can be considered as building blocks of highly crosslinked methylchloropolysilane network, whose structures are unknown at present. Even less investigated was the crosslinking process itself up to now. By combining of I3C and 29Si classical Cross Polarization and Inversion Recovery Cross Polarization Magic Angle Spinning NMR technique experiments one can identify and quantify the various silicon atom sites which are present in these polymers (fig.2-2). Samples which result at the 180 "C level are characterized by a high percentage part of tertiary silicon atom group indicated at -64...-65 ppm in 29Si-NMR-spectrum.Silylene units were also identified corresponding to peaks at 14 ppm if situated in silicon chain and at 24 ppm in cycles. Finally terminal silylgroups are indicated at 35 ppm. A main part of tertiary silicon atoms as branching points origins obviously from condensation reactions involving both silylene and terminal silylgroups. As concluded from additional NMR spectra silylene units {-SiCH$l} are predominantly situated imediately neightboring tertiary silicon groups. The I3C CP-MAS-NMR spectra also useful to label the kind of created building blocks as well as to determine their concentration level. Detailed NMR-spectra analysis is already reported in [4].

Silicori Crirbide Fibers from Highly Recictive Pi)ly(niethql~hlorosil~~ne).s

I1

3. Pyrolysis of poly(methylchlorosilane)s at temperatures > 4OOOC

(-C H&Me&

50

0

-50

i-

-100

-150

(PPW Figure 3-1

29

Si CP MAS N M R results of samples after 450 OC treatment under argon

The increase of reaction temperature up to 450 OC causes considerable changes in silicon backbone of the polysilane which can be indicated as already discussed for Fig. 2-2 and which is clearly monitored by the 29SiNMR- experiments (fig. 3-1). New 29Si-resonancepeaks occur between 0 and -50 ppm chemical shift range corresponding to carbosilane units (CH~),(CH~),Si(Si)4-,.,. A new peak around -1 05 ppm also appears in samples treated above

300 "C due to formation of qarternary silane units Si(Si)4. For each temperature a detailed analysis of 29Si-Cross-Polarization and Inversion Recovery Cross Polarization spectra was done. The 180-450 "C temperature region can be devided into two main conversion steps: (1 ) 180 - 350 "C building up of the polysilane network dominantly proceeds (number of Si-C-

bonds remains constantly ,the number of Si-C1-bonds simultaneously diminishes with an increase of the number of the Si-Si-bonds ). (2)An interesting side reaction is the hydrogenchloride evolution which occurs in the low temperature region of 200-350 "C.

78

Ceramics front Orgonoelement Compounds

We suggest that in this range hydrogenchloride should result from Si-Cl and H-C-bonds.

This route gives rise to the formation of carbosilane units while Si-Si-bonds are preserved. Above 350 "C the polysilane to polycarbosilane transformation happens (number of Si-Sibonds decreases, number of Si-C-bonds increases, CI-content does not vary considerably). The Kumada rearrangement is known to occur in methyl containing polysilanes above 350 "C. But in the present case no silicon sites with Si-H-bonds were identified through IRCP experiments which should be very sensitive. One possible explanation is the immediate consumption of the as-formed Si-H-groups in subsequent reactions like dehydrocoupling or dehydrohalogenation reactions. Methane formation may also take place as shown already in TGIMS-data.

4. Fiber manufacture of poly(methylchlorosilane) For the spinning of the polymer melt it is interesting to examine whether olefines like styrene can be copolymerised with disilanes or intermediates in the catalysed disproportionation process (fig. 4- 1). Starting with a tetrachlorodimethyldisilane/styrene-mixturethe monosilane methyltrichloro-silane was the only gaseous product detected at reaction temperatures up to

220 "C. After two hours of reaction at 220 "C and subsequent cooling down to room temperature solid polymerised styrene containing methylchloropolysilanes were obtained. The yellowish polymers are meltable and soluble in common organic solvents such as toluen or tetrahydrofuran. As calculated from the NMR-peak-intensities the styrene part of the starting mixture is completely incorporated into the final polymer product. But both 29Siand I3CNMR-investigations were unsuccessful to identify bonds between silane and styrene moieties

till now. Nevertheless the as-obtained chlorine containing polymers exhibit very interesting properties as silicon carbide fiber precursors. Continuous polymer fibers were drawn from molten poly(methylchlorosi1ane)-co styrenes under argon atmosphere. The polymer was pressed through a multi-hole spinneret with 200 filaments by argon pessure. Once spun, the

Silicwi Curbide Fibers,frorn Highlv Reactive P ~ ~ l ~ ( m e t h v l c h l ~ ~ m s i l a n e ) . ~79

Figure 4-1 styrene incorporation into methylchloropolysilanes

polymer filaments were immediately stretched to diameters of 30-50 pm by winding on a bobbin placed in the glove box. The green fibers have to be rendered infusible before pyrolysis. In case of the Yajima polycarbosilane the reactivity of the Si-H-bonds formed by the Kumada-rearrangement is used for an oxygen curing step. Crosslinking proceeds via oxygen bridges. Recently electron beam radiation or y-ray treatment are applied under helium atmosphere to cure the polycarbosilane fibers [7]. The reactive poly(methylchlorosi1anes) can be crosslinked by means of gaseous ammonia or alkylamines leading to a superficial silazane network formation (fig. 4-2). The corresponding structure changes of the polymers after curing with ammonia were monitored by 29Si-CP-MAS-NMR-spectroscopy. Spectra exhibit new signals in the -20 to -40 ppm region which are derived from the formation of SiCN groups. After curing processs the preceramic fibers were converted into nitrogen containing silicon carbide fibers by thermal treatment under argon atmosphere resulting in oxygen free S i c fibers. Curing can also be achieved by short-time treatment with moisture containing air. Pyrolysis of such cured green fiber leads to oxygen containing ceramic fiber. 5. Properties of Sic-fibers derived from poly(methylchlorosi1ane)

As known the thermodynamical and kinetical stability of the fiber material is the crux of the matter concerning the mechanical fiber properties and thermal resistance. In case of YajimaNicalon fiber used oxygen curing gives rise to formation of non-stochiometric combined Si-

c- 0 ceramic. The oxygen reacts at around 1050

O C

giving SiO and CO. Fiber tensile strength

80

Cerrtmics,from Orpioelement Compouncl.,

CH3

CH3 I

is,

I -Si

- m,c1

+ CH3

,

CI

/Si

-CH3 \

-N. /Si -CH3

I

NH

I

-Si I 'CH3 /

CH3 /CH3 tJi /;i-m-

/ s'-/i 'CH,

N' iS,

-Si

/CH3

s'i /

/' S ,i

CI

/CH3

tJi ,si-N-I

/CH3

,N,

\

-CH3

\

Figure 4-2 curing of green fibers with gaseous ammonia

drops drastically with increasing temperature. The fibers resulting from poly(methylch1orosilane) after curing either with ammonia or with moisture and subsequent pyrolysis at 1200 "C exhibit a remarkable chemical stability. From 600 "C up to 1400 "C we observe during thermal treatment in air only a small weight loss of moisture cured fibers. Besides the chemical degradation the crystallization of the silicon carbide fibers during thermal treatment is a further reason for damage. It should be possible to move the crystallization tendency to higher temperatures in general by selective incorporation of foreign atoms, for instance nitrogen or boron, into the precursor. Furthermore by specific temperaturehime-regime during pyrolysis such a control of crystal nucleation and crystal growth should be achieved to create nanocrystals. The X-ray diffraction was performed to investigate the influence of nitrogen incorporation onto the crystallization tendency. The samples were prepared by heating up fastly to three temperature levels 1000 "C,1200 "Cor 1500 "C.Then they were kept 15 min at this temperture and rapidly cooled down. Onset of the crystallization begins after the primary formation of amorphous silicon carbide (fig. 5-11, The peak sharpness dimishes up to 1200

"C.But finally at 1500 "C linewith decreasing is obviously. This behavior should result from the retarded crystallization kinetics. Structure changes cause an increase of crystal growth activation energy. Indeed, a density increasing between 1000 - 1400 "C is observed.

Silicori Ccirhide Fibers from Highly Reuctive PoI.v(meth~lchlorosilane)s

81

ammonia-curing

)r .-c

..

v)

S

P,

c

.-c

~.

1

I ooooc

Figure 5-1 X-ray patterns of pyrolysed fibers depending on pyrolysis temperature

Obviously a nanocrystalline solid is formed using the above pyrolysis regime. The micrograph

B (Fig. 5-2) shows fibers after treatment under air at 1400 OC. Some spots on the surface are visible which are attributed to a starting oxide layer formation. Tensil strength is lower as reported for NICALON-fibers what may be caused by the manufacture equipment. Acknowledgement

We are grateful to Dr. F. Babonneau for advice and performance of the NMR measurements. Also we thanks for the financial support of these investigations the Deutsche Forschungsgemeinschaft and the German fonds of the Chemical Industry. References

1. R. Laine, F. Babonneau, J. Chem. Mat. 5 1993,260-279 2. P. Greil, J. Am. Ceram. SOC.1995, 78,835 - 848 3. R. J. P. Comu, M. Enders, S . Huille, L. Lutsen, J.J.E. Moreau, in Applications of

Organometallic Chemistry in the Preparation and Processing of Advanced Materials (Eds.

J. F. Harrod, E.M. Laine), Kluwer Academic Publishers, 1995, p. 185-199,

82

Cerumics ,from Orgunoelement Compounds

4. R. Richter, G. Roewer, U. Bohme, K. Busch, F. Babonneau, H.-P. Martin, E. Muller, Appl. Organomet. Chem. 1 1 1997,71- 106 5 . Z.F. Zhang, F. Babonneau, R.M. Laine, Y. Mu, J.F. Harod, J.A. Rahm, J. Am. Ceram. SOC.

74 1991,670 6. E. Hengge, M. Weinberger, J. Organomet. Chem. 433 1992,71-106 7. Sugimoto, T. Shimoo, K. Okamura, T. Seguchi, J. Am. Ceram. SOC.1995 78, 1849

20 pm

A) Sic-fiber after 1200°C under air

Diameter Tensile Strength Thermal Resistance (in air) Crystallite Size Density oxygen content nitrogen content chlorine content free carbon fig. 5-2 review of fiber properties

20 pm

B) Sic-fiber after 1400°C under air

15-50 pm 500-2 100 MPa 1400°C 1200 "C x-ray amorphous 1400 "C 2 nm 2.5 g ~ m ' ~ -8O"C) attack the siloxane network linking the SAM to the substrate, reducing the structural order of the film and at worst stripping the S A M from the substrate. This fact, taken with the trend shown in Figure 2-4, suggests that thin film oxide deposition on SAMs from aqueous solution should be easily observed with cations of high field strength, i.e. trivalent and tetravalent cations, by using strongly acidic conditions. Previous results have been consistent with this interpretation: TiO, films16 have been synthesized from TiC14 solution at pH -1, and FeOOH films3' have been deposited from FeCl, solution at pH 2.3. Similarly, Alcontaining34and Zr-~ontaining,~ films have been deposited from solutions at pH around 3 and 1, respectively. Also, Bunker et al.I4 have reported synthesis of Sn02 films from Sn4+solutions in strongly acidic condition onto SAMs:

154


In contrast, with low valence cations, e.g. Pb”, Ba2’ and Zn2’, it is difficult to take advantage of a S A M ’ S ability to promote film formation, unless a way can be found to induce oxide formation at lower pH, i.e. at conditions more compatible with the SAM. According to Figure 2-4, Zn should be an intermediate case, and the results reported here on zinc-containing films are consistent with this interpretation. That is, basic conditions were required to produce sufficient precipitates of Zn (hydrated) oxide to form a film; and although compact, uniform films could be formed, they were less adherent than films of e.g. Ti02 and ZrOz, suggesting that the basic environment diminished the adhesion of the S A M and therefore of the entire film. According to this interpretation and Figure 2-4, Y should also be an intermediate case for film deposition onto SAMs from aqueous solutions, but in fact the approach was quite successful (as discussed in 0 2.1.1). The difference appears to be that through the use of urea, it was possible to precipitate oxygen-containingY complexes at moderate temperatures (80°C) and pH (I 6.5), allowing appreciable film formation (though at slower rates than Ti and Zr). In contrast, all of the acidic routes attempted with Zn led to discontinuous, weakly adherent films containing predominantly zinc chloride.

Non-oxide thin films on Self-Assembled Monolayers

3

It is our objective to expand the present bio-inspired approach towards the synthesis of non-oxide thin films, ceramics of similar high technical importance to that of oxides. Our initial goal is the deposition of ceramic-like materials from an organic solution, starting from low molecular-weight precursors. “Ceramic-like material” means that the elemental composition of the as-deposited, amorphous material and the resulting ceramic phase are identical. Therefore the deposited material can be transformed into a crystalline phase without a pyrolysis step. In analogy to the preparation of thin films by decomposition of precursors from the vapor phase (CVD) this process was named Chemical Liquid Deposition (CLD)!2 Silicon-dicarbodiimide is an example of such a ceramic-like material. It was first synthesized by Kienzle et al.63by reaction of tetrachlorosilane with bis(trimethylsily1)carbodiimide in the presence of pyridine, yielding a nanosized white powder: SiCl,

+ 2 MqSi-N=C=N-SiMe3

toluene

+ 4MgSiCl

(1)

n

The carbodiimide group, -N=C=N-, is a pseudo chalcogen, therefore the reaction steps to the polymer are comparable to the reactions between tetrachlorosilane and water to SOz, i.e. substitution and condensation. At 400°C the material transforms reversibly into a phase with anti-cuprite structure, a structure type very similar to cristobalite. Decom osition with loss of nitrogen and dicyan and the formation of Si2CN4begins at

E

920°C.

Synthesis of Oxide and Non-Oxide Inorganic Materials at Organic Surfaces

3.1

155

Experimental

General. All operations with air-sensitive materials were carried out in an argon atmosphere (Schlenk technique). Toluene was freshly distilled from potassium, tetrachlorosilane from magnesium turnings. Bis(trimethylsily1)-carbodiimide was synthesized from hexamethylsilazane and ~yanamide.~' The coatings were characterized by transmission FT-IR-spectroscopy (Bruker IFS 66) and scanning electron microscopy (Cambridge Instruments S 200). Substrates. Silicon wafers (Wacker Chemitronics, 1x1 cm, orientation , p-type/ boron doped) were cleaned by wiping with chloroform, acetone and ethanol and dried in an argon stream. Prior to the deposition, the substrates were cleaned and oxidized with "piranha solution", a 30:70 mixture of hydrogen peroxide and concentrated sulfuric acid at 80 "C for 20 minutes, rinsed with distilled water and dried in an argon stream (denoted here as bare silicon). The synthesis of the surfactant 1-cyano-16-(trichlorosily1)-hexadecane, the preparation of the silicon wafers, the deposition and the in situtransformation of the self-assembled monolayer was according to literature description.66 The quality of the S A M packing and the transformation to amine groups was checked by contact angle measurement (Kriiss G10) and x-ray photoelectron spectroscopy (PerkinElmer ESCA 5400). Silicon-dicarbodiimide fdms. A solution of 5 ml bis(trimethylsily1)-carbodiimide, 5 ml tetrachlorosilane and 10 ml toluene was prepared in a Schlenk tube. A silicon substrate coated with the amine terminated S A M was added. After 6 days at room temperature 20 ml toluene were added. The solvent was then carefully removed by use of a pipet except a small amount. These steps were repeated for 5 times to clean the surface and to prevent contamination during evaporation of the solvent. Finally the solvent was completly removed, the sample was dried in vacuum and stored under argon atmosphere. A bare silicon wafer was simultaneously subjected to identical treatment in a separate tube.

3.2

Results and discussion

self-assembled monolayers were chosen for the deposition of Amine terminated (-NH,) the silicon-dicarbodiimide because of the high reactivity of amines towards chlorosilanes. Depositions should be also possible on the native oxidehydroxide of bare silicon. In fact, the use of both SAM-coated and bare silicon wafers led to film formation. This is indicatea by scanning electron microscopy images. To allow the coatings to be more readily distinguished from the underlying substrates, the surfaces of both films were scratched. The film on the S A M looks much more homogenous than that on the bare silicon (Figure 3-1). A reason therefore might be that the formation of the polymer is induced by the functional groups of the SAM and took place on the surface. In contrast, the film on bare silicon appears mainly to have been formed by adsorption of particles that already exist inside the solution.

156


(a)

Figure 3-1. Scanning electron micrographs of silicon-dicarbodiimide films. (a) Film deposited on mine SAM on silicon. (b) Film deposited on bare silicon.

Synthesis of Oxide and Non-Oxide Inorganic Materials at Organic Surfaces

157

A characteristic vibration band in the IR spectra of carbodiimides is the N=C=N stretching vibration. Therefore, the recorded transmission IR spectrum indicates that silicon-dicarbodiimide was formed on the S A M coated sample (Figure 3-2). CH, stretching and deformation vibration bands can also be recognized, indicating that the deposited material is not fully cross-linked, but contains a significant number of endstanding trimethylsilyl groups. SiI

I

Si

/

s’” 4000

3500

3000

2500

2000

150

1000

wavenumber [cm’] Figure 3-2. Transmission FT-IR spectrum of a silicon-dicarbodiimide film deposited on m i n e SAM on silicon.

4

Summary and conclusions

Previous results on the deposition of oxide thin films on organic self-assembled monolayers (SAMs) from liquid solutions at low temperatures were reviewed. As with biominerals, which served as an inspiration for the present synthetic approach, the formation of the present films depends (with the exception of Y203films) on the presence of a suitably tailored organic layer (here, the S A M ) as the primary deposition surface. In contrast with biominerals, the present films show no texturing nor preferred crystallographic orientation and grow only when spontaneous precipitation takes place in the source solution. This suggests that the films are not formed by heterogeneous nucleation on the organic layer. Also unlike biomineralization processes, the present approach lacks the ongoing participation of the cells of a living organism. This implies that the influence of the SAM must be exerted primarily in the initial formation of the film, and less so as growth continues.

158


For deposition of oxide films from aqueous solutions, the present technique is suitable for oxides of cations that can precipitate in acidic solutions. This arises from the deleterious effects of basic solutions on the order and adhesion of siloxy-anchored SAMs. This appears to limit the present approach to oxides of cations with high ionic strength, i.e. small size and/or high valence. These conclusions were supported by the results on films containing Zn2', which has intermediate ionic strength. Compact thin films of ZnO could be synthesized, using thioacetate or sulfonate SAMs, by heattreatment of films deposited from Zn-containing solutions at pH=l 1 and 35"C, but these films showed less complete coverage and adherence than previously reported films of Ti02,Zr02, and Y203. In extent, the described work has shown for the first time a characteristic influence of self-assembled monolayers on the deposition of a non-oxide phase. Further investigations are necessary to determine whether film formation results here from a surface reaction induced by the functional groups, or by the attraction of pre-formed polymeric particles to the S A M . Improvements in processing and characterization are also important.

5

Acknowledgments

The authors gratefully acknowledge their ongoing collaborations with Prof. A. Heuer and Prof. C. Sukenik, and previous collaborations with Dr. M. Agarwal, Dr. R. Collins, and Dr. H. Shin whose dissertation research was summarized in part here. This work was sponsored by the U.S.Air Force Office of Scientific Research, the Max-PlanckGesellschaft, the U. S. Basic Missile Defense Office, and the Government of Thailand (for a fellowship to S.S.).

References E.G. Bauer, B.W. Dodson, D.J. Ehrlich, L.C. Feldman, C.P. Flynn, M.W. Geis, J.P. Harbison, R.J. Matyi, P.S. Peercy, P.M. Petroff, J.M. Phillips, G.B. Stringfellow, A. Zangwill, J.Muter. Res. 1990,5,852-894. M . Ohring, The Materials Science of Thin Films, Academic Press, San Diego, 1992. D.W. Johnson, Am. Ceram. SOC. Bull. 1985,64, 1597-1602. P.P. PhulC, S.H. Risbud, J. Muter. Sci. 1990,25, 1169-1183. See e.g.: (a) S . Weiner, W. Traub, Phil. Trans. R. SOC.Lon. B 1984,304,425434; (b) A. Linde, A. Lussi, M.A. Crenshaw, Calcif: Tissue Znt. 1989,44,286-295; (c) L. Addadi, A. Berman, J.M. Oldak, S. Weiner, Connective Tissue Research 1989,21, 127-135. P. Calvert, S. Mann, J.Muter. Sci. 1988,23,3801-3815.

Synthesis of Oxide and Non-Oxide Inorganic Materials at Organic Surfaces 7

8

9

10

11

12

13

14

15

16

17

18

19

20 21 22 23 24 25

159

P.C. Rieke in Atomic and Molecular Processing of Electronic and Ceramic Materials (Eds.: I. A. Aksay, G. L. McVay, T. G. Stoebe, J. F. Wager), Materials Research Society, Pittsburgh, 1988, pp. 109-114. P. Calvert in Better Ceramics Through Chemistry IV(Eds.: B.J.J. Zelinski, C.J. Brinker, D.E. Clark, and D.R. Ulrich), MRS Symp. Proc. 180, Materials Research Society, Pittsburgh, 1990, pp. 619-623. B.J. Tarasevich, P.C. Rieke in Materials Synthesis Utilizing Biological Processes (Eds.: P.C. Rieke, P.D. Calvert, M. Alper), MRS Symp. Proc. 174, Materials Research Society 1990, pp. 5 1-60. P.C. Rieke, B.J. Tarasevich, S.B. Bentjen, G.E. Fryxel1,A.A. Campbell in Supramolecular Architecture - Synthetic Control in Thin Films and Solids (Ed.: T. Bein), ACS Symposium Series 499, American Chemical Society, 1992, pp. 61-75. P. Calvert in UltrastructureProcessing of Advanced Materials (Eds.: D.R. Uhlmann, D. R. Ulrich), John Wiley 8z Sons, New York, 1992, pp. 149-157. B.J. Tarasevich, P.C. Rieke, G.L. McVay, G.E. Fryxell, A.A. Campbell in Chemical Processing ofAdvanced Materials (Eds.: L.L. Hench, J.K. West), Wiley Interscience, New York, 1992, pp. 529-542. A.A. Campbell, G.E. Fryxell, G.L. Graff, P.C. Rieke, B.J. Tarasevich, Scanning Microscopy 1993, 7,423-429. B.C. Bunker, P.C. Rieke, B.J. Tarasevich, A.A. Campbell, G.E. Fryxell, G.L. Graff, L. Song, J. Liu, J.W. Virden, G.L. McVay, Science 1994,264,48-55. P.C. Rieke, B.J. Tarasevich, L.L. Wood, M.H. Engelhard, D.R. Baer, G.E. Fryxell, C.M. John, D.A. Laken, M.C. Jaehnig, Langmuir 1994,10,619-622. (a) H. Shin, R.J. Collins, M.R. De Guire, A.H. Heuer, C.N. Sukenik, J. Mater. Res. 1995,10, 692-698; (b) idem., J. Mater. Res. 1995,10, 699-703. P.C. Rieke, B.D. Marsh, L.L. Wood, B.J. Tarasevich, J. Liu, L. Song, G.E. Fryxell, Langmuir 1995, 1I , 3 18-326. M.R. De Guire, H. Shin, R.J. Collins, M. Agarwal, C.N. Sukenik, A.H. Heuer, in Integrated Optics and Microstructures III (Ed.: M. Tabib-Azar), Proc. SPIE 2686, 1996, pp. 88-99. R.J. Collins, H. Shin, M.R. De Guire, C.N. Sukenik, A.H. Heuer, Appl. Phys. Lett. 1996,69,860-862. H. Shin, M. Agarwal, M.R. De Guire, A.H. Heuer, J. Am. Ceram. SOC.1996, 79, 1975-1978. I.A. Aksay, M. Trau, S. Manne, I. Honma, N. Yao, L. Zhou, P. Fenter, P.M. Eisenberger, S.M. Gruner, Science 1996,273,892-898. B.J. Tarasevich, P.C. Rieke, J. Liu, Chem. Mater. 1996,8,292-300. M. Agarwal, M.R. De Guire, A.H. Heuer, Appl. Phys. Lett. 1997, 71, 891-893. M. Agarwal, M.R. De Guire, A.H. Heuer, J. Am. Ceram. SOC.(accepted for publication). S. Mann in Biomimetic Materials Chemistry (Ed.: S . Mann), VCH Publishers, New York, 1996, pp. 1-40.

160 26 21 28 29 30 31

32 33 34 35 36

31 38 39

40 41

42 43

44 45 46 41

48 49 50

51 52 53 54


A. Berman, J. Hanson, L. Leiserowitz, T.F. Koetzle, S. Weiner, L. Addadi, Science 1993,259, 776-779. S. Mann (Ed.), BiomimeticMaterials Chemistry, VCH Publishers, New York, 1996. P. Calvert, P. Rieke, Chem. Muter. 1996,8, 1715-1727. S. Manne, I.A. Aksay, Cum Opin. Solid State Muter. Sci. 1997,2,358-364. J.H. Fendler, Cum. Opin. Solid State Muter. Sci, 1997,2,365-369. M. Maiti, Synthesis ofIron Oxide Thin Films on Organic Templates on Silicon, M.S. thesis, Case Western Reserve University, 1994. H. Shin, Synthesis of Ti02 Films on SelfAssembledOrganic Monolayers on Silicon, M.S. thesis, Case Western Reserve University, 1994. T.P. Niesen, J. Bill, F. Aldinger, in preparation. S. Supothina et al., in preparation. H. Shin, M. R. De Guire, and A. H. Heuer, submitted to J. Appl. Phys. H. Shin, Deposition Mechanisms and Electrical Properties of Ti02Thin Films on Self-AssembledOrganic Monolayers on Si, Ph.D. dissertation, Case Western Reserve University, 1996. H. Jaffe, D.A. Berlincourt, IEEE Proceedings 1965,53, 1372-86. A.H. Fahmy, E.L. Adler, IEEE Trans. Sonics Ultrason., 1972,19,346-349. G.D. Hillman, H.J. Sequin, ZEEE Trans. Sonics Ultrason., 1974,21,49-54. J.M. Hammer, D.J. Channin, M.T. Durn, J.P. Wittke, Appl. Phys. Lett. 1971,21, 358-360. N. Mikoshiba in Guide-WaveAcousto-Optics(Ed.: C.S. Tsai), Springer-Verlag, New York, 1990, pp. 205-233. F.S. Hickernell, IEEE Trans. Sonics Ultrason. 1985,32,621-629. Y. Ohya, H. Saiki, Y. Takahashi, J. Muter. Sci. 1994, 29,4099-4103. U. Lampe, J. Miiller, Sensors and Actuators 1989,18,269-284. H. Nanto, T. Minami, S. Takata, J. Appl. Phys. 1986,60,482-484. W.W. Wenas, A. Yamada, M. Konagai, K.Takahashi, Jpn. J. Appl. Phys. 1991,30, JA4 1-443. K.Kamata, J. Nishino, S. Ohshio, K. Maruyama, J. Am. Ceram. SOC.1994, 77,505508. B.T. Khuri-Yakub, J.G. Smits, J. Appl. Phys. 1981,52,4772-4774. B.H. Choi, H.B. Im, J.S. Song,J. Am. Ceram. SOC.1990,73, 1347-1350. M.G. Ambia, M.N.Islam, M.O. Hakim, J.Muter. Sci. 1994,29,6575-6580. W. Tang, D.C. Cameron, Thin Solid Films 1994, 238,83-87. J.H. Jean,J. Muter. Sci. Lett. 1990,9, 127-129. D. Raviendra, J.K. Sharma, J. Appl, Phys. 1985,58,838-844. (a) A.E. JimCnez-Gonzllez, P.K. Nair, Semicond. Sci. Technol. 1995,10, 1277; (b) A.E. JimCnez-Gonzalez, R. Suzirez-Parra, J Cryst. Growth 1996,167,649-55; (c) M. Robles, J. Taguenamartinez, J.A. Delrio, Thin Solid Films 1997,293,320-326; (d) A.E. Jimtnez-Gonzllez, J. Solid State Chem. 1997,128, 176-180.

Synthesis of Oxide and Non-Oxide Inorganic Materials at Organic Sugaces 55

56 57

58

59

60

161

(a) T. Saeed, P. O’Brien, Thin Solid Films 1995,271,35-38; (b) P. O’Brien, T. Saeed, J. Knowles, J. Mater. Chem. 1996, 6, 1135-1139. K. Ito, K. Nakamura, Thin Solid Films 1996,286,35-36. S.M. Haile, D.W. Johnson, J. Am. Ceram. SOC.1989, 72,2004-2008. The adherence of the films was determined by a simple tape test using transparent tape. The tape was placed on the film’s surface and the force was applied by rubbing over the entire surface. The tape was peeled off, and then both the tape and the sample were analyzed using XPS. Any detection of material from the film on the tape was taken as a sign of lack of adherence. K. Fujita, K. Matsuda, S. Mitsuzawa, Bull. Chem. SOC.Jpn. 1992,65,2270-2271. C.F. Baes, R.E. Mesmer, The Hydrolysis of Cations, John Wiley & Sons, New York, 1976.

61

62

M. Pourbaix, Atlas of Electrochemical Equilibrium in Aqueous Solutions, translated by J.A. Franklin, National Associate of Corrosion Engineers, Houston, 1974. (a) J. Bill, F. Aldinger, Adv. Mater. 1995, 7, 775-787; (b) J. Bill, A. Kienzle, F. Aldinger, German Patent 195.02095.2, 1995; (c) J. Bill, F. Aldinger, 2. Metallk. 1996,87,827-840.

63 64

65

66

A. Kienzle, J. Bill, F. Aldinger, R. Riedel, NanoStruct. Mat. 1995,6,349-352. R. Riedel, A. Greiner, G. Miehe, W. DreBler, H. FueB, J. Bill, F. Aldinger, Angew. Chem. 1997,109,657-660; Angew. Chem. Int. Ed. Engl. 1997,36,603-606. K. Wurm, Ph.D. dissertation, Universiat Stuttgart, in preparation. N. Balachander, C.N. Sukenik, Langmuir 1990,6, 1621-1627.


IV. Characterization


Thermodynamic Calculations in the System Si-B-C-N-0 H. J. Seifert and F. Aldinger Max-Planck-lnstitut fur Metalvorschung and Universitat Stuttgart, lnstitut fur Nichtmetallische Anorganische Materialien Pulvermetallurgisches Laboratorium D- 70569 Stuttgart

Abstract A consistent thermodynamic dataset for the B-C-N-9-0 system was developed by the CALPHAD method of Thermodynamic Optimization. The dataset is used for the calculation of different types of phase diagrams (e.g. isothermal sections, isopleths, partial pressure diagrams), phase fraction and phase composition diagrams. The results are used to simulate high temperature phase reactions and phase stabilities of precursorderived Si-B-C-N ceramics. Preliminary calculations for oxygen containing precursor materials are presented. 1. Introduction Si-B-C-N precursor ceramics can be routinely prepared in a large variety of well defined chemical compositions. Moreover numerous physico-chemical parameters influence the phase reactions during thermolysis as well as phase stabilities of the final product. Thus a purely experimental optimization of materials properties can be a very cumbersome procedure. Therefore, thermodynamic calculations using the CALPHAD method (CALculation of PHAse Diagrams) support the understanding of the materials crystallization behavior and high temperature phase reactions [ 1,2].

After thermolyis the precursor ceramics show amorphous structures which start to crystallize at temperatures higher than 1600 K depending on the specific chemical compositions of the different materials. Typically, nanocrystalline microstructures of Si3N4,Sic, BN and graphite are formed during crystallization and detected by X-ray powder diffraction and electron microscopy, respectively [3]. The thermal degradation of some materials accompanied by significant mass losses can be explained by CALPHAD calcuations as well as the remarkable high temperature stability up to 2300 K of other borosilicon carbonitrides [4,5]. A consistent thermodynamic dataset was developed for the Si-B-C-N allowing the calculation of all types of phase diagrams in the quaternary system (e.g. isothermal sections, isopleths, volatility diagrams) [6]. Quantitative data can be derived from calculated phase fraction diagrams and site occupancy diagrams. Additionally, the simulation was extended to materials containing some oxygen. Therefore, preliminary results for the quinary Si-B-C-N-0 system are presented.

166

Characterization

2. The CALPHAD Method The main idea of CALPHAD is to provide analytical descriptions of all thermodynamic fimctions of state of the particular system in consideration. Advanced sublattice models are chosen which take into account the physico-chemical properties and the crystal structure of stoichiometric and solution phases. As not all the features are quantitatively defined, the models contain adjustable parameters. These are adjusted to experimental data, mostly using the least squares method, after Gauss, in a process of so-called Thermodynamic Optimization [7]. If specific experimental data are missing, estimates derived from comparisons with similar phases may be used. The resulting fit to the experimental data is also a test if the model was well chosen. For computer applications, the analytical descriptions of the state functions of all phases considered are stored in databases. Multicomponent systems can then be calculated efficiently by combining thermodynamically optimized binary and ternary analytical descriptions, in computer databases, and extrapolating to regions of temperature, composition and pressure unexplored experimentally. The amount of experimental work required for the materials development can be drastically reduced [ 1,2]. For thermodynamic optimization and phase diagram calculations, the well established software packages are available [8-113. In this work the programmes BINGSSBINFKT [8] and THERMO-CALUPARROT[9,10] were used. 3. Thermodynamic Data Thermodynamicallyoptimized datasets resulting from this work or taken from literature [12-161 were used for the calculations in the quinary system. The datasets include the descriptions for the stable phases at normal pressure: the gas phase, the liquid phase (metallic and ceramic, respectively), P-boron, graphite, silicon, P-Si3N4,a- and P-Sic, a-BN, B4+&, B3Si, B6Si and B,Si. Oxygen containing phases are B203, Si02 (quartz, tridymite, cristobalite) and Si2N20.Further solid phases were reported in literature but were not documented to be stable under conditions considered here. The descriptions for the gaseous species, e.g. 02,N2, SiO, CO, BO and 55 others, were taken from the substance database (SSUB) provided by the Scientific Group Thermodata Europe (SGTE) [17]. The ternary and quaternary descriptions, except the Si-B-C and Si-B-0 systems, were derived from extrapolating calculations. The Si-B-C system and the quasibinary section B203-Si02section in the Si-B-0 system were optimized.

4. Binary Systems The thermodynamic descriptions of the ten binary subsystems were introduced in the SiB-C-N-0 database. The calculated results for the Si-N and B-C system which are important for the understanding of the phase reactions in the high component materials are reported here. 4.1 The Si-N System The decomposition temperature of Si3N4according to the reaction

Si3N4(s)=3Si(l)+2N2(g) (1)

Thermodynamic Calculations in the System Si-B-C-N-0

167

strongly depends on the atmospheric 1.0 I 21UK 2246K partial pressure of N2. In a closed system o.9with an inert atmosphere a certain partial o.8 pressure of N2 will develop in equilibrium with Si3N4as a function of f O*'total pressure and temperature. Fig. I 0.6shows the development of the N2 fraction .E 0.5C in the gas phase for total pressures of 1 o.4bar and 10 bar, respectively as a function of the temperature. The N2 partial pressure for a given tot'al pressure and 0.1temperature corresponds to the N2 fraction in the gas phase. The partial 0pressure is ficticious as long as no gas 1500 2000 2500 Temperature in K phase exists. Therefore, a small amount of Ar as inert gas was introduced into the Fig. I : The decomposition of Si3N4 as a system. Concerning the phase rule, function of temperature and pressure. nothing is changed: just one component (Ar) and one phase (gas) are added, providing a real meaning for the partial pressure of N2. At a total pressure of 1bar, Si3N4 decomposes at a temperature of 2114 K. A pressure increase to 10 bar raises this temperature to 2305 K. These calculations assume ideal gas behavior at the selected conditions. The pressure dependence of the Si3N4decomposition has to be taken into account for the liquid phase sintering of Si3N4-basedceramics as well as for the heat treatment of precursor derived materials [ 5 ] .

3

4.2 The B-C System

B4+& is the only solid phase with extended homogeneity range in the Si-B-C-N-0 system. Based on experimental crystallographic information [ 181 the compound energy model [19] with the sublattice occupation

(BIIC,BI~)(CBC,CBB,BVaB) was used for the description of the Gibbs energy. The model parameters were adjusted by Thermodynamic Optimization to the most reliable experimental data (phase diagram, enthalpies, and heat capacities). The boron-rich part of the optimized B-C phase diagram in comparison with experimental data is shown in Fig. 2. The B4+&homogeneity range is

* 1400

I

0

B

I

I

I

I

0.05 0.10 0.15 0.20 0.25 Mole Fraction C

I

I

0.30 0.35

Fig. 2: The calculated boron rich part of the BC phase diagram in comparison with experimental data [ 6 ] .

168

Characterization

well described. Experimental investigations at temperatures higher than 2300 K are difficult and only such data not influenced by systematic experimental errors (e.g. documented sample reactions with crucible material) were used for the least squares method (see Fig. 2, usednot used). 5. The Quasibinary B2O3-SiO2System 2000 Only a few experimental data on the technically important quasibinary BZO3SiOz system are documented in the literature. Prolonged heat treatments are 1600 necessary to reach equilibrium and the results can be significantly influenced 1400 by H 2 0 contamination. Only Rocket and Foster [20] investigated the system 8 1200 comprehensively using Pt-capsuled FE samples and analyses by ceramography 1000 and X-ray powder diffraction. Therefore, these data were used to 800 develop the thermodynamic description of the system. The partial ionic liquid 6oo 0 model [2 I J with sublattice occupations

(B'3,Si'4)p(02,Si04",Va,B203,Si02)Q

0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 10

B203

Mole Fradion SO,

SiO,

was used for the liquid phase description. Fig. 3: The calculated B203-SiO~phase The comparison of the calculated phase diagram in comparison with experimental data. diagram with experimental investigations shows good agreement with each other (Fig. N 3). 6. Ternary Systems All ten ternary subsystems were introduced in the database from which the Si-C-N and SiB-C system are described here.

6.1 The Si-C-N System

Fig. 4a shows a calculated isothermal section 0.2 in the Si-C-N system at 1760 K. Additionally, the trace of an isopleth (Graphite 43.4Si,56.6N) is indicated including the c composition of a ceramic, 25.64 Si; 41.02 Mole Fraction Si Fig. 4a-c: The calculated Si-C-N system. C; 33.33 N (at.-%), derived from ,,VT50", a precursor available from Hoechst AG, 4a: Isothermal section at 1760 K with Frankhrt. The calculated diagram of this the isopleth graphite-43.4Si, isopleth is shown in Fig. 4b. At a temperature 56.6N (dashed line) containing of 1757 K, Si3N4and graphite react to form the composition of VT5O-derived Sic and gas phase. Quantitative mass balance ceramic (triangle). information can be derived from the related

-


169

phase fraction diagram shown in Fig. 4c . The calculated mass loss of 28.5 % N2 during the reaction is in excellent agreement with experimental results of thermogravimetric analysis. The graph shows that after the reaction, besides S i c and gas phase, excess carbon remains. The balanced reaction equation Si3N4+3C=3SiC+2N2

(2)

is only valid for the defined Si3N4:Cratio at the intersection of the inner tielines C-SilN4 80 I I I

I

I

I

SI,N,

Gar Liquid

E

50-

E"

40-

a

E

g

U

SIC

8

2114 W.

20001757

!

,

0.1

s

t

B

.I

Graphite,Si,N,. SIC

I5O01 1000 0

-

WN.

I

i

,-

0.2 0.3 Mole Fraction N

I

0.4

,I

0.5

sic

60-

i 3

n 2

Graphite

30-

~

5;

20-

: Graphite

10-

I

/

SIC

I

I

1000

1500

I

I

2000 2500 Temperature

I

3000

[w

4b: The isopleth graphite-43.4Si,56.6N with VT50 ceramics composition.

4c: The phase fraction diagram for the VT5O-derived ceramic (25.64 Si, 41.02 C, 33.33 N; at-%). and N-SIC of the four phase reaction. For precursor ceramics with less carbon content (e.g. NCP200 ceramics) all carbon is consumed during reaction (2) and excess Si3N4 remains. In these cases the remaining Si;N4 may decompose at higher temperatures according to reaction (2) (graphite crucibles are used for the thermogravimetrical investigations) causing a further mass loss. No direct measurements of the temperature of reaction (2) are documented. Our experiments showed significant influence of kinetics and a shift to higher reaction temperatures. However, the mass loss and phase assemblages are in accordance with calculated results.

6.2 The Si-B-C System

The calculated isothermal section at 1500 K of the Si-B-C system is shown in Fig. 5a. The B4+&-phase is in equilibrium with all other phases of the system. At higher temperature some Si is soluble in B4+& and some B in S i c [22-251. These solubilities were modelled using the sublattice descriptions (B 1 I C,B 12)(SiSi,CBC,CBB,BVaB) and (Si)(C,B), respectively. In Fig. 5a the composition 36Si, 12B,52C (at.-%) is indicated corresponding to the Si:B:C ratio of the close-by quaternary composition of the high temperature stable precursor ceramics Si3,0BI.0C4.3N2.0 [26] derived from T2( 1 ) precursor. Additionally, the trace of the isopleth B-Sic is shown. The calculated diagram of this isopleth is shown in Fig. 5b providing information on the phase assemblages as a function of composition and temperature.

170

Characterization @GAS+GRAPHITE+ LIQUID

Si

3000

3

-

2 2500 3 I

P E 2000 P

+ Liqrid + SIC

I500

I00 (I

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.R 0.9 1.0

Sic

Mole Fraction I3

B

Fig. 5a-b: The calculated Si-B-C system. 5a: Isothermal section at 1500 K with indicated 5b: The isopleth B-Sic. composition and isopleth B-Sic. 7. The Quaternary SCB-C-N System Fig. 6a shows the calculated phase fraction diagram for the T2(1)-derived precursor ceramic Si3.0BI.0C4.3N2.0 (9.71B, 29.13Si, 41.47(7, 19.42N at.-%). 70 The BN phase remains stable up to 2586 K, where it decomposes ( 1 bar N2). At a Fso temperature of 1757 K carbon and Si3N4 react, as in the case of VT50 ceramics, to 5 ~ 0 form gas phase and Sic. This (degenerated) reaction is not influenced by the presence of 5 4o BN-phase. From this calculation, a thermal 1 decomposition of the precursor ceramic at f a. 1757 K with significant mass losses is 20 expected. This result is in contradiction to 10 experimental data derived from thermogravimetrical analysis showing a 0 remarkable thermal stability of T2(1) up to looo 1500 2m 2500 3ooo Temperature [KI 2300 K [ 2 6 ] .To understand these conflicting diagrams for Fig* 6a-b: results, the materials microstructure has to be taken into account: High Resolution T2( I)-derived ceramics. Transmission Electron Microscopy 6a: The phase fraction diagram. (HRTEM) and Electron Spectroscopic Imaging (ESI) of numerous high temperature stable Si-B-C-N precursor ceramics [27,28] show nanocrystalline Si3N4-grainseven in materials heat treated at temperatures higher than the decomposition temperature of Si3N4 of 21 14 K (see chapter 4.1). Additionally, all high temperature stable Si-B-C-N materials show the segregation of a turbostratic BN layer along the grain boundaries of Si3N4and Sic, respectively. Carbon is dissolved to a large extent in the turbostratic BN layers as found by ESI method. This

= 2 3


171

microstructure is typical for all silicoboron carbonitrides exhibiting high thermal stability [28]. From this analyses we assume an encapsulation effect resulting in a local pressure increase with the result of phase stabilization as shown for Si3N4in chapter 4.1 (Fig. I). The thermal decomposition of the 1non-stable Si-B-C-N materials originates from reactions (1) and 0(2), respectively. The reaction temperatures mentioned above are given for a total pressure of p= 1 bar. Gnphlt.. SIC. EN Both increase with increasing nitrogen pressure. To illustrate the f effect of increased nitrogen pressure 3on the reaction temperature of Si3N4 and graphite calculated partial 4pressure diagrams, (log(pN2)temperature) can be used [51. To calculate such diagrams a Tempentun pC] particular Si:B:C ratio is fixed and 6b: The partial pressure diagram for B12Si36C52 composition. the chemical potential of nitrogen and the temperature are changed systematically. Fig. 6b shows the resulting log(pN2)-temperaturediagram for the composition Si36BIZC5Z as in T2( 1)-derived ceramics indicated in Fig. 5a. Along the upper line, Si3N4,Sic, graphite, BN and gas phase are in equilibrium. Upon crossing this line, one of the phases disappears by the reversible reaction (2). Above the line, the solid phases Si3N4,graphite and BN are in equilibrium with the gas phase whereas below graphite, Sic and BN are stable beside the gas phase. For the given Si36B12C52 composition some excess graphite remains after the complete consumption of Si3N4. Increasing the N2-pressure from P N 2 = 1 bar to 10 bar, the reaction temperature of Si3N4 and graphite is increased from a temperature of 1757 K to 1973 K. A further effect of increasing the reaction temperature is due to the dissolution of carbon in the turbostratic BN-layers. In this case the activity of carbon is lower than one and will increase the reaction temperature as well.

The partial pressure diagrams form the basis for appropriate heat treatment and sintering of materials to stabilize specific phase equilibria. For sintering of samples with the phase assemblage graphite-Si3N4-BNthe partial pressure of nitrogen should be above the line graphite, Sic, Si3N4, BN, gas. However, for the stabilization of graphite-Sic-BN containing samples the nitrogen pressure range is limited by the the upper and lower line in the diagram. The lower line represents the reaction of BN and carbon to form B4+& and nitrogen [ 5 ] .

172

Characterization

The effects described were proved by experiments and calculation of numerous high temperature stable and non-stable Si-B-C-N precursor ceramics of different composition 8. The Si-B-C-N-0 System The equilibrium phase reactions for an oxygen containing Si-B-C-N material of the composition 14.8B, 14.2C, 47.1N, 0.50, 23.3Si (at.-%), which was found to be thermally quite stable [29], were simulated under He-atmosphere. The reported small CI-content was not taken into account. The calculated phase fraction diagram and the accompanied gas phase compositions are shown in Fig. 7a and b, respectively. At temperatures below 1350 K, Si3N4,BN, graphite, Si2N20and the gas phase exist. The N2 partial pressure is about 9.104 bar (90 Pa). In accordance with the information derived from the partial pressure diagram shown in Fig. 6b, at temperatures of about I

-$-

I

I

I

I

I06 j

I

50

I

I

I

2oMl

2500

3000

I o4

a

-

4

L!

240 4-

g 30

ct 20

d

I

I 0’ 102

10’

loo 10-1

;

102

10

P

I0 3 0

lo00

10-4

1500

2000

Temperature [K]

2500

3000

1000

1500

Temperature [K]

Fig. 7a-b: Calculated diagrams for oxygen containing Si-B-C-N ceramics. 7a: The phase fraction diagram. 7b: The gas phase composition. 1350 K, Si3N4and graphite form SIC and N2 gas species. Fig. 7b shows that the most important gas species below 2000 K are N2, CO, SiO and Si (besides He). With the decomposition of Si2N20,the Si partial pressure increases significantly. The results give a preliminary information on the phase reactions. There are no experimental evidences that Si2N20is formed during the heat treatment and calculations without taking into account this phase are in progress. 9. Conclusions CALPHAD type calculations can be used successfully to explain phase reactions of precursor derived Si-C-N and Si-B-C-N materials. Combination of such calculations with selected experiments are the basis for an efficient materials development. Comparison with available experimental results show that in the quaternary Si-B-C-N system the calculations are reliable. Based on a consistent thermodynamic dataset the most practically useful types of phase diagrams (e.g. isothermal sections, isopleths) and property diagrams (e.g. phase fraction diagrams) can be calculated. The results are valid to find the best physicochemical conditions to develop Si-B-C-N ceramics with defined


173

phase assemblages and properties. The extended extrapolating calculations in the Si-BC-N-0 system were carried out. Further experimental data are required to refine the datasets for oxygen containing phases. However, to use the calculated results for the interpretation experimental data requires the consideration of the materials microstructure, specific crystallization behavior and the kinetics of phase formation and phase reaction. This is true especially for the interpretation of dynamic thermal analysis experiments (DTA, DSC, TG). Moreover, heat treatments are often carried out in an inert atmosphere and evaporating gaseous species are continously removed by flowing gas atmospheres (inert or reactive). This effect deeply influences the phase reactions. Therfore, thermodynamic calculations for the simulation of such processes are in progress. References ( I ) H.J. Seifert, F. Aldinger: Z. Metallkd. 87 ( 1 996) 84 1-853. (2) K. Hack (edt.) The SGTE Casebook, Thermodynamics at Work, The Institute of Materials, Materials Modelling Series, London ( 1 996). (3) J. Bill, F. Aldinger: Z. Metallkd. 87 (1996) 827-840. (4) H. J. Seifert, J. Peng, F. Aldinger: Die Konstitution von Si-B-C-N Keramiken, Proc. Werkstoffwoche '98, 12.-15.10.1998 Munchen ( 1 998) submitted. (5) H.J Seifert, H.L. Lukas, F. Aldinger: Ber. Bunsenges. Phys. Chem. 9 (1998) 13091313. (6) B. Kasper, H. J. Seifert, A. KuDmaul, H.L. Lukas, F. Aldinger: Entwicklung eines thermodynamischen Datensatzes f i r das System B-C-N-Si-0 in Werkstofioche '96,28.-31.05.1996, Stuttgart, Tagungsband (1 996) 623-628. (7) H. L. Lukas, E.-Th. Henig, B. Zimmermann, CALPHAD 1 (1977) 225-236. (8) H.L. Lukas, S.G. Fries: J. Phase Equilibria 13 (1992) 532-541. (9) B. Sundman, B. Jansson, J.-0. Anderson: CALPHAD 9 (1 985) 153- 190. ( 10) B. Jansson, Evaluation of Parameters in Thermochemical Models Using Different Types of Experimental Data Simultaneously. Trita-Mac 234, Royal Institute of Technology, Stockholm, Sweden (1 984). (1 1) G. Eriksson, K. Hack: Met. Trans. B 21B (1990) 1013-1024. (12) J. Weiss, H. L. Lukas, J. Lorenz, G. Petzow, H. Krieg, Calphad 5, (198 I ) 125-140. (13) M. Hillert, S. Jonsson, B. Sundman, Z. Metallkd. 83 (1992) 648-652. (14) L. Dumitrescu, B. Sundman, J. Europ. Ceram. SOC.15 (1995) 239-247. ( 15) J. Grdbner, H. L. Lukas, F. Aldinger, Calphad 20 ( 1 996) 247-254. (16) S.-K. Lim, H. L. Lukas, Thermodynamische Optimierung des Systems B-C-Si und seiner Randsysteme, in: G. Petzow, J. Tobolski, R. Telle (eds.) Deutsche Forschungsgemeinschaft, Hochleistungskeramiken - Herstellung, Aufbau, Eigenschaften, Verlag Chemie, Weinheim ( 1 996) 606-6 16. (1 7) Scientific Group Thermodata Europe (SGTE) Domaine Universitaire, BP66, F38402 St. Martin d'Hnres Cedex, France. ( 1 8) H. Werheit, U. Kuhlmann, priv. com. (1 994). (19) J.-0. Anderson, A. Fernandez Guillermet, M. Hillert, B. Jansson, B. Sundman, Acta Metall. 34 (1986) 437-445. (20) T. J. Rockett, W. R. Foster, J. Am. Ceram. SOC.48 (1965) 75-80.

174

Characterization

(21) M. Hillert, B. Jansson, B. Sundman, J. Xgren, Met. Trans. A 16A (1985) 261-266. (22) R. Telle: Structure and Properties of Si-Doped Boron Carbide in: R. Freer (edt.) The Physis and Chemistry of Carbides, Nitrides and Borides, Kluwer, Dodrecht, The Netherlands (1990) 249-267. (23) H. Werheit, U. Kuhlmann, M. Laux, R. Telle: Structural and Optical Properties of Si-Doped Boron Carbide, in: Proc. 1 Ith. Int. Symp. Boron, Borides and Related Compounds, Tsukuba 1993, JJAP Series 10 (1994) 86-87. (24) P. T. B. Shaffer, Mat. Res. Bull. 4 (1969) 2 13-220. (25) W. D. Carter, P. H.Holloway, C. White, R. Clawing, Adv. Ceram. Mat. 3 (1988) 62-65. (26) R. Riedel, A. Kienzle, W. Dressler, L. Ruwisch, J. Bill, F. Aldinger: Nature 382 (1 996) 796-798. (27) A. Jalowiecki, J. Bill, F. Aldinger, J. Mayer: Composites Part A 27A (1 996) 7 17. (28) T. Kamphowe, J. Bill, H. J. Seifert, F. Aldinger, A. Jalowiecki, U. Alber. J. Mayer: The Thennochemical Stability of Precursor-derived Si-B-C-N Ceramics, Z. Metallkd. 90 (1999) submitted. Baldus, M. Jansen, 0. Wagner, Key. Eng. Mat. 89-91(1 994) 75-80. (29) H.-P.

Photoelectron Spectroscopy as a Tool for Studying Ceramic Interfaces: A Tutorial Friederike C. Jentoft, Gisela Weinberg, Ute Wild, and Robert Schlogl Fritz-Haber-Institut der Max-Planck-Gesellschaft Faradayweg 4-6, 14195 Berlin, Germany 1. Introduction to XPS 1.1 Physical Background and History The photoelectric effect was discovered 1887 by Hertz. Electrons can be released from (solid) matter by irradiation. The photoelectron process is schematically shown in Fig. 1-la. In 19 14, Rutherford stated the basic equation of photoelectron spectroscopy,

I

a

b

--n-o-C

Fig 1-1. (a) Photoelectron Process: (b) X-Ray Fluorescence: (c) Auger Electron Process

which connects the excitation energy (photon energy hv) to the kinetic energy (Ek)of the released electrons and the electron binding energy (Ed: Ek = hv - EB

(Equation 1. I )

If the photon energy of the excitation is known, and the kinetic energy of the electrons is measured, the electron binding energy can be calculated. In practice, the spectrometer work function has to be included in equation 1. It was Siegbahn who predominantly developed the ‘Electron Spectroscopy for Chemical Analysis’ (ESCA, an acronym for X-ray Photoelectron Spectroscopy = XPS) as analytical tool and who discovered the ‘chemical shift’. The technique was established in the late 6Oes to early 70es, and a number of summaries and reviews are given in the literat~re’.~. XPS is a surface sensitive technique. While the X-rays penetrate the specimen, the mean free path of the released electrons is limited. The electron escape depth is determined by the kinetic energy of the electrons and by the properties of the material (Fig. 1-2).

176

Characterization

1

I

1 .o

. o ' r

1

10.0

I

100.0

I

1000.0

Kinetic Energy (eV) Fig. 1-2. Electron Escape Depth plotted versus Electron Kinetic Energy

1.2 Instrumentation A typical configuration of a photoelectron spectrometer is presented in Fig. 1-3. Ultra high vacuum (UHV) conditions are necessary in order to prevent interaction of the photoelectrons with gas molecules. If desired, a clean sample surface can be achieved by various treatments; but even in UHV, clean surfaces can only be maintained for a limited period of time. X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) are being distinguished by the excitation energy. Typical X-ray sources have anodes of aluminum or magnesium, thus emitting photons with an energy of 1486.6 eV (A1 Ka,J or 1253.6 eV (Mg Ka1,2).The line widths are 0.85 (Al) and 0.7 eV (Mg). Monochromators can be used to eliminate contributions of other transitions and to narrow the line width. Helium vapor lamps function as emitters of W radiation, generating photon energies of 21.2 and 40.8 eV. XPS gives information on the core levels while UPS is used for valence band and F e d edge investigations. The electrons are focused with magnetic lenses before they enter the entrance slit of the analyzer which is typically a concentric hemispherical analyzer (CHA) or a cylindrical mirror analyzer (CMA). The hemispherical analyzer can be operated in the Fixed Analyzer Transmission mode (FAT mode, also Constant Analyzer Energy, CAE mode) which allows for quantification; or the Fixed Retarding Ratio (FRR mode, also Constant Retard Ratio, CRR mode) which provides constant relative resolution. Electrons exiting the analyzer enter the detection system which usually consists of a channeltron or channelpiate and a series of amplifying devices.

Photoelectron Spectroscopy as a Tool for Studying Ceramic Integaces

177

Fig. 1-3. Schematic Representation of a Photoelectron Spectrometer with Hemispherical Analyzer

1.3 The Spectrum According to Koopmans’ theorem, the electron binding energy approximately equals the orbital energy of the electrons when relaxation effects are neglected. The primary structure of a photoelectron spectrum, typically presented in units of electron binding energy, thus mirrors the energetic levels (orbitals) of the electrons. The binding energies are mostly specific for a certain orbital in a certain element which enables elemental analysis. If a signal can not be unambiguously assigned to one element, there are usually more signals from other orbitals which allow the elemental identification. Photoelectron spectra always include signals from Auger electrons. The final state of a photoelectron process is an ion with a hole in a core level. In the subsequent relaxation, the hole is filled up with an electron of a higher level. The excess energy of this process can be released in two ways. The energy is either released as a photon which is called X-ray fluorescence (Fig. 1-lb) or a second electron, the Auger electron, is emitted which is called Auger-process (Fig. 1-lc). The two processes are competing, where X-

178

Characterization

ray fluorescence is favored for elements with high atomic numbers and the Auger process is favored for elements with low atomic numbers. The Auger signals differ from the photoelectron signals in their shape and the kinetic energy of the Auger electrons is only dependent on the energy of the three electronic levels which are involved in the process but not on the energy of the incident photons. Secondary structures are observed in the spectrum and they provide additional information. Intrinsic effects originate from the ionized atom. They include core levels which split into doublets due to spin-orbit coupling unless the angular momentum quantum number is zero (s levels). The separation between the two peaks of the doublet can be many electronvolts, increasing with increasing atomic number for a given subshell and also increasing with decreasing quantum number of the orbital angular momentum. The intensity ratio of the doublet peaks is given by the ratio of their respective degeneracy. The splitting of Cu 2p into a Cu 2 p 3 and ~ a Cu 2p1,2 line can be seen in Fig. 5-2. Multiplet splitting (also referred to as exchange splitting), shake-up and shake-off satellites are also intrinsic effects. Multiplet splitting occurs when the emitting atom has unpaired electrons in the valence levels and is strongest in 4f levels of rare earth metals. Shake-up satellites appear when the kinetic energy of the photoelectron is decreased because an electron in the emitting atom is excited to a valence state. Shake-up processes are typical for transition metals and rare earth compounds with unpaired electrons in the 3d or 4f shells. Satellites are of diagnostic value, e.g. Cu(I1) and Cu(1) can be distinguished by their satellite structure but not by their chemical shift. Extrinsic effects, such as plasmon loss signals, occur when the photoelectrons interact with the solid. Apparative parameters can cause ‘ghost’ signals in the spectrum, e.g. additional lines from the source (Korj, KQ, KP) generate X-ray satellites.

2. Chemical Shift and Energy Calibration The binding energy of a specific level of one element can vary by several eV. This shift in the binding energy origins from so-called initial and final state effects. Initial state effects are variations in the oxidation state, lattice site, and, generally spoken, the chemical environment of an atom emitting photoelectrons. Final state effects include relaxation effects, shake-up and shake-off satellites and multiplet splitting. Initial state effects can be described in a simplified charge potential model:

Eb = kQa+ I + V,

(Equation 2.1)

where Eb is the binding energy, Q, is the initial charge on the atom, k is a constant for the respective core level, I is a constant reflecting the reference level, and V, is a Madelung potential term representing the charges on all surrounding atoms. Final state effects are neglected in a ‘frozen-orbital-approximation’ which assumes that the photoionization process is sufficiently fast to exclude contributions from relaxation of the atom (i.e. cation) and the surrounding matrix. Relaxation effects though can be as large as 10% of the binding energy, thus models including the relaxation of intra- and

Photoelectron Spectroscopy as a Tool for Studying Ceramic IntetjGaces

179

extra-atomic electrons provide usually a better match with experimental data. Binding energies resulting from changes i n the valence state can be compensated by relaxation effects. The term ‘chemical shift’ was introduced in analogy to Nh4R spectroscopy and is not consistently used in the literature. ‘Chemical shift’ sometimes refers to any kind of shift in binding energy, but sometimes only to shifts caused by initial state effects. However, due to the chemical shift, XPS does not only allow for elemental but also for chemical analysis. An excellent summary of the variation of binding energies for a number of elements is given in the PE Handbook of Photoelectron Spectroscopy’. As a rule of thumb, the binding energy increases about 1 eV per formal oxidation state. Exact measurement of chemical shifts is closely connected to the energy calibration of the spectrum. Primarily, the binding energy is an absolute value with regard to the vacuum level. For liquids and solids, the work function of the sample must be considered. Samples are grounded to the spectrometer, thus for sufficiently conducting samples such as metals and semiconductors, the Fermi levels equilibrate. The Fermi level can be taken as a reference point, i.e. the binding energy is set zero at the Fermi level.

3. Sample Charging Photoelectrons removed from insulating samples are not easily replaced by electrons from the infinite reservoir of the spectrometer, and the sample may build up a positive charge. A charge can be uniform which just shifts the whole spectrum or it can be differential, depending on the type of sample. There is no common reference level for photoelectrons originating from different areas of a differentially charged surface. Inhomogeneous samples, layered structures, or supported catalysts are good candidates for charging effects. Charges may be constant, or they may vary with time which usually means that the sample is being altered in the X-ray beam or under UHV conditions. The order of magnitude of charging effects is in the eV-range, thus making it difficult to differentiate a charging effect from chemical shifts. Similar problems of finding a reference arise when adsorbed gases are investigated, or (supported) metal clusters of various sizes. or alloys. Charges can be compensated by ‘pouring’ low energy electrons from a flood gun onto the sample; however, if differential charging occurs, only part of the sample can be charge-balanced. Alternatively, internal references can be chosen to calibrate the energy scale such as signals of an inert species present in a series of samples, or signals of gold which has been evaporated on the sample surface, or the carbon Is signal of the omnipresent carbon contaminations. However, non of these techniques is without problems. The Auger parameter is another possibility to correct for charging effects. It combines an XPS-binding energy with the kinetic energy of an Auger transition in adding both the values:

a = Ek(Auger)

+ Eb(XPS)

(Equation 3.1)

180

Characterization

While the XPS-binding energy increases on a positive charge of the sample, the kinetic energy of the Auger electron decreases by the same value, and through the addition the charge disappears. Furthermore, the Auger parameter can be used as a fingerprint for characterizing chemical states, i.e. in cases when initial and final state effects compensate (oxidation states versus relaxation).

4. Specifications of XPS XPS gives elemental and chemical information; it is surface sensitive with a typical probing depth of 3 monolayers; the sensitivity is approximately 0.3%. The data can be quantified, thus surface compositions can be determined. The limit of the spatial resolution is ca. 5 pm. The technique is well established and sufficient support data are available. The preparation of the specimen is generally not very difficult and the amount of sample needed is small (analyzed area typically a few mm2). In comparison to the ion beam techniques, XPS is a priori non-destructive. However, some materials may be altered in ultra high vacuum (reducing conditions) andor by the X-rayirradiation. Results may not necessary be representative for the sample under ambient/operating conditions (pressure gap). Charging effects can be a serious problem in determining chemical states.

5. Copper supported on Aluminum Nitride

Fig. 5-1. Scanning Electron Micrograph of a 300 nm copper film on aluminum nitride before (a) and after (b) exposure to I5 mbar methanol at 625K (back scattering electrons).

181

Photoelectron Spectroscopy as a Tool for Studying Ceramic Interfaces

5.1 Objectives Copper is a catalyst for the selective oxidation of methanol to formaldehyde with oxygen. A model catalyst system with a sharp interface for the copper oxygen interaction was sought to investigate the reaction mechanism. A thin copper film was deposited on Aluminum nitride wafers (AIN, electronic grade). AIN is highly stable towards oxygen and provides good thermal conductivity which allows the heat of the exothermic reaction to escape and therefore overheating of the surface is prevented. However, a rapid failure (after 15. min.) of the model catalyst was observed during operation in methanol conversion at 625 K.

5.2. Characterization by Scanning Electron Microscopy and Catalytic Activity SEM analysis of the AIN shows smooth hexagonal features of approx. 3-5 pm dimension before and after deposition of the copper film (Fig. 5-la). No copper particles can be detected after Cu deposition, and Energy Dispersive X-ray (EDX) analysis confirms that the copper is well-dispersed. The hexagonal structure becomes fuzzy and porous during oxidative treatments. The model catalyst failed rapidly under reaction conditions and the copper film was transformed into copper particles of almost p n size(Fig. 5-lb).

5.3. Quantitative Surface Analysis by XPS Table 5-1. Quantitative XPS analysis of the surface composition of aluminum nitride; content in at%.

Treatment as received after heating after oxidation after ISS 2nd heat. 2nd ox. Treatment __ as received after heating after oxidation after ISS 2nd heat. 2nd ox.

.

Nals 1.6 3.6 7.0 3.6 2.7 4.3 Bls -

--

-

4.6 3.1 3.9 4.2

Cu2p 0.2 0.5 0.3 0.3 0.5 0.4

.

Fls

0.7 0.8 1.5 1.0

01s 29.3 29.0 35.5 35.1 34.8 42.3

Nls 16.0 21.0 16.6 19.2 19.2 13.4

Ca2p 0.8 1.0 1.4

1.2 1.1 1.2

K2p 0.3 0.4 0.8 0.2 0.4 0.4

C__ - l 2-~__ P 2 p___-. Si2s . - A 1 2-~___ AI/N S2p __ __ 0.5 1.5 0.7 19.0 1.2 0.6 0.6 1.9 1.1 29.9 1.4 0.0 0.8 1.7 2.2 28.2 1.7 0.1 0.6 0.5 2.3 33.1 1.7 0.0 0.5 1.0 2.2 32.2 1.7 0.0 0.8 1.0 2.3 28.4 2.1 0.05

Cls 29.5 10.9 0.0 0.0 0.0 0.0

-- .

182

Characterization

The results of the relative quantitative surface analysis of the aluminum nitride substrate as it was received are summarized in Table 5-1. Various treatments were applied subsequently (top to bottom of table, all treatments in UHV if not specified otherwise). Substantial fractions of carbon and oxygen indicate that the typical carbon containing contaminations are present and that the surface is (partly) oxidized. Traces of sodium, calcium, potassium, chlorine, sulfur, phosphor and silicon are found. These elements are typical for sintering additives which are used to form wafers from aluminum nitride powder. The aluminum to nitrogen ratio is close to the stoichiometric ratio of 1. The sample was then submitted to various treatments. Heating can lead to (i) evaporation of volatile surface contaminations (ii) segregation of bulk impurities to the surface (iii) changes in the dispersion of surface compounds, by either spreading (increase in dispersion) or agglomeration (decrease). XPS quantitative analysis is very sensitive to changes in the dispersion. Heating in oxygen can cause (i) everything stated for heating and (ii) combustion and evaporation of surface species. Any treatment can alter binding energies. Ion Scattering Spectroscopy (ISS) allows probing of the surface composition while atoms are being removed layer by layer (ideally) as the sample surface is bombarded with ions. The sample was first heated to 773 K for 20 min. Elements which easily form volatile compounds (carbon, chlorine) show a considerable drop in intensity, also indicating that the topmost layer predominantly consisted of carbon containing compounds. After removal of these contaminations, the intensity of aluminum and nitrogen signals naturally increases. Since the intensities of the signals of the other present elements also increase slightly, they are obviously not surface contaminations but part of the aluminum nitride. Carbon is not detected anymore on the surface after treatment at 1 bar oxygen and 773 K for 2 hours, showing that the carbon containing species were burnt off. Since the combustion of carbon contaminations is highly exothermic, the surface temperature has possibly exceeded 773 K. The decrease in the aluminum and nitrogen content of the surface and the increase in oxygen and the metallic impurities can be explained by the formation of oxidic species of the impurities on the aluminum nitride surface. Concomitantly, an increase in the aluminum to nitrogen ratio is observed. The composition after ISS (2 keV He) reflects which elements were in the topmost layer and were thus removed. The subsequent heating step again leads to an enhancement of the surface concentration of the listed impurities. This is consistent with the presence of remainders of sintering additives in the bulk which diffuse to the surface upon heating. The low electronegativity elements are easily oxidized, leading to a fairly high oxygen content on the surface after another oxidation step. XPS quantitative analysis shows that (i) aluminum nitride contains numerous impurities and (ii) is not inert towards oxygen as a consequence of the presence of low electronegativity metals.


183

5.4. Interpretation of XPS Binding Energy Shifts The signal of the copper impurity was taken as a reference for the energy calibration of the spectra. The two peaks seen in Fig. 5-2 are the Cu 2pjI2and, towards higher binding energy, the Cu 2pI12signal, respectively. The expected intensity ratio of 2pv2 to 2~112is 2: 1; but here it is obscured by satellite contributions. After the treatments in oxygen, the copper binding energy was set to the value of copper (11) oxide while in all other cases it was set to the binding energy of metallic copper. The presence of Cu(I1) is indicated by the weak broad shake-up satellite between the two peaks in spectra c and f. The described method of referencing gives a consistent picture of copper binding energies throughout the various treatments. Heating does not alter the binding energy while oxidative treatment increases the binding energy. The ion scattering experiment leads to a decreased Cu binding energy, i.e. either the oxidized form of the copper is removed from the surface and copper of a lower valence state present in the bulk becomes exposed, or the copper is reduced as a result of preferential sputtering of oxygen. The series of aluminum signals in Fig. 5-3 referenced to copper as described shows some unexpected trends: the binding energy seemingly increases i n the first heating step, and seemingly decreases in 4*104 the first oxidation step. The nitrogen signal, see Fig. 5-4, shifts i n a similar manner. Reasons for untypical binding energy shifts can be (i) incorrect binding energy referencing, e.g. as a 3*104 consequence of differential charging, or (ii) binding energy changes due to a change in oxidation state are compen960 940 sated by relaxation effects, e.g. if the structural environment of the emitting atom is also ig. 5-2. Cu 2p signal of the copper impurity in aluminum nitride (a) AIN as altered. A r~fa'ence received; (b) after heating to 773K in UHV; (c) after heating to 773K in I bar problem is present in the oxygen; (d) afier He sputtering; (e) after second heating to 773 in UHV; (0 here, indicated by afier second oxidation at 773K in I bar oxygen

binding energy (eV)

184

Characterization

the parallel shifts of the A1 and N signals. The assumptions on which the copper reference is based may be faulty. The binding energy of single copper atoms as an impurity in aluminum nitride does not necessarily correspond to the binding energy of metallic copper, and an oxidic copper species formed upon oxidation probably consists of very small particles whose binding energy differs from the binding energy of bulk copper(I1) oxide. Alternatively, charging may be the reason for the observed inconsistencies. The oxidic copper species may be present in the form of islands on the substrate surface. AIN itself as a semiconductor is not expected to build up a charge, but the copper containing islands may be charged, so that there is no universal reference and specifically, copper is not suitable as a reference for Al and N signals. The carbon signal which is often taken as a reference does not provide more insight since the carbon contaminations are gone 4*104 after the first oxidation (Fig. 5-5). Aluminum or nitrogen would not be good references for charged oxide islands on v) Q the surface; and all other 0 impurities are basically afflicted with the same 2*104 problems as copper. Further evidence for charging and differential charging can be taken from Fig. 5-5. Spectrum a, which represents the untreated sample, yields correct C and K binding 80 75 energies after referencing (solid line). The potassium signal is shifted to higher binding 'ig. 5-3. Al 2p signal of aluminum nitride (a) AIN as received; (b) after energy after the oxidation heating to 773K in UHV; (c) after heating to 773K in I bar oxygen; (d) after treatment (spectrum a, He sputtering; (e)after second heating to 773 in UHV;(0 after second dotted line, and spectrum oxidation at 773K in 1 bar oxygen b). The shift has to be due to increased charging of the K environment since a chemical shift can be excluded. Additionally, the decrease in resolution of the two K peaks, the 2p3/?line and the 2pIl2line, indicates differential charging of the potassium containing species in different locations on the specimen. Although signal shapes and binding energies have to be interpreted with great care

70 binding energy (eV)

Photoelectron Spectroscopy as a Toolfor Studying Ceramic Interfaces

185

when charging occurs, it is possible to extract further information from the spectra. The nitrogen Is signal in Fig. 5-4a of the untreated AIN sample indicates the presence of two chemically different species, mostly likely nitride ( ~ 3 9 7eV) and NH groups ( ~ 4 0 0 eV) which terminate the material at the surface. The signal at 400 eV has disappeared after heating (spectrum b), which is consistent with the evaporation or reduction of the NH containing species. A shift of the peak at =397 eV towards higher binding energy after heating and towards lower binding energy after oxidation (Fig. 5-4a, b and c) is not of chemical origin but based on charging and reference problems as described (consistent with the shifts of the Al signal). The oxidation step leads to an asymmetric change of the line profile: Contributions on the high binding energy side of the peak are diminished which becomes obvious from the difference spectrum (inset in Fig. 5-4). Residual NH species with slightly higher binding energy than N in nitride were removed by the oxidative treatment. The aluminum 2p spectra shown in Fig. 5-6 can be interpreted in context with the results of the quantitative analysis. The original material shows A1 predominantly i n a nitride environment (triangles). Heating and sputtering cleans contaminants off the surface, increasing the Al intensity in the spectrum. After the second oxidation treatment (solid line) the signal is not only enlarged but also broadened. The difference spectrum (dashed line) gives a rough estimation of the partition of a new Al species of higher binding energy. Table 5-1 shows that the AI/N ratio is above 1 and increases with each oxidation step as does the oxygen content. It is thus inferred that an aluminum species of oxidic nature is formed on the surface of the AIN.

2*1o4 v)

9. 0

1*104

400 398 396 39

0 400

395 binding energy (eV) Fig. 5-4. N Is signal of aluminum nitride (a) AIN as received; (b) after heating to 773K in UHV; (c) after heating to 773K in I bar oxygen

186

Characterization

4000

I

I I

.!

I

c 1s

2000

0

300 290 280 binding energy (eV) Fig. 5-5. C Is signal of aluminum nitride (a) AIN as received; (b) after heating to 773K in UHV and heating to

773K in I bar oxygen

shift: 0.6 eV

80

75 70 binding energy (eV)

Fig. 5-6. A1 ?p signal of aluminum nitride (triangles) AIN as received; (solid line) after second heating to 773K in 1 bar oxygen; (dashed line) difference spectrum


187

6. Conclusions Aluminum nitride, electronic grade, was not suitable as a substrate for the investigation of methanol oxidation because (i) the copper forms agglomerates under reaction conditions, maybe due to a change in adhesion conditions when impurities emerge from the bulk and (ii) the AIN shows oxygen activity. XPS has been extremely helpful in analyzing the composition of aluminum nitride and in understanding the problems associated with using aluminum nitride as a substrate for copper to form a model catalyst. At the same time, it was shown how difficult the referencing and the interpretation of binding energies can be when a composite material is investigated. In-situ modification of the specimen by heating, oxidation and sputtering was an important tool to distinguish surface and bulk impurities and to investigate the stability of the material in different environments.

7. References 'Practical Su$ace Analysis: Vol. I , Auger and X-ray Photoelectron Spectroscopy, (Eds.: D. Briggs, M.P. Seah), John Wiley & Sons, Chichester, 1990. ' C . DefossC in Chemical Industries I S , Characterization of Heterogeneous Catalysts (Ed.: F. Delanney), M. Dekker Inc., New York, 1984. 'R.S. Swingle, W.M. Riggs, Critical Reviews in Analytical Chemistry 1975, 5. 267. 4K. Siegbahn, C. Nordling, A. Fahlmann, R. Nordberg, K. Hamrin, J. Hedman, G. Johansson, T. Bergmark, S.E. Karlsson, I. Lindgren, B. Lindberg, ESCA Atomic, Molecular and Solid State Structure studied by means of Electron Spectroscopy, Almqvist and Wiksells, Uppsala, 1967. 5J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer Cooperation, Physical Electronics Division, 1992.

Corrosion: No Problem for Precursor-Derived Covalent Ceramics? K.G. Nickel Eberhard-Karls- Universitat Tubingen Institut fur Mineralogie, Petrologie und Geochemie 0-72074 Tubingen, F.R.G.

Abstract The few data on the corrosion behaviour of the new percursor-derived covalent ceramics allow only a very preliminary assessment. Thermal stability experiments indicate a reluctant crystallisation of amorphous phases, in particular when they contain boron. These data are not pertinent to the active oxidation behaviour, which is probably very similar to pure Si3N4 and Sic. Experiments up to 1600°C in air do show a very promising passive oxidation resistance, which is certainly much better than standard sintered non-oxide ceramics. However, TGA data do not prove a superiority over highpurity Si3N4.Si-C-N-types seem to be equivalent to pure Si3N4and for boron-containing types the stoichiometries of the reactions indicate low to zero weight gain during scale growths. Envisaged corrosion problems to be investigated include the influence of H20 and externally applied impurities. The new materials are probably susceptible to molten salt corrosion: the role of boron is unknown, but those whith free carbon are likely to show strong degradation. 1. Introduction

There are but few data on the corrosion of precursor-derived covalent ceramics, simply because they are too new to have received much attention in the corrosion community. Pure thermal stability has been addressed as extremely high for some types [ 11, but this should not be confbsed with the interaction with a hostile environment.

For non-oxides air is thermodynamically already a hostile environment but the proclamation of high hopes (,,resitant to high-temperature oxidation"[2]) based on a small number of experiments in air needs verification in detailed studies. The importance of this statement is easily envisaged if one realises that corrosion resistance is not simply a material constant: Corrosion is a system property. In active corrosion physical parameters related to the gas flow (e.g. velocity, laminar or turbulent character) play an important role. Furthermore real-world applications of ceramics at high temperatures will face a multitude of environments with widely changing compositions, which may not be constant during operation.

Corrosion: No Problem for Precursor-Derived Covalent Ceramics?

189

In this paper we will not undertake another review of basic principles and data from related ceramics like Si3N4and Sic. Such reviews have been presented in recent books and handbooks [3-81. In the following we seek to identify the level of performance and the risks for the new percursor-derived covalent ceramics on the basis of this knowledge. 2. Thermal stability A special case of corrosion would be the simple reaction to a thermal attack without additional chemical interaction. In the case of a crystalline body this may be a phase transformation, which can be detrimental to a component as is well know e.g. for the ap-transition in cristobalite or the m-t-transition in Zr02.

In pure Silicon Nitride we have an a- and a P-modification, a transforms into p by dissolution-reprecipitation processes, often seen in sintering and densification. No fatal component failures due to this are reported. In Sic there are numerous modifications[9], most of them related to layer stacking. Again those will usually be of little concern unless fibers or whiskers are concerned, which may be weakened by phase transformations. Thermal stability can also mean the decomposition of a phase. In SIC or C the upper temperature limits are so high that it is difficult to investigate them. Following thermodynamic calculations and background data reported in the JANAF tables[ 101 C should decompose only at = 3600"C, Sic at =3000°C, eventhough pertitectic decomposition of Sic at temperatures down to = 2500°C has also been reported. Nitrides too can have very high decomposition temperatures: BN at = 2300 -2500°C[101 and Si3N4at about 1865°C. Those values denote the temperatures at which 1 atm. N2pressure is developed by the substance. From this point-of-view reports on phases in the system Si-B-C-N to be stable up to 20OO0C[I ] do not seem to be record-breaking. However, precursor-derived ceramics are amorphous and have stoichiometries where detrimental structural and chemical changes may occur on crystallisation. [ 1 11 and For the carbonitrides the stoichiometry of the amorphous product is Si3+xN4Cx+y thus in terms of possible crystallisation products there is always C in addition to Si3N4 and Sic. As will be outlined below Si3N4and free carbon are not stable together at T 2 1405°C and hence a crystallisation-inducedpartial decomposition is predicted.

To make fair comparisons between established high-purity ceramics and precursorderived material we first have to look at the crystallisation behaviour of amorphous substances in the Si-(B-)C-N system. Sic from CVD-processes usually comes directly in crystalline form. Precursor-derived Sic is amorphous but shows already at P100O"C crystallisation[l2] and studies in oxycarbides indicate likewise that at least at T 2 1200°C the crystallisation of Sic proceeds[ 131.

190

Characterization

From oxidation studies on CVD-derived Si3N4 [14] we know that amorphous Si3N4is very sluggish in its crystallisation behaviour. After 10 h at 1550°C the first isolated ctSi3N4spots develop, but XRD-detectable amounts of Si3N4 develop only weakly at 1600°C and significantly at 1650°C in 10 h[ 151. In the middle between those two we have the precursor-derived Si-C-N-ceramics where micro/nanocrystalline Si3N4and Sic appear at = 1400-1500°C [ 161. In ceramics from polyborosilazanes on the other hand ordering effects are evidenced only at 1700°C and strong crystallisation at 18OO0C[171. The crystallisation of Si3N4at 1750°C in this system was also confirmed by Baldus et al. [ 181 who reported even higher crystallisation temperatures, sometimes exceeding 1900°C. In this respect there are significant new and higher values of crystallisation temperatures for the new percursor-derived materials. The question arises why we then have reports of substantial mass loss in the stationary noble gas atmosphere with Si3N4destructed at = 1450°C [l]. The reason is to be sought in the experimental condition. In graphite crucibles we have an additional parameter: excess carbon. And with this we induce the destruction of Si3N4by the reaction Si3N4+ 3 C a 3 Sic + 2 N2 (2) Just like reaction { 1} it is dependent on P(N2) but the temperature for I bar of N2 is much lower: the calculation yields = 1405°C for the equilibrium temperature. The effectivness of this reaction is well known from the sintering of Si3N4-SiCcomposites 1191. Therefore the thermal-stability-experimentsdo show that in Si-C-N-ceramics reaction { 2) takes place at about the thermodynamically predicted temperature. Its suppression by oxidation [2] indicates an internal pressure build-up of several bars. In boron-containing systems the initial Si3N4-crystallisationwithout decomposition may indicate that the carbon is not present in the form of a reactive free carbon: boron may keep carbon in some other (amorphous?) form to suppress or to retard the kinetics of reaction (2). Thus a fair comparison of thermal stability with simple Si3N4has to use the 1865°C value and this is close to the temperatures for most of the polyborosilazane-derived materials. There is a danger lingering behind the definition of thermal stability via crystallisation: most of the studies do not include systematic time studies and thus it is not clear, whether a long heat treatment at lower temperatures would not have shown the first signs of crystallisation / decomposition. Furthermore, it is a very old experience that the crystallisation of amorphous substances is influenced by numerous parameters[20]. Therfore we consider it very dangerous to take the observed crystallisation behaviour as a fixed value. The issue of thermal stability is linked to a corrosion type usually addressed as


191

3. Active Oxidation

Active oxidation is defined here as the partial or complete transformation of the ceramic into gaseous species with according mass losses, the name ,,oxidation" is only a historic phrase, it does not need to involve oxygen. To see the principle of it we may refer to the simple decomposition reaction Si3N403Si+2N2

(1)

which was implicitly already mentioned, because it defines the stability temperature of Si3N4to be 1865°C at 1 bar N2. '%N1

Wh'4 a1 I bai

t Nz

dl

1

bd1

t 0

Fig. 1: Equilibrium calculations for Si3N4 with or without external N2 supplied

This reaction can of course be calculated by hand, given the thermochemical data[ 101. Using the thermochemical program ChemSage[2 11 to include all possible gas species (e.g. Si, Si2, Si3, N) we may calculate the resulting equilibrium, if you put Si3N4into an ,,inert" atmosphere(Fig. 1a) at 1 bar. It is obvious that Si3N4needs a certain N2-pressure and if the external environment does not supply it, it will produce it by itself: it will decompose just enough Si3N4into Si to obtain the equilibrium. Therefore the activity of Si will always be 1 and Si3N4vanishes at a temperature corresponding to 1 bar N2, i.e. 1865°C.

Supplying external N2, as is done in Fig. 1b, will change the role of N2 and Si: now no Si will be present and P(N2) will be very close to 1 bar. Only at 1865°C Si3N4 vanishes again to yield Si. As a net effect noble gases do by no means create an ,,inert" environment for the nitrides. But what is the measurable effect of this situation? This depends strongly on the experimental situation. In a TGA-experiment, where Si3N4is put into a closed volume with a noble gases (as e.g. done in the experiments of Riedel et al [I]) there will be no measurable effect at all, as the total amount of N2 needed to induce the equilibrium

192

Characterization

pressure is very low at total pressures < 1 bar. Only above this value it will be noted as the system keeps to a total pressure of 1 bar and will extract the additional N2 by bleeding from the valve. This will change dramatically if a different set-up is used. If there is a way to remove the produced N2, the decomposition will proceed. This may be realised by a streaming atmosphere. Within this situation the volume flow of the streaming atmosphere becomes a key parameter along with some other physical conditions, e.g. if the flow is laminar or turbulent, the interdiffision coefficients of the gases etc. We have pointed out previously [22] that partial pressures of gas species produced in active oxidation as low as bars can cause very substantial corrosion rates when the appropriate physical surrounding conditions are met. For this reason it must be emphasised that the experiments on thermal stability do not create an assessment of the risks from active oxidation. These can only be gained from detailed studies. Furthermore, if we have active processes, e.g. the preferential leaching and evaporation of boron, there is a change in bulk chemistry from the Si-B-C-N-system towards Si-C-N. Since there are strongly differing crystallisation temperatures it is envisaged that the first can crystallise after some time, and then reaction {2} could lead to failure. A good hint towards the true active oxidation behaviour is nonetheless given in thermal

stability tests. It relates to the oxygen impurities of the precursor material. The mass loss observed and clearly attributed to oxygen removal from silicocarboboronitrides [ 11 starts from T = 1300°C. Thus in an environment with low P(02), at which no protective layer is formed, we will have substantial loss by active oxidation. Like their pure carbidic and nitridic base ceramics the precursor derived materials will have the same problems: An application in the active regime will not make much sense. The classical theorectical treatment of the boundary between active and passive regimes, i.e. the determination of the T-P(O&condition, at which a protective layer is developed or not, was done by Wagner[23] for silicon and was successfully applied to silicaforming ceramics[24]. A key parameter for the calculation of such boundaries is the interdiffusion of gas species in a gas boundary layer. For the new ceramics we do not expect that it will be substantially different to other nitride and carbide ceramics. Clearly this has to be veryfied in experiments. Once we have sufficient oxygen in the environment (roughly l o 6 to and 1600"C, respectively), we enter the regime of

bars at 1100

4. Passive Oxidation Here some very promising data from the new ceramics have been given [2] [25]. The passivation of silica-formers is usually attributed to the slowness of oxygen transport through a growing layer.


193

The details of it, whether in Si3N4duplex scales with oxynitrides develop and to which extend we have reaction control instead of diffusion control is still in debate [26-301. But it is clear that pure Si3N4is much more resistant to oxidation then other silicaformers. From the few data it seems that silicocarbonitridesbehave as good as pure Si3N4[2] and that the boron-containing types are allegedly even more resistant [25]. The first insights into this indicate that the development of a complex scale with BN at the interface to the material is involved. Because of arguments in the paper of Jansen et a1 [25] and personal communication [G. Passing, Bayer AG, see also his contribution to this volume) we have done some thermodynamic calculations. These indicate (Fig. 2) that indeed a scale with vertical development should exist if a strong oxygen potential gradient is present. The assumption of such a gradient is very likely from the knowledge of other silica-formers. The calculations for Fig. 2 were performed for a bulk S i 0 2 + B203 composition SiBCN3 with the condition that equilibrium is obtained (Fig. 2a) S i 0 2 + C +BN or that the Sic-formation due to reaction (2) is Si3N4 + SIC + BN Si3N4 + C + BN b suppressed (Fig. 2b) and that Si2N20is not formed in both cases. The calculated activity of Si3N4in case (a) is 1, Fig. 2: Schematic scale structure for Si-B-C-N-cerarnics predicted by the calculated amount of thermodynamic calculations with reaction {2) (a) or with its Si3N4is very low. It is clear suppression (b) from the graph that in both cases we should have BN+Si02 at the material interface and B203 replacing BN at positions further outwards, where the P(0,) oversteps the equilibrium of the reaction Si02 + B203

2 B N + 1.502-B203+N2

(3).

These predictions shed a light on the measurability of this oxidation by TGA-experiments. The net reaction equations for Si3N4and the new ceramic are then 1/3 Si3N4+O2 e Si02+ 413 N2

(4)

SiBCN, + 2.25 O2e Si02+ % B203+ CO + 1.5 N2 ( 5 ) . A mass balance shows that during Si3N4-oxidationwe have weight gain of app. 13 g for

every mole (60g) of Si02 produced. In SiBCN3we will have a weight gain of only 2 g for 95 g of Si02+B203resulting. Even substantial layer growth on specimen with several cm2of exposed surface will record in the pg-range!

194

Characterization

If additional evaporation of B from the scale is taken into account, we may end up with a zero-weight gain during scale growth or even a weight loss. Thus it is no question that oxidation rate constants (Kp) in terms of [(weight change / area)2/time]will show extremely low values, even if the yield of scale or penetration is high. A thorough investigation needs clearly the determination of scale thickness plus structure as a function of time. Thus as yet there is no prove for a strong superiority of the new materials over the oxidation resistance of high-purity Si3N4. Nonetheless the experimental evidence for a good behaviour of the new materials up to 1600°C in air [2, 251 is certainly impressive. High-purity Si3N4 begins to exhibit problematic features like bubble formation and scale spallation at T21600"C [ 14, 3 11. It is not known whether these features can be suppressed by boron. It will be an area of highly interesting studies. Water is a part of most atmospheres of interest. While it has now been demonstrated that water by itself does increase oxidation rates of silica-formers only moderately and that its main effect is to be the carrier for impurities [32, 331 it still needs consideration when assessing the behaviour of the new materials.

H 2 0 is soluble in glass melts [34], it influences the crystallisation behaviour of silica, controls the volatilisation of oxides from it [35] and in particular there is a strong interaction with boron [36-381. It is almost certain that the scales of boron containing carbonitrides will react strongly to different water contents of the atmosphere. 5. Hot Corrosion All Silica-formers are prone to show a degrading reaction with molten salts (,,hot corrosion"), which are known to form in a number of combustion environments. Model substance for this process is Na2S04. Detailed descriptions of basic processes have been reviewed by Jacobson et al. [39, 401. A main destruction mechanism is the formation of Na-silicates from silica: Fig. 3: SEM-surface view of a Si3N4(NC 132) after 30 min. ofNa2S04-treatmentat 900°C [41]

Na2S04+ x Si02 CI Na20.(Si02), + SO3 { 5)

Both the sulfate and the silicate are liquid at T > 884°C. In recent studies[42] we have shown that the interaction of silica-forming non-oxides and sulfate melts involves a process, where the non-oxide acts as reducing agent. The effectivity of this process is


I95

strongly enhanced by surface tension dynamics of the two liquids involved: A characteristic surface morphology is created (Fig. 3), which helps in consuming the silica and provides sites for bubble formation and pit growths [41]. Precursor-derived ceramics should act likewise. When they contain boron we will have top scale compositions in the Na-B-Si-0-system, i.e. a classical glass forming system with low melting points and viscosities[43]. Apart from the hot corrosion degradation by itself it is completely unknown whether this will induce crystallisation of the amorphous precursor-derived ceramics. A well documented effect of hot corrosion enhancement comes from studies in S i c with free carbon [44]. Hence it is almost unavoidable to conclude that those precursorderived ceramics, which do contain free carbon, will also be very susceptible to hot corrosion. We thus have good reason to to be cautious towards the hot corrosion resistance of the new materials and investigations into it must be made before applications in combustion environments can be considered.

References 1.

2. 3.

R. Riedel, et al., Nature 382, 796-798 (1996). R. Riedel, H.-J. Kleebe, H. Schonfelder, F. Aldinger, Nature 374, 526-528 (1995). R. E. Tressler, M. McNallan, Corrosion and Corrosive Degradation of Ceramics, Ceramic Transactions (American Ceramic Society, Westerville, OH, U.S.A., 1990), vol. 10.

4. 5. 6. 7. 8. 9.

10.

11. 12.

13. 14.

K. G. Nickel, Corrosion of Advanced Ceramics - Measurement and Modelling, NATO AS1 Series E (Kluwer Academic Publisher, 1994), vol. 267. Y. G. Gogotsi, V. A. Lavrenko, Corrosion of High-Performance Ceramics (Springer Verlag, Berlin, F.R.G., 1992). L. Gmelin, Silicon. F. Schrlider. Ed., Supplement B 5dl: Silicon Nitride (SpringerVerlag, Berlin, ed. 8th, 1995), vol. 15. L. Grnelin, in Gmelins Handbuch der Anorganischen Chemie . Silicon. (Verlag Chernie, Weinheim, 1986) pp. 325-395. K. G . Nickel, P. Quirmbach, in Technische Keramische Werkstofle J. Kriegesmann, Ed. (Deutscher Wirtschafisdienst, Ktiln, 1991) pp. Chapter 5.4.1.1., p. 1-76. T. F. Page, in Physics and Chemistry of Carbides, Nitrides and Borides R. Freer, Ed. (Kluwer Academic Publish., Dordrecht, NL, 1990), vol. NATO AS1 Series E: 185, pp. 197-2 14. M. W. Chase. et al., J.Phys.Chern.ReJData 14 (1985). R. Riedel. in Processing of Ceramics, Part 2 R. Brook, Ed. (VCH, Weinheim, 1996), vol. 17B, pp. Chapter 1 1 : 1-50, J. Lucke. S. Lauterbach, G. Ziegler, Entwicklung von prtikeramischen Polymeren f3r die Herstellung von Siliciumcarbid, F. Aldinger, H. Mughrabi, Eds., Werkstoff-Woche '96, Stuttgart (DGM Frankfurt, 1997). G. D. Soraru, G. D'Andrea, R. Camprostini, F. Babonneau, G. Mariotto, J. Am. CeramSoc. 78,379-387 (1 995). T. Narushima, R. Y. Lin, Y. Iguchi, T. Hirai, J.Am.Ceram.Soc. 76, 1047-1051 (1993).

196 15. 16.

Characterization T. Hirai, K. Nihara, T. Goto, J.Am.Ceram.Soc. 63,419-424 (1980). M. Friess, J. Bill, F. Aldinger, D. V. Szabo, R. Riedel, Key Eng.Mat. 89-91, 95-100 ( 1994).

17. 18. 19. 20. 21. 22. 23. 24. 25.

26. 27. 28. 29. 30.

31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

41. 42. 43. 44.

0. Funayarna, T. Kato, Y. Tashiro, T. Isoda, J.Am.Ceram.Soc. 76, 717-725 (1993). H.-P. Baldus, M. Jansen, 0. Wagner, Key Eng.Mat. 89-91,75-80 (1994). K. G. Nickel, M. J. Hoffmann, P. Greil, P. G., Adv.Ceram.Mut. 3, 557-562 (1988). W. Vogel, Struktur und Kristallisation der Glaser (VEB Deutscher Verlag f i r Grundstoffindustrie, Leipzig, 1965). K. Hack, Met.Trans. 21 B, 1013-1023 (1990). K. G. Nickel, H. L. Lukas, G. Petzow, in The SGTE Casebook. Thermodynamics at Work K. Hack, Ed. (The Institute ofMetals, London, 1996), vol. 621, pp. 163-175. C. Wagner, J.Appl.Phys. 29, 1295-1297 (1958). K. G. Nickel, J.Europ.Ceram.Soc. 9,3-8 (1992). M. Jansen, et al., Erforschung multintirer Nitride und Herstellung von nitridischen Pulvern und Werkstoffen, F. Aldinger, H. Mughrabi, Eds., Werkstoff-Woche '96, Stuttgart (DGM Frankfurt, 1997). L. U. J. T. Ogbuji, J.Am.Cerum.Soc.75,2995-3000 (1992). L. U. J. T. Ogbuji, E. J. Opila, J.Electrochem.Soc. 142,925-930 (1995). L. Ogbuji, A Comprehensive Look At The Passive Oxidation Of Silicon Nitride, Microscopy of Oxidation 3, Cambridge, UK (1996). K. L. Luthra, J.Am.Cerum.Soc.74, 1095-1103 (1991). K. L. Luthra, in Corrosion of Advanced Ceramics - Measurement and Modelling K. G . Nickel, Ed. (Kluwer Academic Publishers, Dordrecht (NL), 1994), vol. NATO AS1 E 267, pp. 23-34. K. G. Nickel, E. Butchereit, Cerum.Eng.Sci.Proc. 18 (1997). E. Opila, J.Am.Ceram.Soc.78, 1107-1110 (1995). E. Opila, . (New London, NH, 1995). H. Scholze, GlastechmBer. 32, 81-88 (1959). A. Hashimoto, Geochim.Cosmochim.Actu56,s 11-532 (1 992). H. Scholze, Glustechn.Ber.32,278-281 (1959). H. Franz, J.Am.Ceram.Soc.49, 473-477 (1966). R. Brlickner, J. Navarro, Glastechn.Ber. 39,283-293 (1966). N. S. Jacobson, J.Am.Ceram.Soc. 76,3-28 (1993). N. S. Jacobson, J. L. Smialek, D. S. Fox, in Corrosion of Advanced Ceramics Measurement and Modelling K. G . Nickel, Ed. (Kluwer Academic Publishers, Dordrecht, 1994), vol. NATO AS1 Series E 267, pp. 205-222. C. Berthold, K. G. Nickel, J. Europ.Ceram.Soc. (submitted). C. Berthold, K. G. Nickel, Key EngMat. 132-136, 1588-1591 (1997). G. W. Morey, J.Soc. Glass Technol. 35,270-283 (1951). N. S. Jacobson, J.Am.Ceram.Soc.68,432-439 (1985).

Solid State NMR Studies for Ceramic Characterization Klaus Miiller Institu t fur Physikalisc he C hem ie Pfaffenwaldring 55, D-70569 Stuttgart, Germany

Introduction Non-oxide ceramics containing silicon, carbon and nitrogen are of great interest for a variety of technological application^.'-^ The introduction of heteroatoms like B, Al, etc. gives access to quaternary systems which are known to exhibit unusual material properties, e.g. Si-B-C-N ceramics are prominent candidates for high temperature application^.^ In this connection it has been shown that the thermolytic conversion of suitable polymeric precursors represents a very promising route to produce crystalline ceramic materials.’ S i c and Si3N4 thus can be obtained by the therm ol ysis of pol ys ilanes and pol yc arbosilanes or pol ysil azanes while ternary Si-C-N ceramics are accessible from carbon-containing polysilazanes. In this contribution we focus on the formation of Si-C-N and Si-B-C-N ceramics by the thermolysis of polysilylcarbodiimides and boron-containing polysilylcarbodiimides. The various steps during the thermolysis are followed by multinuclear solid state NMR techniques. In particular, we address to the temperature range between the polymeric precursor and the amorphous ceramic which is almost not accessible by other experimental technique. In the past it has been shown that such NMR methods are well suitable for the evaluation of the local structure in amorphous organic and inorganic material^.^'^

Experimental Materials. Methylvinylpolysilylcarbodiimide 1 (see Figure 1) was synthesized by treating methyldichlorosilane with cyanamide. The boroncontaining polysilylcarbodiimide 2 was obtained by the hydroboration of inethylvinyldichlorosilane with BH3*S(CH3)2 which afterwards was polymerized with bis(trimethylsily1)carbodiimide. Further details about the synthesis can be found elsewhere.* The various samples which were used for the solid state NMR investigations were prepared as follows: A sample was subject to the standard temperature program used for the ceramization process. At a desired temperature point the program was

198

Characterization

stopped and the sample was quenched to room temperature. With this procedure samples were prepared which cover the thermolytic process up to 1000 "C with temperature intervals of 100 degrees. NMR measurements. The NMR experiments were done on a Bruker CXP 300 spectrometer at a static magnetic field of 7.05 T using a Bruker 4 mm magic angle spinning (MAS) probe. 29Si and 13C NMR experiments were performed at 59.60 and 75.47 MHz, respectively, and at sample spinning rates between 5 and 6 kHz. Depending on the amount of protons in the sample either cross-polarization or single pulse detection were used. The I'B NMR spectra (single pulse spectra, 45 degree excitation) were recorded at 96.29 MHz with spinning rates between 10 and 12 kHz. 'H NMR spectra were taken at 300.13 MHz at a spinning rate of 10 kHz. The chemical shift values are given relative to trimethylsilane (I3C, 29Si,'H) and BF3.OEtz ("B), respectively.

n

Fig. 1. Chenlical structures of the polymeric precursors studied here.

Results and Discussion In the following we present multinuclear solid state NMR investigations on the thermolysis of polysilylcarbodiimides and boron-containing polysilylcarbodiimides for ceramic preparation. It has been shown that such NMR techniques are very suitable for the evaluation of the local structure and their changes during the various steps of cerarnizati~n.~-*~ This particularly is true for the temperature range where only amorphous samples are available and techniques like X-ray studies usually fail for the determination of the local structure. Depending on the specific

Solid State NMR Studies for Ceramic Characterization

199

composition of the polymeric precursor various nuclei can be monitored during the NMR investigations. In the present case I3C, 29Si, 'H, "B and ' N NMR experiments might be performed. The I5N nucleus, however, requires isotopic enrichment and so far has not been considered. In such multinuclear NMR studies the structural information primarily is obtained from the observed chemical shifts of the corresponding NMR signals. With the known chemical shift values from well-defined model compounds or structural components a reliable structural analysis of the present ceramic systems thus should be In addition, other spectral editing techniques like depolarisation experiments might be used " ~ the interpretation of "B during the I3C and 29Si NMR s t ~ d i e s . ' ~For NMR data, additional techniques like satellite spectroscopy or nutation experiments are available. lS.l6

Fig. 2 Experimental 2ySiand I3CNMR spectra recorded for sample 1.

In the following we report on 13C, 29 Si and 'H NMR studies of polysilylcarbodiimides and 2. For the boron-containing precursor 2 also 11 B NMR experiments have been performed. To begin with, we discuss the results for boron-free polysilylcarhodiimide 1 which are summarized in Figure 2 and 3. The 29Si NMR spectrum of the polymeric precursor

200

Characterization

(Figure 2) obtained before any heat treatment exhibits a single line at

- 34.8 ppm, as expected for this compound. Upon heating to 200 "C a second signal at -21 ppm shows up which can be attributed to a SiCZ(N=C=N)Z-group and which is due to the cross-linking of the vinyl groups. Further heating to 400 and 500 "C creates SiCN3-units (- 16.1 ppm). Here, most probably cyanogen or acetonitrile evaporate from the sample. It is rather likely that SiC(N=C=N)s-units also are present in the sample which can be taken from the broad spectral component in the high field region (< 35 ppm). Above 500 "C these intermediate structures disappear. Now, SiN4-units evolve which reflect regions with amorphous Si3N4. Significant changes in the 13C NMR spectra are visible at temperatures 2 300 "C. The spectrum at 300 "C exhibits a broad signal at about 24 ppm which can be understood by the build-up of aliphatic methine and methylene groups as a result of the cross-linking of the vinyl groups. This assignment can be done on the basis of depolarisation experiments, as shown in Figure 3. The application of a short contact time (TCP = 40 ps) should result in inverted signals from CHz-groups while for a long contact time (TCP = 1500 ps) positive signals from CH3-groups and inverted signals from CHz-groups are expected. In addition, in the latter experiment the signals due to CH-groups should vanish. From this, it is concluded that the broad signal at 24 ppm is a superposition of aliphatic CH- and CH2-groups. The further heat-treatment is accompanied by the decomposition of the olefinic groups which eventually is followed by the transformation of the methyl groups. Above 500 "C carbon exclusively exists in a graphite-like form which can be deduced from the broad signal at about 116 ppm. The latter result is confirmed by the 'H NMR spectra (not shown) which exhibit at higher temperatures an additional signal in the aromatic region. It is assumed that the aromatic signal assigns those protons which are used to saturate the boundaries of the graphite-like structures mentioned earlier. It should be noted that aliphatic protons although highly diluted - are present up to 1000 "C. Forthcoming studies towards the crystalline ceramic will show whether hydrogen will remain in the sample at higher temperatures as well.


20 1

4 0 p ~ , ~ n = 3 7 ~ ~ CPPI w - . CPPI ,T = 1500 ps, r, = 37 ps T , =

I _

r,

= 3000 p,r, = 0

200

PPm

CP

0

Fig. 3 Experimental I3C NMR spectra (depolarisation experiments) recorded for sample

-1 at 300 "C.

In summary, the thermolysis of precursor 1is characterized by a crosslinking of the highly reactive vinyl groups between 200 and 300 "C. Above this temperature, the hydrocarbon components are decomposed and build up graphite-like structures along with Si3N4-regions. It is found that the major changes in the local structure have taken place below 600 "C. Above this temperature only marginal effects are visible during the I3C and 29Si NMR studies. Similar experiments have been performed for the boron-containing precursor 2.Representative spectra are given in Figures 4 and 5. The 29Si NMR spectra (see Figure 4) indicate the presence of SiCz(N=C=N)z- and SiC3(N=C=N)-units at lower temperatures. With increasing temperature SiC(N=C=N)3-units evolve (signal at -35 ppm at 400 'C). At elevated temperatures these intermediate structures are transformed into SiN4units which is in line with amorphous Si3N4-regions, as discussed previously. It is found that the major structural changes again are completed at about 500 "C. That is, the structure of the starting polymer is destroyed at this temperature which also can be deduced from the corresponding 13C NMR investigations. Interestingly enough, it was not possible to record I3C NMR spectra above 500 OC for sample 2.Although at present the exact reason for this observation is not known, one might discuss several potential explanations. Thus, it might be possible that the carbons possess very long relaxation times (> 1 hour). Another reason might be found in relaxation effects due to the interactions with the

202

Characterization

neighboring boron nuclei which possess a strong quadrupolar moment. Finally, there is a chance that paramagnetic centers evolve during the thermolysis which causes a very effective relaxation of the carbons along with very broad and almost undetectable lines.

"c

200

0

PPm

- 200

200

I

PPm

0

Fig. 4 Experimental 29Si and I3CNMR spectra recorded for sample 2.

The 'H NMR spectra, given in Figure 5, exhibit at low temperatures signals due to aliphatic protons. Above 500 "C again a signal at 7 ppm shows up which, as before, can be attributed to aromatic structural units, i.e. protons at the boundaries of graphite-like structures (see above). The 11 B NMR spectra recorded for the samples above 500" C are typical for quadrupolar nuclei with a non-integer spin and a large quadrupolar coupling constant of about 2.5 MHz. This value usually is observed for boron with a trigonal coordination, such as for hexagonal BN. At lower temperatures the "B NMR spectra indicate a much smaller quadrupolar coupling constant. This is in line with a higher symmetry of the charge distribution around the boron nucleus which actually is expected for a tetrahedral coordination. In this case the fourth binding site of the boron nucleus might be occupied by the nitrogen atom of a neighboring carbodiimide group. It is worth mentioning that the transformation from


203

boron of coordination 4 to boron of coordination 3 is not a sudden process but occurs over a larger temperature range. Again, the "B NMR experiments indicate that the major structural changes have occurred below 600 "C. Summing up, the final amorphous ceramics consists of Si3N4-regions, trigonal coordinated boron, most likely BN. The nature of the carbon is still open although from the 'H NMR experiments there is evidence for the presence of graphite-like structures as for sample 1.

I

I

20

I

,

0

PPm

,

I

-20

I

,

200

0

PPm

- 200

Fig. 5 Experimental 'Hand "B NMR spectra recorded for sample 2.

Finally, it should be noted that X-ray and neutron scattering experiments have been performed on these systems at temperatures above 1000 O C up to the crystalline ceramic. These experiments clearly prove that the local structures of amorphous ceramic, as evaluated from the present multinuclear NMR work, is maintained in the crystalline state.l7

204

Characterization

Acknowledgment We gratefully thank J. Schuhmacher for his help during present NMR investigations. We are indebted to Dr. M. Weinmann, Dr. J. Bill and Prof. F. Aldinger for the collaboration in this work. Financial support by the Deutsche Forschungsgemeinschaft, JST and the Fonds der Chemischen Industrie is gratefully acknowledged.

References [ l ] M. Mitomo and Y. Tajima, J. Ceram. SOC. Japan 1991, 99, 10141025. [2] K. Niihira, J. Ceram. SOC.Japan 1991, 99,974-982. [3] F. Wakai et al., Nature 1990,344,421-423. [4] R. Riedel et al, Nature 1996,382,796-798. [5] J. Bill and F. Aldinger, Adv. Mater. 1995, 7,775-787. [6] C. A. Fyfe, Solid State NMR for Chemists, CFC Press, Guelph, 1983. [7] K. Schmidt-Rohr and H. W. Spiess, Multidimensional Solid State NMR and Polymers, Academic Press, London, 1994. [8] M. Weinmann, R. Haug, J. Bill and F. Aldinger, J. Schuhmacher and K. Muller, J. Organomet. Chem., 1997,541, 345-353. [9] G. R. Hatfield, K. R. Carduner, J. Mater. Sci. 1989,24,4209-4219. [lo] J. Lipowitz, H. A. Freeman, R.T. Chen, E.R. Prack, Adv. Ceram. Mater. 1987, 2, 121-128. [ 111 C..Gerardin, et. al., Bull. Magn. Reson. 1990, 12, 84-88. [12] G.D: Soraru, F. Babonneau, J.D. Mackenzie, J. Mater. Sci. 1990,25, 3856-3893. [13] D. G. Cory, Chem. Phys. Lett. 1988,152,431 [ 141 X. Wu and S. Zhang, Chem. Phys. Lett. 1989,156,79 [ 151 K. Herzog, J. Peters, B. Thomas, C. Jiiger, Ber. Bunsenges. Phys. Chem. 1996,100, 1655-1657. [ 161 R. Janssen and W.S. Veeman, J. Chem. SOC.Faraday Trans. I 1988, 84,3747-3759. [ 171 J. Durr, J. Bill, M. Weinmann, F. Aldinger, to be published.

Solid State NMR Studies of Organically Modified Ceramics M. Templin, U. Friedrich, U. Wiesner, H. W. Spiess Max-Planck-Institut fur Polymerforschung, Postfach 3 148, 5502 1 Mainz, Germany

Recently the study of organic-inorganic hybrid materials has become a field of increasing interest."-31These materials are often referred to as an organically modified ceramic (ORMOCER).[41They combine organic as well as inorganic components on a molecular or nano level in the final composite. This intimate mixture leads to a class of advanced materials with unusual features. The potential of these hybrids is related to the combination of properties typical for inorganic compounds like hardness with the favorable mechanical properties of synthetic polymers. Such innovative materials promise new applications in many fields such as mechanics, biology and electronics. For example, such composites can be used for immobilization of enzymes"] or as thin-film coatings on polymers, thereby enhancing the abrasion resistance up to values typical of conventional glass.[41 At this early stage in the development of the field, the design and synthesis of organically modified ceramics with predictable properties is difficult. A deeper understanding of the relationship between the structure and dynamics on the molecular level and macroscopic properties is therefore absolutely essential and necessary. There is thus a need for modern characterization methods that can reveal the behavior of the organic and inorganic components as well as their interplay. In the present study we have therefore investigated the influence of different amounts of aluminum alkoxide as hardening component on the reaction behavior of an organically modified alkoxysilane by characterizing the final materials with advanced solid-state NMR techniques. NMR spectroscopy is a very versatile characterization tool with an unique selectivity[61and is therefore predestined for the structural elucidation of these amorphous hybrids.[7281 In Fig. 1 a schematic drawing

206

Characterization

of such an organic-inorganic composite is shown together with the different nuclei, which are used in our solid-state NMR investigations. 27

Q \ si,

AI-,29Si-NMR

-A'

8 structure and dynamics of the inorganic part

I

\

Aluminum increases surface hardness

'H-,''C-NMR

8 structure and dynamics of the organic part organic groups improve flexibility Fig. 1. Schematic drawing of an organic-inorganic composite together with the nuclei used in the present NMR studies .

The main goal of the present work is to show that the combination of NMR results from different nuclei can lead to a comprehensive understanding of the structural aspects of organic-inorganic composites. Since the organic unit at the alkoxysilane contains a polymerizable group, not only the degree of condensation on the inorganic side, but also the conversion behavior of the reactive organic group under the chosen reaction conditions is addressed. The organic-inorganic hybrids were synthesized by the well known sol-gel process,['] which offers an excellent route for the preparation of metal0x0-based macromolecular networks. The composites were synthesized as recently described using different ratios of (3-Glycidoxypropyl)trimethoxysilane, (CH~O)~S~(CH~)~OCH~CHCHI-O, to aluminum sec-

-

Solid State NMR Studies for Organically Mod@ed Ceramics

207

butoxide, AI(OBuS)3,as the chemical precursors. The obtained xerogels were afterwards characterized by solid-state NMR. As depicted in Fig. 1 'H and I3C NMR were used for the characterization of the organic groups. For solids IH spectra usually result in broad unresolved peaks under magic angle spinning (MAS)" with moderate spinning speed, containing mainly information about molecular dynamics. For structural analysis of the hybrids on the organic side I3C spectra are therefore the main source of information. Representative I3C MAS spectra, recorded with cross polarization (CP),'"] for GLYMO and a mixture with 1 mol-% AI(OBus)3are shown in Fig. 2, together with the assignment of the main peaks. The composite obtained from pure GLYMO still shows the two peaks of the carbon atoms of the epoxy ring at 44 and 51 ppm. There are no new signals compared to the corresponding spectrum of the precursor (not shown here). These results demonstrate that the reactive organic group, the epoxy ring, is stable under the reaction conditions (slightly acidic water and heat treatment at 130°C for 45 min) of the experiment.

1 ?

394

100 : 0

A

Jk& 99 : 1

l

90

~

l

80

'

70

l

60

~

50

l

'

40

l

30

~

l

20

~

10

l

~

l

~

l

0

[PPml

Fig. 2. I3C CP MAS NMR spectra of composites of pure GLYMO and a 99:l mol-% mixture of GLYMO and Al(0Bu')) together with the assignment of the main peaks; the carbons corresponding to peaks I, I1 and 111, correspond to different reaction paths shown in Fig. 3.

'

l

'

208

Characterization

Already the addition of 1 mol-% of AI(OBus)3to the sol changes the I3C CP MAS spectrum of the final xerogel in a dramatic way (see Fig. 2, lower spectrum). The two peaks of the epoxy ring disappear and several new peaks show up. There are no further changes detectable in the spectra when higher amounts of AI(OBus)3 are added. This shows that the aluminum acts mainly as a catalyst for the epoxy ring-opening reaction. The large width of all peaks indicates a broad distribution of chemical shifts, confirming the rigid amorphous character of the hybrids. The various products of the epoxy ring conversion, indicated by the appearance of new signals at low field, will now be discussed in more detail. In general there are three possible pathways for the reaction of the epoxy group. First, the ring can be opened by methanol, produced by the hydrolysis of the silicon alkoxide, resulting in a methylether (I) (Fig. 3). Instead of methanol also water can convert the epoxy group, thereby yielding a diol (II). The third possibility is the reaction of the oxirane with another epoxy ring resulting in oligo- or poly(ethy1eneoxide) derivatives (In).

‘9

-0--Si-oWo I

5)

Fig. 3. The three possible reaction pathways for the conversion of the epoxy ring.

In the recorded I3C CP MAS spectrum, signals corresponding to all three reaction products are identified (see Fig. 2, lower spectrum). Quantitative results for the different reaction pathways were deduced by performing a

Solid State NMR Studies for Organically Modified Ceramics

209

single pulse experiment (spectrum not shown here). This reveals that at least 25 % of the epoxy groups are converted by reaction with methanol or water. This high number of end groups and knowledge of the aqueous reaction conditions used leads us to the conclusion that there are mainly oligo(ethy1eneoxide) derivatives in the hybrid materials, rather than poly(ethy1eneoxide) as suggested in earlier studies.[13,141 Information about the inorganic part can be obtained by measuring 27Al and '9Si NMR. Single pulse (SP) 27A1 MAS NMR is performed to determine what kind of aluminum species are present in the materials. From the spectra (not shown here) it is apparent, that besides small amounts of fivefold-coordinated aluminum mainly tetrahedrally and octahedrally coordinated aluminum species are present in approximately equal amounts. The amount of tetrahedral compared to octahedral coordination decreases a little with increasing aluminum content in the composites. In alumino-silicate gels the fourfold coordinated aluminum is normally ascribed to network aluminum, which is bonded to silicon units (Si-O-Al)."S1 The sixfold-coordinated aluminum can be ascribed to aluminumoxohydroxo complexes, A10,(OH),(H20)z. Further information about the structure of the inorganic side can be gained from 29SiNMR. In Fig. 4 representative 29SiCP MAS spectra are shown for three different ratios (100:0, 90:10, 60:40 mol-%) of GLYMO to Al(OBu')3 together with the assignment to the T-groups nomenclature. Continuous changes in the spectra can be observed with increasing amounts of aluminum alkoxide in the composites. This is in sharp contrast to the behavior on the organic side, where even traces of aluminum are enough to change the structure completely, and demonstrates that the conversion of the reactive organic groups has only a minor influence on the structural development of the silicon units. The addition of Al(0Bu')' to GLYMO leads to three main changes in the '9Si CP MAS spectra. The peaks become broader, intensitiy increases in the area of purely siliceous TI- and T'-groups and the maxima of the T'and T'-groups are shifted to lower fields by nearly 4 and 1 ppm, respectively. The significant broadening observed in the 29Sispectra with increasing amounts of Al(OBuS)3 indicates a broad distribution of environments in the second coordination sphere of silicon due to the amorphous nature of these composites. The wide range of chemical neighborhoods in the second coordination sphere is attributed to the effect

210

Characterization

of Si-0-A1 bonding. The existence of such connectivities is also supported by the low field shift of the peak maxima, because it is known that the chemical shift of silicon atoms with aluminum in their second coordination sphere (Si-0-A1) is moved in this direction."61 R\

'

1

-40

~

-50

1

~

-60

1

-70

~

1

-80

'

1

'

1

-90

[PPml

Fig. 4. 29SiCP MAS NMR spectra of composites with three different molar ratios of GLYMO to AI(OBU')~(100:0, 90:10, 60:40)together with a representation of a T2-group.

From all these considerations it can be suggested that the aluminum must be distributed in a rather homogeneous way throughout the material, otherwise it would be difficult to understand that the 29Si spectra are so strongly disturbed by the aluminum addition. Moreover, the observed increase in intensity in the chemical shift area of purely siliceous TI- and T2-groups is consistent with the appearance of T'- and T'-species such as RSi(OA1)2(OSi) or RSi(OAI)z(OH) in the network, which have their

Solid State NMR Studies for Organically Modified Ceramics

211

resonance frequencies in this area. A simple decrease of the condensation degree of the silicon units as a result of the presence of aluminum cannot be excluded, however. There are two possible reasons for such a decrease. First, the incorporation of aluminum leads to a hardening of the matrix and therefore the mobility of the silicon units is reduced, which is needed for a high degree of condensation. Second, the octahedrally coordinated aluminum acts as a network modifier and thereby decreases the condensation degree. There is another interesting feature in the 29Sispectra that we would like to point out. At -46.5 ppm a new peak arises with increasing amounts of AI(OBU’)~.The chemical shift of this group lies between that of purely siliceous TO- and TI-units. A more remarkable behavior of this group is shown in Fig. 5.

I

Si-0-AI

”/ \

L-

less mobile

Single Pulse

-40

-50

-60 -70 [PPml

-80

-90

Fig. 5. 19SiCP and SP MAS NMR spectra of a 80:20 mol-% mixture of GLYMO and AI(OBU’)~.The recycle delay in the SP experiment was 30 s.

From the comparison of the spectrum recorded with cross polarization and that obtained by a single pulse experiment with a recycle delay of 30 s it is obvious that the new peak at -46.5 ppm is missing in the case of the single pulse experiment, which is in contrast to the rest of the peaks. Only when the recycle delay is longer than 10 min this new peak is appearing slowly in the single pulse spectrum. Obviously the spin-lattice relaxation time T I is more than one order of magnitude longer than what is usually

212

Characterization

obtained for T-groups (20-40~).[~’ It indicates that this group is in a more rigid environment than the other T-units. Such a hindered mobility can be explained by silicon units which are localized in pockets of aluminumoxohydroxo complexes. At the same time the presence of three aluminum atoms in the second coordination sphere around this T3-group [RSi(OA1)3] would explain the observed low field chemical shift. The combination of the various results from the NMR investigations of different nuclei leads to a structural model for the composite of GLYMO with A1(OBuS)3(see Fig. 6).

Fig. 6. Structural model for the GLYMO/AI(OBus)~system.

The main features of the structure are the two different aluminum coordinations (sixfold and fourfold), the three different surroundings of silicon in the second coordination sphere (only silicon, only aluminum or both together) and the links of oligo(ethy1eneoxide) on the organic side, which lead to a higher network density in the hybrid material. In summary, this work demonstrates the power of multinuclear solid-state NMR studies for a better understanding of organic-inorganic hybrid materials on a molecular level.

Solid State NMR Studies for Organically Modi$ed Ceramics

2 13

ACKNOWLEDGMENT M. Templin and U. Friedrich thank the Bundesministerium fur Bildung und Forschung for financial support. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

14. 15. 16.

P. Judeinstein, C. Sanchez, J. Muter. Chem. 1996,6, 51 1 U. Schubert, N. Hiising, A. Lorenz, Chem. Muter. 1995, 7, 2010 D. Avnir, Acc. Chem. Res. 1995,28,328 H. Schmidt, H. Wolter, J. Non-Cryst. Solids 1990, 121,428 M. T. Reetz, A. Zonta, J. Simpelkamp, Angew. Chem. Znt. Ed. Engl. 1995,34,301;Angew. Chem. 1995,107,373 H. W. Spiess, Ber. Bunsenges., Phys. Chem. 1997, 101, 153 M. P. J. Peeters, W. J. J. Wakelkamp, A. P. M. Kentgens, J. NonCryst. Solids 1995, 189,77 F. Babonneau, Muter. Res. SOC.Symp. Proc. 1994,346,949 C. J. Brinker, G. W. Scherer, The Physics and Chemistry of Sol-Gel Processing, Academic, London, 1990 M. Templin, U. Wiesner, H. W. Spiess, Adv. Muter. 1997, 9, 814 J. Schaefer, E. 0. Stejskal, J. Am. Chem. SOC. 1976,98, 103 1 S. R. Hartmann, E. L. Hahn, Phys. Rev. 1962, 128,2042 R. Kasemann, H. K. Schmidt, E. Wintrich, Muter. Res. SOC.Symp. Proc. 1994,346,915 R. Nass, E. Arpac, W. Glaubitt, H. Schmidt, J. Nun-Cryst. Solids 1990, 121,370 A. D. Irwin, J. S. Holmgren, J. Jonas, J. Muter. Sci. 1988,23, 2908 S. Ramdas, J. Klinowski, Nature 1984,308, 521

Characterization of Amorphous Materials by Diffraction Methods P.Lamparter Max-Planck-Institut fur Metallforschung SeestraBe 92, D-70174 Stuttgart, Germany

Abstract A brief outline is presented of the application of X-ray and neutron diffraction for the investigation of the structure of amorphous solids. Wide angle scattering as probe of the atomic short range order and small angle scattering for inhomogeneous structures is addressed. Structural modelling is illustrated.

1. Introduction Figure 1 shows the scattering function of an amorphous Si37C32N31 ceramics plotted versus the wave vector transfer (or scattering vector) q = 4n(sinO)/h, where 2 0 is the scattering angle and h is the wavelength. The structure factor S(q) is in principle the scattered intensity after several corrections and normalization. In the context of the structure of amorphous materials we usually distinguish among two distance-scales which are associated with two scattering regimes: the small angle scattering (SAS) regime and the wide angle scattering (WAS) regime. The WAS regime contains the information about the atomic scale structure, the short range order. That means it probes phenomena on a scale of distances from 1 A up to something like 20 A. The SAS regime contains the information about inhomogeneities in an amorphous material. The inhomogeneities may occur on quite different scales of size, say from 10 8, up to 1000 8, or more, and accordingly, we may have to measure the small angle scattering A-'. If the down to very small scattering vectors q, e.g. down to q = inhomogeneities are not too extended they may be referred to as medium range structure. In the intermediate scattering range around q = 0.1 8,-' to 0.5 A-', corresponding to distances of the order of several 10 A, both

Characterization of Amorphous Materials by Diffraction Methods

215

scattering regimes overlap and their distinction refers rather to the specific instruments employed for their investigation than to structural phenomena.

I

w

3;

Small angle range

wide angle range

inhomogeneities

atomic structure

r = 10...1000

r = 1...20

A

R

1

10-~

lo-’

5

10 =-

I -
denote averages of the scattering lengths bi of the atomic species i over the atomic fractions Ci. The total pair correlation function G(r) is obtained by Fourier transformation of the structure factor: 2

G(r) = - q [ S(q) - 1 ] sin (qr)dq .

n: 0

Besides the G(r)-function, the pair correlation function g(r) = 1 + G(r)/4m and the radial distribution function RDF(r) = 4m2pog(r)are in use which, however, contain the same physical information. From the aerea of the peaks of €2DF(r)-functions coordination numbers are determined. The total experimental G(r) of a system with n components is the weighted sum of n(n+1)/2 partial pair correlation functions Gij(r):

In case of a ternary system (n=3), e.g., we need 6 partial Gij(r)-functions for the structural description of the system (note that Gij(r) ) = Gij(r)). The partial pair density distribution functions

describe the number of j-type atoms per unit volume at distance r from an i-type atom at r = 0, where po is the mean atomic number density of the material.

2.2 Contrast variation Note, that the weighting factors Wij depend on the scattering lengths bi of the components. This can be used to derive the partial correlation

2 17


functions Gi,(r) by application of a contrast variation method. Several independent diffraction experiments have to be performed where the scattering length of at least one of the components is different. The straightforward evaluation of the three partial Gi,(r) from three diffraction experiments has been carried out for many binary systems hitherto, mainly for so called metallic glasses [2]. For the six G,,(r) of a ternary system we would need six experiments, which in practical work is hardly feasible. What can be usually obtained from contrast variation experiments is rather some information on the partial functions than their complete evaluation. In case of covalent amorphous systems, such as Si-C-N ceramics, we benefit from the fact that the well defined distances in individual atomic pairs i-j often appear quite well resolved in the total G(r)-functions. Out of the different possibilities to achieve a contrast variation the two major methods are illustrated in the following.

2.2.1 X-rays arid neutrons,

Si37C32N31

The X-ray scattering lengths of the elements are proportional to the atomic number, whereas for neutrons they vary in a non-systematic way. Thus, the different ratios of the scattering lengths of the components of a sample for X-rays and neutrons, respectively, provide a contrast.

I

SIN

I

I

”

I

1

Si-Si

0.05

I

I

Si-C

Si-N

C-C

N-N

C-N

0.15

0.21

0.11 0.19

0.29

I

I

I

n

N

-

-

‘ 0 s n W L

0.31

c3

0.23 0.26

0.04

0.06

0.10

0

0

1

2

3

4

5

7

6

I

I

I

8

9

10

[A1 Figure 2. Amorphous Si&32N3, : Total pair correlation functions G(r) from X-ray and neutron diffraction and weighting factors Wi, of the partial Gij(r).

11

218

Characterization

Figure 2 shows the total pair correlation functions GX(r)and G"(r) of an amorphous Si37C32N31 ceramics from X-ray and from neutron diffraction [4]. These total G(r) are composed of six partial Gij(r) according to the weighting factors Wij which are listed in Fig. 2. With X-rays we observe two peaks: the Si-N distance in SIN4 terahedra at rSiN = 1.75 8, and the Si-Si distance between neighbouring tetrahedra at rsjsi = 2.98 A. The C-C, N-N and C-N correlations are not detectable with X-rays due to their small weights Wij. With neutrons we see again the Si-N peak with about the same weight and a peak reflecting the N-N distance between the N atoms of a SiN4 tetrahedron at r" = 2.81 8,. In addition, a peak at rcc = 1.38 A reveals direct C-C neighbours occuring in an amorphous graphite-like carbon phase in the Si-C-N ceramics. 2.2.2 Isotopic substitution, Si02-Li20 The isotopic substitution method makes use of the fact that the stable isotopes of an element have more or less different scattering lengths for neutrons. In some favourable cases even isotopes with a negative scattering length exist, which of course is very beneficial for a contrast variation study. In a study on Si02-Li20 glasses [5] Li isotopes were employed. Figure 3(a) shows the g(r) of glassy Si02 without Liz0 from neutron diffraction. The first peak at 1.6 8, reflects the Si-0 correlation with 4 oxygen atoms around Si in Si04 terahedra, and the second peak at 2.6 8, belongs to the 0-0correlation. One of the questions of the investigation was the Li-0 coordination when 40 at% Liz0 was added to the glasses. The curve g6G in Fig. 3(a) was obtained from a sample prepared with the isotope 6Li, which has a positive scattering lengts, b6Li = +O. 18*10-'2cm. The Li-0 contribution comes in with a positive weighting factor, WGO= +0.15, and accordingly with a positive contribution around r = 2 8,. The curve ,g, was obtained from a sample where the isotope 7Li was used, which has a negative scattering length, b7h = -0.22.10-12cm. Now we observe a negative Li-0 peak in g(r), corresponding to the weighting factor Who = -0.33. In the difference function gd = 1.74(gsh - g7~)/0.74in Fig. 3(b) the Si-Si, 0-0 and Si-0 correlations cancel out, and we are left with the Li involving correlations. Now we observe the isolated Li-0 peak at 1.9 8, from which the coordination number of 4 oxygen atoms around Li could be derived.


8

sid

2 19

I

00 +4 - -

A

!'\

A 3 -

-

-

-

-

-

-

97~i

1

1.5

2

2.5 r

3

3.5

4

[A1

0,

2 -

-1

- a) 1

1.5

. 2

I

I

I

2.5

3

3.5

r

4

[A1

Figure 3. Amorphous S O 2+ 40% Li20: a) total g(r), (-) without Li20; (- - -) with 6Li,(- -) with 7Li. b) difference function (see text).

.

2.3 Structural models 2.3.1 General remarks The information provided by pair correlation functions G(r) is restricted mainly for two reasons. First, they present on a scale of atomic distances a one-dimensional description of a 3-dimensional structure. They tell us the distances, but not directly the bond angles between the atoms. Secondly, they present statistical averages of structural parameters, such as coordination numbers, but do not give information about the site-to-site variation of the local order. To answer such questions we have to construct 3-dimensional models which are consistent with the experimental data. There are quite different approaches for this task, and in the following one specific method is selected as an example.

2.3.2 Reverse Monte Carlo model, A1203 Figurc 4(a) shows the experimental G(r) from X-ray and neutron diffraction with amorphous A1203 [6].The first sharp peak at r A l 0 = 1.8 A corresponds to the coordination of 4 oxygen atoms around A1 in tetra-

220

Characterization

hedral A104 units. In the range of the second peak a nice contrast is observed due to the different Wij. Neutrons probe mainly the 0-0 correlation at roo = 2.8 A and X-rays mainly the A1-A1 correlation at rA]Al= 3.2 A.

I

I

I

-6= n

neutron 0

L

0

n

6 1.80

0.21

4.1

3.20

0.55

6

v L

I

(3

x-ray 0

0

0

I

I

I

I

I

2

4

6

8

10

r

120

[A1 0

I

I

I

2

4

6

I

I

I

8 1 0 1 2 r

[A1

Figure 4. Amorphous A1203: a) total pair correlation functions, (0) experimental, (-) RMC model. b) partial correlation functions from RMC; distances bj, pcak widths bij, coordination numbers Zij.

With the Reverse Monte Carlo (RMC) method [7] the atoms of a starting cluster are moved in a random way, within some reasonable constraints, until the agreement of the Gm(r)-functionsof the model cluster with the experimental G(r)-functions is optimized. Using modern computers this can be reached within weeks with clusters of several thousand atoms after some million displacements per atom. The solid lines in Fig. 4(a) show the result of the RMC simulation. The agreement with the experiment is very good. The partial correlation functions, calculated from the RMC cluster are shown in Fig. 4(b) together with the corresponding structural parameters. From the 3-dimensional cluster those structural properties can be established, which are not provided by the experimental one-dimensional G(r)-functions. Fig. 5 shows as an example the histogram of the ZA]O coordination number. We state that the number

Characterizationof Amorphous Materials by Diflaction Methods

22 1

of 4 oxygen atoms around Al, as found by experiment, turnes out to be an average of a distribution ranging from 3 to 5 0-atoms.

0.8

I

I

1

I

I

I

1

2

3

4

5

6

>r V

C

al

&

0.4

2

Lc

0 0

'A10

Figure 5. Amorphous A1203: Histogram of thecoordination number ZAlo from RMC model.

3. Small angle scattering and medium range order 3.1 Theoretical If the system is not homogeneous, but contains fluctuations of the composition andor density on a medium range scale beyond the scale of atomic distances, say beyond 10 A, a small angle scattering effect may be observed [8,9], provided a sufficient contrast exists. In case of the separation of a phase p, the scattering regions, in a matrix m the integrated intensity Q, the so called invariant, is: Q = J I ( q ) q 2 d q = 2 n ( A ~ ) 2~ ( 1 -v),

(5)

where v is the volume fraction of the phase p, and the scattering contrast Aq = q m - qpis the difference of the scattering length densities of p and m. If the scattering regions are not diluted, but their volume fraction is substantial they are positionally correlated. In this case the scattering intensity I(q) can be written as:

222

Characterization

Isp(q) is the single particle scattering as observed if the regions were independent, and the structure factor S,(q) describes the modification due to the distance correlations. A simple, but in practical work very useful approach for the evaluation of a measured small angle curve according to Eqn. (6) is the application of hard sphere models for the functions S,(q) and ISpm~91.

3.2 Pores in amorphous A1203 Amorphous A1203, produced by anodic oxydation of Al, exhibits a strong small angle scattering (SAS) effect as shown in Fig. 6(a) [6]. In this

4

3 n U

= 2

first second neighbours

1 0

0

0.02 0.04 0.06 0.08 0.1

(4 LA-’]

0 0

200 400 600 800 1000 r

[A1

Figure 6. Amorphous A1203:a) small angle neutron scattering from pores, (0) measured intensity I(q), (- -) single particle scattering Isp(q), (-) structure factor S,(q). b) distance correlation function C(r) of the pores.

-

quite simple case the scattering regions are pores in the amorphous matrix, i.e. the scattering contrast is caused solely by density fluctuations (the more complicated case of concentration fluctuations is described in the study on Si-C-N ceramics [ 11). The scattering length density of the pores is zero, and the contrast Aq between the regions and the matrix is given by the scattering length density of the A1203matrix. From the invariant Q of the SAS intensity, according to Eq. ( 5 ) a volume fraction of about 8% pores in the material is obtained.


223

The strong diffraction peak in the SAS curve indicates that the pores do not exist independently from each other, but are positionally correlated. Figure 6 (a) shows the decomposition of the measured curve I(q) into the scattering from a single isolated pore, Isp(q),and the structure factor S,(q) according to Eq. (6). In this case it was done in a rather empirical way by fitting through the measured I(q) curve a hard-sphere Isp(q)-functionfor a diameter of 140 A. Fourier transformation of S,(q) yields the distance correlation function C(r) of the pores, as plotted in Fig. 6(b). This function tells us that the pores are correlated with a prefered distance at 340 A, and that also second neighbours at about 620 A are indicated.

4. Conclusion For a comprehensive characterization of amorphous materials by diffraction methods the investigation of the wide angle- as well as of the small angle scattering regime is important. Contrast variation techniques provide detailed structural information and are essential for the construction of 3-dimensional models on the basis of experimental data. References [ 11 J. Durr, S. Schempp, P. Lamparter, J. Bill, S . Steeb, and F. Aldinger,

[2] [3] [4] [5]

[6] [7] [8] [9]

these proceedings. P. Lamparter and S . Steeb in: Materials Science and Technology,Vol.l, Structure of Solids, V. Gerold, edt., VCH, Weinheim 1993, pp. 217-288. T. E. Faber and J. M. Ziman, Phil. Mag. 11, 153 (1965). S. Schempp, J. Durr, P. Lamparter, J. Bill, and F. Aldinger, to be published. H. Uhlig, M. Hoffmann, P. Lamparter, F. Aldinger, R. Bellissent, and S . Steeb, J. Amer. Ceram. SOC.79,2833 (1996). P. Lamparter and R. Kniep, Physica B 234-236,404 (1997). R. McGreevy and L. Pusztai, Molecular Sim. 1,359 (1988). A. Guinier and G. Fournet, Small - Angle Scattering of X - Rays, New York John Wiley & Sons, London Chapman Hall 1955. 0. Kratky and 0. Glatter, Small Angle X-Ray Scattering, Academic Press, London, New York 1982.

X-ray and Neutron Diffraction Investigation on Amorphous Silicon Carbonitrides J. Diirr, S. Schempp, P. Lamparter, J. Bill, S. Steeb, and F. Aldinger Max-Planck-Institut f ir Metallforschung, SeestraBe 92, D-70 174 Stuttgart, Germany

Abstract X-ray and neutron scattering with amorphous Si24C43N33 ceramics was performed using contrast variation by isotopic substitution of ""'N by "N. The atomic structure was studied in the wide angle scattering regime. The peaks of the total pair correlation functions can be attributed to two phases, namely amorphous Si3N4 and amorphous carbon. No peaks indicating Si-C or C-N bondings were observed. The temperature dependence of the medium range structure of amorphous Si24C43N33 ceramics was investigated in the small angle scattering regime. Contrast variation revealed that the scattering signal is caused by two phases of amorphous Si3N4 and amorphous carbon which are separated already in the as-prepared ceramics. Annealing causes a coarsening of the two phases.

1. Introduction For a detailed understanding of the structure of amorphous ceramics the investigation of both the wide angle scattering and the small angle scattering regime is essential. In the present work amorphous Si24C43N33 ceramics, produced by pyrolysis of a polysilycarbodiimide polymer, were investigated. Contrast variation by combination of X-rays and neutrons and by the isotopic substitution of natural nitrogen and the isotope 15N allowed the identification of the contributions of the individual atomic pairs to the wide angle scattering and the characterization of the inhomogeneities giving rise to the small angle scattering effect. For the details of the experiments and the data evaluation we refer to the original papers [l-31. The equations and definitions for the description of the structure of amorphous materials were presented in the preceding article of this Volume [4] and in the review [ 5 ] .

X-ray and Neutron Diffraction Investigation

225

2. Experimental The Si24C43N33 ceramics were produced by pyrolysis of a polysilylcarbodiimide polymer as precursor [ 1, 61. In order to achieve a low oxygen contamination, the polymers were neither crosslinked nor densified but directly pyrolysed at 1100°C in argon atmosphere. The heat treatment yielded no dense bodies but black sponge like solids of amorphous Si-C-N with a bulk density of 1.84 g/cm3. After pyrolysis further anneals were carried out under argon atmosphere for two hours at 1200 "C, 1300 "C, and 1400 "C, respectively. X - ray diffraction showed that the material was still amorphous after the 1400 "C anneal. The chemical analysis of the ceramics yielded an average composition of Si24C43N33. The densities, determined by mercury porosimetry, yielded values for the atomic number density. po which increased during annealing from po = 0.067 for the as-prepared ceramics at 1100 "C up to po = 0.073 at 1400 "C. For the neutron scattering experiments samples with natural nitrogen, natN,and samples with the isotope ' N were prepared. 3. Atomic structure from wide angle scattering The Faber-Ziman [7] total pair correlation functions G"(r) and G"(r), obtained by Fourier transformation of the structure factors are shown in Figs. 1 (a,b). The total G(r) functions are weighted sums of six partial pair correlation functions, which describe the contribution of the individual Gij(r) atomic pairs i-j. The weighting factors Wij of the Gij(r) are listed in table 1. Table 1: Faber-Ziman weighting factors Wu of the partial pair correlation functions Gu(r); for X-rays the Wu are averaged over the measured range of the scattering angle.

Sample

Si-Si Si-C Si-N

C-C

N-N C-N

Si-C-natN/l100 "C, x

0.215 0.262 0.231 0.083 0.064 0.145

Si-C-natN/llOO"C, n

0.021 0.121 0.127 0.173 0.193 0.365

Si-C-"N/1300 "C, n

0.029 0.158 0.125 0.214 0.134 0.340

226

Characterization

6.0 4.0

2.0 0.0 2.0 70.0

52.0 h

$0.0 2.0

0.0 2.0

0.0 -2.0

0

2

6

4

8

10

0

2

4

6

8

10

r[AI

r[AI Figure 1. Total pair correlation functions, a) from X-ray scattering, b) from

neutron scattering. and I5N, respectively, The contrast between the neutron curves with is not strong, but still useful, whereas the contrast variation between X-ray scattering and neutron scattering yields very different pair correlation functions. With X-rays two distinct peaks at 1.73 8, and 2.97 A can be observed, whereas the neutron G(r) exhibit four distinct peaks at 1.38 A, 1.74 A, 2.43 and 2.81 A. The amplitudes and areas of the peaks were determined by Gaussian fitting. The differences of the peak areas, caused by the contrast variation, in combination with the weighting factors Wij of the partial Gij(r) in table 1 were used for the identification of individual atomic pairs i-j which cause a certain peak in the total G(r). The peak at 1.38 8, is not visible with X-ray scattering, so it might be caused by a C-C or a N-N correlation. However, as the amplitude with neutron scattering is slightly larger for 15N than for natN it can be concluded that this peak is caused by a C-C correlation since the corresponding C-C weighting factors in table 1 show the same tendency.


227

At 1.74 8, a peak is visible with X-ray scattering as well as with neutron scattering. Because of the weighting factors in table 1 this peak can not be due to a Si-Si, a C-C or a N-N correlation. Considering that its amplitude with neutron scattering is nearly the same for 15N and natN,but smaller than with X-rays, it can be attributed to a Si-N correlation. The peak at 2.43 8, can be discussed in the same way as the peak at 1.38 A,i.e. it should be caused mainly by a C-C correlation. The peak at 2.81 8, is not visible with X-rays, so it might be caused by a C-C or a N-N correlation. In this case the amplitude with neutron scattering is smaller for 15N than for natN.Obviously this peak is mainly caused by a N-N correlation. The peak at 2.97 A,not visible with neutrons, is caused by a Si-Si correlation.

:>

I

Si - N Q.73 A, X-ray)

N - Si ( 1.74 A, neutrons)

I

I

\\

C - C (1.38

A, neutrons)

1

I

I

I

1100

1200

1300

1400

T["C 1

Figure 2.

Partial coordination numbers

Z,(jaround i).

The partial pair coordination numbers zj, determined from the fitted Gaussian curves are presented in Fig. 2 in dependence of the annealing temperature. The average coordination number ZsiN of about 4 N-atoms around Si increases slightly from 1100°C up to 1300°C and then decreases between 1300 "C and 1400 "C. The corresponding coordination number ZNsi of about 3 shows the same temperature dependency. The coordination numbers GCincrease distinctly with increasing annealing temperature. The number of nearest C-C neighbours at 1.38 8, increases from 2.6 at

228

Characterization

1100 "C to 3 at 1400 "C , and the number of second neighbours at 2.43 8, increases from 3.9 at 1100 "C to 5.7 at 1400 "C. It is important to note that the pair correlation functions in Fig. 1 (a,b) do not show any peaks which were expected in case of direct Si-C and/or C-N bondings. This feature already indicates that the inhomogeneous structure of the amorphous Si24C43N33 ceramics consists of a system which is phase separated into the two amorphous phases Si3N4 and carbon. In the context of this question a comparison of the observed peak positions up to 3 8,with the distances in related crystalline phases is revealing. The peaks at 1.74 8,, 2.81 8, and 2.97 8, are in very good agreement with the corresponding distances in crystalline Si3N4, Si-N: 1.71 to 1.79 8,, N-N: 2.80 to 2.94 A and Si-Si: 2.72 to 3.15 8, [8]. The coordination numbers Z s i N and ZNSi match well with those of crystalline Si3N4, where Z s i N = 4 and ZNSi = 3. Therefore it can be concluded that there exists an amorphous phase within the investigated materials with SiN4 tetrahedra as structural units. The results indicate that this phase exists already after pyrolysis at 1100°C. The decrease of Z s i N and ZNSi between 1300°C and 1400°C suggests that within this temperature range the thermal decomposition of the Si3N4 hase starts. The C-C peaks at 1.38 and 2.43 8, are in good agreement with the distances in graphite-like carbon [9] at 1.42 8, and 2.46 8, and are distinctly smaller than those in diamond [101 at 1.54 8, and 2.52 A. Furthermore, the coordination numbers at 1400 "C, &C = 3.9 and &C = 5.7, are close to those in graphite, namely 3 and 6. This suggests that there is a phase of amorphous graphite-like carbon within the Si-C-N ceramics. The structural units of this carbon phase might be carbon rings or chains. For the increase in the C-C coordination numbers with increasing temperature there are three explanations. From SAS it is known that the carbon segregates have diameters of 20 8, at 1100°C and 40 8, at 1400°C. The number of carbon atoms at the surface of the carbon segregates, where a small number of bondings to the Si3N4 phase is expected, becomes smaller with increasing annealing temperature, compared to the number of carbon atoms enclosed in the segregate volume, which are expected to have three direct carbon neighbours. The second possibility is that a small hydrogen content of about 6 at%, as detected with neutron scattering, is bonded to carbon. In this case the removal of the hydrogen atoms by annealing should increase the C-C coordination numbers. The atomic densities po of the investigated samples increase from 0.067 up to 0.073 with increasing annealing temperature. Compared with the atomic densities of crystalline Si3N4 (PO = 0.103

1


229

and graphite (PO = 0.113 A-3)[l 11 the densities of the Si-C-N ceramics are rather low. In combination with the C-C coordination numbers this suggests, that the carbon phase is very dilute, i.e. there is a kind of atomic scale porosity with pores below 30 radius and thus not visible with mercury porosimetry. During annealing the atomic scale porosity of the carbon phase decreases and the atomic densities increase. A densification of the carbon phase might be one more explanation for the increase of the C-C coordination numbers with increasing annealing temperature.

4. Medium range stucture from small angle scattering The small angle scattering (SAS) cross sections in Figs. 3 (a,b) reveal that the Si-C-N ceramics are inhomogeneous. Towards very small q values the scattered intensities show a linear increase in the log - log plot. This effect is caused by surface scattering of the powder particles according to Porods law and was subtracted in the further discussion. At larger q values a scattering effect follows which is due to the inhomogeneities in the bulk and which depends strongly on the annealing temperature. The peak in the SANS curves indicates an interference effect, which means a rather high volume fraction of the scattering regions within the samples. With increasing temperature the observed signals are shifted to lower q values. This indicates that the size of the regions is growing with increasing temperature. It is obvious, that the Si-C-15N/1100"C sample yields a SAXS signal which is different fom that of Si-C-natN/llOO"C sample. This was not expected for X -ray scattering, and shows that the two samples, which stem from different batches, are not in the same structural state. However, after an additional anneal at 1300 "C the Si-C-l'N sample showed the same SAXS signal as the Si-C-natNA100 "C sample. From this it was concluded, that both samples were now in the same structural state and thus can be employed as a pair of samples for the identification of the type of the inhomogeneities by isotopic substitution as shown later. The runs of the SANS signals of this pair of samples (Si-C-"N/1300 "C and Si-C-natN A100 "C), in Fig. 3(b) are distinctly different, i.e. a distinct contrast variation due to isotopic substitution is observed. The runs of the SANS curves of the Si-C-""N samples in Fig. 3(b) are comparable to those observed in the SAXS measurements (Fig. 3(a)). The parameters of the medium range structure were determined by fitting to the experimental SAS curves a model of distance-correlated hard

230

Characterization

J

I' 0,Ol

I

I

I

I

0,1

Figure 3. Small angle scattering cross sections, a) X-rays, b) neutrons.

spheres for the scattering regions as described in [1,3]. This yielded the average diameter D of the regions as well as the invariant Q of the scattering function as a measure for the volume fraction. The results are plotted in Figs. 4 (a,b). We state that the invariant remains nearly constant over the whole tem erature range, whereas the diameter increases at the same time from 20 to 40 A. This implies that the ceramics are already phase separated after their production at 1100 "C and that the inhomogeous medium range structure shows a coarsening at higher temperatures. It should be noted that the hard sphere model does not tell directly the chemical nature of the scattering regions. In the following the contrast variation was used for their identification. Different two phase systems were discussed as possible candidates which are listed in table 2. Concerning the system Si3N4 in Si-C-N we note that the chemical composition Si24C43N33 can be subdivided in a formal way into the two phases (Si3N4)8.~3+ NO.67c42.4. Because of this the material also can be treated as a two phase system, namely amorphous Si3N4 with a volume

8:

X-ray and Neutron Diffraction Investigation 1100

1200

1300

23 1

1400

fraction of 37.2% (PO = 0.103 l/A3), and amorphous carbon with a volume fraction of 62.8% (PO c- 0.046 l/A3), which are M X-rays,Si-(;hllN dispersed in each other. For each --cneutrons Si-C-WN candidate the contrast, i.e. the difference of the scattering length densities of the regions and the matrix, Aq, was calculated. For the unknown atomic densities of amorphous Si3N4, S i c or C regions the values of the crystalline phases were used. The scattering lengths of Si, C and N were taken from [ 121. 1100 1200 1300 1400 Table 2 shows the results of the T["CI contrast variation: The different Figure 4. a) invariant Q of the SAS, ratios (Aqa/AT)b)derived from the b) diameter of the Si3N4regions. SANS and SAXS experiments with Si-C-"atN, produced at 1100°C and Si-C-"N, annealed at 1300 "C are compared with the values as expected for the different types of two phase systems. Comparison shows that two possibilities are consistent with the experimental data: A kind of silicon - nitride in a Si-C-N matrix, or amorphous Si3N4, dispersed in an amorphous carbon matrix. In the case of silicon - nitride in a Si-C-N matrix the SAS signal would be caused by the formation and growth of regions within the material. This would be correlated with an increase of 4-

1_1

Table2: (AqJAqb) derived from combination of two experiments a and b with Si-C-N and expected values for different combinations of regions in a matrix.

Experiment Sic in Si-C-N C in Si-C-N Si3N4 in Si-C-N Si3N4 in carbon

1.9 0.5 0.8 1.7 1.7

6.0 9.8 1.o 8.4 8.5

3.1

19.7 1.3 5.0 5.0

232

Characterization

the volume fraction, occupied by the regions, and thus an increase of the invariant with increasing annealing temperature would be expected. However, this behaviour is not observed in Fig. 4a. As the invariant Q remains nearly constant over the whole observed temperature range, the volume fraction of the heterogeneities and Aq remain constant. This behaviour can only be explained with the two phase model: Apparently the material consists of an amorphous Si3N4 phase, dispersed in an amorphous carbon phase. The volume fractions of the two phases do not change by annealing. The two phases are separated from each other already at 1100 "C, where the material was produced. With increasing temperature a coarsening of the inhomogeneous structure takes place. This phase separation might be the reason for the high thermal stability of the amorphous state of the Si-C-N ceramics. The transition of each of the two phases from the amorphous into the crystalline state, i.e. the establishment of a long range ordering, will be hindered by the presence of the other phase.

5. Conclusions Amorphous Si24C43N33 ceramics were investigated by X-ray and neutron wide angle and small angle scattering. Using the contrast variation between X-ray and neutron scattering with isotopic substitution the individual pair correlations could be identified and partial pair coordination numbers were determined. The Si-Si, Si-N and C-C correlations are similar to those in crystalline Si3N4 and graphite, respectively. The number of Si-C and C-N bondings within the amorphous ceramics is very poor. Amorphous Si24C43N33 is phase separated into the two amorphous phases Si3N4 and carbon on a length scale of about 20 8, at 1100 "C, which shows a coarsening upon annealing up to 40 8, at 1400 "C. The increase of the coordination number in the carbon phase indicates that annealing causes a purification and densification of the carbon phase.

Acknowledgements We wish to thank the European Union which supported this work by its Human Capital and Mobility Programme for Large Scale Facilities. We are grateful to A. Soper and R. Heenan for the help with the neutron scattering experiments and to A. Kienzle for support with the sample preparation.

X-ray and Neutron DiffractionInvestigation

233

References J. Durr, S. Schempp, P.Lamparter, J.Bil1, S. Steeb, F. Aldinger; Proc. of Xm"ISRS, Hamburg 8.-12.09.96, to appear in Solid State Ionics, (1997) J. Durr, P. Lamparter, J. Bill, S. Steeb, F. Aldinger, Proc. of NCM 7, Cagliari 15-19.09.97; submitted to J. of Non Cryst. Sol.; 1997 J. Durr, PhD thesis work, University of Stuttgart, (1997) P. Lamparter, this Volume P. Lamparter, S. Steeb, Materials Science and Technology, Vol. 1, Structure of Solids, V.Gerold, edt., VCH, Weinheim 1993, pp. 217-288 A. Kienzle; PhD thesis work; University of Stuttgart; (1994) T. E. Faber, J. M. Ziman; Phil. Mag. 11,153 (1965) S. N. Ruddlesden, P. Popper; Acta Cryst., lJ, 465, (1958) P. Trucano, R. Chen; Nature, 258, 136, (1975) T. Hom, W. Kiszenick, B. Post; J. Appl. Cryst., 8,457, (1975) Alfa, Finest Inorganic Research Chemicals and Metals; Johnson Matthey, Karlsruhe; 1995/96 V. F. Sears, Neutron scattering lengths and cross sections; Neutron News; Vol. 3, No. 3, 1992


V. High Temperature Mechanical Properties and Characterization of Grain Boundaries


Compression Creep Behaviour of Precursor-Derived Ceramics Giinter Thurn and Fritz Aldinger Max-Planck-Institut f i r Metallforschung and Universitat Stuttgart, Institut f i r Nichtmetallische Anorganische Materialien, Pulvermetallurgisches Laboratorium, Heisenbergstr. 5,70569 Stuttgart, Germany The creep behaviour of precursor-derived Si-C-N and Si-B-C-N material was investigated in the temperature range between 1400°C and 1550°C at compressive stresses between 30 and 250 MPa. It was shown that the temperature dependence of creep behaviour of such materials is very low. At high temperatures the creep resistance of precursor-derived Si-C-N material after 100 hours creep deformation is higher than that of the state-of-the-art silicon nitride ceramics. The Si-B-C-N precursor-derived ceramics show a lower creep resistance than the Si-C-N materials, but it could not be distinguished whether this was due to a different processing method or due to the different chemical composition.

1 Introduction Silicon nitride and silicon carbide are the state-of-the-art ceramics for high-temperature applications, because of their high chemical and mechanical stability at elevated temperatures. It was shown, however, that the creep properties of silicon nitride depend on the grain boundary phase which is necessary to promote densification during liquid phase sintering [I]. It is assumed that the deformation in compression tests occurs by a solution reprecipitation mechanism through the grain boundary phase, while cavity formation is the major deformation mechanism under tensile load. Equivalent results can be suspected for liquid phase sintered silicon carbide. It is possible to process Si-C-N ceramics without grain boundary phases by thermolysis of polymer precursors [2, 31. Since the creep behaviour of sintered silicon nitride is determined by the grain boundary phase, the precursor-derived ceramics are suspected to have higher creep resistance than liquid phase sintered ceramics.

2 Experimental Procedure To analyse the compression creep behaviour of polymer-derived ceramics, three different types of materials were investigated: PVSI-derived material with the chemical

' Polyvinylsilazane VT50, Hoechst AG, Frankfurt, Germany

238

High Temperature Mechanical Properties and Characterization of Grain Boundaries

composition SiC,,,N,3, PHMS'-derived material with the chemical composition SiC,,N, and boron-containing T2( 1)l-derived material with the chemical composition Si, oB,,oC, ,N, ,. The Si-C-N materials were produced by plastic forming of crosslinked and ground powder, whereas the T2( 1)-derived powder was compacted by cold isostatic pressing. Both types of powder compacts where thermolysed at 1050°C. The porosity of the PVS- and PHMS-derived material were 4% and 14%, respectively. Details of sample preparation are presented elsewhere [3,4]. The specimen for the creep experiments were cut and ground to a height of 7.5 mm with a cross sectional area of 1 . 5 1.5 ~ mm'. A square cross sectional area was chosen to facilitate the preparation of the specimen. The creep tests were performed in air at temperatures of 1400"C, 1500°C and 1550°C. The temperature was measured with a PtRh( 10)/Pt thermocouple which was placed 2 mm beside the specimen. The load was applied by a spiral spring which was preloaded by a spindle. The compression of the spiral spring was measured by a linear potentiometer and the load of the specimen was calculated from the elongation with an accuracy of f l N. The strain of the specimen was calculated from the displacement of the load pads: Two Sic scanning pins connected the upper load pad with the housing of an inductive strain gage which was placed under the specimen outside the furnace in the cold part of the testing machine. The displacement between the upper and the lower load pad was measured with a third scanning pin which connected the sensor of the strain gage with the lower load pad. Since oxidation of the test samples was observed at intermediate temperatures before a protective layer of silica was formed at higher temperatures, the samples were heated quickly to the testing temperature by using the maximum power of the furnace in order to avoid oxidation. The heating ramp for tests at 1550°C is shown in Fig. 1. I600

t

1400 1200 1000

%

800

$

600 400

Figure 1: Heating ramp for a compression creep test at 1550°C.

200

"0

20

40

60

80

100

Time(min)

120

140

160

180

+

During the heating a clamping force of about 2 N was applied to the specimen. After the test temperature was reached, a waiting period of 30 min was necessary to get the strain measurement system into thermal balance. After the waiting period the testing

PolyhydridomethylsilazaneNCP200, Chisso Corp., Tokyo, Japan PoIyorganosiIazane

Compression Creep Behaviour of Precursor-Derived Ceramics

239

stress was applied within 30 seconds. The strain measurement started after the testing stress was reached. A schematic diagram of the experimental procedure is shown in Figure 2.

f g

0

v)

Figure 2: Experimental procedure of a compression creep test. Timet

+

3 Results 3.1 Creep Resistance For the analysis of the creep results the Norton power-law relation

i: = A(t). on . exp(

"R) . T

was taken, where (T is the creep stress, n the stress exponent, Q the activation energy, R the gas constant, and T the temperature in K. The time dependence is considered by A(t). Figure 3 shows creep curves of the PVS-derived material at a compression stress of 100 MPa at 1400°C and 1500°C.At both temperatures, the strain rate decreases with time in very much the same way. Even after 6.105s stationary creep is not observed. The increase of the strain rate of the 1500°C/100 MPa specimen after approximately 2.10' s is an artefact due to an increase of the room temperature in the testing laboratory, because the air condition was switched off during the weekend. The opposite effect with a decrease of strain rate is observed after 4.10' s when the laboratory was cooled down again. From the strain rates after 100 hours creep deformation an activation energy Q=260 kJ/mol was calculated. Figure 4 shows the influence of compression stress on the strain rate. From the strain rate at 100 hours creep deformation a stress exponent of n=0.7 was determined.

240


10-5

v

2 lo-'

Figure 3: Temperature dependence of the strain rates at a compression stress of 100 MPa for the PVS-derived material.

Y

10-5

t 10" . n d

9

v

10'

Figure 4: Stress dependence of the strain rates at 1500°C for the PVSderived material.

G

'E!

vl

10-8

I o3 Time@)

101

10'

102

I 03 Time@)

I o4

106

+

1o4 +

Figure 5: Temperature dependence of the strain rates at a compression stress of 100 MPa for the PHMS-derived material. 105

106

compression Creep Behaviour of Precursor-Derived Ceramics

24 1

The results of creep tests of PHMS-derived material at a compression stress of 100 MPa and different temperatures are shown in Figure 5. From the creep curves at 1400°C and 1500°C after 100 hours creep deformation an activation energy of 250 KJ/mol was determined. Increasing the temperature from 1500°C to 1550°C the creep rate decreases from 3.2~1Ois.'to 1.7.10-*s-I. The compression creep behaviour of the T2( 1)-derived material is shown in Figure 6 . The specimen tested at 1500°C and 100 MPa fractured after 2 minutes. At lower temperatures (1400°C) and stresses (30 MPa) the fracture of the specimen occurred after 4.5 hours. At 1400°C the creep rates of the boron-containing ceramic were somewhat higher than those of the two Si-C-N ceramics. Fracture occurred at relatively low strains, probably due to the different oxidation behaviour as compared to boron-free materials investigated.

0

Time (s)

-

IWC.30MPa

Figure 6: Creep curves of the Si-B-C-N ceramic derived from T2(1) precursor.

3.2 Oxidation Resistance After 300 hours creep deformation at 1500°C with a stepwise increased compression stress from 100 MPa to 300 MPa a dense passivating oxide layer was observed on the PVS-derived material. Due to SEM studies a multiple oxidation layer with an overall thickness of approximately 10 pm was formed (Fig. 7). PHMS-derived material at 1500°C also forms a dense passivating oxidation layer with a thickness of 2.2 pm (Fig. 8). However, after the creep test at 1550°C no oxidation layer was found on the surface of the specimen, but an area of high porosity near to the surface indicating a decomposition of the specimen (Fig. 9). With the T2( 1)-derived Si-B-C-N materials, bubbles were observed at the surface after the creep test at 1400"C, indicating that an oxidation layer develops which has no passivating effect (Fig. 10).

242


Figure 7: Fracture surface with oxidation layers of a PVS-derived creep specimen tested for 300 hours at 1500°C and compression stresses of 100,200, and 300 MPa.

Figure 8: Fracture surface with oxidation layers of a PHMS-derived creep specimen tested for 100 hours at 1500°C and 100 MPa compression stress.


243

Figure 9: Fracture surface near the lateral surface of a PHMS-derived creep specimen tested for 100 hours at 1550°C and 100 MPa compression stress.

Figure 10: Lateral surface of a T2(1)-derived creep specimen tested for 4.5 hours at 1400°C and 30 MPa compression stress.

244


4 Discussion The creep deformation of the precursor-derived materials investigated under compression is characterised by rather high initial strain rates. However, whereas e.g. at 1500°C and 100 MPa the strain rate of the T2(1)- derived material decreased only slightly (Fig. 11) with time until the sample ruptured at a strain of some 2.3 % (Fig. 12), the strain rate of the PVS- and PHMS-derived material decreased continuously with loading time down to some lo's-' and neither stationary creep nor rupture were observed for both materials even at a loading time of some 167 h. 0.035

1

0.030 T2( ])-Derived Materia

0.025

t 0.020

HMS-derived material

$ 0.015 VI

0.010

VS-derived material

0.005 0.000 0

100000

200000

300000

400000

Figure 11: Comparison of the creep strain of the three different materials at a temperature of 1500°C and 100 MPa compression stress.

Tirne(s) +

[

OOmw PVS-Derived Material A

0

PHMS-Derived Material

I Figure 12: Comparison of the strain rates of the three different materials at a temperature of 15OOOC and 100 MPa compression stress.

It is interesting to note that the total compression strain remained below about 1.5% and 3% for the PVS- and PHMS-derived material, respectively. Although the deformation of both the PVS- and PHMS-derived materials varied somewhat from specimen to specimen during the first hours of creep at longer loading times both material types behaved very much the same way.


245

The main reason for the high values of the strain rate in the initial deformation state and their decrease with loading time is probably a superposition of creep deformation and time-dependent shrinkage. Since the maximum processing temperature of the asthermolysed materials had been 1050°C the deformation at higher temperatures is partly due to further densification of the sample. The mechanism of this obviously stressindependent process is however not clear yet. The early fracture of the T2(1)-derived material was probably caused by the substantially higher porosity of this material since cold isostatic powder compaction resulted in much lower green densities compared to warm pressing used for the production of PVS- and PHMS- derived materials. Since the temperature dependence of the creep deformation revealed a rather low activation energy of 250 kJ/mol which is only about one-forth of values typical for liquid-phase-sintered silicon nitride these preliminary results indicate that amorphous precursor-derived materials reveal an interesting potential for high-temperature materials. The observed decrease of the creep rate between 1500°C and 1550°C of the PHMS-derived material pretending a negative apparent activation energy for the creep may be due to a beginning crystallisation of this material at the temperature range under investigation as was observed with TEM investigations [ 5 ] . The oxidation behaviour of the materials investigated is very much controlled by the formation of silica layers at the surface. Since there are no grain boundaries and intergranular oxide-type phases there is no interdiffusion of oxygen into inner parts of the sample along grain boundary phases being the main oxidation process of liquidphase-sintered ceramics at high temperatures. The different behaviour of the boroncontaining T2( 1)-derived material is probably due to the formation of boron oxide which degasses at temperatures under investigation. In summary one can say that the preliminary results show very interesting properties indicating precursor-derived amorphous covalent materials to be a new class of high temperature materials.

Acknowledgements For critical editing we like to thank Axel Forderreuther.

References [ 13 W. E. Luecke, S. M. Wiederhorn, B. J. Hockey, R. F. Krause Jr. and G. G. Long, J. Am. Ceram. SOC.,78 (1995) 2085-2096. [2] J. Bill and F. Aldinger, Z. Metallkd., 87 (1996) 827-840. [3] J. Seitz and J. Bill, J. Mater. Sci. Let., 15 (1996) 391-393. [4] T. Nishimura, R. Haug, J. Bill, G. Thurn and F. Aldinger, submitted to J. Mater. Sci.. [5] J. Bill, J. Seitz, G. Thum, J. Durr, J. Canel, B. Z. Janos, A. Jalowiecki, D.Sauter, S. Schempp, H. P. Lamparter, J. Mayer and F. Aldinger, phys. stat. sol. (a) 166 (1998) 269-296.

Nano-, Micro-, and Milliboundaries in Silicon Nitride Based Ceramics Pavol Sajgalik Institute of Inorganic Chemistry, Slovak Academy of Sciences, Dubravska cesta 9, SK-842 36 Bratislava. Slovakia

1. Introduction Silicon nitride is still a perspective engineering material for high temperature application. Silicon nitride based ceramic materials are intensively studied more than 30 years, though till today there is a lot of unknown in design of these materials for particular applications. Silicon nitride is a covalently bonded compound with an exceptional tensile strength of 1.5 GPa (this is value for a-Si3N4 whisker) and quite poor fracture toughness of 1.5-2.0 MPa . m", [l]. The brittleness and problematic growth of large single crystal of silicon nitride make these crystals useless in engineering application. The polycrystalline silicon nitride based materials found wider area of application. The intensive research of this materials brought an improvement of mechanical properties at room as well as high temperatures. While the polycrystalline silicon nitride prepared under extreme conditions without any sintering additives have the fracture toughness slightly higher comparing to the single crystal, i.e. 3.0 - 3.5 MPa . m", the tensile strength is much lower, 0.5 GPa [2]. On the other hand, conscious design of controlled microstructure [3] resulted into the silicon nitride based ceramics with excellent mechanical properties, fracture toughness of 9.0 - 9.5 MPa . mIn which is substantially higher comparing single crystal and bending strength of 1.1 GPa, the value is close to that of single crystal. Why the polycrystalline silicon nitride ceramics has substantially higher fracture toughness comparing single crystal? What is different? Polycrystalline silicon nitride consists of grains of beta silicon nitride, which can be considered single crystals with the relating properties. The other phases are the grain boundaries and triple pockets of secondary phases used as sintering additive and/or the phases used in sake of reinforcing or microstructure forming effect. Phases the triple points consists of are usually oxide glasses and/or some crystalline silicates, mechanical properties of these are usually poorer as silicon nitride single crystals. The reinforcing phases, mainly silicon carbide have the mechanical properties in the same level as silicon nitride. From this point of view the simple sum of properties should result in the material with less quality as the corresponding silicon nitride single crystal. But the last mentioned quantities pointed out presence of synergy effect which results in improvement of mechanical properties of the polycrystalline silicon nitride ceramics. The interfaces among particular phases are probably responsible for this synergy effect. In the silicon nitride based ceramics these interfaces with respect to the scale can be categorised into three categories: nano-, micro-, and milli-boundaries between the phases present. The effect of particular boundaries on mechanical properties of silicon nitride based ceramics is discussed.

-

Nano-, Micm- and Milliboundaries in Silicon Nitride Based Ceramics

247

2. Experimental The seven types of silicon nitride based materials have been prepared in present study and they are listed in Table 1. Table 1 Composition ofthe ceramic materials used in present study

Material

Si3N4 wt%

Sintering Additives wt% A1203

y203

A B1

82* 72'

4.6$ 4.60

3.4' 3.4@

B2

72'

4.60

3.4@

C D El E2

72* 92*

4.6' 4.6'

P-Si3N4 Whiskers

Sic Platelets

Amorphou s SiNC

wt%

wt%

wt%

1o+

20 (as received) 20 (BN coated)

3.4' 3.4' CD/C Layered Composite D/CD Layered Composite

208

UBE SN E10; A16, Alcoa; H.C. Stuck, Fine; ' prepared at Pulvermetallurgischeslaboratorium, MPI Stuttgart by pyrolysis of NCP 200; LC-12, H.C. Stuck; F l u b AG, Germany;@Tecsnabexport,Russia

*

'

'

The starting mixture A have been prepared by attrition milling of starting powders in isopropanol for 4 h. The dried powder was sieved through 25 pm sieve and then formed in CIP at 630 MPa into the green compacts with dimensions of 1.4 cm x 2.0 cm x 5.5 cm. These were gas pressure sintered at 10 MPa of nitrogen and 1900 C for 3 h. The starting mixture B1 and B2 have been prepared by attrition milling of silicon nitride and sintering additives for 4 h, Table 1. After addition of whisker suspension, the whole suspension was simultaneously stirred, ultrasonicated and dried, in order to avoid the whisker sedimentation and agglomeration in the starting mixture. Dense disks with the diameter of 5 cm and thickness of 0.5 cm were prepared by hot pressing at 1750 "C for lh at 30 MPa in the nitrogen atmosphere with 20 kPa overpressure. The BN coated whiskers introduced to the starting mixture in case of sample B2, were prepared in house by dip coating method. The procedure and the result of BN coating of P-Si3N4whiskers is described elsewhere, [4]. The starting mixture C have been prepared by attrition milling of silicon nitride and sintering additives for 4 h, Table 1. After addition of Sic platelets, the whole suspension was homogenised for 90 h in plastic bottle with the silicon nitride spheres. Than the suspension was dried and treated by the same procedure as starting mixture A. The starting mixture D have been prepared by attrition milling of silicon nitride and sintering additives for 4 h. Than the suspension was dried and treated by the same procedure as starting mixture A in order to prepare the dense ceramics. Layered composite was prepared by sequencing the layers of dried starting powder C and D (Table 1) into the rubber mould. To get the layer composite the CIP of 250 MPa was applied, the green body was gas pressure sintered at 1900 C for 3h at 10 MPa of nitrogen. The three layer composites D/CD were cut from the dense sintered bodies after GPS.

248


The dense samples were cut into the bars of 3 mm x 4 mm x 45 mm and polished into 15 pm finish. These were subjected to the 4-point bending test with 40120 outerlinner span. Microstructures were investigated either on the thin foils by TEM or on the polished and plasma etched surfaces by SEM. The local chemical analysis was performed by EDX. 3.

Results and Discussion

3.1 Nano- boundaries 3.1.1. SiC/Si3N4Nanocomposites The ceramic nano/micro- composite was prepared from the composition A, Table 1. Mechanical properties of this composite are in Fig. 1.

Fig. 1 Mechanical properties of monolith SisN4 and Sic- SinN4 nano/micro-composite

As can be seen from Fig. 1, the mechanical properties of nano-composite are higher comparing to the relative monolith. The difference between two ceramic materials from the chemical point of view is presence of Sic phase in the nano-composite. The TEM micrograph of this composite is shown in Fig. 2. The TEM-investigation showed that the microstructure is homogeneous without any observable macro-defects. The microstructure has a bi-modal distribution of the grains with the mean size of fine fraction below 1 pm and large grains with lengths up to 10 pm. The RGB elemental map of the same area showed that composite consists of Si3N4and Sic micro-grains. Some Si3N4 grains contain small Sic inclusions. A detailed TEM investigation at higher magnification revealed that the majority of Si3N4 grains contain randomly distributed Sic nano-inclusions of size from 5 to 50 nm, Fig. 3. The elemental EDX analysis of Si, N, C and 0 confirmed that the nano-inclusions is SIC with a thin oxygen rich surface layer. Elemental EDX analysis of the another Si3N4 grain containing Sic inclusions showed that the Sic nano-inclusions of this grain are oxygen free. Bright field TEM images of Si3N4micro-grains with the oxygen free Sic nano-inclusions contain the strain contrast joining the Sic inclusion and grain boundary, an example of such a strain contrast is shown also in Fig. 4. No chemical inhomogenities along these lines were observed. The energy-filtering TEM investigation shows that the SiC/Si3N4 nanocomposite consists of different kind of grains. These can be classified as follows: Si3N4micrograins, Sic micro grains, Si3N4micro-grains with Sic nano-inclusions, schematic of such a structure is shown in Fig. 5. The last mentioned micro grains can be divided into

Nano-, Micro- and Milliboundaries in Silicon Nitride Based Ceramics

249

two main groups: Si3N4micro-grains containing oxygen free Sic nano-inclusions and Sic nano-inclusions with oxygen rich surface layer, respectively. No other grains or areas with different chemical compositions, e.g. free carbon, were found.

Fig 2: TEM-bright field image of Sic/ Si,N4 nano/micro-composite

Presence of several kind of grains with different thermal expansion coefficients (3.5-5.0 x 10-6/Kfor S i c and 2.6-3.4 x 1076/K for Si3N4) should lead to the existence of residual stresses. Presence of Sic inclusions should produce the residual stresses inside of Si3N4 grains for the same reason. These tensile stresses within the Si3N4grain are oriented toward the centre of the Sic inclusion as it is schematically shown in Fig. 6. Based on the TEM observation the boundaries between the micro-grains in the present composites can be divided into six categories: 1. Si3N4- Si3N4;2. Sic - Sic; 3. Sic - Si3N4;4.Si3N4(withSic inclusions)-Si3N4(with Sic inclusions); 5 . Si3N4(withSic inclusions) - Sic; 6. Si3N4(withSic inclusions) Si3N4,Fig. 5 .

250


Fig. 3 TEM bright field image of Si3N4 grain containing Sic inclusions and Si-, N-,C-, 0-maps of the same area.


25 1

Fig. 4 TEM bright field image of three Si3N4 grains with SIC inclusions and Si-, N-, 0-maps of the same area.

252


Strength of particular boundaries must be different because of neighbourhood of various kinds of grains with different thermal stresses as it was mentioned above. The influence of thermal expansion coefficients of grain boundary amorphous phases with respect to their chemistry is neglected because of simplification of the problem. As it is schematically demonstrated in the Fig. 6 presence of Sic inclusions with higher coefficient of thermal expansion comparing to Si3N4grains causes tension at their grain boundaries. Similarly the boundary between Si3N4 and Sic micro-grains should be under tension

Fig. 5 Schematic ofthe nano/micro-structure of nano/micro Sic/ Si3N4composite prepared in present study

These statements lead to the conclusion that SiC/Si3N4nanolmicro-composite contains “stronger” and “weaker” grain boundaries between the micro-grains as a results of their thermal stresses. From this point of view, only the boundary between two Sic or Si3N4 grains can be considered as a strong one, i.e. this is affected only by their structure (thermal expansion coefficient). The majority of the micro-grain boundaries within the composite are of type 4, Fig. 5 . The present analysis must take into account two types of Sic inclusions mentioned above. Si3N4grain with Sic inclusions containing a glassy surface layer is supposed to be stress free because of relaxation of thermal stresses within the oxygen layer. The indirect confirmation of t h s statement is the presence of strain contrast only in the Si3N4grains with the Sic inclusions without oxygen layer. These strain contrasts are suppose to be attributed to the subtle deformation of the lattice of Si3N4grain due to presence of the Sic inclusion. This kind of “sub-grain boundaries” was also reported in the work of Niihara, [ 5 ] . Such a contrast lines were not found in grains containing Sic inclusion with the oxygen (glassy) layer. Based on this assumption the stress status of the most micro-grain boundaries will be dictated by the occurrence of oxygen (glassy) layer on the surface of Sic inclusions. In the work of


253

Pan et. al., [6] the Occurrence of this oxygen rich (glassy) phase is only attributed to the large Sic inclusions. The present study shows that the glassy phase is also connected to the very small Sic inclusion with the diameter of 5-10 nm, Fig. 3. The bonding between Si3N4 micro-grains with inclusions coated by oxygen rich (glassy) layer can be considered to be similar to those of pure Si3N4micro-grains. The grains with oxygen rich layer coated Sic inclusions are not depicted in Fig. 5.

Fig. 6 Schematic shown the crack propagation through weak boundary

It is believed that the strength of the micro-grain boundaries plays an important role in influencing the fracture behaviour of the composite and consequently the fracture toughness. Weak boundaries of type 4 will promote the inter-granular fracture in the nano/micro composite. This behaviour is schematically shown in Fig. 6. Their occurrence will contribute to the prolongation of the crack path. From this point of view number of weak boundaries (volume fraction) is an important parameter. Their volume fraction will be depended on the volume fraction of Si3N4 micro-grains with Sic inclusions without any glassy interlayer. The fracture toughness 6.9 MPa . m” of presented microhano composite is rather high for such a fine microstructure shown in Fig. 1. This value is comparable to 7.3 MPa . m’ of substantially coarser microstructure of silicon nitride based “micro” ceramics prepared with the same additives at the same gas pressure sintering conditions, [7]. It is supposed that this rather high value is reached by intensification of inter-granular fracture and prolongation of the crack path which is a consequence of increased volume fraction of weak boundaries in the present nano-composite.

’

254

High TemperatureMechanical Properties and Characterization of Grain Boundaries

I-hgh strength of 1.2 GPa and Weibull modulus of 19 of prepared microhano-composite give an evidence that the size of technological defects is undercritical. Sic fine precipitates in the microstructure serve as the grain growth inhibitors, as it was pointed out in [8] and these are responsible for the homogeneous particle size distribution. The refinement of the microstructure and homogeneity of the microstructure lead to the high values of strength and Weibull modulus. For comparison, the strength and Weibull modulus of silicon nitride “micro” ceramics prepared by the same way were 987 MPa and 8.7, respectively, [7]. The relative high difference in Weibull modulus shows that flaw concentration and its size in the “micro” ceramics is substantially higher comparing to SiC/Si3N4nano/micro composite. The conclusion of this part can be drawn as follows: The number of oxygen-free Sic- Si3N4nano-interfaces (volume fraction of Sic oxygen free nano-inclusions) should be as high as possible because of their weakening effect on the micro-boundaries in the SiC/Si3N4 nano/micro-composite. The presence of Sic nano/micro grains within the structure hinders the abnormal grain growth and makes the structure homogeneous which increase the Weibull modulus of the composite.

Fig. 7 Microstructure of the P-Si3N4 whisker reinforced Si3N4ceramics

3.2 Micro- boundaries 3.2.3 p- Si3N4whisker reinforced Si3N4ceramics The microstructure of p- Si3N4whisker reinforced Si3N4ceramics with 4-point bending strength of 930 MPa and fracture toughness of 7.2 MPa . m“ prepared in present study is shown in Fig. 7. The elongated p- Si3N4whiskers are introduced in order to make the dissipation of the crack tip energy more effective and thus increase the fracture toughness of the composite. The extent and effectiveness of dissipation is depending on the toughening mechanism present during the mechanical failure. The type of the toughening mechanism is depending on the “strength” status of the interface between reinforcing whisker and matrix. This interface must be strong enough for keeping the good strength of the composite but on the other hand weak enough in order to control the crack path along the reinforcing whiskers and thus to introduce the reinforcing


255

mechanisms as whisker pull out, crack bridging and crack deflection. To design the interface in such a way that above mentioned conditions are fulfilled is rather complicated, mainly for silicon nitride ceramics reinforced by p- Si3N4whiskers.

Fig. 8 BN coated P-Si3N4 whisker by “dip coating”

The problem is the p- Si3N4whisker growth during the hot pressing. The interlocking of the p- Si3N4whiskers into the matrix is a consequence of this grain growth. The wlusker pull out is than hindered and the toughening effect of whisker application is less effective. Moreover, whisker impingement diminish the whisker tensile strength, [ 9 ] . To protect the whisker against the morphological changes during the hot pressing the thin surface coating of whiskers by BN was performed. Fig.8 shows the micrograph of BN coated whisker. Enhanced pull out was observed in the ceramics reinforced by BN coated whiskers, Fig. 9. The fracture toughness of these composites increased from 7.2 to 8.9 MPa . m”*. This result show that the chemistry of interface determines its “strength and thus fracture behaviour of the composite. To find out the “optimum” composition of the interface is rather difficult among others also because of the presence of metallic impurities which are solute within the glassy boundary phase. Their presence is hardly detectable, because of low concentration, but after crystallisation of the triple points these impurities are segregated into objects localised at the interfaces between the whiskers and matrix and their local concentration is high. The whiskers used in present study contained the Fe, Cr and Ti as main metal impurities. These were not detected by EDX in the glassy triple points, but after devitrification of glass phase, the spherical objects of size up to 100 nm were identified, [lo]. The elements identified in these objects by EDX were the same as impurities introduced along with the p- Si3N4whiskers. Presence of areas with high concentration

256


of impurities locally decreases eutectic temperature, these areas at the high temperature loading are predominantly melted and create the cavities which are responsible for degradation of high temperature properties.

Fig. 9 Enhanced pull out of BN coated whiskers introduced into the S&N4matrix improved the fracture toughness from 7.2 to 8.9 MPa .m".

The complexity of the grain boundary engineering, as it was partly shown above, discriminates this part of ceramics design for the wider industrial application. But it seems to be one of the most perspective way leading to substantial increase of ceramics fracture toughness. 3.3 Milli- boundaries 3.3.1 Layered silicon nitride based composites The layered ceramic composites with the composition E l and E2, Table 1 were prepared in present study. Presence of layers with the different composition (C and D, Table 1) results in the conserving the residual stresses within the composite after cooling. The stresses are a consequence of different thermal expansion coefficients and Young's modulus of particular layers, [ 111. The stress status is stepwise changed at the layer boundary. While the residual stress in the silicon nitride layer (layer C) is compressive with the value of 50 - 110 MPa, the residual stress in the layer containing Sic platelets (Layer D) is tensile with the value 70 - 120 MPa. Presence of this kind of stresses can be documented in the schematic shown in Fig. 10. Fig. 10a shows the inden-

Nano-. Micro- and Milliboundaries in Silicon Nitride Based Ceramics

257

tation impression localised at the layer boundary. The cracks are anisotropic, the crack in the layer D is slightly longer comparing the crack in the layer C. After bending test, the crack in the layer under tension (layer D) cross the whole layer, whle the length of crack in the layer C, which is under compression is almost not changed.

PRIOR TO BENDING

AFTER BENDING b)

a)

-

I

------,-

T residual 0 compression

II

' t-

residual compression1

applied stress

-.+

Fig. 10 Schematic of the presence of residual stresses in individual layers of a three layer Si3N4 based composite

The stress status of the adjacent layers influences the mechanical properties of the composite, as it was pointed out in our previous work [ 11-15]. The example of their role can be as follows: Three layer composite El, with the layer sequencing C/D/C has a bending strength of 1042 MPa and fracture toughness measured by ISB (indentation strength in bending) This composition has two Si3N4 outer layer under method of 8.2 MPa . compression and only the layer Si3N4+SiC,~in the middle of the composite is under tension. When this sequencing of the layers is vice versa, i.e. composite D/C/D was subjected to the 4-point bending test, the strength of 524 MPa and fracture toughness of 6.2 MPa . mln were obtained. This substantial difference in mechanical properties shows the potential of this kind of composites, the properties can be changed within wide range of values. Decreased sensitivity to the flaws of this kind of ceramics was also documented, [ 12, 151. 4-point bending strength of layered ceramic composite consisting of 4 layers with fine and coarse microstructure was 785 MPa whde the strength of monoliths with the composition relating to the particular layers was lower 540 and 470 MPa, respectively, [12]. These quite low values of strength are caused by presence of pores filled by large elongated p- SisN4grains, the size of the flaws observed was up to 100 pm. The increase of strength of layered composite comparing to the monoliths can be explained by the schematic shown in Fig. 11. The stress concentrated on the largest defect C T ~resulting from bending test can be diminished by the value of A q which is

258


relating the compressive stress in the layer containing the defect. This composite material is not following the Griffit’s relation between the fracture toughness, stress and flaw size, this material is less sensitive to the flaws as it was shown above. The layered composite seems to be a perspective materials utilising the positive effects of reinforcing at the nano/micro level and simultaneously open the possibility to design the properties on the larger scale. The mechanical properties, bending strength and fracture toughness of these composites can be influenced besides toughening mechanisms acting at the nano/micro-structural scale also due to residual stresses introduced with the layers of different thermal expansion coefficients and Young’s modulus.

MONOLITH

LAYERED COMPOSITE

Ao

k . op a-

applied stress op

--++

k . CJP- ACJ

-applied stress op

Flg. 11 Schematic explaining decreased sensitivity of the layered ceramic composite comparing to the relative monolithic ceramics

4. Conclusions The presented results show that design of strong, reliable and crack resistant silicon nitride based materials requires precise engineering at the all levels of size scale, i.e. from nano- to milli- or even higher. The fault at a particular level can have the fatal consequences on the mechanical property of the engineering material. The reinforcing, toughening effect developed at the particular level of size can be completely destroyed by presence of the flaw of comparable or higher size. The conscious design and composition of interfaces of silicon nitride based ceramics seems to be the most important step in production of really high-performance materials. All size levels, from nano- up to micro- must be under control in order to avoid cancellation the reinforcing, toughening effect.

5. Acknowledgement Author acknowledges Dr. Joachim Bill for supply of amorphous SiNC powder used in present study, Prof. Jan Dusza for measuring of mechanical properties, Drs. P. Warbichler and F. Hofer for the TEM investigations, Dr. Sadananda Sahu for suply of


259

BN coated silicon nitride whiskers, the Powder Metallurgical Laboratory, Max-PlanckInstitut for Metals Research, Stuttgart, Germany for the possibility to carry out some experiments used in present study, Alexander von Humboldt Foundation for the financial covering of author's stays at PML.

6. References 1. 1.E Reimanis, H. Suematsu, J.J. Petrovic, T.E. Mitchell in Ceramics Transactions (Eds.: B.W. Sheldon and S.C. Danforth) The American Ceramic Society, Westerville, Ohio, USA, 1994, Vol. 42, pp. 229-236. 2. I. Tanaka, K. Igashira, T. Okamoto, Y. Miyamoto, M. Koizumi, J. Am. Ceram. SOC. 1989, 72, 1656-60. 3. W. Dressier, PhD Thesis, 1993, Universitiit Stuttgart, Germany 4. S. Sahu, S . Kaveclj, C. IlleSova, J. Madejova, I. Bertbti, J. Szdpvolgyi, J. Europ. Ceram. SOC.1997, submitted for publication 5. K. Niihara, J. Jpn. Ceram. SOC.1991, 99, pp. 974-982. 6. X. Pan, J. Mayer, M. Ruhle, J. Am. Ceram. SOC.1996, 79, 585-90. 7. J. Dusza, P. Sajgalik in Deformation and Fracture in Structural PMMaterials, Vol. 2, (Eds.: C.Parilhk, H. Danninger, J. Dusza, B. Weiss) IMR-SAS, KoSice 1996, pp. 61-73. 8. P. Sajgalik, D. Galusek, J. Muter. Sci. Lett. 1993, 12, pp. 1937-9. 9. K. Rajan, P. Sajgalik, J. Europ. Ceram. SOC.1997,17, pp. 1093-97. 10. K. Rajan, P. Sajgalik, prepared for publication 11.P. Sajgalik, J. Dusza, Z. LenEeS in Engineering ceramics '96: Higher reliability through Processing (Eds. G.N. Babini, M. Haviar, P. Sajgalik), NATO AS1 Series, Kluwer Academic Publishers, Dordrecht, Boston, London 1997, pp. 301-10. 12.P. Sajgalik, Z. LenEeS, J. Dusza, J.Muter. Sci. 1996,31, pp. 4837-42. 13. P. Sajgalik, 2. LenEeS, J. Dusza in Ceramic Materials and Components for Engines (Eds. D.D. Yan, X. R. Fu, S.X. Shi) World Scientific, Singapore, New Jersey, London, Hong Kong 1995, pp. 198-201. 14. P. Sajgalik, J. Dusza, Z. LenEeS in Ceramic Processing Science and Technology (Eds. H. Hausner, G. L. Messing, S. Hirano), Ceramics Transactions, Vol 51, The American Ceramic Society, Westerville, Ohio 1995, pp. 603-7. 15. P. Sajgalik, Z. LenCeS in Ceramic Microstructures '96: Control at the Atomic Level (Eds.: A. P. Tomsia, A. Glaeser), Plenum Publishing Corporation, New York, London, Washington, D.C., Boston 1997, to be published

Plastic Deformation of Covalent Bonded Ceramics at High Temperatures Sadahiro Tsurekawa Japan Science and Technology Corporation (JST), Nagoya 456, Japan Present Address:Departmentof Machine Intelligence and Systems Engineering,Faculty of Engineering, Tohoku University, Sendai 980-77,Japan

Kouichi Kawahara and Hideharu Nakashima Department of Materials Science and Technology, Graduate School of Engineering Sciences, Kyushu University, Kasuga 816, Japan.

Abstract A review is given on our recent investigations concerningthe high temperature deformation mechanism of covalent bonded ceramics, titanium carbide (Tic) and silicon carbide (Sic). Particular attentions have been paid to the effect of nonstoichiometry and to the effect of polytypes and addition of sintering aids on their high temperature deformationbehavior for T i c and Sic, respectively. The observedphenomena are discussed from the viewpoints of the characteristics of dislocation motion, then the importance of understanding the nature of dislocation motion even for ceramics is emphasized.

1. Introduction Covalent bonded ceramics have been expected for structural materials at high temperatures. However, the rack of reliability coming from brittleness has restricted the practical applications so far. At high temperatures, grain boundary (GB) sliding and diffusion become so easy to occur that superplasticity-likebehavior would be attainedeven on ceramics having very small grains, generally speaking less than Ipm in grain size. For such case, however, a problem arises. That is, on the contrary to occurrence of a large elongation, high temperature strength would often remarkably decrease. For structural materials, actually, it is craved to have a large elongation without strength decrease. On the other hand, at high temperatures, not only the GB sliding and the diffusion but also the dislocation motion must be thermally activated. Then, if the ability of plastic deformation due to dislocation motion is possible to be increased, the reliability for ceramic materials would increase. Therefore, it is of importance to know the dislocation motion giving rise to plastic deformation. For ceramic materials, however, investigations from such a viewpoint have been scarcely conducted so far. Covalent bonded ceramics are classified into two distinct groups depending on their mechanical properties: in one group, relatively high hctility is obtained at high temperatures and in the other group, the hctility is limted even at high temperatures.

Plastic Deformation of Covalent Bonded Ceramics at High Temperatures

26 1

For instance, transition metal carbides such as titanium carbide belong to the formercase and silicon carbide or silicon nitride does to the latter case. In this paper, firstly, high temperature deformation mechanism in titanium carbide will be stated. Particular attention will be paid to the effect of nonstoichiometry. Secondary, high temperature deformation behavior of silicon carbides, in particular, the effects of polytypes and additions of B and C as sintering aids on their properties will be reviewed. They are being discussed in connection with characteristics of the dislocation motion.

2. Titanium Carbides (TIC) 2.1 Characteristics of Tic An interesting characteristicof T i c is that it has a wide nonstoichiometry over the range of C/Ti atom ratio, x, from0.5 to 1.O without changing the crystal structureof rocksalt type [ 11. The nonstoichiometry is known to come from the vacant carbon sites. In addition, in the lower carbon region of x from 0.50 to 0.71, an ordering of carbon vacancies has been reported to occur below 1 lOOK [2].

2.2

High Temperature Deformation Behavior

Figure 1 [3] shows the temperature dependence of the initial part of the stressstrain curve for single crystals with various Cmi atom ratios. At lower temperatures the yielddrop occurs in all crystals, however, it tends to become less evident as x &creases except It should be noted that Tic,,,, shows the largest yield drop, the highest temperature where the yield drop disappears and the highest flow stress compared with the other compositions in the lower temperature region. Once, Williams showed that the critical resolved shear stress, z,, decreased monotonously with decreasing x in the range from 0.79 to 0.95 [4]. Our result shown in Fig. 2 [3] also shows the same x - dependence in the high x region below ca. 1800 K. However, as x further decreases beyond 0.85, z, increases. Passing through the peak at x = 0.75 it decreases again. As temperature rises, the peak in 7, around x = 0.75 lowers with the remarkable decrease in the flow stress and finally disappears at 2270K. Above ca. 1900K a new peak appears at about x = 0.9. This composition corresponds to the highest melting Figure 1. Temperature dependence of yielding point. behavior for TiCx compressed at a strain rate of The decrease in z, would probably correspond to decrease in the 6 x lo4 s-I.

262


Peierls barrier for dislocations to overcome, and the increase in z, in the low xregion is assumed to be explained as an orderhardening, though the ordering of carbon vacancieshas been reportedso 300 far only below x = 0.71 [2]. la

a

f

2.3 Identification of Carbon Vacancy Ordering Figure 3 [5] shows the diffraction patterns obtained by in-situ observations using a heating stage in a transmission electron microscope. These observations reveal that a diffuse scattering occurs over the whole range of nonstoichiomet~ indicating the existence of short-range Order (SRO) Of carbon vacancies. The diffuse scatteringbecomes weaker with the increase in temperature and CiTi atom

TiC0.59

(211)

TiC0.75

2o

k

0.5 0.6

0.7

0.8

0.9

1.0

X F@re 3. x-dependence of critical resolved shear stress for TiC.Open and solid marks indicate the presence and the absence of yield drop, respectively.

(411)

TiC0.86

(221)

Figure 2. In-siru observation of electron diffraction patterns from TiCx with increasing temperature.


263

ratio. Furthermore, temperature dependence varies with x, for instance, for x = 0.86 the diffuse scattering is very weak at 1270K, while forx=0.75 it is still strong even at 1370K, which was the highest temperature attained by the heating stage. This observation suggests that the orderhardeningshould be negligible at high temperatures for high carbon composition. For x = 0.59, extra superlattice spots are observed. The superlatticespots were confirmed to correspond to the Ti$-type long-range order (LRO) having R3m structure [6, 71. The xand temperature dependences of diffractionpatterns strongly support our assumption on the order hardening mentioned above.

2.4 Strain Aging Effect due to Carbon Vacancy Ordering Figure 4 [5] shows the aging effect studies for identifying the cause of the short-range orderhardening. The temperature, 1270K, employed in this experiments is so low that the dislocationsannealed - out during aging should be negligible, but high enough for diffusion of carbon atoms through dislocations. In the case of Tic,,,, on the underload aging for a very short time, the stress-straincurve continues in a similar manner to that of no deformationinterruption and the yield drop is not reproduced. However,on the full unloading for a very short time, the yield drop recovers partially. As the aging time is prolonged, the yield drop magnifies. Therefore,the aging effect observed would probably result from a dislocation locking. On the other hand, in the case of Tic, 95 without order, the yield drop does not recover by aging. From these results, the aging effect on Tic, 75 is confirmed to come from the short - range order. The probable explanation of dislocation locking is preferential ordering around edge dislocations; that is, Sumino mechanism [8]. It is well known that an ordering use to induce a volume change. In fact, the ordering in T i c induces a volume expansion [7]. A strain field introcClced by the volume change would interact with a strain field of dislocations. Thus, the ordering should preferentially occur on one side of edge dislocations ( tension side in this case ) and lock them. This locking would be probably responsible for occurrence of the peak in z, at an incomplete ordering of because such preferential ordering aroundedgedislocationsis not expectedin a well - ordedcrystal.

2070Kfor 1.Bks

e (Yo) Figure 4. Strain aging effect on the yielding behavior in (a) TIC,,,, measured at I370K and (b) Tic,, ,.at ( 1270K.

264


3. Silicon Carbides (Sic) 3.1 Slip Systems in 6H SIC Single Crystal

Figure 5 [9] shows stress - strain curves for 6H S i c single crystals deformed along two distinct directions. In the case of 2000 (a), C axis normal to basal plane was inclined by ca. 30 e r e e to compression direction, then the basal slip could be easy to occur. On 2 1500 the other hand, in the case of (b), since C axis F b was parallel to compression direction, the i basal slip should never operate. One can see i 1000 that the specimen (a) yielded at about 70 MPa and showed the steady state deformation followed by the work - softening, while the 500 specimen (b), in which the basal slip was restricted fractured at ca. 2200 MPa without I 0 showing any plastic strain. This results 0 5 10 15 20 suggest that non basal slip systems such as the Strain, c (%) pyramidal slip one are very difficult to be Figure 5. Stress-strain curves for 6H-Sic single activated even at high temperatures in 6H crystals deformed along two distinct directions. single crystal. 2500

.-I

F

'

3.2 Effect of Polytypes on High Temperature Deformation Behavior Silicon carbideis well known to have many polytypes. However, they can be classified into two groups; one is &Sic with cubic structure and the other a - S i c with hexagonal structure. In p - Sic, ( 1 11 ) is the major operative slip system at high-temperaturesas well as that in Tic, so that p - SIC satisfies the von Mises criterion [ 101, which is closely related to the ductility of polycrystalline materials. On the other hand, the predominant slip system in a - S i c is (OO01) basal slip; there areonly two independent slip systems in this case. Unless the pyramidal slip systems operate, a - S i c does not satisfy the von Mises criterion, sot that it should be poor in ductility even at high temperatures. As described above, ' in fact, it is very difficult for the non basal slip to operate even at high temperatures in a (6H) - Sic. Then, it is expected that . there is much difference in deformation . behavior between a and p - Sics . polycrystalline materials. Figure 6 [I I ] shows an example ' of stress-strain curves of 01 and p - Sics compressed at 2060 K. The samples were pressureless sintered with addition of 0.37 E mass % B and 3.1 mass % C. The polytypes were analyzed to be 96.4 % 6H in Figure 6. Comparison of stress-strain curves for a a - SIC and 98.1 % 3C in p - Sic. and p - Sics with B+C as sintering aids.


Contrary to the expectation, it seems no appreciable difference in the stress - strain curves between them. The flow stress shows a peak and subsequently decreases monotonously. At the peak stress, some cracks were formed even in p S i c which should satisfy the von Mises criterion, resulted in decrease in the flow stress. In spite of the early crack formation, the specimen could be much deformation by compression without fracture. TEM observation revealed that many stacking faults and microtwins were present ubiquitously [ll]. In particular, these defects were noticeable in p - Sic. It is suggested, therefore, that these defects must have been the cause of the unexpectedresult, i. e., { 11 1 } slip in p - S i c should be constrained to operateby these dense stacking faults and microtwins. Consequently, the ductility of p -Sic probably became as low as that of a - S i c , then p - Sic would show deformation behavior similar to that of the a - S i c . Figure 7 [ 111 shows X-ray diffraction patterns of the a and p - Sics. Figure 7 (a) and (b) are obtained before deformation, Fig. 7 (c)after deformation of 28 % at 2 190 K and Fig. 7 (d)after annealing at 2190 K for 7.2 ks without

265

(a) P -Sic Belore delormalion

(b)

--L-

o -Sic

(111)

Belore delorrnalion

(200)

3C and 6 H

(C)B-SiC Alter delormalion

I

and4H

(d) &Sic Aller annealing al 2190K lor 7.2kr

Figure 7. X-ray diffraction panerns of (a) a - S i c and (b)-(d) PSC (a) and (b) before deformation. ( c ) after deformation of 28% at 2190K. and (d) after annealing at 2190K for 7.2ks without deformation.

Figure 8. High resolution image of b-Sic deformed by 28% at 2 I90K.

266


deformation. After deformation of p - Sic, some peaks corresponding to 6H and 4H a polytypes are observed, though no such peaks occur before deformation. According to a stability diagram for S i c polytypes, 6H structure is the most thermally stable one above ca. 2270 K, so that the polytypic transformation of 3C 6H could occur during heating at 2190 K. From Fig. 7 (a), however, there appears no definite sign of a structure occurring even after the annealing; namely, p structureis retained. This evidence strongly supports that the deformation would be probably responsible for the occurrence of hexagonal structures in the original p - S i c . Figure 8 [ 1I] is a high-resolution image of p - S i c deformed at 2190 K. The image reveals that the hexagonal structures exist in the form of twin-like layers. The crystal structurebetween arrows, for example, can be characterizedby stacking sequenceof 6H structure (...ABCACB...). Such regions must have given rise to the reflections corresponding to a type structures by the x - ray diffraction. Therefore, such partial structure change h e to deformation must be also cause of no significant difference in deformation behavior between a and p - Sics +

3.3 Stress Oscillation on Stress - Strain Curve Figure 9 [12] shows a temperature - dependence of stress-strain curves of

pressure-less s i n t e d p - S i c with additions of B and C deformed at a strain rate of s-’, which is ca. two order lower than that employed for the compression tests 2x m . . . . . . . . . shown in Fig. 6. It is noted that in ,800. B-Sic c.2XlO % ’ . spite of a lower strain rate in Fig. 9 comparing with that in Fig. 6, observed flow stresses were higher in 2 lm. the case of Fig. 9 than in Fig. 6 f 1000because of difference in size of ‘b asamples for compression tests, 2 m m x 2 m m x 3 m m in size for Fig. 6 and 1 mm x 1 mm x1.5 mm 0 1 2 3 4 5 ( I 7 a 9 1 0 for Fig. 9. E (Yo) None of cracks were Figure 9. Stress-strain curves compressed at various p r O h d after deformation Of ca. temperatures shown in figure at a strain rate of 10 5% in these cases. It should be ,0-5 s - ~ . notice that a regular oscillation of laT-2170K B-SiC flow stress is seen on stress - strain lm curves. This oscillatory behavior 21°00. depends markdy on strain rates as seen in Fig. 10 [12]. Theamplitude r*ZXlO’s ‘ . ‘O 8ooof the oscillation on the stress-strain a. curve increases with decreasing the strain rate and the period decreases with decreasing it. The oscillatory 0 1 2 1 4 5 6 7 8 9 1 0 behavior is similar in appearance to c (Yo) that observedin metals such asnickel Figure 10. Stress-strain curves compressed at 2170K at or copper with a low stacking fault various strain rate shown in the figure.

.

.


267

energy. In the case of metals, it is known for the oscillation to occur owing to the dynamic recrystallization hring high-temperature deformation. From the analogy of metals, it is also suggested that the dynamic recrystallization would take place in silicon carbide during deformation . The assumption was supported by the facts as follows. SEM observations revealed that grains changed their shape from acicular, which is characteristicof pressureless sintered p - Sic, into equiaxedafterdeformation[ 121. Furthermore, such a change in structure (polytype) h e to deformation as mentioned previously was scaresly observed when the stress oscillation appeared [ 121. In the case of metals, the decrease in stress on stress - strain curves with oscillation should be event when the decrease in dislocation density h e to the dynamic recrystallization prevails comparing with increase in it h e to deformation. It is believed, however, that the reverse is probably true for the case of covalent bonded materials: increase in dislocation density, which results in decrease in the average of dislocation velocity, must be responsible for the stress decrease, but the recrystallization during deformation,following the decreasein the dislocation density, probably contrubutes to increase in the flow stress [ 121.

3.4 Effect of Addition of Sintering Aids on Plastic Deformation Behavior According to our previous EELS analyses, an enrichment of sintering aids of B and C at grain boundaries was recognized[131, so that the added B and C were suggestedto solve into the grain interior. Thus, it is expectedthat B + C W S i C might be solution hardened by the sintering aids. The comparison of the stress - strain curves between HIPed p - S i c without any sintering aids [ 131 and B and C added one exhibits that there is considerable difference in high temperature deformation behavior between them, as shown in Fig. 1 1 [14]. Contrary to the expectation,S i c without any additives shows an extremely high strength, being ca. twice stronger than B + C added one. In addition, a regular stress oscillation barely appears on the stress - strain curves for p - S i c without sintering aids. Figure 12 shows weak - beamimages of dislocation structuresafter deformation at2170Kandatastrainrateofca. I X ~ O -s -~' : Fig. 12(a)and(b)areSiCswithB+C and without sintering aids, respectively. TEM micrographs reveal that the dislocation structures developedduringdeformation are surprisingly differencebetweenthem. In Fig. 12 (b), the stacking fault contrast is I T-2170K seen to be ribbon-like, indicatingthat 1200 p-SiC(no additives) i0= x 1 0 ' 5 ~ i the width of dissociated partial dislocations is relatively narrow. 2 On the other hand, in Fig. 12 (a),the 'o partial dislocations are so widely g dissociated that the stacking faults are 2 400 . observed being plate - like. Consequently, it is confirmedthat the addition of B andC should give rise to 0 2 4 6 0 10 12 drastic decrease in stacking fault Strain, c (%) energy. The difference in the Figure 11. Comparison of stress-strain curves for PSiCs deformation behavior between them without additives and with addition of B+C.

( -

'

268


Figure 12. Weak-beam images of PSiCs (a) with addition of B+C and (b) without additives, deformed to 10%at 2170K.

shown in Fig. 11 would be probably understoodin connection with the stacking fault enegy and the reconstructed core of Shockley partial dislocations as follows. Figure 13 [ 151shows the structureof the 30" partial dislocations on ( 111 ] slip plane of Sic: the open marks represent silicon atoms and solid ones carbon atoms, and the location of a pair of dissociated partial dislocations are indicated by the arrows. The core of one of the partials consists of all Si atoms while the core of the other partial does of all C atoms. Owing to the difference in core structures, the mobilities for dislocation glide motion must be very different for the two partials. In other words, the dislocations having Si - Si bonded core , Si dislocations, migrate much more easily than those having C - C bonded core, C dislocations,because a C - C bond is much stronger than a Si - Si bond ca. 3.7 eV for C C bondenergy and ca. 2.3 eV for SiSi [ 161. Therefore, in B and C 0 0 &Sic with a lower stacking fault Figure 13. Reconstructed core structure of 30" partial energy ,preferentialmotion of the Si dislocations on [ 1 I1 slip plane. Open and solid marks dislocations probably occurs, are silicon and carbon atoms, and the partials are resulting in the partial structure located along the arrows.


269

(polytype) change p 01 [15]. Then, the &formation should be controlled by the Si dislocations. On the other hand, in the S i c without additives, immobile C dislocations should be necessary to move because of its higher stacking energy [14]; therefore, this is most likely to result in the fact that the S i c without additives has about the twice strength of B and C added one . Also, the high stacking fault energy in S i c without additives is probably relevant to the absence of the stress oscillation and the structure change [ 141. -+

Acknowledgment The authors wouldlike to express hearty thanks to coworkers in Kyushu Univ. who shared interest with us and to emeritus Prof. H. Yoshinaga (Kyushu Univ.) who encouraged us. Financial support from JST is gratefully acknowledged.

References 1. E. K. Storms, The Refractory Carbides, Academic Press, New York, 1967, p. 1. 2. for example, V. Moisy-Maurice, N. Lorenzell, C. H. De Novion and P. Convert, Acta Metall., 1982, 30, 1769-1779. 3. S. Tsurekawa, K. Kuwayama, H. Kurishitaand H. Yoshinaga, J. Japan Inst. Metals, 1989, 53, 20-27; Muter. Trans., JIM, 1989, 30, 1016-1026. 4. W. S. Williams, J.App1. Phys., 1964, 35, 1329-1338. 5. S. Tsurekawa, H. Kurishita and H.Yoshinaga, J. Nucf. Muter., 1989, 169, 291298. 6. S. Tsurekawa, H. Kurishita and H.Yoshinaga, Lattice Defects in Ceramics, JJAP Series 2, 1989, 47-56. 7. S. Tsurekawa and H. Yoshianga, J. Japan Inst. Metals, 1992, 56, 133-141. 8. K. Sumino, Sci. Rep. RITU, 1958, 10, 283. 9. K. Kawahara, S. Tsurekawaand H. Nakashima, to be submitted to ScriptaMetall.. 10. R. von Mises, Z. ang.Math. Mech., 1928, 8, 161. 11. S. Tsurekawa, Y. Hasegawa, K. Sato, Y. Sakaguchi and H. Yoshinaga, Muter. Trans., JIM, 1993, 34, 675-681. 12. S. Tsurekawa, Y. Hasegawa, H. Yoshinaga and Y. Ikuhara, J. Japan Inst. Metals, 1995, 59, 263-270. 13. S. Tsurekawa, S. Nitta, H. NakashimaandH. Yoshinaga,Inrerface Science, 1995, 3, 75-84. 14. K. Kawahara, S. Tsurekawa and H.Nakashima, submitted to J. Japan Inst. Metals. 15. P. Pirouz and J. W. Yang, Ultramicroscopy, 1993, 51, 189-214. 16. W. Lambrecht and 0. K. Anderson, Phys. Rev. B, 1986, 34, 2439.

Correlation Between Deformation Mechanisms and Microstructural Evolution in Silicon Nitride Ceramics J.A. Schneider and A.K. Mukherjee University of California, Davis, California, USA

Abstract Changes of phase composition and morphology were investigated in silicon nitride both before and after compressive deformation testing. Si3N4specimens have been consolidated that retained the equiaxed, metastable a-phase. Use of additives such as MgAl,04 promote formation of a high viscosity additive system which rapidly pulls the grains together by capillary forces during the liquid phase sintering prior to the phase transformation. If additives are used such as Y,O, which promote formation of a low viscosity additive phase, the phase transformation from a to P-phase occurs prior to completion of the liquid phase sintering resulting in elongated grains. Equiaxed initial structures of the a-phase are noted to transform to an equiaxed P-phase microstructure following the elevated temperature compression tests. Compressive deformation tests indicate different flow stress dependencies for the equiaxed and elongated morphology. Limited deformation is noted in the elongated morphology with a flow stress of n = 1 indicating viscous flow of the grain boundary phase. Enhanced deformation is noted in the equiaxed morphology with a flow stress tending toward n = 2 indicating grain rearrangement. A recent model has proposed an interface control mechanism that may be correlated with enhanced deformation in Si,N4. These studies are in agreement with this model and indicate an equiaxed morphology is required to permit grain rearrangement in response to applied stress.

1.0

Introduction

Silicon nitride (Si3N,) based materials are of interest in both high temperature structural applications as well as chemically corrosive environments. However, the mechanical properties that make this ceramic desirable in various applications also make it very difficult and costly to conventionally machine into complex shapes. Thus, methods to produce net shape components are of interest. One method that is commercially attractive in metal-based systems is the use of enhanced or superplastic

Correlation Between Deformation Mechanism and Microstructural Evolution

27 1

(SP) deformation. Empirical correlation of the elevated temperature forming conditions with the microstructure indicate enhanced deformation is most likely to be observed in materials with a fine, equiaxed grain morphology. Oxide additives are commonly used in consolidation of the covalently bonded Si,N, which results in a microstructure of a and P-phase Si3N, grains embedded in a grain boundary phase. These oxides are selected based on their ability to react with the inherent oxide coating of SiO, on the Si,N, particles to form a liquid at the consolidation temperature. The resulting grain boundary phase is initially amorphous but maybe crystallized to some extent by subsequent heat treatments. If the 2 phases of Si,N, form unconstrained, the a-phase displays an equiaxed morphology [ 1,2] and the P-phase an elongated morphology [3,4]. The phase transformation is reconstructive and occurs by a solutionheprecipitationmechanism, irreversibly progressing from a to Pphase. Different morphology microstructures can be obtained by using different viscosity forming oxide additives. Based on Kingery’s [5] model of solutiodprecipitation liquid phase sintering, Jack and Hampshire [6] correlated the rate controlling mechanism with the viscosity of the additive system for the densification and phase transformation processes.

2.0 Experimental Procedure 2.1 Material Preparation Ube Industry SN-El0 Si,N, powders with a reported average particle size of 500 nm and > 95% a-phase were used in this study. Two oxide-additive systems were introduced: Y,O, alone, and Y,O, in combination with MgAl,O,. The powder containing 5 wt% Y,O,and 5 wt% MgAl,O, was provided by IMRA Materials in Japan. A spray-drying technique was used to produce spherical granules with an average 45 pm diameter. The powder containing 5 wt% Y,O,was provided by NGK Corporation in Japan and was prepared by wet ball milling. The Y,O,and MgAl,O, additives had average reported particle sizes of 1.6 pm and 0.3 pm, respectively. The specimens were consolidated using PAS equipment of Sodick Co. of Japan, mechanisms of which are discussed elsewhere [7,8,9]. Effective processing parameters for consolidation of Si,N, with additives in less than 5 minutes are 2023 K under a load of 60 MPa in air [7,8,9].

2.2

Compressive Deformation Testing

The specimens were deformed under compression loading in an air furnace at 1673 to 1773 K with moly-disilicide heating elements. Temperature was measured with an Pt-Ph thermocouple and controlled by a

212


programmable temperature controller to within 2" of the set temperature. The push rods were made of alumina and were 50 mm in diameter. To evaluate the effects of atmosphere, a nitrogen purge was installed around the test specimen for several tests. Com ression tests were conducted over a range of strain rates, from 5 x 10-?to 1 x lo's-', at constant temperature to determine the flow stress sensitivity. A MTS servohydraulic mechanical tester was used which has both stroke and load controlling functions with an automated data acquisition system. Constant strain rate compression tests were terminated at predetermined deformation stages of 70%, 40% and 10% true strain.

2.3 Microstructural Evaluation Phase content and texturing of specimens were determined from x-ray diffraction (XRD) analysis using Cu-Ka radiation. The method of Suzuik and Kanno [ 101 was used to determine the a and P-phase contents. Densities of the consolidated specimens were measured using the Archimedes' immersion method. Initial grain morphology was obtained from scanning electron microscopy (SEM) fracture surfaces of the specimens. Specimens were prepared for transmission electron microscopy (TEM) by ion beam thinning for electron transparency. Phase content of specific grains was obtained using selected area diffraction (SAD) in the TEM. A thin layer of carbon was deposited on the surface to minimize charging under the electron beam for both SEM and E M .

3.0 Results Characterization of the S$N4 powders used indicated a range of 100-700 nm with an average partxle size of 225 nm. XRD analysis indicated 98% a -phase. Use of Mg in the oxide additive systems has been reported to increase the viscosity of the grain boundary phase [ 1 11. High retention of the a-phase (89 %) was obtained at relatively high densities (95-96 % TD) for the MgA1204/Y,03additive system. The lower viscosity system of Y203produced specimens with textured P-phase grains as evidenced by the high ratio of the (210) to (101) intensities in the XRD analysis of Figure 1. Elongated grains can be observed in the TEM image in Figure 2. In comparison, no texturing was observed in specimens with the higher viscosity system of Y203and MgA120, as evidenced by similar intensities of the (2 10) and (102) peaks in the XRD analysis of Figure 3. The equiaxed a-phase grains are shown in the TEM micrograph in Figure 4. Based on the equiaxed morphology and high density achieved with the Y203and MgA1204additive combination, a series of constant strain rate

0.0 20

Figure 3

-

3 C

a 0.5 C

.E

1 . 0 ) ,.

30

28 (degrees)

35

40

45

50

PAS Consolidated Si3N4with 5 wt% Y,03 & 5 wt% MgAl,04 Additives. XRD Analysis Figure 4. TEM of Eqmaxed Grain Morphology

25

PAS Consolidatad

SN0116%41

. . . I . . . . I . . . . I . . . . I . . . . , . . . . ,

PAS Consolidated Si,N4 with 5 wt% Y203Additives. XRD Analysis Figure 2. TEM of Elongated Grain Morphology

28 (degrees)

. . . . I . . , . . I . . . . I . . . . l . . . . , . . . . (

Figure 1.

1.0,

s

t4 4

w

3

?.

F

h

b %

a

$

a

0

s.

3

n

214


compression tests were used to evaluate the material flow stress characteristics. The effect of temperature on the steady state flow stress conditions during deformation was evaluated at an initial strain rate of 5 x s-'. These tests were intentionally terminated at 70%true strain. An equiaxed morphology is still noted after deformation testing to 70%true strain as shown in the TEM photograph in Figure 5.

33% a-phase 70%true strain

I 0.1 pm I Figure 5. TEM Micrograph of Grain Morphology of Compression Tested Specimen (Si,N4 w/ 5wt% Y,O, & 5 wt% MgAl,04) The flow stress dependency was evaluated by holding temperature constant at 1773 and 1723 K. Steady state flow behavior was observed over a range of strain rates from 5 x s-' to 5 x lom5 s-'. Since steadystate flow behavior was observed after 30%true strain, these tests were intentionally terminated at 40% true strain. Evaluation of the flow stress dependency for the specimens with Y,O, and MgAl,04 additives indicated different slopes for the two temperatures as shown in Figure 6. However, since excess decomposition was noted in the specimen tested in air at 1773 K, the calculated true stress value underpredicts the actual specimen geometry. This decomposition was not observed at 1723 K. Thus the flow stress dependency is noted to tend toward a value of n = 2. No decomposition was noted in the


275

compression specimens containing Y,O,additives with a flow stress dependency of n = 1 as shown in Figure 7.

4.0

Discussion

To evaluate the effect of grain morphology on the deformation characterization of Si,N,, different morphology specimens were required. Liquid phase sintering models were used as a basis for producing such Si,N4 specimens. Modeling of the liquid phase sintering model for nitrides by Kmgery [5] considered 3 stages. This model predicts that theoretical densities of up to 95% can be obtained by particle rearrangement and mass transport via solution-reprecipitation. Final elimination of close porosity is noted to require longer times. Further interpretation of the 2nd stage of Kingery's model was approached by Hampshire and Jack [6] to consider solution vs diffusion control in the 2"d stage of Kingery's model for the Si3N4system. For additives such as MgO that react with SiO, to form a h g h viscosity glass, complete densification could be obtained prior to transformation. Whereas for additives such as Y,O,which react with SiO, to form a lower viscosity glass, complete phase transformation would be predicted to occur prior to densification. Results similar to this study have been reported by Bowen, et. al. [ 121 with the MgO additive system. Persistence of the glassy grain boundary phase around both the a and P-phases of the consolidated and compressed specimens is observed in the TEM micrographs of Figures 1,3, and 5. Various models [ 13-17] have been proposed to describe the deformation of this type of microstructure. Of these models, all but Wakai [ 171 predict a stress dependency of n = 1. These models consider the rate controlling deformation mechanism to be either viscous flow of the glassy phase or mass transport of material along the glassy phase. In these models, the only barrier is the rate of diffusion in the glass or liquid grain boundary phase. Wakai's model [ 171 considers the case where an interface reaction between the grain surface and glassy grain boundary phase becomes the rate controlling mechanism. For the interface reaction control, two cases are described by Wakai in the step model [ 171 which correspond to different growth patterns in crystals [ 181. The flow stress dependency is based on the density of favored molecular attachment sites at the interface. As favored kink sites become inaccessible or are filled, molecules can only attach to less favored step and ledge sites. Since nucleation of sites is required to roughen the atomically flat surface to again produce the favored kink sites, there is a delay in the precipitation of material to increase the P-phase content. This delay causes the reaction at the interface to become rate controlling. This

216


solutiodprecipitation process also corresponds with the phase change mechanism in the Si3N4system. The specimens produced with the high viscosity additive system of MgAI,04 and Y,O,displayed a equiaxed morphology both before and after compressive deformation testing. Retention of an equiaxed grain morphology has been correlated [191 with the combined sliding and rotation of grains in response to an applied load. As the stressed material is driven into solution, accompanying grain rotation would limit and possible occlude the subsequent precipitation of molecules to the more favored attachment sites. As molecules attached to less preferred sites or smoother interfaces, roughening occurs. Continued rotation would then promote equiaxed attachment and maintain the equiaxed grain shape. Large amounts of deformation can be achieved as the grains slide and rotate in response to the applied load. A flow stress dependency tending toward n = 2 is consistent with the enhanced ductility observed. Transfer of strained mass is accomplished by the solution-precipitationprocess and maybe enhanced by the 01 to P-phase transformation which occurs simultaneously [20]. In contrast, steady state flow stresses were only observed within a very narrow range of conditions for the elongated specimens. A flow stress dependency value of n = 1 was obtained from the slope of the strain rate vs true stress plot in Figure 7. This suggests that viscous flow of the grain boundary phase is rate controlling. With an elongated microstructure, deformation of the specimen would cause grains to slide past one another to accommodate the imposed strain. Favored attachment sites would be in the c-axis direction [21] which further promote the existing texture [22]. Thus only a limited amount of strain (10%)is be accommodated prior to specimen failure since only sliding of the grains occurs.

5.0

Summary

Si,N, specimens have been consolidated that retained the equiaxed, metastable a-phase. Use of additives such as MgAl 0,promote formation of a high viscosity additive system which rapidly pubs the grains together by capillary forces during consolidation by liquid phase sintering as modeled by Jack and Hampshire [6]. If additives are used such as Y,03 which promote formation of a low viscosity additive phase, the phase transformation from a to P-phase occurs prior to consolidation. Compressive deformation tests indicate different flow stress dependencies for the equiaxed and elongated morphology. Limited deformation is noted in the system with elongated morphology with a flow stress dependency of n = 1 indicating viscous flow of the grain boundary phase. Enhanced deformation is noted in the equiaxed morphology with a flow stress tending toward n = 2 indicating grain rearrangement. Recent

211

Correlation Between Deformation Mechanism and Microstructural Evolution lo4

-

b cc a?

c

zc

n=1.35 @I773 K

'5 tj

-

10.4

-

n=2.07 @I723 K

lod

I

I

I

I

I

S

B

B

1

,,

I

10

loo

1000

Figure 6. Temperature Effects on the Flow Stress Dependency of Si3N4with 5 wt% Y,O,and 5 wt% MgAl,04 Additives 104

c.

% cc a? c

zc

'E

tj 101

Equiaxed Grains wl 5 wt% Y,O,

-

-

and 5 wt% MgAI,O, c1

Elongated Grains w l 5 wt% Y203 104

I

1

I

I

I

#

t

8

1

I

10

loo

I

I

I

I

, . I .

Figure 7. Flow Stress Dependency of Si,N, with 5 wt% Y,03 and 5 wt% MgAl,O, Additives vs Si3N, with 5 wt% Y,O, Additives

1000

278


models have proposed an interface control mechanisms that may be correlated with enhanced deformation in Si3N4[171. These studies are in agreement with this model and indicate an equlaxed morphology is required to permit grain rearrangement in response to applied stress.

6.0 Acknowledgments The authors wish to thank Prof. K.Yamazaki at the University of California, Davis and Dr. K.Shoda of Sodick Co. in Japan for the use of the PAS equipment and Mr. C.J. Echer of the National Center for Electron Microscopy, Lawrence Berkeley Laboratory, for assistance with the TEWAEM. This work has been supported in part by the DP Materials Initiative, with Dr. Wendell Kawahara as the Sandia National Laboratory collaborator, and by the Ceramics Program of the Division of the Materials Research Grant #NSF DMR-9314825 monitored by Dr. Lisolette J. Schioler.

References Kijima, K.,Kato, K.,Inoue, Z., Tanaka, H., J. Mat. Sci., 1975, 10, 362-363. Morgan, P.E.D., J. Mat. Sci., 1980, 15, 791-793. Turkdogan, E.T., Bills, P.M., Tippett, V.A., J. Appl. Chem., 1958, 8, 296-302. Inomata, Y., Yamane, T., J. Cry. Growth, 1974, 21, 317-318. Kingery, W.D., J. Applied Physics, 1959, 30, 301-306. Hampshire, S . , Jack, K.H., Special Ceramics 7 (ed. D.E. Taylor, P. Popper), B.C.R.A., 1981, 31, 37-49. Schneider, J.A., Risbud, S.H., Mukherjee, A.K., J. Muter. Res., 1996, 11, 358-362. Schneider, J.A., Mukherjee, A.K., Yamazaki, K.,Shoda, K., Materials Letters, 1995, 25, 101-104. Schneider, J.A., Mishra, R.S., Mukherjee, A.K., Proc. Second International Symposium on Advanced Synthesis and Processing, Ceramic Transactions (ed. R. Spriggs, 2.Munir, K. Logan), 1996, 79, 143-151. Suzuik, K.,Kanno, Y., J. Jpn. Ceram. SOC., 1984, 92, 101-102. Kaplan-Diedrich, H., Eckebracht, A., Frischat, G.H., J. Am. Ceram. SOC., 1995, 78, 1123-1124. Bowen, L. J., Weston, R.J., Carruthers, T.G., Brook, R.J., J. Mat. Sci., 1978, 13, 341-350. Stocker, R.L., Ashby, M.F., Reviews of Geophysics and Space Physics, 1973, 11, 391-427. Raj, R., Chyung, C.K., Acta Metall., 1981, 29, 159-166.


19)

279

Pharr, G.M., Ashby, M.F., Acta. Metall., 1983, 31, 129-138. Dryden, J.R., Kucerovsky, D., Willkinson, D.S., Watt, D.F., Acta. Metall.., 1989, 37, 2007-20 15. Wakai, F., Acta Metall., 1994, 42, 1163-1172. Elwell, D., Scheel, H.J., Crystal Growthfrom High-Temperature Solutions, Academic Press, 1975, Chapter 4. Grifkins, R.C., Superplastic Forming of Structural Alloys (ed. N.E. Paton, C.H. Hamilton), 1982, TMS-AIME, 3-26. Rossignol, F., Rouxel, T., Besson, J.-L., Goursat, P., Lespade, P., J. Phys. III France, 1995, 5 , 127-134. Chen, I.-W., Mat1 Sci. and Engr., 1993, A166, 51-58. Schneider, J.A., Mukherjee, A.K., to be pub. Ceramic Interfaces Control at the Inteface Level, 1996, Berkeley, CA, June 24-27.

Fabrication of Fine-Grained Si3N4and S i c T. Nishimura, M. Mitomo and H. Emoto* National Institute for Research in Inorganic Materials, 1-1 Namiki, Tsukuba, Ibaraki 305 Japan 'Denki Kagaku Kogyo Co., Ltd., 3-5-1 Asahimachi, Machida-shi, Tokyo 194 Japan

Abstract

Effect of particle size distribution of raw powders and type of additives on the liquid phase sintering and grain growth of Si3N4 and Sic ceramics was investigated. High temperature type, P-Si3N4 powder was used to eliminate the effect of phase transformations. Fine and homogeneous Si3N4 powder with average particle size of 280 nm was hot-pressed. Fine-grained Si3N4with average grain size of 210 nm were obtained. The Si3N4 ceramics were superplastically deformed at 1500 "C because of fine and stable microstructure. Fine-grained Sic ceramics with average grain size of 110 nm were prepared by hot-pressing fine P-powder with average particle size of 90 nm at 1750 "C. The Sic ceramics were deformed superplastically at 1700 "C. M icrostructural stability was examined by annealing at higher temperatures than sintering temperatures, i. e. static grain growth. The fine microstructures without appreciable static grain growth could be obtained only by low temperature sintering of fine and homogeneous raw powders. 1. Introduction

Ceramics are hard and brittle at room and high temperature, and these properties make it difficult to conduct plastic forming like in metals. But when microstructures become fine in nano-scale, a ceramic has superplasticity at high temperature. In 1984, Wangand Rajr'l found the superplastic oxide ceramics containing glassy phase. Based on the

Fabrication of Fine-Grained Si3N4 and Sic

28 1

constitutive equation of superplasticity (eqn. (l)), fine grains are one of important necessities for superplasticity.

where E‘ is strain rate, o applied stress, d grain size, n and p are both constants from 1 to 3 and A is a temperature-dependent and diffisionrelated constant. Superplastic non-oxide ceramics were fabricated by Wakai et al. in 1990L2]. The material was Si3N4/SiC composites. The superplastic Si3N4 related materials reported until now are a’@’SiA10N[31,alP-Si3N4[41and p-Si3N4[5,61. These superplastic materials consist of fine grains. During superplastic deformation of Si3N4/SiC composites[21,a’/P’-SiAlONr3]and U / P - S ~ ~ N applied ~ [ ~ ] , stress gradually increased with the increase of strain. This is called strain hardening which result from grain growth during deformation (dynamic grain growth). This phenomena can be understood by the eqn. (1) that “o” must be increased with the increase of “d” under constant 6. ” condition. Therefore, in order to obtain large deformation at high strain rate, stable microstructure is desirable, which means no dynamic grain growth occurs during deformation. Recently, superplastic Si3N4 consisting of rod shaped grains were reported. But unusual strain hardening was observed at strains greater than 50 % elongation[71. This suggests that fine and stable microstructure is still important to obtain superplasticity . Therefore, fine-grained monolithic Si3N4 was fabricated by sintering fine and homogeneous P-type powder at lower temperature for short timeL6]. The a powders have not been used in this investigation because phase transformation occurs during sintering which induces abnormal grain growth. On the Sic, superplastic-like behavior was reported by Carry and Mocellin in 19841g1. However, the deformation temperature was as high as sintering temperature of > 1900 OC. Fine-grained P-Sic was fabricated by low temperature sintering of fine powder to avoid phase transformation during sintering. The Sic was also deformed superplastically at 1700 ‘C without strain hardening[’]. In the present work, the effect of particle size of raw powders and the type of additives was investigated in relation to liquid phase sintering temperature and microstructural stability of Si3N4 and Sic. “

282


2. Experimental procedure 2.1 Fabrication of SisN4 Fine P-Si3N4powder (designated NBF in the present work) was prepared by grinding and centrifugil sedimentation of a commercial submicrometer P-Si3N4 powder (designated NBC, SN-P21FC grade, Denki Kagaku Kogyo Co., Tokyo, Japan). After the powders were mixed with sintering additives, hot-pressing was conducted in N2 flow at 20 MPa. In order to investigate stability of microstructures (static grain growth), hot-pressed samples were heated at 1800 "C in NZ. 2.1 Fabrication of S i c Fine p-Sic powder was a commercial p-Sic powder prepared by gas phase reaction (T-1 grade, Sumitomo-Osaka Cement Co., Tokyo, Japan). The powder contains a large amount of free carbon, i.e., 4.15 wt. %, which was detrimental to sinterin$"]. The carbon powder and large agglomerates in the powder were eliminated by centrifugal sedimentation. The recovered Sic powder was used for sintering (designated CF). As a reference, commercial submicrometer p-Sic powder (designated CC, Betarundum Ultrafine grade, Ibiden Co., G i h , Japan) was used. The sinteringadditives, 7 wt% A1203 , 2 wt% Y2O3 and 1 wt% CaO were selected to obtain lowest sintering temperature. The Sic powders mixed with the additives were hot-pressed in Ar. The sintering temperature and soaking time were controlled to obtain over 97 % of the relative density. The hot-pressed samples were heated at 1850 "C in Ar to investigate the static grain growth behavior. 3 Results and discussion 3.1 Feature of powders Figure 1 shows scanning electron micrographs of NBF and NBC. Particle size distribution measured by a sedimentation method was summarized in Table 1. The NBF is homogeneous powder with average particle size of 280 nm. Larger particles than 500 nm does not exist in NBF because they were removed by the centrifbgal sedimentation. The NBC has broad particle size distribution. It contains large particles than 1000 nm.


283

Figure 1. SEM micrograph of Si3N4raw powders: (a) NBF powder and (b)NBC

powder. Table 1. Particle size distribution of SLNd Dowders. NBF Powder 280 Particle size (nm) 50 % 70% 330 450 90 %

NBC 510 650 1110

The characteristic of Sic powders were listed in Table 2. Figure 2 shows scanning electron micrographs of CF and CC powders. The CF consists of fine and homogeneous particles with average grain size of 90 nm. The CC consists of both fine and coarse particles. The fine particles are almost same size as CF and size of coarse particles is about 1000 nm. The average particle size is obtained as 280 nm as a result of the wide particle size distribution. Both powders consist of almost pphase. Free carbon content of CF was reduced to 1.88 wt% by the sedimentation process explained in 2.2. Table 2. Characterization of Sic Dowders. Powder Average particle size (nm) p content (wt. %) Free carbon (wt. %)

CF 90 98 1.88

cc 280 97 0.16

284

High Temperature Mechanical Properties and Characterizationof Grain Boundaries

Figure 2. SEM micrograph of Sic powders: (a) CF powder and (b) CC powder.

3.2 Densification and grain growth An additive system of superplastic Si3N4 reported previously was Y2O3 and A1203[2~475~71. In order to optimize fabrication processes, the effect of type of additives was investigated. The NBF powder with a number of additive systems were hot-pressed. Giving examples, the density of a material sintered with 5 wt% Y203and 2 wt% A1203at 1700 "C for 5 min was 96.3 %. Denser material, 99.6 %, was obtained by sintering with 5 wt% Y2O3 and 2 wt% MgO at 1700 "C without soaking. As a result of other type of additive systems, Y203 and MgO system was chosen from the point of view of low temperature and short time sintering and restraint of grain growth. The both NBF and NBC powders with 5 wt% of Y2O3 and 2 wt% MgO were hot-pressed to investigite the effect of particle size of raw powders. The heating was stopped when the relative density became over 90 % without holding. The sintering condition was set in order to obtain high density without gain growth. As shown in Table 3, dense ceramics with relative density of 99.6 % could be obtained from the NBF at 1700 "C. The relative density of material from NBC hot-pressed at 1800 "C was 93.5 %. The sintering temperature of the former is about 100 OC lower than that of submicrometer powder. Scanning electron micrographs of plasma etched surfaces of hot-pressed samples from NBF

Fabrication of Fine-GrainedSi3N4 and Sic

285

and NBC are compared in Figure 3. The microstructure from NBF is fine and homogeneous. The average of equivalent grain diameter of the materials from NBF is 210 nm. Comparing the diameter with particle size of raw powder, grain growth hardly occurred, because the sintering was conducted at low temperature (1700 "C) without soaking time. This specimen was superplastically deformed under compressive stress at 1500 "C at the strain rate of 1.4 x /sed6]. The microstructure of the material from NBC is homogeneous but the average of equivalent grain diameter is 680 nm. Smaller particles than 500 nm were probably consumed by grain growth of larger particles" 'I. The hot-pressing conditions for Sic to attain high density >97 % are listed in Table 4. The CF densified at about 150 "C lower than CC without phase transformation. The microstructure of the material from CF is fine and homogeneous (Figure 4 (a)). Average grain size is 110 nm. Considering the particle size of raw powder, grain growth rate during sintering was slow. This material was deformed under compressive

Figure 3. Microstructure of hot-pressed Si3N4from (a) NBF and (b) NBC.

Table 3. Characterization of hot-pressed Si3N4 ceramics. Powder NBF Hot-pressing temperature ("C) 1700 Relative density (%) 99.6 Average of equivalent grain diameter 2 10 (nm)

NBC 1800 93.5 680

286


Figure 4. Microstructure of hot-pressed Sic from (a) CF and (b) CC.

/s at 1700 OC. The microstructure of stress at the strain rate of 5 x the material from CC is coarser and heterogeneous (Figure 4 (b)). The average grain size is 510 nm. Higher grain growth rate of CC is attributed to wider grain size distribution and higher sintering temperature[']. It has been reported that Sic ceramics can be fabricated by liquid phase sintering at about 2000 OC. Microstructure of materials from p powders showed phase transformation from j3 to a and induced abnormal grain growth['21. Though CF is p powder, the grain growth was restricted and homogeneous microstructure was obtained because of low sintering temperature and without phase transformation.

Table 4. Characterization of hot-pressed Sic ceramics. cc Powder CF 1750 "C for 15 min 1900 "C for 30 min Hot-pressing Relative density (%) 97.2 99.0 Average grain size (nm) 110 510 3.3 Static grain growth behavior Figure 5 shows microstructure of annealed Si3N4at 1800 OC, which is a little higher temperature than sintering or deformation temperature, for (a) 0.5 h and (b) 4 h. Comparing Figure 3 (a), grain diameter of annealed sample shows a little growth only at early stage of annealing. Though


287

Figure 5. Microstructure of Si3N4annealed at 1800 "C for (a) 0.5 h and (b) 4 h.

grains grow slowly, the homogeneity of the microstructure is still retained even after the heat treatment for 4 h. The grain size of Si3N4from NBC increased with the increase of heat treatment time at 1800 OC["]. The driving force for grain growth is so low that the microstructural stability of a material from NBF is high, because of narrow grain size distribution and the absence of phase transformation. Figure 6 shows microstructure of annealed Sic at 1850 OC for (a) 6h and (b) 12 h. The growth of large elongated grains in small matrix grains

Figure 6. Microstructure of Sic annealed at 1850 "C for (a) 6 h and (b) 12 h.

288


is shown in an annealed material for 6h. The microstructure of an annealed material for 12 h consists of also fine matrix grains and elongated grains but both are elongated. Primary phase of heat treated material was still p even after heat treatment for l2h[I3]. The annealing temperature is fairly higher than sintering temperature. And superplastic deformation was attained as low as at 1700 "C. The results suggest that microstructural stability is very high at about 1700 OC.

3.4 Dynamic grain growth behavior The dynamic grain growth is induced by plastic deformation at high temperature, which is different from static grain growth. But strong effect of heat treatment time on strain hardening was reported in superplastic deformation['931. A fine-grained LiAlSiZO6 was deformed under tensile stress. When the deformation rate was high (6.7 x /s), the strain hardening did not occur, but it occurred at low strain rate Is)[']. Superplastic deformation of a'/P'-SiAlON was (1.7 x conducted under compressive stress. The strain hardening was observed at lower strain rate than 7 x /s, but it was not at high strain rate /s)[~]. These results suggest that long time exposure is more (1 x effective for dynamic grain growth than induced strain. The materials developed in present work showed no appreciable dynamic grain growth['41,which is a reason that Si3N4and Sic showed superplastic deformation at low temperature of 1500 OC and 1700 OC, respectively. The results of static grain growth were very useful for the development of superplastic ceramics. 4. Summary

(1) Effect of particle size of raw powders and type of additives was investigated to optimize sintering temperature and time for fabrication of dense Si3N4 and Sic ceramics without grain growth by hotpressing. (2) The ceramics consisted of fine and homogeneous grains. Stable microstructure for static grain growth was confirmed by heat treatment at higher temperature than sintering. (3) Absence of phase transformation was also a reason for low rate of static and dynamic grain growth during sintering and plastic

Fabrication of Fine-Grained SijN4 and Sic

289

deformation. The present fine-grained Si3N4and S i c ceramics deformed at the higher strain rate than 1 0-4/s at 1500 and 1700 OC, respectively. (4) The informations on low temperature sintering, grain size minimization and driving force for grain growth are useful for the development of superplastic ceramics. References

1 . J.-G. Wang, R. Raj, J Am. Ceram. SOC.1984,67,399-409. 2. F. Wakai, Y. Kodama, S. Sakaguchi, N. Murayama, K. Izaki, and K. Niihara, Nature 1990,344,421-23. 3. I-W. Chen, and S.-L. Hwang, J. Am. Ceram. SOC.1992, 75, 1073-79. 4. P. Descamps, D. Beugnies, F. Cambier, Fourth Euro-Ceramics (Ed.: A. Bellosi), Gruppo Editoriale Faenza Editrice S. p . A., Faenza, Italy 1995, vol. 4, p. 251. 5. P. Burger, R. Duclos and J. Crampon, J. Am. Ceram. SOC.1997, 80 879-85. 6. M . Mitomo, H. Hirotsuru, H. Suematsu, and T. Nishimura, J. Am. Ceram. SOC.1995, 78,21 1 - 14. 7. N. Kondo, F. Wakai, T. Nishioka, A. Yamakawa, J. Mater. Sci. Lett. 1995,14, 1369-7 1 . 8. C. Carry and A. Mocellin, Materials Science Research, (Ed.: R. E. Tressler and R. C. Bradt), Plenum Press, New York, USA, 1984, vol. 18, p. 391. 9. M . Mitomo, Y.-W. Kim, H. Hirotsuru, J. Mater. Res., 1996, 11 160 1-4. 10. Y.-W. Kim, H. Tanaka, M . Mitomo, S. Otani, J. Ceram. SOC.Japan 1995,103,257-261. 1 1 . H. Hirotsuru, M . Mitomo, T. Nishimura, J. Ceram. Soc. Japan, 1995, 103,464-469. 12. N. P. Padture, J Am. Ceram. SOC.1994, 77,519-23. 13. Y.-W. Kim, M . Mitomo and H. Hirotsuru, J Am. Ceram. SOC.1995 78,3 145-48. 14. T. Nishimura, Y. Bando, M . Mitomo, H. Suematsu, Fourth EuroCeramics (Ed.: A. Bellosi) Gruppo Editoriale Faenza Editrice S. p. A., Faenza, Italy, 1995, vol. 4, p. 265.


MAX-PLANCK-INSTITUT FUR METALLFORSCHUNG PULVERMETALLURGISCHES LABORATORIUM STUTTGART

Program Workshop on

Grain Boundary Dynamics of Precursor-Derived Covalent Ceramics

10th - 16th November 1996 Schlorj Ringbergmegemsee GERMANY

F. Wakai Japan Science and Technology Corporation (JST) F. Aldinger and J. Bill Max-Planck-Institutfur Metallforschung November 1996

292

Program of the workshop

Monday, 11th November Opening F. Aldinger, MPI f. Metallforschung, Stuttgart

Globalization of Research Science and Technology Policy Shift in Japan Learn from Japan Research Development Corporation M. Kawasaki, JST, Tokyo, Japan

-

-

Resemh between Nationality and Internationality W. Hasenclever, MPG, Miinchen, Germany Globalizationof Resemh in Industry M. Unger, Hoechst AG, Frankfurt, Germany Globalizationand the G e m Fedelal Research Management The Materials Technology Program ,,MaTech" J. Roemer-Miihler, BMBF, Bonn, Germany

-

-

Excitements in Japanese-Gem Resemh Coopenitions G. Petzow, MPI f. Metallforschung, Stuttgart, Germany

Keynote Lectures Precursor-Derived Covalent Ceramics J. Bill, MPI f. Metallforschung, Stuttgart, Germany Ceramics Superplasticity F. Wakai, JST, Nagoya, Japan Question Time about MPI-MF/JST Joint Resemh Program


293

Tuesday, 12th November 1996

Ceramics from Organoelement Compounds Ceramisation Process of Polycarbosilane and its Selected Polymer K. Okamuru, Osaka Prefecture University, Osaka, Japan Single S o m e Precursors for the Prepamtion of Highly Stable Multinary Nonoxide Ceramic Materials M. Junsen, Institut f. Anorg. Chemie, Bonn, Germany Pmcessing Phase Pure SiC-fibers fmm Precursors R. k i n e , University of Michigan, Ann Arbor, USA Multicomponent Silicon and Bomn himnitrides A Novel Generation of Advanced Ceramics R. Riedel, TH Darmstadt, Germany

-

The Polymer-to-Cemnic Conversion Process in SiH2CJ32 and SiOCH2 Polymer Precursors to Silicon Chbide and Silicon Oxycadide L. Interrunre, Rensselaer Polytechnic Institute, New York, USA Highly Reactive Chlommethylpolysilanes as Effective Sic-fiber Precursors G. Roewer, TU Bergakademie Freiberg, Germany Ultra-High Tempem& Stable Ceramics fmm Inorganic Polymers

M. Weinmunn,MPI f. Metallforschung, Stuttgart, Germany Pmcessing and High Tempem& Mechanical Properlies of C e d c s fmm Organosilicon Polymers G.D. Soruru, Universita di Trento, Trento, Italy Bomn Polymers and Materials M. Sneddon, University of Pennsylvania, Philadelphia, USA Oxide and Non-Oxide Ceramics fmm Organosilicon Polymels G. Ziegler, H.-J. Kleebe, D. Suttor, Universitat Bayreuth, Germany

294


Synthesis Principles and Processing of Oxidic Ceramic Materials Derived from Metal Organic Compounds G. Miiller, Fraunhofer-Institut, Wiirzburg, Germany Synthesis of Oxide Inoqanic Materials at Organic Swfaces

M.De Guire, MPI f. Metallforschung, Stuttgart, Germany Non-Oxide Thin Films by Bioinspined Reactions Th. Niesen, MPI f. Metallforschung, Stuttgart, Germany


Wednesday, 13th November 1996

Precursor-Derived Ceramics in Industry Industrial Applications for Siliconboron Carbonitride Ceramic Materials G. Passing, Bayer AG, Leverkusen, Germany Precursor-Derived Ceramics from an IndusMal View H . Biider, Robert Bosch GmbH, Stuttgart, Germany Precursor-Derived Ceramics in Aerospace Applications W. Vogel, Dornier GmbH, Friedrichshafen, Germany

Characterization The Potential of Surface Analysis for the Investigation of Growing Interfaces F. Jentoj?, Fritz-Haber-Institut der MPG, Berlin, Germany

Comsion Behaviourof Ceramics K. Nickel, Universitat Tubingen, Tubingen, Germany

295

296


Thursday, 14th November 1996

High Temperature Mechanical Properties and Characterization of Grain Boundaries The Role of Grain Boundaries on the Creep Behavior of Silicon Nitride S. M. Wiederhorn, Nat. Inst. of Stand. & Technol., Gaithersburg, USA

Characterization of High-Tempemtm Ploperties of Pmcmor-Derived Ceramics by Compression Cleep Tests G. Thurn, MPI f. Metallforschung, Stuttgart, Germany Mechanical Spectsoscopy of Materials at High TempemM. Weller, MPI f. Metallforschung, Stuttgart, Germany Nano-, Micm-, and Milliboundaries in Silicon Nitride Based Ceramics P. Sajgalik, Institute of Inorganic Chemistry, Bratislava, Slovakia Plastic Deformation of Covalent Bonded Ceramics d i n g High Tempemtums S. Tsurekawa, JST, Nagoya, Japan Creep FracaUe in Fibmus Micmstructms I. Beyerlein, S. L. Phoenix, R. Raj, Cornell University, Ithaka, USA Deformation and Phase Tmnsformation of Silicon NiMde and Silicon Wide I-Wei Chen, University of Michigan, Ann Arbor, USA Viscoplastic Forming of Si3N4-based Ceramics T. Rouxel, ENSCI, Limoges, France Micmstructural Development of Superplastic Silicon Nitride E. Sato*, N. Kondo, F. Wakai * Inst. of Space and Astronautical Science, Kanagawa, Japan Correlation Between Deformation Mechanisms and Microstructural Evolution in Silicon Nitride Ceramics A.K. Mukherjee, University of California, Davis, USA


Fabrication of Fine-Grained Si3N4 and SIC T. Nishimuru*, M. Mitorno*, H . Emoto * Nat. Inst. for Research in Inorganic Materials, Ibaraki, Japan Characterizationof Celamics by Imaging, Diffmction and Spectnoscopic Techniques M. Riihle, MPI f. Metallforschung, Stuttgart, Germany

297

298


Friday, 15th November 1996

Theory and Modelling of Materials Bond Order Potentials for the Atomistic Simulation of Covalent Materials D.G. Pettifor, University of Oxford, Oxford, England

The Ab-initio Molecular Dynamics Method A Short Review of Applications to Materials Science Problems P. Ballone, MPI f. Festkorperforschung, Stuttgart, Germany Computer Simulations of Silicon NiMde and Silicon CarboniMde Ceramics P. Krol2, TH Darmstadt, Darmstadt, Germany Thennodynamic Calculations in the B-C-N-O-Si System H.-J.Seifert, MPI f. Metallforschung, Stuttgart, Germany NMR Characterization of Polymer-Derived Ceramics F. Babonneau, Universite Pierre et Marie Curie Tour, Paris, France Solid State NMR Studies for Ceramic Characterization K. Miiller, Institut fur Physikalische Chemie, Stuttgart, Germany Characterization of PolymeFto-Ceramic Convelsion Process M. Sasaki, Muroran Institute of Technology, Hokkaido, Japan Solid State NMR Studies of Organically Modified Ceramics M. Templin, U.Friedrich, U.Wiesner, H. W. Spiess MPI f. Polymerforschung, Mainz, Germany Diffmction Methods for the Characterization of Amorphous Materials P. Lamparter, MPI f. Metallforschung, Stuttgart, Germany

X-Ray and Neutrron Diffraction Investigations on Amorphous Silicon CahniMdes J. Diirr, MPI f. Metallforschung, Stuttgart, Germany TEM-Characterizationof Precursor-Derived Ceramics J. Mayer, MPI f. Metallforschung, Stuttgart, Germany

Precursor-Derived Ceramics: Synthesis, Structure and High-Temperature Mechanical Properties

Mechanical Properties of Ceramics

Mechanical Properties of Ceramics

Structure and Properties of Ceramics

Clathrochelates.. Synthesis, Structure and Properties

Clathrochelates: Synthesis, Structure and Properties

Cellular Ceramics: Structure, Manufacturing, Properties and Applications

Carbon nanotubes synthesis structure properties and application

Carbon Nanotubes: Synthesis, Structure, Properties and Applications

Mechanical Properties of Ceramics and Composites: Grain and Particle Effects

Mechanical Properties of Ceramics and Composites: Grain and Particle Effects

Metal Complexes and Metals in Macromolecules: Synthesis, Structure and Properties

Nanocrystalline Ceramics by Mechanical Activation

Nanomaterials: synthesis, properties, and applications

Polyynes: Synthesis, Properties, and Applications

Amines: Synthesis, Properties and Applications

Metallomesogens: synthesis, properties, and applications

Structure and Properties of Polymers

Quasicrystals: structure and physical properties

Nanocrystals:: Synthesis, Properties and Applications

Nanocrystals: synthesis, properties and applications

Nanomaterials: Synthesis, Properties and Applications

Foldamers: Structure, Properties, and Applications

Structure and Properties of Composites

Perovskites: Structure, Properties and Uses

Mechanical Properties and Performance of Engineering Ceramics and Composites IV (Ceramic Engineering and Science Proceedings)

Mechanical Properties of Engineered Materials

Mechanical Properties of Complex Intermetallics

Mechanical Properties Of Engineered Materials

Mechanical Properties of Solid Coatings

Mechanical Properties of Complex Intermetallics

Precursor-Derived Ceramics: Synthesis, Structure and High-Temperature Mechanical Properties

Mechanical Properties of Ceramics

Mechanical Properties of Ceramics

Structure and Properties of Ceramics

Clathrochelates.. Synthesis, Structure and Properties

Clathrochelates: Synthesis, Structure and Properties

Cellular Ceramics: Structure, Manufacturing, Properties and Applications

Carbon nanotubes synthesis structure properties and application

Carbon Nanotubes: Synthesis, Structure, Properties and Applications

Mechanical Properties of Ceramics and Composites: Grain and Particle Effects

Mechanical Properties of Ceramics and Composites: Grain and Particle Effects

Metal Complexes and Metals in Macromolecules: Synthesis, Structure and Properties

Nanocrystalline Ceramics by Mechanical Activation

Nanomaterials: synthesis, properties, and applications

Polyynes: Synthesis, Properties, and Applications

Amines: Synthesis, Properties and Applications

Metallomesogens: synthesis, properties, and applications

Structure and Properties of Polymers

Quasicrystals: structure and physical properties

Nanocrystals:: Synthesis, Properties and Applications

Nanocrystals: synthesis, properties and applications

Nanomaterials: Synthesis, Properties and Applications

Foldamers: Structure, Properties, and Applications

Structure and Properties of Composites

Perovskites: Structure, Properties and Uses

Mechanical Properties and Performance of Engineering Ceramics and Composites IV (Ceramic Engineering and Science Proceedings)

Mechanical Properties of Engineered Materials

Mechanical Properties of Complex Intermetallics

Mechanical Properties Of Engineered Materials

Mechanical Properties of Solid Coatings

Mechanical Properties of Complex Intermetallics

Recommend Documents